You are on page 1of 434

Topical Drug

DRUGS AND THE PHARMACEUTICAL SCIENCES


A S@ri~"s
of Texthoks and Monogrilphs

Edited by
James Swrrbrt~k
&hool af Ph#rmacy
University of North G"aro/tna
Chap& Hfll, North Carolina

Volume 1. PWARMAGQKtNETICS, Mifo GibalclriandDonktldPerrier


lout sf print1

Valumr?2. GQOO MANUFACTURING PRACTICES FOR


PHARMACEUTICALS: A PLAN FOR TOTAL QUALITY
CONTROL, Sidney N. Wif//g,Murray M. Tuckwman, and
WiIIiam S Hitchings I t / tout of prlntl

Volume 3, MIGRQENGAPSULATION, edit& by J. R. Nixon

Volume 4. DRUG METABOLISM: CHEMICAL AlVD EliOCWEMlCAL


ASPECTS, Bernard Tsta and Peter Jmnw

Volume 5. N E W ORUGS: C)ISCOVEF(V AND DEVELOPMENT,


edited by Alan A. Rubin

Volume 6, SUSTAINED AND GONTRQLLED RELEASE DRUG BELIVERY


SVST E MS, edit& by Jowph R. Robinson

Volume 7. MOOERN PHARIVIACEUTICS, &t@d by Gilbert S.


13anker and Christopher T. Rhoda

Volume 8, PRESCRIPTION ORUGS IN SHORT SUPPLY CASE


H I STQ R I ES, Michael A, Schvvartr

Volume 9, ACTIVATED CHARCOAL: ANTIDOTAL AND OTHER


MEDICAL USES, David O. Cooney

Volume 10. CONCEPTS IN DRUG METAEIOLISM (in two parts), @titfed


by Peer Jtlnner and Bernard Testa

Volume I I . PWARNIACEUTtGAt ANALYSIS MODERN METHODS


(rn two parts), edited try Jamm M/, Munssn

Volume 12. TECHNIQUES OF SOLUBlr t fZAT10N Q F DRUGS,


edited by Samuel H. Valkowsky
Volume 53, ORPHAN DRUGS, d i W b y F r d E Kareh

Volume 14, NOVEL DRUG DELIVERY SYSTEMS: FUNDAMENTALS,


DEVELOPMENTAL CONCEPTS, BIOMEDICAL ASESNIENTS,
Vie wemien

Volume 15. PXARMACOKl NETIC5, W o n 4 Edition, Revised and Expanded,


Mi/@Giba/di and Donald Perrier

Volume 16, GOOD MANUFACTURING PRACTICES FOR PHARMACEUTICALS:


A PLAN FOR TOTAL QUALtTV CONTROL, acond Edition,
Revised and Expanded, Sidney W, WiIIig, IWurri3y M. Tuckeman,
and Wi//i@mS. firi'f6hings I V

Volume 17. FORMULATION OF VETERINARY DOSAGE: FORMS, edltdhy


Jaek Blodingw

Volume 18. DERMATOL061CAL FORMULATIONS: PERCUTANEOUS


ABSORPTION, B r i ~ n Barw

Volume 19. THE CLINICAL RESEARCH PROCESS IN THE PHARhltACEUTfCAL


INDUSTRY, edited by G a v M &taran

Volume 20. MICRQEN@APSUM"TONAND RELATED DRUG PROCESSES,


Patrick bt. D e w

Volume 21. DRUGS AND NUTRIENTS: THE INTERACTfVE EFFECTS,


@dig& by Daphne A. Roe and 7: CoIin Campbe//

Volume 22, BtOTECHNOLQGY OF INDUSTRIAL ANTIBIOTICS,


Erick J, Vandammtz

Volume 23, PHARMACEUTICAL PROCESS VALIDATION,


edited by Bernard 7: Lotus and Rclbea A. Na&

Volume 24. ANTlCANCER AND INTERFERON AGENTS: SYNTHESIS


AND PROPERTIES, Etdited iety Rapliisel M. Onenbrite and
Gmrg-e h3, But/er

Volume 25. PHARMACEUTICAL STATISTICS: PRACTICAL AND


CLINICAL APPLICATIONS, SnfordB~ltclm

Volume 26. DRUG BVNAMtCS FOR ANALYTICAL, GLfHiCAL, AND


B I 0LOG t CA L CHF,hnt STS, Benjamin J. Gudrinowiez, &rmw T:
Vounkin,Jr., md Michaei J. GUdzin~wicz
Volume 27. MODERN ANALYSIS 01"ANTlBf3rQTiGS,t?ditdby
Adodan Aa&/os
Volume 28, SOLUBIL1TV AND RELATED PROPERTIES, Kenneth C.Jmes
Volume 29, CONTROLLED DRUG DELIVERY: FUNDAMENTALS AND
APPLICATIONS, Second Edit ion, Revised and Expanded,
edited by Joeph R. Robrnson a d Viment N. li. Ltm
Volume 30* NEW DRUG APPROVAL PROCESS: CLINICAL AND
REGULATORV MANAGEMENT, edited by RiGnard A,
Guarino
"Vlume 31, TRANSDERMAL CONTROLLED SYSTEMIC
MEDICATIONS, d i t e d b y Yie Vy, Chim

Votume 32. DRUG DELtVERV DEVICES: FUNDAMENTALS


AND APPLICATIONS, edited by Praveen Ty/@
Volume 33, PHARMACOKINETfGS: REGULATORY * tNOUSTRiAL *
ACADEMIC PERSPECTIVES, edited by Peter G,W@//ingand
Francis L. S 7-ti%?
Vofume 34. CLINlCAL DRUG TRiALS AND TRrBULATfONS, edited by
A//@nE: Cztto
Votune 35. TRANSDERMAL DRUG DELIVERY: DEVELOPMEMTAL
lSSUES AMD RESEARCH tNITiATfVES, edited by Jonathan
Nadgr&N and Richard N. Guy
Volume 36. AQUEOUS POLYMERIC COATINGS FOR PHARMACEUTICAL.
DOSAGE FORMS, ~ditedby James W. McGinity
Volume 37. PHARMACEUTICAL PELLETIZATf ON TECHNOLOGY,
edited by / w c Ghebr&Se//asie
Volume 38. GOOD LABORATORY PRACTICE RECULATIQNS, edited by
A/& f Nirsh
Volume 39, NASALSYSTEMICDRUG DELIVERY, Yie W. Chien,
I(enneth S 14". Su, and Shyi-Feu Chang
Votume 40. MOOER M PHARMACEUTICS, Sconcl Edition, Revised and
Expanded, edited by Gi/bsrt.S Banker and Chriaopher T: Rhodes"
Valumtj 41, SPECIALIZED DRUG DELlVERV SVSTf MS: MANUFAC-
TURING AND PRODUCTION TECHNO LOGY, edit& by
Praveen Tyle
Volume 42- TOPICAL DRUG DELIVERY FORMULATIONS, editgd by
David W. Q&ome a d Antan W. Amann
DRUG STAB1 LITY: PRINCIPLES AND PRAGT lCES, Jens X
Garsenmn

PHARMACEUTICALS STATISTEGS: PRACTICAL AND GLlMt-


CAL APPLICATIONS, Second Edition, Revisd and Expanded,
Sanford Bolmn
Drug De

edited by
David W. Osbome
Anton H. Amann
The Upjohn Company
Kalamazoo, Michigan

informa
healthcare
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742

© 1989 by Taylor & Francis Group, LLC


CRC Press is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S. Government works


Version Date: 20130312

International Standard Book Number-13: 978-1-4822-0682-1 (eBook - PDF)

This book contains information obtained from authentic and highly regarded sources. While all reasonable
efforts have been made to publish reliable data and information, neither the author[s] nor the publisher can
accept any legal responsibility or liability for any errors or omissions that may be made. The publishers wish to
make clear that any views or opinions expressed in this book by individual editors, authors or contributors are
personal to them and do not necessarily reflect the views/opinions of the publishers. The information or guid-
ance contained in this book is intended for use by medical, scientific or health-care professionals and is provided
strictly as a supplement to the medical or other professional’s own judgement, their knowledge of the patient’s
medical history, relevant manufacturer’s instructions and the appropriate best practice guidelines. Because of
the rapid advances in medical science, any information or advice on dosages, procedures or diagnoses should be
independently verified. The reader is strongly urged to consult the drug companies’ printed instructions, and
their websites, before administering any of the drugs recommended in this book. This book does not indicate
whether a particular treatment is appropriate or suitable for a particular individual. Ultimately it is the sole
responsibility of the medical professional to make his or her own professional judgements, so as to advise and
treat patients appropriately. The authors and publishers have also attempted to trace the copyright holders of all
material reproduced in this publication and apologize to copyright holders if permission to publish in this form
has not been obtained. If any copyright material has not been acknowledged please write and let us know so we
may rectify in any future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmit-
ted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented,
including photocopying, microfilming, and recording, or in any information storage or retrieval system, without
written permission from the publishers.

For permission to photocopy or use material electronically from this work, please access www.copyright.com
(http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Dan-
vers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and registration for a
variety of users. For organizations that have been granted a photocopy license by the CCC, a separate system of
payment has been arranged.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only
for identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com

and the CRC Press Web site at


http://www.crcpress.com
This book is desi-ed to provide the pharmaceutical formulator with.
the fundamental understanding necessary to prepare efficacious
topical drug delivery formulations, In the past the abiEty to formu-
late topical creams and ointments has been described as both "artFt
and "ma@c." The demanding expectations of topicals include (I)
formulations that have both chemical and physical stability for at
least two years, ( 2 ) fornufations that have components bath, alone
and in combination that are nonirritating, nonsensitizing, and non-
allergenic, (3) fornutations that are at least cosmetically acceptable
and preferably eosmetiealrly e l s e n t , and ( 4 ) formulations thrat are
efficacious because of their abifity to release anidlor deHver them-
peutie levels of a drug. Note that these expectations are for an in-
herently multiple-phase system (usually an emulsion or suspension)
t h t contains components having a wide range af polarities and
physical properties.
With such demanding expectations, it is of little surprise that
mast scientists, w h n challenged vvith the assiwment of formulating
a new topical drug, select a tmditional farmulation taken from one of
the m n y formulato~iesavdlable, TMs tactic will ultimtely provide
a suitable formulation (usually d t e r appropriate, often f ~ u s t m t h g ,
adaptations ar ff"fne-tuninglf") However, the maazation- that drugs
can be delivered transdermally ta systemically therapeutic levels (as
evidenced by commercially available nitroglycerin, and scopolarxline
patch systems) and that changes in vehieIes can s i ~ z i c a n t l ychange
efficacy has permanently altered -I he way phmmaceutical scientists
view topic3als. No longer can we consider creams and ointments as
inert d m g carriers. Rather, we as formulators are requimd to
desim veMcles t b t target follicles or other specgic skin ~ e @ o n s ,
v e ~ e l e sthat optimize drug-skin and prodrug-skin interactions, or
vehicles that release drugs for sustained periods of time, Traditional
formulations can seldom meet these additional demands.
Therefore, a symposium was organized as part of the 6Lst Galoid
and Surface Science Symposium held in Ann Arbor, Micfipn, June
24, 1987. The symposium, entitled Topical Drug Delivery Formulations,
was designed to bring together a c a d e ~ cand industrial scientists who
had (1) devellaped techniques that aid in selection of an optimized
formulation, ( 2 ) futhered current understanding of how vehicle corn-
ponents and drugs interact with the skin, or (3) investigated non-
traditional drug dlegvery formulations. B y using this smpasiurn as a
car@,additional mate~ialwas included that would provide the formula-
tor with state-of-the-art understanding in three areas: ( I ) how to
d e s i w stabigty and in vitro testing to obtain optimized formuhtions
before conducting clinical investigations; ( 2 ) the role of the skin as an
immune response organ and barrier ta percutaneous absorption, and
how this barrier can be omp promised using enhancers; and (3) how
nontraditional topical drwg deiivery formulations can be utifized a s
pharmaceuticals. By combining the latest insights with established
principles, we sincerely hope that this book will be a useful tool in
formulating the next genemtion of highly effective nontratditional topi-
cal drug delivery formulations.

David WbV,Osborne
Anton iJ. Amann
Preface

Contributors

Part I : PRINCIPLES OF SKIN AND FORMULATEON


INTERACTIONS

1 Practical Considerations for Topical Drug Formulations


With and Without Enhancers
Eugene R, Cooper and Dinesh C . Patet
I, Introduction
II. VeMcle B e s i p
If 1. Evaluation of VeEcle Performance
IV . Formufatian Example : Desonide
References

it The importance of Epidermal Lipids for the Stratum


Corneum Barrier
Peter M. Ebias
I. Evaltrtion. of Cument Ganeepts of Stratum
Corneum Structure
If.. Epidermal Differentiatian
III . Composition of Stratum Corneurn Lipids
IV. Lipid 1Biosyntfiel;ie Gmdients in the Epidermis
V. Clinical and Patjrsophysiob@e Implications of
the Two- Compartment Model
References
vi Contenfs

3 Stratum Corneum Structure and Transport Properties


S t i g E , Friberg, Ibrahim
H. Kayati, Marion Margosiak,
David Mi. Qsborne, and Anthany J . I . Ward
I. Introduction
11. Lipid Organization
111. Sumarry
References

4 Penetration Enhancer Incorporation in Bilayers


Anthony J . I. Ward and Regina Tallon
I, Introduction
If. Experimental Backaound
111. Experimental
IV. Results and Discussion
References

5 The Influence cf Skin Surface Lipids on Topical


Formulations
David W. Osborne and Douglas A Hatzenbuhler
I. Intmduction
If. Skin and Surface Lipid Excretion
111. Composition of Skin Surface Lipids
IV. Fallacies Gancerning Skin Surface Lipids
V. Model Skin Surface Lipids
VI, Interactions Between Skin Surface Lipids and
Topical Farmulations
VII. Concluding Remarks
References

6 Immunology of the Skin


David W. Lynch and Lee K , Roberts
I, Introduction
11. Structure and Histob@cal Or-nimtion of the Skin
111. Inflammation and Immune Responses in the Skin
IV. Skin-Associated L p p h o i d Tissues
V. Immunocompetent Cells Within the Skin
VI. Effects of Agents that Deplete Skin of k n g e r h a n s
Cells on Immune Responses to Antigens
Encountered Through the Skin
VII. Conclu~uaion
References
Contents

Part f l : DRUG FQRMULATIQM SELECTION AND TESTING


FOR TOPICAL USE

"7~ompuitational Methods for Prodrug or Drug Analogue


Selection Optimized for Percutaneous Delivery.
David W , Osborne
Introduction
1,
ICI.Description of the Models
111, Comparison of Predicted and Experimental Flux
Values
IT. Use of the Predictive Models
V . Contour Plots
Vf , Additional Considerations
References

8 fr(inet;ic Considerations in the Design of Surfactant-


Based Topical Formulations
Anthony J . I , Ward
I. Introduction
11. Experimental B a c k ~ o u n d
f 11. Results and Discussion
fV. Summary
References.

9 The Use of Phase Behavior and Laboratory Robotics


for the Optimization of' Pharmaceutical Topical
Formulations
David Mr. Osfaorne
I. Introduction
I . Determination of Component Ranges
111. Optimization Techniques
IV. Closing Remarks
References

10 Generat Gansideratior-ts for Stability Testing of


Topical Pharmaceutical Formulations
Gary R, Dukes
I, Xntraduction
X 1. Preformulation Stage
IT1, Parmulation Development Stage
IV . P~oposedProduct
V. New Product Stage
V1, Established Product Stage
VII. Product Revision Stage
VTI1 , Conclusion
References
viii Con tents

11 In Vitro Testing of Topical Pharmaceutical Formulations


Ronald C. Wester and Howard 1. Waibach
I, Introduction
11. Basic Study Design
111, Factors
IV . ConcIuslons
References

12 Drug Delivery from Topical Formulations : Theoretical


Prediction and Experimental Assessment
William J. Addicks, Gordon L . Flynn, Norman I), Weiner,
and Rane L, Curl
I. Introduction
11. In Vitro Release from Vehicles
111. Applications of Theory: In Vitro Release Studies
IV. Diffusion Studies Involving Skin
References

13 The Use of Solubility Parameters of Drug and Vehicle


to Describe Skin Transport
Kenneth B. Stoan
I. Introduction
11. General Applications of Regular Solution Theory
to Partitioning Processes
111. Appfication of Remiar Solution Theory to
Percutaneous Absorption Processes
X V. Conclusion
References

14 Use of an Epidermal Cutture for Cutaneous Toxicity


Studies
F . L. Vaughan
1, Introduction
XI. Developing Cultivation Proeeduws
IXI. Stratification and Differentiation of Keratinacytes
Cultured on Nylon Membranes
XV. Experhentation with Sulfur Mustard in Cutaneous
Toxicity Studies Using the Epidermal Culture
V. Summry and Conclusions
References
Contents

Part I l l : hlONTRADlTfOMAI, TOPICAL DRUG DELIVERY


FORNULAT tQMS

15 The Microsponge: A Novel Topicar Programmable


Delivery System
Sergio ioachl and Martin K a l z
I, The Need for Better Skin Delivery Systems
11. Microsponge Technolow
III, Microsporrge Sptlresis
IV , Physical Charactedzation of Microsponges
v, Pra@axnmable Parameters
VE. Release Mechanisms
VX1. Safety S u b s ~ n t i a t i o n
VIII, Increased Efficacy and Aesthetic Appeal of Drug
Formulations
Possible AppKcations for Mcrosponge Systems
Advantages for Industry
References

16 Liposome-Based Vehicles for Topical Delivery


Paul S , Usler
I. Xntroductian 321
11. tvLiposomola~'t 327
111. Formulating Topical Products e t h Liposornes 332
IV , Rc;searckt. Studies 331
V . Future Directions 344
References 346

17 Surfactant Association Cof lioids as Topical Drug Delivery


Vehicles 349
David W , Qsborne, Anthony J , I , Ward, and
KiIian J , O%~er'l;l,
I. Irrtraduction 349
TI. Description of Surfactant Association Collbids 350
Elf. Formulation of Topical Surfactant Aissoeitltfon.
Colloids 355
IV . Drug Delivery. from Topical Surhctant Association
Colloids 372
V, Concluding Remarks 316
References 377
x Contents

18 Gel Dosage Forms: Theory, Formulation, and Processing


Lorraine E . Pena
I. Theory
11. Formuktion
111, Processing
References

19 Using Silicones In Topical Products


Michaei S , Starch
I. Introduction
11. Types of Silicones Used in Topical Products
and Their Properties
111. Formulating Silicones into Conventional Topical
Formulations
'EV . Nontraditional Topical Formulations Based on
Silicones
References

Index
Contributors

.
William .J Adbieks* University of Michigan College of Pharmacy,
Ann. Arbor, Michigan
Eugene R . Cooper Sterling Drug, Inc. , Malverne, Pennsylvania
Rane t. Curl University of Michigan College of En@neering, Ann
Arbor, Michipn
Gary R , Dukes The Upjohn Company, KaXamazoo, Michigan
Peter M , EZias Veterans Administration Medical Center, University
o f California School of Medicine, San Francisco, CaHfornia
Gardon L. FIy nn Cygnus Research Corpomtion, Redwood City,
California
Stig E. Friberg Clarkson University, Potsdam, New Vark
Dougfas A . Hatzanbuhler The Upjohn Company, Kalamazoo, Michigan
Martin Katr Advanced Polymer Systems, Inc ., Redwood City,
California
lbrahim H , KayatI Clarkson University, Patsdam, New York
David H . Lynch Immunex Corporatian , Seattle, Washington
Howard t , Maibath University of Galgomia Sckroal of MeCLicine , San
Francisco, California

*Current affiliation : E , I , du Pont de Memours and Company,


Wilrnin@on, Delaware
xii Cont~ibutors

Marion Margosiak Clarkson University, Potsdam, New York


Sergio Nacht Advanced Polymer Systems, Inc., Redwood City,
California
Kilian J. OVJeill University College Dublin, Dublin, Ireland
David W. Osborne The Upjohn Company, Kalamazoo, Nlichi-n
Dinesh C . Patel Theratech, Inc. , Salt Lake City, Utah
Lorraine E. Pena The Upjohn Company, Kalamazoo, Michien
Lee K . Roberts University of Utah School of Medicine, Salt Lake
City, Utah
Kenneth B. Sloan University of Florida, Gainesville, Florida
Michael S , Starch Dow Corning Corporation, Mdland, Michigan
Regina Tallon University College Dublin, Dublin, Ireland
Paul S. Uster Liposome Technolo=, Inc., Menlo Park, California
F, L. Vaughan University of Michigan School of Public Health, Ann
Arbor, Michimn
Anthony J , I . Ward* University College Dublin, Dublin, Ireland
Norman D. Weiner University of IVlichipn College of Phamacy, Ann
Arbor, IVlieMgan
Ronald C. Wester University of California School of Medicine, San
FrancLeo, California

"Current affiliation : Clarkson University, Potsdam, New York


TopScal Drug Delivery
PRINCIPLES OF SKIN
AND FORMULATION INTERACTIONS
Practical Considerations for Topical Drug
Formulations With and Without Enhancers

EUGENE R. COOPER Sterling Drug, Xnc, , Malveme, Pennsylvania


DINESH C , PATEL Theretteeh, Tne, , Salt L a b C i t y , Utah

I INTRODUCTION

The purpose of topic& &sage forms is to conveniently deliver dmgs


across a loeafized area of the skin. To develop an ideal dosage form
one must t&e into account the flux of drug across skin, retention
of the dosage form on the skin's surflace, the reservoir capacity of
the dosage form, and the patientsf aceeptslrbiXity of the formulation.
The problem of formulating a drug is complex because of the wide
diversity of drug solubnity in vehicle components and the vast range
in cutaneous Buxss (six orders of mamitude).
The objective of this chapter is to simplify- and make more effi-
cient the development of optimum topic& dosage forms. There is af-
ways a certdn amount of empifidsm to this process, but a str;sltew
based upon fundament& principles of thermodynamics and diffusion
can greatly reduce t l ~ etime required to develop good topic& dosage
forms. Another topic to be addressed in this chapter is how to com-
pare two different: topic& dosage forms. In some organizations,
physical and chemical stabnities, plus the formulatorts opinion of
aesthetics, are d l that is required of a formulation. The sdence
and technolow avdable today make this approach sbsolets, and one
must have a quantitative measure of delirreq to develop a competi-
tive topied dosage form.

if, VEHICLE DESIGN


A, General Considerations
The purpose of dissoldng a drug in a solvent or mixture of sol-
vents i a to facutate the transport of d m g to the surface of the
2 Cooper and Patel

skin. Most drugs are erystdline, and the mere rubbing of sm&X
erystds on the surface of skin. will result in very p m r transfer to
the skin, Delivery from volatile solvents, such as ethanol, and ace-
tone, wnl trmsfes small portions of drugs into the outer layers of
the skin, and the flux will quickly drop off. In fact, m y transpart:
study conducted with such, solvents, will more than likely measure
the itissolution rate of the drug, rather than the transport proper-
ties of skin,

B , Thermodynamic Factors
The drt.idng force for transport is the chemical ptentiitl padient,
and within a single phase, this reduces to the concentration gradi-
ent. To create this gradient within the skin, one normdfy dissolves
the drug in a solvent or vehicle and establishes a cer-tdn concentra-
tion of drug in the outer surface of the skin, Different vehicles
can provide different concentrations of d m g at this interface and,
thus, the d r i ~ n gforce and, hence, the flux will be a function of
the vehicle. In generd, one does not know the concentration of
d m g in the outer surface of the skin, but there is a simple tktermo-
dynamh relationship to compare the concentration of drug in the
skin for different vehicles.
The chemical poten tiat or activity is the continuous vaAable
across interfaces f 1) and, thus, one has the fiirlowing equation at
the skin -vehicle boundary :

where av is the tactidty of the d m g in the vehiele, and as is the


acti.vlty of the &rug in the skin (assumed to be a; lipid medium for
most situations). Thus, to produce equivalent actidty in the skin
with different vehicles, one only has to ensure equal actidties in
the vehicles. The actidties are usually w ~ t t e nas

where y is the aclidty coefficient and C is the concentration. Thus,


for equivdent aetidtg for vehicle 1 and 2 , one has

To estimate the concentration in vehicle 1 to @ve equaf actidty to


vehicle 2 , one has
PmeticaE Considemtians for Use of Enhancers

The simplest way to estimate 7.2 and y l is to measure the solubsity,


where S is the solubBfty.

Thus, E q . 3 becomes

Another waty to state Eq. 5 is that egud fractions of the sola-


bi_lity will provide flop equal a c t i ~ t i e sin various vehicles. In most
applications one wishes to optimize the flux in a formulation, and,
the simplest winy to do this is to work with saturated vehicles. An
example of equd flux at saturation for markedly different saXubitj:ties
is shown in Pimre 1, and further use of %hi@ con~eptwill be dealt
with later by way of a practical example. Despite that the solubn-
it-y of sdicylie acid is 100 times grezzler in propylene glycol than in
water, the fittxes mcross human epidermis are nearly the same.

C, Finite Doses and Vehicle Evaporation


Except for endosed systems, such as transdermlll patcckes, one is
limited lo a very thin f s m , Product application of a c r e m is usual-
ly limited ta a few milligrams per square centimeter, because appli-
cation of mare material will result in a sticky f%m that will easily be
removed by clothing and such. The consequences of thin films can
be easily seen by consideAng the solution of the appropAate diffu-
sion equations, I f a finite dose of volume 7i7 i s applied to surface
of area A and it is assumed that the vehicle does not penetrate or
evaporate, the equations .for the flux J, are
Cooper and Patel

375

150

1
P
125
Y
.u
-5i
100
0% Salicylic Acid in
4 75 Propylene Gfycal

Figure I s 1 Penetration of ~ d i c y l i cacid acro8s human epidermis in


d t r o from saturated solutions.

C ( x , O) = O, C ( h , t ) = O f 81

where h is the thickness aP the stratum corneum, E) is the diffusion


coefficient, CV is the concentration in the vekicfe, R ia the partition,
coefficient between skin and She vehicle, C i s the concentration in
the skin. These equations reffect the assumptions that diffusion in
the vehicle is much faster than in the skin, and that sink eondi-
tions exist below the skin, The solution of Eqs. 6 through 9 can
be obtained from the existing solution far heat conduction (2) and
is

expr-an2
J = -13
n
2 2 f loll
x = h n=l cos a, f B + B + R]
Practical Considerations far Use of Enhancers

where

a is a r m t of a tan a =
n

and h, is the thickness of the vehicle. The dimensionless param-


eter, T, is used fur a general solution in terms of number af lag
times, Thus, a p a p h of ffux versus T can be used for any diffu-
sion ewfficient and membrane thickness.
A plot of the r a t h J/Js versus r is @ven in F i p ~ e2 for sev-
eral vdues of 8 , For f3 < 0.1 the flux is near the steiady-state
vdue for many lag times. For f3 I, however, the maximum flux
is only 60% of the steady-state value, and it decays rapidly after
two lag times. For example, consider the application of 5 mg/cm2
of cream contdning 20%oil and l"k,salicyHc acid. After evaporation
one is left with an oil faxn about 20 vrn thick (corresponding to 1.
mglerntif). Here, R I, and & is unity if we use h 10 For
a molecule like sdicylic acid that penetrates well ( s ~ Q &lag time),
one cannot reach steady-state flux, nor maintdn i t , without a thick-
e r film or a solvent in which sdicylic is more soluble (e. g, , poly -
ethylene glycol) ,
This type o f behador is observed in d v o fmonitufing urinam
excretion) for benzoie acid penetration fmm a lipophilic matrix (31,
The mathematic& descdptfon of the 2n vrivu situation is sirnifiar to
that described here, but it is more involved because of the additim
of whole-body pharmacokineties ( 4 ) .
A. simple in vitro test to determine the effcjlct of film thickness is
useful before conducting axz in viva study so thsrt one can know the
film thickness required tu obtdn steady. state, Evaporation and
penetration of the vehicle eornpsnents that are not extremely volatne
will complicate the andysis but will serve to keep the concentration
higher and, thus, m&ntain the flux more in a plateau re@on, al-
though the diecay rate will be faster when the reservdr i s depleted,

I), Penetration Enhancers


The performance of a topical dosage form is linked to the Bux of
d m g acmss the skin, unless one has reached beyond the plateau on
6 Cooper and PateZ

o i a 3 4 5 6 z 8 9 10
7

Figure 1.2 Effect of thickness on mtaneous flux.

the dose-response curve. Such response is easUy observed in the


effect of togiedly applied nonsteroidaf anti-inaarnmatory a ~ n t on s
p i n e a pigs irradiated with ultradolet light. The blanching of the
erythema can reach a mmimum, after which higher concentrations
prodde no further benefit, Human skin is a much greater barriter
and, thus, enhanced penetration i s of greater vzlfue for human skin
applications.
Penetration enhancers can be of great vdue, and formulations
induding enhancers should genesally be prepared as an option, to be
evaluated in the in viva system. Enhancers can be calegodzed con-
veniently in terms a f the type of drug to be delivered. Far polar
molecules, surface rr~tiveagents ( 5 , 6 ) are v e q effective enhancers.
For lipapk2ic molecules, dimethyl sulftoxide ( 7 , 8 ) is the elassie en-
hancer, but other agents such as Xa~rocapram(Azone; 9) &ad polar
lipids (10) are very. effective. A simple model for viewing the ef-
fects of enhancers on skin is to regard the barfier as consisting of
two parallel pathways. The polar pathway is thought to be hydrated
protein that is quite sensitive to conl"armational ehanges induced by
surfactants, herat, and the like. A marked contrast between the
effects of surf"rtetanls on polar versus nonpolar molecules serves to
mustrate this point (6). Enhancers for the nonpolar pathway are
thought to frluidize "te lipids (11). This concept is quite reasonable
Prae tical Considerations for Use of' Enhancers 7

when considering diffusion prmcesses in a continuous medium. The


for"mu1ation of these enhancers is not simple because they often in-
teract with emolfients and such, that are put into a cream and can
be rendered ineffective. Because of potential iirstatisn problems
with enhancers, they are most practicat for short-term use. Par
prolonged application, more attention will have to be placed on reg-
ulating cutansous irritation.

111, EVALUATION QF VEHICLE PERFORMANCE


A, In Vitra Evaluation
As descrjibed in earxer sections one can use thermodynamics and
elementary diffusion theory as p i d e s for optimizing the flux. In
the find andysis, however, one should measure the penetration sf
the drug front the actual vehicles. A vadety of experiimental. meth-
ods (descdbed elsewhere in this book) are avdfable for determining
cutaneous flux in vitro, and they are quite simple to use. These
in vitm techniques are an inviifuable wide to the formulation of
topied dosage firms and should be a standard laboratory tool.
Without a measurement of penetration, there is ~ r t u d l yno way to
compare different dosage forms, When employing enhancers, jt is
absolutely necessary to make these measurements.
We prefer to use human skin, when possible, although animd
skin is acceptable, to compare different formulations , Our experi-
ence has been that for comparing vehicles there is good correlation
between human and animal skin, but sometimes animal skin will e v e
a false-positive result when studying skin penetration enhancers.
Although there are a number of diffusion apparatus avdable, we
prefer the one shown in Fiwre 3 (12) because it is relatively inex-
pensive and occupies very little faboratory space,
Typicfit d8t8 from in d t r o experiments are depfcted in If""i~;ure
4
for penetration from large reservoir systems in which the efhet of
an enhancer is shown. Were, the effect of a lipid Buidizing agent
.
is shown for the enhancement of lipophilic drugs (10) The lag
time is an important factor to minimize for treatment of acute symp-
toms such as pmfitis and burns, For treatment of less acute
symptoms, this f a ~ t o ris of less importance, Although for transport:
across homogeneous membranes the lag time is only a i'unction of
the diffusion, coefficient and not the partition coefficient, it is a
function of partition coefficient for heterogeneous membranes ( 13).
The fag time may not be well defined in the presence of enhancers,
espectidly if the membrane b a r ~ e rproperties change over a reason-
able time pefiod. The enhancer, however, should shoden the on-
set time for pharmacolo@c response,
For d m g s that penetrate rapidly, one ean use a vehicle with a
lower partition coefficient to sustain the deliveq, rather t h m have
8 Cooper and Patel

Figure 1.3 Diffusion apparatus.

pulsed delivery. Again the in d t r o method is required to determine


the actutinl Buxes, &though the theory can be used as a v i d e .
Transport acposs diseased skin can patentidly be quite different
f I d ) , and. one must consider this possibaity, part-ieularly if the skin
i s thickened or braken,

6. In Vivo Evaluation
The most quantitative in d v o evduations of toplcdly applied drugs
have been, done with skin grading, in which blanching of norm&
skin by steroids or the blanching of erythema by nonsteroidal anti-
inBammatow agents is measured. En these models, the onset of re-
Praetieat Considerations for Use o f Enhancers

Figure 1.4 Effect of oleic acid on salicylic acid (1%)in d t r o pene-


tration across human epidermis,

sponse is related ts lag t i m e , and shorter lag t-imrss arc: ernr red as
good, For disease states, it 3s not clear whether a large, pulsed
deliveq is better, or a slow, prolonged delivew is preferred. AG-
curate knowledge of vehicle delivery rates correlated with clinicFz3
results could lead to new vehicles with very different delivery pro-
files. These systems, in turn, could be evduated clinically to de-
termine the best delivery rate, The methods and skill eer-t&nly ex-
ist to mswer same of these key questions, which undoubtedly will
be mswered as the dosage forms become more contmllable for de-
livery rate,
Topied treatment of skin cancer, wafls, batdness, and so on,
may be possible, not only because of new drugs, but &so because
better vehicles can be designed to enhance delfveq. The skin is
an accessible tissue of large surface area and should be a .rY.iable
port of entry for drugs, Vehicles with supesior delivery will un-
doubtedly play a substantid prole in increasing the usefulness of
this port of entry,
10 Cooper and Patel

Table 1.1 Vehicle Composition

Vehicle 1 Vehicle 2
Ingredient (wt%) (wt%l

Sorbitan stearate 1.2 1.2


Polysorbate 60 3.6 3.6
Pmpylene glycol 11.9 0
Capryliclcaprie triglyceride 0 11.3
Water 83.3 83.3

IV. FORMULATION EXAMPLE : DESONIDE

It is often desirable ta change key ingredients in a formulation, and


this substitution can occur without any loss in vehicle performmce.
A s an example consider the substitution of capryliclcapric triglyc-
eride for propylene glycol in a formulation for the steroid desonide.
The reason for such a replacement is to move from a hydrophilie
base, such as propylene glycol, to a Upophilic base. I t is also ex-
pected that, under oeclusion , the triglyceride system ( 15,16) would
be less irritating than the propylene glycol vehicle.
The experimental vehicles saturated with desonide are listed in
Table 1. According to thermodynamics, the fiux should be the same
from both systems, unless there is an effect on the barrier proper-
ties of the skin. The fluxes across hairless mouse skin and the
vehicle solubilities (17) are &en in Table 2 , from which it Is ob-
served that the fluxes are indeed the same (within experimental
variations) , and the solubilities are only slightly different. When

Table 1.2 Desonide Solubility and Flux Across Hairless Mouse Skin

SoIubMty Flux
Vehicle (mglml) (mgfhr x 104)

1 Propylene glycol 0.40 2 0.02 8.3 2 4


2 Caprylicfcapric triglyceride 0.53 2 0.03 13 +8
Prae tical Considerations for Use o f Enhancers li

these two systems were compared in the rrasaconstdctor assay ( 18)


there was no difference in biolo@;icd response between them, Under
closed -patch irritation studies, however, the gropylene glycol -based
vehicle was more irdtating, as expected (1v ),

It may not &wags be possible to carry each different vehicle


through all the steps just descfibed. One part that can always be
conclueted, is the saturation step. If the solubilities are tenfold or
greater in dif hrence , one might encounter reservoir Umit ations , as
discussed earlier, for their Elms. Unless there are toxicity eon-
eerns, one should seek to find a very soluble vekiele to prodde a
long-lasting reservoir, The in vitro penetration step may have to
be omitted, but it is; highly recommended that it be included be-
cause, Tor example, solvents such as polyethylene glycol (10,19-21)
or glycerol have a tendency to retard penetration compared d t h
pmpylene glycol.

REFERENCES

A . Katchalsky , P. F , Curran, in NonequiEibrium Themodynam-


ics in Biophysics, Harvard Uuriversity Press, Cambridge, Mass.,
p. 113, 1965.
.
H, S . Carslaw, and J , C Jaeger, Conduction of Neat in Sol-
ids, Bx-Eord University Press, Oxford, Eng, , p. 128, 1959.
T J , Franz, psrsonetf communlcatlion.
I

E. R . Cooper, and B. Berner, tP, Pharm, Sci., 74:1100, 1885.


f". R . Bettley, and E , Donoghue, Nature, 27: 185, 1960.
E , a. Cooper, in Solution Behavior of Surfactants: Theoretical
.
a n d Applied Aspects ( K . L. Mittal, and E. J. Xtendler, eds.)
Plenum Press, N e w York, p, 1505, 1982,
S . K , Chandrasekaren, P. S. Campbell, and A . S. Michaels,
A I C N E J , , 23:810, 1977,
.
S K . Chandrasekaren, Drug Dev, Ind. Plxarm. , 8:82'7, 1983,
.
R E3, Stoughton, Arch, DermatoE, , 2 2 8 : 474, 11382,
E . R, Cooper, J. Plzarm, Sei, , 73: 1153, 1984,
E . R Cooper, in Percutaneous Absorption: Principles and
Practices ( R . Bronaugkr, and H , 1 . Maibach, eds .), Mareel
Dekkes, New Uork, 1985.
E , R e Cooper, and E , W. Merdt , e7, Conlmiled Release, I : 161,
1984.
3%. M, Barrer, in Diffusbn in Bafymers ( 3 . Crank and G. S,
Park, eds.). Academic Press, New York, p. 165, 1968,
. .
B W Barry, Dermatological Formulations Percutaneous A b -
sorption, Marcel Dekker, New York, p. 132, 1983.
Technical Bulletin, Kay Fries, Ine ,, l-tockleigh, Hew Jersey.
12 Cooper and Patel

16. Technical Bulletin, Henlrel , Waboken, New Jersey,


17, f), Patel,D. Welsh, and M, Bakes, J , Sac, Cosmet. Chen,
(i-n press),
1-8, A , W , MeKenzie, R. B. Stoughton, Arch, Dermatol., 86:608,
1962,
19. J. A. Faueher, E , 33, Coddasd, and R , Mulkami, 3 , A m , Oil
Chem. Soe,, 56: 776, 1899.
20, A . A , BeXmonte, and W, Tsai, J. Pharm, S c i , , f57:517, 1978.
21, J , L , Zatz, CTFA Cosmel, J., 15: 6, 1983,
The Importance of Epiderma
Lipids for the Stratum Corneum
Barrier
PETER M. ELIAS Vetemns Administration Medical Center, and
Univemity of California SchooZ of Medicine, San F~aneisco,Caltfornia

I. EVOLUTION OF C U R R E N T CONCEPTS OF
STRATUM GORHEUM STRUCTURE

The of$ image o f m m m a m stratum corneurn, based on routine his-


tolo@cd preparations, is one of a loclsefy bound layer in rragous
.
stages of disorganization and shedding (Table 1) This misleading
image, refuted by phy sieoehemical ( 1), fmzen-section ( 2- 4) , and
freeze-fracture (5-7) studies, led to the long-held ~ e v vthat the
barrier resides at the stratum granulosum (SGf -stratum cornam
( S C ) interface (8). New appreciation for the intsgdty of the SC
initidly led to the concept that the TUB thickness of this layer Is
funct~orrrtllycompetent ( 1,9) , a dew that ando@zed the stratum
corneum Po a homageneous film ('3pXstic wrap" hypothesis; 91, a
*euv supported by x-ray diffraction studies that demonstrated. high-
ly ordered lipid structures in the SC (10). The homogeneous fBm
a n d o w dews percutaneous tmnsport as occurring transcelllularfy,
without regard to membrane, intereeflufar, or intraeelfuIas cornpart-
rnents (9,111, and assumes that the SC i s d e ~ t d i z e d(-lee., meta-
bolically ine&, with perrneabnity phenomena revlated solely as a
passive membrane ; 9) ,

A, The Two-Compartment Model


Studies outlining the eddsnce supporting a two-compartment model
are outlined in Table 2 , In 1968, Middleton first reported cer/t&n
str~ghtforward,yet elegant, studies that tended to refute the ha-
moeneous film concept (11). Whereas I r v h Blank had shown years
Table 2.1 Evolution in Understanding of the Stratum Corneum

1, Disorganized, nonfunctional layer in various stages of shedding


2. Homogeneous film-the "plastic wrapnp'hodel
3. Lipid-protein compartmentalization-the "two-compartmentfbodel
4. Metabolically active tissue with ongoing modulations in structure
and composition

earlier that organic solvent extraction destroyed the water-holding


capacity of SC (12), Middleton showed that pulverization was as ef-
fective as solvent extraction, thereby providing evidence that SC
lipids might be organized as osmotically active membranes within the
.
SC Shortly thewafter, the existence of separate lipophilic versus
hydrophilic pathways of percutaneous absorption was suggested from
physicochemicezl studies ( 13).
The destruction of stratum corneum during processing far rou-
tine histofogical and ultrastructural preparations obscured further
advances until the early 1970s, when the cells of the SC were shown
by frozen sectioning to comprise tightly arrayed, polyhedral struc-
tures in vertical, interlocking coXumns (2- 4). The initial awareness
of lipid-protein sepegation to specific tissue compartments came
with freeze-fracture replication, which showed, for the first time,

Table 2.2 Lines of Evidence for Lipid-Pmtein Compartmentalization


in the Stratum Corneum

Pulverization destroys the water-holding ctipacity of SC ( 11) .


Hydmphnic versus lipophilic substances cross separate SC path-
ways (13).
Freeze-fracture reveals lipid lamellae in SC interstices (5-7).
Frozen sections display neutral lipids in SC interstices (6).
SC can be dispersed into individual cells with organic solvents (14,
15).
Isolated SC membrane sandwiches account for most S@ lipids (16).
X-ray diffraction shows ordered lipids in isolated SC membranes (17).
Catabolic enzymes coloealize with lipids in SC interstices (18-20).
Importance of Epidermal, L i p i d s

the presence of multiple, broad lamellations in the interstices of


several types of mammalian, keratinizing epithelia (5-7). Several
other lines of indirect evidence also supported such structural het-
erogeneity, including lipid histochemistry of frozen sections ( 6) , as
well as the ability of SC to be dispersed by certain organic solvents
1I , . Definitive eddence for the compax7tmentdizatjion of lipids
came with the isolation of SG membrane '%sandwiches," containing
trapped interceflular gpids, These preparations ( 2.6) :( 1) comprised
about 50%lipid by weight, and accounted for over 80% of SC lipid;
( 2 ) displayed the same lipid profile as whale S C ; (3) contained the
same broad lamellae on freeze-fracture as are present in the inter-
stices of whole SC; and ($) generated the same ordered x-ray Qt-if-
fraction pattern previously ascribed to the "interfilarnentous lipid
matsixty (17). More recently, the coloe~lizationof lipid catabolic en-
zymes to SC membrane domains, both by ultrastructurd eytochemis-
t r y and by enzyme biochemistry, can be considered further e-idenee
for the stmctural heterogeneity of mammalian SC 118-26).

B. Localization of the Barrier


The locdizatiorr of the barrier continued to be debated: Is the en-
tire SC functionaI1y competent, or does the principd barder reside
in the lower layers? Tracer perfusion studies, as early as 1969,
showed that water-soluble molecules, injected into the dermis, do
not reach the SC (21,22). Outward percolation halted in the outer
SG , at intercellular sites engorged with discharged lamellar body
contents (5). These studies, although admittedly employing tracers
considerably larger than the water molecule itself, pointed to the
presence of a basrires in the outer S G . Direct eGdence of the bar-
rier eapab2ities of different layers of the 5 6 came with the recent
isolation of intact s h e t s of parcine stratum compacturn, after prior
enzymatic stripping of the stratum disjunct~mt (23). Whereas these
studies did not address the eapadty of the stratum disjunetum to
contribute to b a r ~ e rfunction, they did demonstrate the function&
integdty of the stratum compacturn.

It , EPIDERMAL DiFFERENTIATIOM

The ultimate goal of epiderntzl differentiation $8 tke production o f


t he stratum coxmeurn, a process more correctly caUed carnification
rather than keratinizat-ion,

A. Major Structural Components


Modern research in epidermd cellular and molecular biology now rec-
ognizes four distinct cellular events that occur duA-ing the process
of cornification , ineluding keratiniza tion, the synthesis of the prin-
d p d fibmus proteins of the keratinoeyte ; keratohyalin deposi-.tion,
associated with synthesis of f'histidine- rich pmtein ," stratum cornrt-
um basic protein, or fgaggrin ; formation of highly cross-linked ,
insoluble peripheral envelope of the cornwcgte, composed of two or
more prewrsor proteins, including involucrin and keratolinixr ; and
the generation of neutrd lipid-en~chedintercellular domdns, result-
ing from the secretion of distinctive stmctures termed lamellar bad-
ies (membrane-cornling granules, Qdland bodies, Xreratinosomes, ce-
mentsomes; 24).
Although the precise role of these components in protective func-
tion rerndns to be discovered, the insoluble stratum corneum enve-
lope appeas to provide a ri@d stmeturd eetoskeleton for the cor-
nified cells, a scaffold for insertion of keratin filaments, and a high-
ly- resistant barrier to external chemicd assault. Additionagy, in
conjunction with ext racellular lipids, the peripheral envelope may. se-
lectively r e p l a t e the permeability of the cornified cell to water, hy-
.
dmphobie electrolytes, and nonelectmly-les ( If) Although both
keratin fgament s and keratohyain -derjivecX prot eins are major e p i d e ~
mail differentiation products, their function remains conjectural.
Clearly, the filament-matdx complex in the cornified cells imparts
stmcturat and chemical intepity, acts as a filler to inddent ultra-
violet radiation, and acts as an absorptive "span@" for water and
other sm&T hydrophobic molecules , Whether the cornifled eeU, inte-
rior also represents a potentid pathway or reservoir for substances
in transit across the stratum earneum i s less clear.

El . Lamellar Bodies
The lamelliar body, a 0 . 2 to 0 . 3 ~ n ndiameter, ovoid, secretory or-
gmelle, which is considered the centrd actor in the formation, of
the irttercellular "'mortar," is synthesized pdmady within the spi-
nous eel1 m d then displaced to the apex and periphery of the gran-
ular cell ( 2 4 ) . Xn response to an unknown signal, it fuses with the
plasma membrme , secreting its contents into the intercellular spaces,
thereby generating an expanded intercellular compartment that con-
stitutes from 1Q% to 40% of the total volume of this tissue (25).
Thus, secretion is one of two cellular events associated wBh lamel-
lar body ~txwytosis;fusion also occurs, and the ""splicingt' of addi-
tional organelle membrane into the plasma membrane may contribute
a large reservoir of surface area that could explain the stratum
comeurn" remarkable water- holding capacity ( 25) ,
Lamellar bodks appear to contain three types of matedds:
first, sugars, in the form of glycosphingolipids and possibly glyco-
proteins ; second, free sterols and phospholipids; m d tl.rird, a se-
lective array of hydrofgtic enzymes possibe charged with degrading
Impartance of Epidermal Lipids 17

"PrroBarr~er"Lipids: Convers~onof "Pro-Barrier" Ltplds to


Non.Pol8r Products (Lipases, Glrcos~dases)+
Barrier Function

Galaboiffi Enzymes:
Acid P8esphotase.
Dasquamatioa
Proteaser, t~pases, 2) Dsgradation of &oe.LLprd tnterccll~lar
Glycosldases Specks (A& Phos$hatase, Prateares)

Figure 2.1 Model that summarizes avaLlable data about dynamic


changes in the stratum corneum interstices that result from the
secretion of lamellar body lipids and hydrolytic enzymes.

intercellular materials (26,27; Fig. 1). After outward migration into


the stratum corneum, dramatic changes occur in both the morpholog-
ic appearance and histochemical reactions of the secreted material:
in thin sections and h freeze-fracture replicas, large planar mem-
branes gradually replace the short disks of the initially secreted
material ( 5- 7,20). A family of lipid catabolic enzymes, including
steroid sulfatase, sphingomyelinase, phospholipase A , acid Iipase,
and possibly, glycosidases contribute to the ultimate demadation of
residual polar lipids (gIycosphingolipids, phospholipids, and choles -
terof sulfate), leading to the formation of these broad membrane bi-
layers (18-20; see Fig. 1).

111. COMPOSITION OF STRATUM CORNEUM


LIPIDS

Despite the 4-decade-old observation that stratum corneum lipids are


generally nonpolar and enriched in cholesterol, the full panoply of
mammalian stratum corneum lipids has only recently been unraveled.
The stratum corneum is virtually devoid of phospholipids and is se-
leetively ensiched in ceramides, free sterols, and free fatty acids,
with smaller quantities of glycolipids , sterol esters, triglycerides ,
cholesterol sulfate, and hydrocarbons present as we11 (28-30; Table
3). Yet, there is a gradient vdthin the stmtum corneum itself:
whereas the stratum compactum still contains demonstrable levels of
phospholipids, glycosphingolipids , and cholesterol sulfate, only the
last persists into the stratum disjunetum (20; Fig. 2). This change
Table 2.3 Composition of Mammalian Epidermal Lipids

Lipid type Living layers (%) Stratum corneum (%I

PhosphoIipids 40 Trace
Sphingolipids 10 35
Cholesterol 15 20
Triglyce~ides 25 Trace
Fatty acids 5 25
Other -5 10
Totals : I00 100

@Glycosphingotipids
A Phospholipids
Il Cholesterol sulfate

c
Inner SC
SG Outer SC
LAYER
Figure 2.2 C h a n e s in lipid composition that occur during stratum
corneum trmsit. Note that smalX amounts of phospholipids (dia-
monds) , glyeolipids (circles) , and cholesterol sulfate (squares) re-
main in the inner stratum eorneum (SC) (stratum compactum),
whereas only cholesterol sulfate persists in the outer SC. SG , stra-
tum granulosum.
in composition presumably reflects ongoing metabolic activity, fur-
ther dispefling the notion of the stratum corneum as an inert tissue
( 20; see Table 1). Despite the paucity of phospholipids, glyco-
sphinogaGpids, and cholesterol sulfate, these constituents apparently
can arrange themselves into membrane bBayers, possibly by exploit-
ing the ampkipathie properties of sphhgolipids (25,29--313, which
apparently are capable of extensive hydrogen bonding. Yet, the
hydrophobic, long-chain bases (31,32) , the long-chain , fully satu-
rated fatty acids (28,29,31,31; Table 41, and the endchment of
sphingolipids in linoleic acid ( 3 1,33- 35) , all make spkingolipfds par-
ticularly good candidates for epidermal water-proofing (Table 5).
Hence, sphingolipids are bipolar, possessing both a relatively hy-
drophilic terminus, capable of intermolecular bonding that can form
ordered membrane stmctures, and highly hydrophobic moieties that
could. be highly water-repellent. Stilil unresolved is the fate of
these lamellations during the first stages of shedding: X s membrane
bilayer break-up a prerequisite for shedding? Do further changes
in composition or the physicochemical properties of these bilayers
lead to desquamation?

1'4. LtPlD Bf OSYNTHETtG GRADIENTS


I N THE EPIDEWMtS

Certain striking features of the modulations that occur in lipid com-


position during epidermal differentiation suggest that the nonpolar
mixture that ultimately resides in the stratum carneum is impareant
for barAer function, Yet, a lipid analflic approach provides only
indirect evidence about the function of et raturn cornelxm Upids ; fur -
tlzermore, it can not define the role, if any, of neutral lipids, par-
ticularly free fatty acids, free sterols, and the smdler quantities of
dkanes , sterol esters, triglycerides , and cholesterol sulfate in bar-
rier function,
In the early 1980s, we noted that the skin of both rodents and.
primates synthesized abundant chalesterol and other nonsaponifiable
lipids (36,371, In fitct, the synthetic actidty of the skin rivalled
that of the liver and gas"tointrtstiaa1 tract, the two major putative
sites af sterologenesis ( 38) , Moreover, in contrast with hepatic and
gastrcthtestinal synthesis, cutanmu8 sterologenesis was not in-
a u e n ~ e dby circulating sterol levels ( 3 8 ) , suggesting an unusual
degree of independence from systemic regulation. To determine
whether or not barrier functiian can innuenee epidermd lipid syn-
thesis, we have employed a -functional, rather than a purely analfl-
i c d , approach. The basic s t r a t e m is first, to perturb epidermtll
barrier function, typictally with an organic solvent such as acetone
Tablet 2.4 Fatty Acid Composition of Lipids from Human Abdomen
Stratum Comeum (mol%)

C no, Free fatty acids Sterollwax esters Ceramides

14: 0 3.8

Trace

Trace

Total 1OO. 0 100.2 100.0

Source: Modified from Ref. 28.

Tabla 2.5 Evidence for Rale of SphingoHpids in Barrier

Carriers of extremely long-chain, saturated fatty acids


Replacement of long-chain by short-chain fatty acids in marine mam-
mals
Principaf mpositories of linolelc acfd (essenaal fatty acid deficiency)
Removal reqdred to completely break the barrier
Importance of Epidermal Lfpids

TIME (hr)
Figure 2.3 The time course of epidermd lipid biosynthesis and bar-
rier function after treatment with acetone. Note that transepidermal
water loss (TEWL) , totd nonsaponifiable lipid biospthesfs (TNS) ,
and cholesterol (C) synthesis exhibit a parallel return toward normal
over 24 h r (from Ref. 39, reprinted with permission).

and, then, to correlate barrier status with both an assessment of


lipid replenishment in oil red Q- and nile-red-stained frozen sec-
tions, and with rates of lipid biospthesis (from tritiated NzQ) in
the same samples. These results have shown ( 1) that the epidermis
is a major site of sterol synthesis, acwunting for about 30%of total
cutane~usstemlop~@nesis (38); and (2) that both epidermal sterol
and fatty acid synthesis are stimulated by perturbation of the per-
.
meability barrier (39,40) Morwver , such stimulation is localized to
treated sites, is limited to the epidermis (39-41), corrects as bar-
rier function returns to normd ( 39; Fig. 3) , and correlates, first,
with removal and, subsequently, with repletion of lipids in the stra-
tum corneum (39,41). (3) Epidermal sterologenesis is stimulated in
essential fatty acid deficiency in relation to the defect in barrier
function, (i-e., lipogenesis rates normtclize with occlusion despite
persistence of the underlying deficiency state; 42). (4) Synthesis
is normalized when an impermeable membrane is applied to the per-
turbed skin (39-41). (5) Finally, the relationship of epidermaf.
Iipogenesis to barrier function is underscored further by the lack of
Table 2.6 Evidence That Barsier Function (Water Loss) Regulates
Epidermal Lipogenesis

Dietary or solvent-induced barrier disruption stimulates epidermal


lipopnesis ,
Extent of lipid biosynthetic rates parallels severity of barrier defect.
Normalization of lipid biosynthetic rates parallels barrier recovery.
Occlusion with impermeable, but not vapor-permeable membranes af-
t e r barrier disruption blocks acceleration of lipid biosynthesis.

modulations in cutaneous sterologenesis in response to either vitamin


D deficiency or excess, despite the known impedance of the cutane-
ous free sterol, 7-dehydroeholestero1, for vitamin D synthesis (43).
In more recent studies, the role of the water molecule itself as
the regulatory signal has been explored in greater detail (41; Table
6). Whereas occlusive membranes blunted the expected burst in
lipid biosynthesis that fbllows barrier dismption, application of semi-
pemeable membranes did not block synthesis, retard the rate of re-
turn of normal barrier function, nor impede the return of stainable
Iipid in the stratum corneum. These studies lend strong support to
the concept that the rate of water loss itself may be the regulatory
signal for epxdermal apogenesis,
Although the fomgoing body of evidence demonstrates the ca-
pacity of the epidermis to synthesize lipids in response to barrier
requirements, they leave unanswered the sites of synthesis In the
epidermis and the relative importance of specific SC lipids for bar-
rier function. The rapid rate of return of barrier function to nor-
mal [i.e,, by 15-20 h r , even after exhaustive delipidization aceom-
panied by transepidermal water loss (TEWL) rates > 1000 ppml(cm2/
h r (39,40)] suggests that lipid synthesis may not be limited to the
basal layer but may extend to, or even be accelerated in, the outer
layers of the viable epidermis. Indeed, in both in vivo and in organ
culture, the rates of epidermal lipogenesis in the SG of nwnatal mouse
epidermis approach those in the subjacent basal;/spinous layers, even
under basal conditions (44; Fig. 4), that is, in the absence of bar-
rier perturbation. The importance of this observation is underscored
further by the simultimeous, preupitous decrease in protein, DNA,
and C02 generation in the same layer (44). Whether the SG IS the
slte of accelerated synthesis after barrier perturbation is currently
under investigation,
Certainly, the foregoing studies establish that nonsaponifiable
lipids (i.e., sterols and possibly hydrocarbons) are important for
Importance of Epidermal Lipids

Figure 2.4 Synthesis of virtually dl epidermd lipids is accelerated


in the stratum granulosurn in comparison with the bas& 'tayer,

barrier function, Although the role of splzingolipids has not been


assessed by the metabolic approach, the observation that fatty acid
synthesis Is d s a r e p l a t e d by barrier requirements (40) i s consistent
with a rofe for acylated lipids, including sphingollipids , in barder
function. Recently, we assessed the importance of relatively polar
St2 lipid species (i,e, , sphingalipids and free stemts) compared with
nonpolar species (1.e., Eree f ~ t t yacids, sterol esters, and hydra-
carbons) for barrier function ($5). With use af a highly nonpolar
organic solvent, petroleum ether, instead of the mare bipolar s d -
vent, acetone, we found that removal of highly nonpolar species
done produced a significant break in the harder, but TEWL rates
never exceeded 150 mg/(cm~/hr). Whereas petroleum ether removed
large quantities of nonpolar lipids, it left the polar species in place.
In contrast, acetone treatment caused profound barrier defects, as
more and more Xipfd was removed. Moreover, whereas petroleum
ether removed only nonpolar lipids, acetone removed a much larger
proportion of polar species., These experiments show I l l a linear
relationship between lipid content and barrier function; (2) that non-
polar lipids, in the absence of sphingolipids , provide a "first -lineg'
of barmier function; and (3) that relatively polar lipids appear to
provide a more profound level of barrier integrity and cohesion.

V. CLINICAL AND PATHOPHYSIOLOGIC IM-


PLICATIONS OF THE TWO-COMPARTMENT
MODEL

The two-compartment model of the stratum corneum has stood the


test of several pwdictions. First, in situations for which barrier
function is defective, such as in essential fatty add deficiency,
there is a demonstrable abnormcility in the lamellar body secmtory
system that leads to inadequate intercellular lipid deposition in the
stratum corneum (46; Table 7). Moreover, lipophilic substances ap-
pear to traveme intercellular, rather than trmsceIlular, routes in
pawing across the stratum corneum (47), consistent with sequestra-
tion of lipids to intercellular domains. There is also evidence that
quantitative differences in Iipid content are more accurate predictors
of regiond variations in skin permeability than either stratum cor-
neum thickness or cell number (Table 83, that is, neither the thick-
ness nor the number of cell layers in the stratum corneum correlate
.
with permeability (48) These findings explain both the observation
that lipophilic agents, such as topical steroids, readily traverse fa-
cial stratum comeum (10%-20% lipid by weight) and the poor barrier
properties of the palm and sole stratum corneum, (1%-2% lipid by
weight). This model also may explain why eczema occurs most read-

Table 2.7 Importance of Stratum Gorneum Lipids for Barrif?r

Solvents and detergents destroy barrier while extracting lipids.


Barrier properties of different skin sites [face, leg, abdomen,
palms) is related directly to lipid content.
In patholo@cd conditions in which the barrier is defective, there is
a decrease in lipid content (e. g., essential fatty add deficiency).
Importance of Epidermal Lipids 25

Table 2.8 Regional Variations in Upid-Weight Percent and Mstribu-


t2on of Major Lipid Species

Site

Abdomen Leg Face Plantar


(n-4) (n=4) (12-3) (n=3)

Lipid weilfht ($) 6.5 2 0.5 4.3 1 0.8 7.2 10.4 2.0 F 0.6
Polar lipids 4.9 1 1.6 5.2 11.1 3.3 t 0.3 3.2 1 0.89

Neutral lipids 77.7 25.6 65.7 1 1.8 66.4 1 1.4 60.4 t0.9

Source: Ref. 28.

Table 2.9 Potential Practical Applications Based upon Current


Knowledge

Therapy of dry skin conditions


severe -to- mild
loeaEzed-to-~neralized
seasonaI
Therapy of aging skin
prevention of further damage
reversal of preeldsting damage
Therapy of defective skin barrier
palms and soles
burn wounds, blisters, stasis ulcers, bedsores
Mmipulatioxt of skin barrier
enhmcement of percutaneous drug delivery
reinformment of barrier (occupational and warfa= considerations)
ily on the pdms and soles , (i. e . , these sites would be most suscep-
tible to further depletion from exposure to hot water, detergents,
or solvents). Furthermore, the two-compartment model also pos-
sesses important implications for desquamatian . That lamellar body -
derived lipids r e p l a t e desquamatian is supported by numerous stucl-
ies that link abnormal desquamatian. to inherited, lipid metabolic dis-
orders and to drug-induced ichthyoses, Because of the limited
scope of this redew, the reader who desires further information
about the role of lipids in desquamation is referred to several; recent
redews (49,50). Finally, Table 9 points to several areas far poten-
tial exploration based upon. the kregoing information.

ACKNOWLEDGMENTS

This work was supported by NXH grant AM 19098 and the Medical
Research Serdce, Veterans Administration, We appreciate the typ-
ing assistance of M r . Bil Chapman and Ms. Sally Michael,

REFERENCES

A . M , Kligman, and 15, Christophers, Arch, Dernatol, , 88: 702,


1964,
E , Ckristophers, J , Invest, Bermatof, , 56: 165, 1971.
I , C. MacKenzie, Nature, 222: 881, 1969.
I), N , Menton, and A . Z .Eisen, J , Ultrastructure Res. , 35:
241, 1971.
P , M. Elias, and R . S, Fdend, J , Celt BioI., 65: 180, 1975,
. .
P M, Elias, J , Goerke, and D Friend, J , lrrvest. Bermatol. ,
69:535, 1977,
P . M , Elias, N . $3. WlcNutt, and f). Fdend, Anat, Ree. 1 8 9 ;
577, 1971.
.
H Stupel, and A , S z a i l f X , Die Wirkung von Waschmitteln auf
die Haul, Heidelberg, Huthig , 1957.
R . J , ScheupXein, and I. H . Blank, Physiol, R e v , , S1:702,
1971,
(I;. Swmbeek , Ac ta Dermalo-Venereal, , 39(supgl. 43) :00, f 959.
J . B. Middleton, B r . J , Dermalol, , 80: 437, 1968.
I,11. Blank, J , Invest, Dematol, 21 3 259, 1953,
A. S. Michaels, S. K , Ghandrasekaran, and J. E . Shaw, 6.
Am, Xnst, Chem, E n g , , 21:'385, 1975.
P , M, Elias, Int. J. Dermatol., 20: 1, 1981.
W, P , Smith, M e S. Christensen, S. Naeht, and E . H , Gans,
J , Invest, Beymatot, , 78: 7, 1382,
Importance of Epidermal LQids 27

8 . Grayson, and P , M. Elias, J , Invest. Demalol., 78: 128,


1982.
P , M . Elias, L. Bonar, S. Grayson, and N. P , Baden, J , In-
v e s t , DermaEol,, 80:213, 1983,
P , M , Elias, M . L . Williams, M. E . Maloney, J. A . Bonifas,
B, E t Bmwn, S, Grayson, and E , H . Epstein, J r . , J, Glln.
Invest. , 74: f 414, 1984,
C , K , Menan, S, Grayson, and P. N. Elias, J , Invest, Der-
matot, , 86: 591, 1986,
P , M. EZias, G , EC. Menon, 5 . Crayson, and B, E , Brown, J .
Invest, Dematoh. (in press), 1988.
E. Scfireiner, and K . WoIff, Arch. RXin. Exp. Dematot. 235:
78, 1969.
.
C A. Squier , J. UZtrastruclure Res. , 43: 160, 1913,
1D. A. Bowser, and R . J . White, Br, d , Dematol, 132:1, 1985,
G. P. Odland, and K . Hcslbrmfr, Curr. Probl. Dematol., 5):29,
1981.
P, M, Eliaes, J . Invest, Demaloh, , 88: 44, 1983.
.
R . K , Freinkel, and T, N Traczyk, J . Invest. Dermatol. , 85:
2 9 5 , 1985.
S, Crayson, A, @. Johnson-Winnegar, X3, U. Wintroub, E . I-1.
Epstein J r , , and P , M, Elias, J, Invest. Dermatol'. , 85: 289,
1985,
M. A . Lampe, A . L , Burlingame, 6, Wh.itney, M. L. Williams,
B , E. Brown, E . Roitman, and P , M. Ellas, 2 . Lipid Res., 24:
120, 1983.
P. M. EIias, 1 3 , E. Brown, I?, 0, F r i t ~ e h ,R , J , Goerke, G . M.
Gray, and R , S, White, J. Invest, Dermatol. , 73: 339, 1979,
@. IM. Gray, and El, d , Yardley, J. Lipid Res., 16: 441, 11975,
6 . M , Gray, and R . J. White, J , Invest, Dermatoi, , 70: 336,
1918.
P . W . Wertz, and C). T. Downing, J , Lipid Res., 2 4 : 7 5 9 , 1983,
6. ELI. Gray, Ft. J , White, and J , R . Majer, Bisehim. Biophys,
Acta 528: 122, 1978,
P , W , Wertz, and f), T, Downing, S , Lipid Res., 28:753, 1982,
P , A. Bowser, D. H, Nmteren, R , J, White, U, IM, T * Routs-
muller, and C. Prottey, Biochim. Biophys, Aeta, 834:419, 1885,
K. R , Feingold, M. H, Wliley, 6. MaeRae, S. R . Lear, A. H.
Moser, G . Zsfgmond, and M, D. SIperstein, Metabolism, 32: 75,
1983.
K . R , Feingold, M. M. Wiley, A . H e Mifoser, S. R , Lear, and
M. D, Siperstein, J. Lab. Cline Med,, 100: 405, 1982,
I(;. R. Feingold, B. E . Brown, S . R . Lear, A. H e Moser, m d
P , Nf. EXias, J , fnvest, BermatoZ,, 81:365, 1983,
(2, K . Nenon, K , R , FelngaId, A , Eli. Moser, E3. E , Brown,
and P. M , EXias, J, L@id Res, , 26: 418, 1985,
28 Elias

G , Gmbauer, K , R. Feingotd, and P. M. Eliar;, 6, Lipid Res.,


28: 746, 1987,
G , Gmzbauer, P I M, Elias, and IC. R . FeingoZd, elin, Res,
(In Press) 1988.
K . R . Tieingold, Z1, E. Bmwn, S , R. Lear, A . H, Moser, and
P . M. Egas, J, Invest. Dematol., 87:588, 1986.
K, R, Feingold, M. L. Wnliams, S, PSlai, G , KI, Menan, B, P,
Halloran, D. ID, Bikle, and P. EM. Elias, Biaehim, Blophys.
Acta, 930:193, 1989,
D, J, Monger, M . E , W2lianns, I(;. R, FeingoId, B , E, Brown,
and P. M. Elias, 6, Lipid Res. 2S1: 603, 1988,
C , Gmbauer, K, R. Feingold, and P. Nf, EIiaa, b, Invest,
.
Bematol, , 88: 492, 1981 f abstract)
P. M, Elias, and B. E , Brown, Lab. Xnvest,, 39;574, 1978,
M. K . Nernanic, and P , M. Elfas, J . Histockem. Cytaehem,,
28: 513, 1980,
P . M. EXias, E . R. Cooper, A. Kore, m d B . E . Bmwn, J,
Invest. De~matol., 76: 297, 1981.
Stratum Corneum Structure
and Transport Properties

STfC E. FRIBERG, IBRAMIM X , KAYALI, and MARIQM MARGOSIAK


Glarksan University, Potsdam, N w Vork
DAVf D W , OSBQRNE The Upjohn Company Kalamazoo, Michigan
ANTHONY J , 8 , WARD* University College Dublin, BubEfn, Ireland

1. INTRODUCTION

The outermast layer of the skin, the stratum corneurn, has an es-
sential role as a harder against the transport of water and of chem-
ic& and biolo@cd agents ( 1,2). This barder is functionally the
sane, for short pesads, in bath lidng and dead s k h ( 3 1 , vvith 1i-
pids behg responsible for the essentid function (4,5). The bio-
logiicd and medied importance of this cannot be overstated, and the
investigations into the i n d i ~ d u a lstructure and over& organization
of the straturn earneum lipids has been extensive (6-12).
These investigations and the scientific examinations remain an
ongoing process (131, but recent investigr;ltions d3ow a limited dis-
cussion of the relationship between structure and transport of water
through the stratum cornearn. This chapter wiU describe the lay-
ered stmcture and redew its importance far tr~nsdermal,transport.

il. LIPID QRCAN1ZATION

The basis for the discussion is the layered structure proposed by


Elifts ( 13) and andy-zed by Dawning C 14) and by us ( 15). The es-
senti& experimental. hformation, was suggested hy electran. miem-
graphs (13) that disclosed the kyered stmcture. Bunt on this in-
farmatian Elfas (13) pmposed the layered stmeture shown in Figure
3 . 1 . The polar head gmups o f the lipids (Table 3.1) are gathered in
layers with the nonpolar chains pointed in opposite direc-t;tons form-
h g layers of methyl groups in the plane where the hgdrocar;bon ends
+Current affiliation: Crarkson Univer~ity, E"otsdam, New York
Wl Cytoplasm Plasma ,Mbrane

Fatty Acid
' - - n

Figure 3.1 Layered structure of epidermal lipids as proposed by


Elias ( 13).

meet. White (16) has recently made a very extensive x-ray inves-
tigation into the stratum corneum lipid strmcture, finding a layered
structure only at high temperatures. At physiolo@cal temperatures
solid phases were also present and the structure was complex.
Our discussion will be limited to the features of the layered
structure, which are more complex than indicated by Figure 3 . 1 .
Low-angle x-ray diffractometry shows that in a model such as Elias"
(13) not all of the lipids are positioned with their polar gmups lo-
calized in the polar layer (see Fig. 3 . 1 ) . Some of them are actually
located in the region between the methyl groups, In addition, the
low-angle x-ray diffraction patterns provide information with which
to estimate penetration by water into the space between the lipid
molecules. Order parameters of lipid groups by nuclear magnetic
resonance (PJMR) (17) are used to confirm the x-ray results and to
provide information on the mobility of groups within the lipid mole-
cules.
The results of low-angle x-ray diffraction and NMR, as well as
their interpretation, are comparatively new to the dermatological
constituency; therefore, the methods used and the interpretation of
results will be briefly discussed.

A, Low-Angle X-Ray Diffraction


The low-angle x-ray diffraction pattern from a lamellar structure is
unique and, at the same time, directly and unambiguously inter-
preted. A structure with dtemating layers of electron-rich (polar
Stratum Corneum Structure and Transport 31

Table 3.1 Composition of Model Epidermal Lipid

Purity Wt%in
Component Source ( %) mixture

PE Avanti Polar Lipids


CholestePyl Research Plus
Sulfate
Cholesterol Fisher
Triolein Sigma
Free fatty acids Sigma
myristic Sigma
linoleic Sigma
oleic Sigma
palmitic Sigma
pafmitoleic Sigma
stearic Sigma
Qleic acid Sigma
Palmityl ester
Squalene Aldrich
Prist ane Aldrich
Ceramides Sigma

groups) and electron-poor material (nonpolar parts) wiU reflect x-


rays according to Braggls law and give a series of reflections as
shown in Figure 3 , 2 . Hence, the appearance of diffraction lines in
.
the ratio 1:2: 3 : 4 . . means that the structure is layered.
The distinction between a erystdline and a liquid crystalline
stmcture is made on the basis of the x-ray diffraction at higher
angles. A crystalline structure gives rise to a series of sharp re-
flections in the range 3.5 to 5 because of the crystdine packing
of the hydrocarbon chains (16). Conversely, a liquid crystal has
no short-range orderSng of the hydrocarban chains and, hence,
displays a single diffuse reflection at about 4.5 A.
The informatian given by low-angle x-ray diffraction on the lo-
cation of individual molecules within the bilayer is well illustrated
Electron rich

Electron poor

Figure 3.2 X-rays are reflected off of the parallel planes of the
layered structure. These reflections may then be converted to the
interlayer spacing, d , by using Braggts equation.

by recent results from an analysia of stratum oorneum lipids (18).


The interlayer spacings versus water content for aifferent combina-
tions of lipids from the stratum corneum spectrum (18) w e shown
in Figure 3.3,
The basic feature of the x-ray diffraction evaluation is the loca-
tion of an added substance, as directly reflected & the change of
interlayer spacing, as a function of the amount of substance added
(Fig. 3.4). Loeation of the added substance between the lipid chains
(see Fig. 3.4 (a) leads to no interlayer apacing change, whereas lo-
calization to the space between the methyl group layers (Pig. 3.4(b))
(or in the aqumus layer) leads to a strong increase in spacing.
With knowledge of this =lationship, a cursory glance at the re-
sults in Figure 3.3 may be given a straightforward (but ermneous)
interpretation, It would predict that the layered structure of the
stratum corneum free fatty acids plus water will accommodate added
phosphatidy~ethanoIamine, mostly between the methyl groups, and
that a subsequent addition of cholesterol w i l l be localized to the
same site,
Stratum Gorneum Structure and Transport

Figure 3 . 3 Interlayer spacing as a function of water content for


the systems: (A) partidy saponified free fatty acids; (B) phos-
.
phatidylethanolamine + A ; ( C ) cklesterol + B
Friberg e t at,

Added Substance

Mde Fraction Added I Total Host

Figure 3 . 4 Interlayer spacing as a function of mole fractbn added:


(a) the added substance is located between lipid chains; (b) the
added substance is located between the methyl group layers.
Stratum Corneum Structure and Transport

Figure 3.5 Extrapolation to zero water concentration for the curves


from Figure 3,

It cannot be overemphasized that this interpretation bads to er-


Mtneous reslzlts, because the curves in Figure 3.3 reflect the influence
of both the water and the added substance, The correct information
is obtained only after an extrapolation to zero water concentration
has been made (Fig. 3 . 5 1 . The extrapolated curve C has identical
extrapolated interlayer spacing with curve A , whereas curve B
shows a higher one. Hence, the correct interpretation is that the
added cholesterol is, by no means, located between the methyl group
layers in the structure. On the contrary, all cholesterol resides en-
tirely between the lipid chains (Fig. 3.6 (b)). In addition, the added
Figure 3 . 6 ( A ) The added phosphatidylethanoXamine being locmlized
between the methyl v o u p layers, (E3) The added cholesterol brings
the added phosphaticfy.lethstnolam-ine into the space between the chains,

cholesterol bdngs the added phosphatidylethanolamine that was ear-


lier located between the methyl group layers (see Fig, 3,CiA) into
the space between the chains (see Fig, 3. SB) , The high vt-ilues of
the interlayer spacings after addition of cholesterol (see curve C ,
Fig. 3 . 3 ) are a consequence of reduced water penetration into the
lipid parts when compared with conditions for curves A and B .
This type of information is definitely usefix1 to claAfy the organ-
ization-functionaf. relationship for the strfllurn corneum lipids. The
primary source of the information is the x-ray diffraction results,
but their interpretation it? based on an assumption of no change in
arder of the amphiphilic chains nor in their tilt. Nuclear magnetic
resonance offers a means to confirm the interpretation of the results
from the low-angle x-ray diffraction and, in addition, it provides
direct information about the order of the indiddual molecules and
groups involved.

B, Order Parameters OF Lipid Groups


A deuterated, proton site on the lipid molecule @ves rise to a quad-
rupolar splitting of the NMR signal through the interaction with elec-
Stratum Gorneum Structure and Transport 37

tric field gradients (19- 20). The quadmpole splitthg , Av , for an


unoriented lamellar dispersion is directly proportional to the order
parameter S

where x is the quadrupole coupling constant and S is the order pa-


rameter as~ociatedwith a C-r) bond,
For molecules deuterated in different positbns, a spectrum will
resuXt that comprises a series of splittings centered about a eentral
frequency. The diff~rencein the degree of order in the molecule
is reflected in the size of the order parameters and may be ex-
pressed as an order pmfiile for which Si is considered as a functhn
of position in the molecule.
The location between the hydrocarbon chains of the lipids @ves
an order parameter of about 0.1 to 0 . 2 for groups attached. ta a
long-chdn compound. Compounds located between the methyl group
layers show order parameters approximately one-tenth of this value
(23-1 *
The N M R speetmm of perdeuterated pdmitie acid is a good ex-
ample, The eentrsrl p e a (Fig, 3.1) is from palmitic acid localized
between the centrd methyl gmup layers. These pdmitic acid mol-
ecules show little order, their motion is essentiafjty isotmpie. The
sedes of split signals arise from pdmitic acid being located in the
organized amphiphnic layer, The order parameter for the latter is
about 0.15.
This layered structure and its order have the most pmfszlnd in-
fluence on the transport properties of lipids and water.

C. Transport in Layer4 Structures


The diffusim in Xiquid c v s t a l s is highly anisotropic (C2-25)- The
essential iilformation for the prablem of transport across the stratum
cgtrneum is that the diffusion parallel to the layers is fast; the same
mamitude as in a liquid, Conversely, perpendicularly to the layers
the diffusion is one or two magnitudes lower.
A layered. stmcture, such as that ;In tke lipid part of the stra-
tum corneurn, is not a perfectly organized array of layera parallel
to the skin surface but, instead, a series of dislocations afways oc-
cur as shown schemalie&ty in FQuse 3.8, Hence, the pertinent dif-
fusion saefficient is the gross one for a partidly organized lamellar
stracture ,
The permeability of water through reconstituted stratum comeurn
was determined earlier (5). The method Xends itself to investiga-
Friberg et al,

The deuteraum W M R ct Ptimat,~. 0 3 5 iOte*c ~ c l d

Figure 3.7 The NMR spectrum of perdeuterated palmitic aeid added


to the system oleic acid-sodium oleate.

tions of any lipid model combination, and we have recently applied


it to lipid compounds by use of the following process: Fragments of
stratum corneum , obtained either by scraping or from trypsinized
skin obtained at autopsy, were lipid-depleted using chlomformimeth-
an01 (2: 1 vol) and, subsequently, reduced to individual cells with a
ground-glass homogenizer. These cells (about 20 mg) were suspend-
ed in ether and pipetted onto a phosphate-buffered solution ram
(pH 7.2) confined within a Teflon O-ring with 1 . 3 em diameter. Li-
pid (1-5 mg in 15 ml ether) was then layered onto the cells, and
the films were stored at QQC for at least 1 2 h r , after which the re-
constituted stratum corneum tablets were used in the permeability
studies. The lipid model gave reconstituted stratum corneum tablets
with sufficient mechanical properties to be used in the diffusion cell
according to Blank ( 2 6 ) .
These results, surprisingly, showed the lipid madel .lo give
fluxes through the tablet at the same level as the ones found for
the extracted lipids (5). Table 2 shows the values from studies in
which extracted stratum corneum cells from different persons were
Stratum Corneum Structure and Transport 39

Table 3.2 The Amount of Water Transferred per Unit Time, Ex-
pressed as mg/(cm2/hr) per Unit Membrane Thickness

Reconstructed SC disks Flux [mg/lem2/hr)l

Subject A SC with 20%lipid model 0.86 2 0,05


Subject B SC with 20%lipid model 0.71 ; 0.09
Subject C SC with 5.8% partially saponised FFA 0.99 t 0.03
Subject C SC with 16%partiaUty saponified FFA 0.87 2 0.W
Smith's value for calf SC with native lipid 1.0

In each of the above cases, the flux was determined for water pene-
trating through at least two tablets. The values shown are the mean
2 standard d e ~ a t i o n .

Cell htereelkriar Lamelkrr Liid Pdnt of m-ation

Figure 3.8 The current concept of the atratum corneum as a two-


compartment system, in which ceIls can be analogized to bricks, and
intemellular lamella to mortar.
combined with the lipid model to give flux values sirnaar to those
found by Smith et al, (5) who used the extracted lipids,
A s a matter of fact, the sane flux vafues were also obtdned
when the lipids were limited to the p m t i a g saponified fatty adds,
Table 3 . 2 shows us no s i p s i c a n t difference between these vafues and
those fmm tablets with more complex combbations of lipids present,
These results appear to indicate that stmeturd organization of
the lipids is the imporLant element to prevent fast water transport
through the stmcture, The difference between the v&aes fmm a
f&ty acidlwater-layered structure and the one contdning all the li-
pids of the stratum comeum was insignificant,
The majar difference was found when comparing water transport
thrmglr. layered stmctures of different fatty acid-sow combinations
without the pratehs. The system of water-fatty acids with saturat-
ed hydrocarbon e h h s stood out only in the comparison 1/27]. In
fact, a structure with o d y saturated hydrocarbon chains gave no
barrrier whatsoever to water transport; the vrifues of evaporation
rates were identical d t h those from an unprotected water surface!
This result agrees Ath emly results on evaporation retardation by
monsmolecular layers (28) that showed cracks in it cry.stalline mno-
layer to reduce the retardation barfier to water evaporation by 958.
A cornpadson of these results with the symptoms sf the essential
fatty acid deficiency syndrome is an illustrative exercise. Lack of
unsaturated fats in the diet gives rise to dry cracked skin with h-
creased trrtansdermd transport. In our opinion these results , corn -
bined d t h the diffusion results from the lipid model structure and
the wea-known phase diagrams of water-amphiphse systems ( 29),
show that the lipid hydrocarbon chains in the stratum cornearn need.
moblBty (i ,e ,, t a t d crystaHiza~onm s t be prevented) to function
effii3ctively as a barAer to water transport,
Crystaine packing of the lipids means a bdtUle stmcture in
which deformations lead to open cracks: the consistence of a liquid
e q s t d , on the other hand, is that of soft butter and will tolerate
the necessary bending and stretching of the skin Athout losing the
basic layered stmeture. Conversely, our results from evaporatbn
af water from layered stmcturea (27) reveated that a mixture of
unsaturated and saturated ch&ns results in a slightly enhanced bar-
rier to transpbrt, compared with the values fmm a strtlcture of un-
saturated chains only, This result is interesting in relation to the
recent results by White et d. (16) demonstrating the presence of
both crystdline and liquid chains in the stratum earnearn lipids.
With the layered structure as a background, it follows that a
substantid increase of transport through the stratum corneum would
be obt&ned if the long-rmge order of the lipid organization were
bmken down to form an isotropic liquid, The diflfixsion now would
Stratum Corneum Structure and Transport

not be restdcted to paths along the layers, but would be fast in


all three directions, The overall diffusion coefficient is thereby in-
creased more t h m one order sf mapitude when the system beeomes
Xiqdd.
However, the complete transition'ta m fsotropic Equid is not
necessary to ensure enhanced trmsport. Ewn an fncrease of the
f'rddity ( e , 8. , a srnaXl disordedng) will give enhmced trmspor"l, as
shown by our results (211, This means that substances that in-
crease disorder in a H q ~ dcrystalEne system are potential transder-
rnal t r a n s p o ~cmhancers, Such subst~ncesare well known in the
colloid chemistry ; they are called h y d ~ a t m p e s ,
Table 3 . 2 shows structures of selected hydrotrc)lpes and of trans-
dermd transport enhancers, The s i d larity of structure is s t ~ M n g ,
and a few experimental methods to determine the action and mecha-
nism of hydrotropes may be of interest for the area of transdermal
transport,
A s an example, a hydrotrope in the form of the manosoap of
the 5-carboxy- 4-hexyl- 2-eyclohexane- 1-yl o~tanoicacid (NaBG) .
Its function fs well nlustrated in Figure 3,9. The curves show the
order parameter of the hydrocarbon chains in the liquid csystaX af-
ter addition of the hydrotrope a r of a surfactant (30). Addition of
the hydrotrope caused a disordering, whereas the added surfactant
cause6 no change whatsoever,
The reason for the enhanced disorder is found in the eonforma-
tion of the acid, when present in the Equid crystal (31). The acid
(Fig, 3.10) is present with d1 of its polar groups at the interface.
The narrow loop and the short chdn protruding into the hydmcar-
ban space of the amp hSpMle leave a considerable sagace vacant for
the ori@nal hydrocarbon ehrjEins to become disordered,
The comparison between molecules that cause enhanced trans-
dermal transport and the hydrotropes is of interest, but it is eer-
tdnly incomplete, In a preGous study (323, it was shown that NADA
(see Fig. 3.9) appears Po function as a selective hydrotrope, desta-
biagng only bilager systems that are formed from ampEpzles with
relatively small-sized head p a u p s , Hence, NaDA did not enhance
percutaneous permeation fused at 10 w t % concentration) because it
appeared to be ineffective at destabiHdng epidermal bilayers formed
from Hpids contkning medium-sized head groups. The study fur-
ther fmpEed that a nonirdtating transdermal deEveq enhancer sys-
tem is feasible, if a selective hydrotrope was found that would de-
stabil_ize the bilayers formed by the medium-sized head groups of
epidermal faipids , while not disrupting t able cell membranes that are
formed fmm large- sized p hosphofipid head poups, Work in addition
to tGs preliminary study should prove useful in the search for
unique transdermal enhancers. Note that the essential transtlerm&l
Friberg et ul.

Figure 3.9 Adding the compound in Figure 10 to a Iamellar liquid


crystal caused disordesing (01, Addition of surfactants gave no
change ( * I ,

Figure 3.10 The structure of 5-carboxy-4-hexyl-2-eyelohexme-


I-yl
octanoic acid,
Stratum Corneum Structure and Transport 43

Table 3 . 3 Ckrernicd Structures of Selected Hydrotropes and of


Selected Transdermal Trmsport Enhancers

isopropy l a1 coho1 prapyt ene gf ycol

0
CH3\ II CH,\
N-CH S=O
CH,/ CH<

dimethyl formamide dimethyl sul Foxide oleic acid

(CH2I,, CH,
44 Friberg et al.

enhancer is to facilitate transport of the drug and, thus, the com-


bined influence of the drug molecule and the transport enhancer
must be taken into account.

111. SUMMARY

The structure of the iipid component of the stratum corneum was


described as a lamellar liquid crystal. It was shown that a change
to a predominantly crystalline structure would cause enhanced trans-
dermal transport because of cracks in the structure, and a compar-
ison with the enhanced transport during the essential fatty acid de-
ficiency syndrome was made.
Transdermal transport enhancers act in the opposite direction,
and a comparison of t h e i ~molecular structure was made with that of
hydrotropes that cause destabiEzation of a liquid crystal through
enhanced disorder of its hydrocarbon chains.

This research was supported in part by Dow Chemical, Kdland,


Michigan, and by The Upjohn Company, Kdamazoo, Michigan.

REFERENCES

A. W, Kliegman, in The Epidermis, Chap. 20, (W. Montagna,


ed,) , Academic Press, New York-Landon, 1964.
P. M. EBas, and D. S. Friend, J. Cell. Bio?. , 65:185, 1975.
6. E. Burch, and P. Winsor, Arch. Intern. Med., 74:437,
1944,
P. M. ELias, N. S. NIcNutt, and D. 8. Friend, Anaf. Rec.,
289: 577, 1977.
W. P, Smith, M. S. Christenson, S. Nachet, and E. ZI. Gans,
3. Invest. Dematol, , 78:7, 1982.
P. 101. Elias, J. Goerke, and D. S. Friend, J . Invest. Der-
matol. , 89: 535, 1917.
O. M. Gray, and R, J , White, 3. Invest. Dermatol., 70:XX,
1977.
O. M. Gray, and H. J. Yardley, J. Lipid Res., 26:435, 1975.
E . 6. Bligh, and W. J. Dyer, Can. J. Biochem, Phys., 37:
911, 1959.
W. Abraham, P, W. Wertz, and D. T. Downing, J. Lipid Res.,
26: 761, 1985.
A. W. Ranasinghe, P. W. Wertz, D, T. Downing, and 3. C.
Mackenzie , J. Invest. Dermatol, 88: 187, 1986.
Stratum Gorneum Structure arid Transport 45

P. M, Eliias, J. Invest, Dematol., 80:44, 1983,


P. M, Exas, jfnt, J, DematoE,, 20:1, 11381.
TZ, W. Wertz, 111. Abraham, L. Landmmn, md D, T, Downing,
9 , Invest. BermatoE,, 87:582, 1986.
S , E, Friberg, and 2). VV, Osborne, J, Dispersiion Set, Technol,,
6:485, 1985,
S. H. White, D, Mirejovsky, and G . 1. King, Biochemistry, 27:
3725, 1988,
J, Seeag, 9. Rev. Biophys. 1 0 : 3 5 3 , 1917,
S, E l Fdberg, If, Suhdmi, and L. B e Goldsmith, in Stratum
Corneum Lipids in a Model Structure, (in press).
A, J. 1. Ward, S . E, Friberg, D, W, Larsen, and S , B,
Rmanavare, Langmuir 1:24, 1985.
J , H, Davies, f-fiochim,Bioptzys. Acta, 737:117, 1983.
UnpubBshed results,
G , Lhdblom, Acta Glzem, Seand., B35:61, 1381.
M. £3, Schneider, W. K, Chan, and VV. VV, Webb, d . Biopfiys,
Soe. , 43: 157, 1983,
W. C. 11. V a z I a, M, elegg, and E>. Hallman, Btachemistry,
24: 781, 1885.
G . GMdicMmo, D, d e Faario, G. A. Raulied, md M, Terenzi,
Mot. Cryst. t t q . Cryst., 133: 1, 1986,
I. H, Blank, J, Maloney, G, E. Alfred, 1, Simon, m d C, Apt,
S, Invest. Bermatol. , 84: 188, 1984,
S, E, Fdberg, and 1, Rayali, J , Pharm. Sci,, (in press),
V . K, ZaMer, ed, , Retardation o f Evaporation b y Maemlayem:
Transpart Processcrs, Academic Press, Hew York , 1362,
P, Ekwa12, in. Advances in Liquid Grystats, (0,H, Brown,
ed. ), Aeademie Press, New TYlork, 1925,
S. E, Friberg, S . B, Rananavare, and D, W, Oabome, I ,
Colloid Interface Set. , 108: 487, 1986.
11". E"l;tlim, and S, E, Fdberg, J, Colloid Interface Set., 97:2 6 ,
1984,
2). W, Osborne , Colloids Surfaces, 313: 13, 1988,
Penetration Enhancer Incorporation
in Bilayers

ANT HOhlY f , f . WARD* and REG[MA TALLON University College


Dublin, Dublin, Ireland

The use of agents to enhance the penetration of dmgs through the


skin has been tvidely studied, although phenomena assodated with
the modification of the barfier function and the moleeufar nature of
the incorporating molecules rem&n undear , An understandkg of
these Interactkns, however, is fundamental. to the design of topic&
de1iver;y vehicles and their modes of interaction with the skin,
This chapter explores the use af a simple surfactant system to
model some of the features a f the lipaphi%ccomponent of "re stratum
carneum (5C). For this, the efioice of model is determined by the
head group nature of those lipids that are predominately rteutrd
( 1 , 2 ) to about 75% to 80- the S@ compared with about 10%phos-
pholipids. This type of Upid interacts with the aqueous endranmerrt
through hydrogen-bonding hteractions. A similar situation a ~ s e s
with simple surfactants of the d k f l polyoxy-ethylene glycol ether
(CnEOm) type that form assockted stmelures when mked with wa-
ter ( 3 ) . Particular attention will be foeused on the lyotrapic meso-
phase formed when the amphiphnic mdecules assockte h t o birnolec-
ular arrangements separated by aqueous hyers f Fig. I), known as
the lamellar phase, Note that this phase comp~sesrand0rnl.y dis-
persed lamellar domgns and should not be confused with vesicles
and liposonres, dthough these structures &so m&ntah bimolecular
arrangements of the ampkiphses. Indeed, many physicochernic&
p r o p e d i e ~are andogous between the two types of systems, but
some dsferences exist in observed b e h a ~ a r(see Sect, XI,A) because
of differences in isotropy, Furthermore, the lamellar phase d w s

"Current affiliatbn: Clwkmn University, Potsdarn, New Uork


Ward and TaEEsn

Figure 4.1 The arrangement of surfactant molecules in the lamella^


liquid cmy.stA phase.

not suffer from some of the inherent instabaities encountered in


vesicles.
The choice of enhancing agent systems is based apon those com-
monly used and studied; namely, ofeyl alcohol, pmpylerre glycol, and
dodeeyf (rhexaydm- 2N-iizepin- 2-diaione lauracapram ; Azons) , which
has been regarded as having some potentiafly novel features (41,
The behador of oleyl dcohol md Lauroeapram, both s-ingly and in
csmbinati.orr with propflene glycol, has been investigated when in-
corporated lExl the lamellar phase of n-dodecyl_pentao~ethyleneglycol
ether ( 6 125&05),
To study this type of system, techniques sensitive ta the dimen-
slions of the associated strmctures and the time sealles of the sys-
tem's dynamics are required, The most pawerful techniques avstiia-
bEe for such studies of anisotropic phases are those of smd-angle
x-ray diffraction and nuclear magnetic resonance (NMR), There-
fore, a brief background and description of these methods will be
given as tan aid to understmdhg the presented data,

It, EXPERIMENTAL BACKGROUND


A, Nuclear Magnetic Resonance
Nuclear magnetic resonance (NMR) is arwably the most powerful
method for investigating the dynamics and structure of asssaciated,
Pene trctlion Enhancer Incorporation r'n Bilay ers 49

surfactant and lipid systems. The basis of the power of these tech-
niques is that only the nuclei, which are at resonance, will be di-
rectly observed, This allows observation of selected parts snd fac-
ets of the system, u s u d y by making use of the constituent nudei
of the components (i,e., no perturbation of the system 'by bcorpo-
r&ed foreign ""pmbet' molecules). Substitution of pmtons by deu-
te~iumatoms elither in the aqueous fraction or h the lipid component
has relatively mhas and predictable effects on the system" behaaor
and has been used to great advantage (5-18) in the investigation
of baayer phases formed by lipids and surfactants in the presence
of water.
m i k e the more commonly eneountemd hydrogen (&) and car-
bon ( 1 3 6 ) nuclei, the deutedum ( 2 ~ nucleus
) has a spin quantum
number, I = 1, wit k an associated quadrupoxar moment. This means
that there are three nondegenerate nuclear spin enerm levels (Fig.
2 ) - In an isotropic endronment in which the effects of electric field
gradients at the nudeus are averaged to zero, a shgle msonance is
obt?e~pred. The presence of electdcd field gradients h an axiaIly
symmetdc field, however, leads to changes in the spin e n e r w levels
(see Fig. 21, which leaves the mi = 0 --+ +I and m = -1 -+ 0 transi-
tions with different energies, This has the effect of producing two
&sorption p e a s at resonance separated by a frequency- difference
d v , the size o f which depends on the size of the hteraction with
the electricd field gradient as

where x is the quadmpolar coupling constant that has a value of


167 kHz for deutesum attached to earban (19) and 8 is the angle
between the principal. axis, being the vector of the C - % ) bond and
the magnetic field. The case of the nudeus interacting with a field
gradient of axid symmetry has been shown to be applicable to those
normdilly studied in laneflar surfactant and lipid systems.
Xn molecules contttinhg sever& d e u t e ~ u matoms, a doublet split-
ting will be observed for each atom, resecting the degree of dynamic
freedom that atom has izl the molecule, Assignment of splittings to
padjleular deuterium atoms in a mojecule must be done by observttian
of selectively deuterated molecules. Deuterium N M R may, therefore,
be used to probe the dynamic stmcture of the chdns, as a function
of position, composition of phase, and temperamre. An order pm-
ffke of the lipid chains, defined by the order parameter, SC]D, can
be derived from the quadimpolar splitting8 f E q . 2) of the ZW[ nuclei
in the different segments.
Ward and Tallon

ZEEMAN QUADRUPOLAR

Figure 4.2 Nuclear spin energy levels for a spin = 1 nucleus (e.g.,
2 ~ showing
) the allowed transitions (a) for an isotropic and (b) an
axially symmetric electrical field gradient,

Here the bar in Eq. 2b represents the averaging over all possi-
ble orientations of the randomly dispersed lamellar domains. Many
studies have now been carried out to derive order profiles in Iipid
and surfactant systems in lyotropic lamellar phases (5-18). The
form of the profile (Fig. 3) is remarkably consistent for a wide va-
riety of systems showing highest order (SCD values) in the seg-
ments of the amphiphilie chains cIasest to the head group. A rapid
decrease in order, as expressed by Scf), is seen in the methylene
segments near the terminal methyl group, which is located at the
center of the bilayer. The shape of this type of profile is a result
of the packing constraints in the associated bilayer structure, re-
ducing the canformational freedom of the individual chains ( 11,18),
Penetration Enhancer Incorporation in Bilayers

Headgroup Terminal Methyl

Figure 3.3 The typical form of order profile of hydrophobe chains


observed for surfactants and lipids in lamellar phases.

Perturbations of this structure by incorporation of solubilizates or


changes in composition or endronment can, in principle, be studied
through the resulting effects on the order profile.
Many Iyotmpic lamellar systems of surfactants and phospholipids
may be oriented between flat @ass plates (i.e., the effect of the
surface field of the glass is to orient bilayers so that the amphiphile
chains are peqendicular to the surface). The osentation of the
principd axis of the bilayer, called the director (see Fig. I ) , can
be determined relative to the mametic field. Were, the avera@ng
of the directors in the sample is removed, and the order parameter
c m be derived fmm the angular dependence of the quadrupolar
splittings ( 14- 16).
Ward and TaEton

F1, SmatI-Angle X-Ray Diffraction


Qne of the most important methods of identifying and characteddng,
lyatmpic mescrphases is x-ray diffraction at low angles (20,21), The
wen-know B ragg relation, namely-,

n, X = 2d sin 8 133

where m fs the order of resection of x-rays with wavelength X from


a r e p l a r structure with repeat distmce d . For nickel-filtered
copper radiatian of wavelength 0,2542 nm, Eq , 3 gives angles of re-
flection of between 0.4 and %.ti0 for the range of repeat lengths
commonly encountered in bnayer systems (2- 15 nm) ; hence, the
term small or low-angle x-rtiy diffractbn. Lamellar systems are
..
characterized by a sedes of reflections in the ratio I: 2: 3: 4. , ear-
responding to the one-dimension& repetition with first-order ( n = 1)
spachg (d) equivdent to the thickness af the water and birnol~leeular
layer (see Fig. I) .
If the partial spedfic volumes (v) are Mown, it is possjble to
calculate the thicknesses of the water layer, d ~ ~ and a ,lipid bilay-
e r , d l , as

The measured d-spacings ape u s u d y plcctted as a f u n c ~ o nof C /


( I - 6 ) where C is the concentration of water (wt$) and the bifay-
er thickness is obtdned by extrapolation to C = 0. The slope of
the plot is related to the ratio of the pmtial specific volumes*
Geornet~cconsiderations show that the area-pcilar group (A) in in-
terface of the IameBrrr phase can, be dedved fmm the fallowhg re-
lat-lonship:

where V 1 is the molar volume of the lipid and N is the Avogadm


number.
A, result of the relatively simple geomet~cpropaflies of the
lyotropic phases is to &ow the dist&bution of components ~ t h h
the system to be inferred from the d-spacbgs, The experPimexlt4lf
d-spacing (dexp) in a lamelar system can be separated into con-
Penetration Enhancer Incorporation in Br'Eayers 53

tributions from the bilttyer (dbn) and the solvent layer (ds) as

where Qtlanis the volume fractrion of the apid in the system,


A plot of log deXp against -log +bil should be linear with unit
slope if there is na penetration of the solvent into the bBayer re-
gion , If penetration occurs, however, the s1ope of the plot d in-
crease. The percentage penetration (p) can. then be estimated by
ca2cuXathg the difference bet ween the expedmental d-spacing and
that expected far nonpenetration (dcd) given as

where mp is the volume solventllipid ratio.


Phase identeication rests primady on the ratios of the spadngs
of the dfffrtnction lines; thus, unlike the equaf spacings of the lanel-
lar lines, the hexagonal phase, for example, exhibits line ratios of
1: (3)-112: (4)-112: (7)-1/2 ...
, and the cubic phase oflen has the
ratio 1: (3/4)If2:(3/8)2/2: (3/11)1/2 , ...
Ift . EXPERIMENTAL

Phase diagrams were constructed by o b s e r ~ n gthe equBibrated state


of the samples of different compositions prepared by weight. The
presence of anbisolmpic phases was determined by observation under
polarized Iight microscopy. Thorough m k h g of the samples was
achieved by vortexhg and centrSugalion ant2 sample homogeneity
was establishedi .
The NMR spectra were o b t d e d at a resonance frequency of
13.71 MHz for deuterium and 22.49 MHz for 1 3 using~ a multinuclear
pulsed. PJMR spectrometer (JEOL PX9OQ) operating in the Fou&@r
transfarm mode,
n-Dadecy1pentaozryethy"r.eneglycol ether was from N&ko Ltd.
(Japan) and used ~ t h o u tfurther purlBcation, n-Mkanes were from
A l d ~ c hLtd, and were passed through &urnbe before use; deuledum
(99.58) was o b t b e d from fl"luarochem Ltd ,
Ward and Tallon

IV, RESULTS AND DISCUSSION


A, Surfactant-Water Systems
The phase behavior of binary mktures of GnEOm-water have been
extensively studied and presented as temperature-composition dia-
grams (3). A combination of deuterium NMR and vapor pressure
measurements indicated that a simple two-state derscription of the
water in the systems was ~ldequate(22-24). The obseyed quadru-
polar splitting of the water, Avobs, could then be written simply as

Av = f Av + fiSOAviSO
obs b b

where fiso represents the fractions of water in the bound or iso-


tropic "free" state and A v ~are the corresponding quadmpolar split-
tings when in state i, Because one of the fractions is regarded as

Swfactant/Water (Mole Ratio)

Flgure 4.4 The dependence of quadmpolar splitting of water in the


lamellar phase of C12EOg--water (3: 2 w/w) at 298 K upon the amount
of water.
Penetration Enhancer Incorporation in BiZayers 55

being essentially simBar to that of free liquid water, AvisQ is ex-


pected to be zero and E q , 9 will reduce to

Here, x is the quadmpdar coupling constmt m d Sb is the order


parameter, Equation IOa t&es inta account the complex avera@ng
through the random dispersion of the lamellar domains. Traiues of
Sb derjved fmm the observed splittings are usually in the range
of 10-2 to 10-3. A plot of bvOb~against the mole ratio af surfic-
tmtlwater (Fig, 4) shows the linear povYtion predicted from E q , l a
for the C 12EQg-water system studied.
It may be concluded from these observations that the quadm-
polar splitting of the water in the lamellar phase depends on

1. The degree of headgroup solvatbn as manifested by fb


2, The amount of dynamic order in the bound sites as expressed
in the order parameter, Sb

The utility of these facts has recently been made use of in a


study of the interaction of commonly used hydrotrapic agents with
surfactant bsayers (25).

B. Surfactant-Water-Enhancer
The pa&iaf phase diagrams of the host surfactant-water system
containing laurocapxram (Azone) , oleyl alcohol, and propylene glycol,
respectively (Fig. 53, show that the amaunt of pmpylene glycol
that can be accommodated is much smaIter tkm the other two oils,
It can d s o be seen that the heorparatioion of either oleyl alcohol or
Azoxle hcreases the ability of the lamellar phase to aceommodate
water, with the maimurn water content rising from about 50Un the
host system to about 85%in the presence of the enhancer. This is
not observed for soltzb.ilized prepylene glycol for which, in fact, the
water cizpacity of the bilayer phase Is reduced on ineavoratiaion,
Such an enhancement in the water retaining capadty of a bntsyes
system is of Importance in the consideration of penetration enhance-
ment, because it has been shown that the water content of the skin
affects delivery mtes in several cases f 26-28).
Some bformation about the state af surfactant hydration in the
presence of these agents may 'be derived. from the variation of the
quadrupdar splittings of the surfactant - I%HIW $3 system. Studies
made of samples with a f k e d surfstctantlwater mole ratio in the
WATER

fa)
01L

OLEY L ALCOHOL \ \

WATER 632E05
(b)
Figure 4.5 (a) Phase diagram at 298 K for the system C 121E05-
Gzone-water, (b) Partid phase diagrams showing the extent of the
lamellar phases o f the system C laEOg-water with oleyl alcohol and
propylene gXycaX at 298 K .
Penetration Enhancer Incorporation in Bilayers

0 0.5 1.0
f
(a) Moles OiIlMoie Surfactant

Figure 4.6 Quadmxpolar splitting of [ 2Ei]H20 in lamellar C l S 0 5 -


water (3: 2 w/w) at 298 K as a function of solubfiizate content: (a)
A r n e ; (b) oleyl alcohol; (c) propylene glycol.

presence of fncreasing added solubilizate are Blustrated in Figure 6.


Any changes in Avobs must be a result of changes in fb or A v ~
because the total water content is constant. The decrease in AVobs
in the presence of propylene glycol is simfiar to that found for the
C12E04-water system ( 2 5 ) , and can be discussed in terms af
changes in the composition of the hydration layer. A replacement of
about 2 mol of bound water for each 1 mol of propylene glycol in-
corporated is implied.
The b e h a ~ o rof the systems containing either Xaurocap~amor
oleyl rilcohol, however, is quite different with Avobs showing large
increases with increasing incorporation. This may be interpreted
&san increase in the fraction of water molecules contributing to the
bound fraction or as an increase in the order parameter of the bound
water molecules, or a combination of both phenomena. The observed
increase appears to be too large to be a result of an effwtive in-
crease of fb alone, if it is assumed that a maximum of 11 mol of wa-
ter per mole of surfactant may contribute to this fraction. Xt is
more likely that the increase arises, in part, from an increase in
the order of the bound water fraetion. Consideration of the struc-
tures of both oleyl alcohol and Azone reveals then also to be of an
amphiphifie disposition, and they would be expected to heorgorate
between the surfactant molecules forming the bilayers, Such mode
of solubilization of long-chained alcohols in lipid and surfactant bi-
Ward and TalZon

(b) Moles OitJMole Surfactant

MOLES PG
MOLE c 1 2 E 0 5

Figure 4.6 (C0ntitinuld).


Penetration Enhancer lncorparation in BfEayers 59

layers is well known (21) and, generally, leads to increased lamel-


lar-phase stability.
The orientation of the Azone molecules when solubilized in the
bilayer can be inferred from the behavior of the 1 3 chemical
~ shihs
of the ring carbonyl carbon atoms (Fig. 7). A comparison of the
136: shifts in the larndlar phase with Azone dissolved in polar sol-
vents such as alcohols or polyethylene glycol and in nonpolar media
such as alkanes indicates that the ring is located in the polar en-
vimnment at the bililyer-water interface.
Inclusion of weWy amphiphilic molecules in the bilayer, there-
fore, provides furlher hydrogen-bonding type interactions at the

0 20 40 60 80 300
W A T E R CONTENT f w t . %)

Figure 4.7 Chemical 136 shifts of the carbonyl carbon of Azone in


different media.
Ward and Tallon

Figure 4.8 Interlamellar layer d-spacings of the system 6 12E05-


Azone-water.

interface with the aqueous compartment, This fact coupled with the
increased structure arising from a greater number of trans config-
urations in the hydrophobic region-a necessary result of packing
of alkyl chains in this fashion---&a lead to more order in the region
of the polyoxyethylene chains to which the water molecules are
bound. The consequence of this is to increase Sb and Avb.
Small-angle x-ray diffraction measurements yielding values of
the characteristic d-spacing (Fig, 8) also indicate structural
changes in the system as the oil content is varied at fixed mole ra-
.
tios of Azonelwater The expected linear increase in d-spacing is
found with increasing water content (Fig. $), dthough the slope of
the plot decreases with an increasing Azonefsurfactant ratio. If the
surfactantlwater ratio is kept constant, an initial increase in the d-
Penetration Enhancer Incorporation in Bilayers

A Z O ~ ~ / C , , E ORatio
~

Figure 4.9 Interlamellar layer d-spacings of the system Cl$ZQ5-


water (surfactmtlwater B , 70: 30; *, 55: 45 wlw) as a function of
Azone content.

spacing compared with the Azone-free system is found, which Is fol-


lowed by a decrease at larger Azone contents. These data, when
taken in conjunction with the 1% shift results (see Fig, 7) indicate
that Azone does not completeily penetrate between the surfactant
chains at low Azone concentrations, The degree of penetration in-
creases from about 79% to 100%at 10% w/w) of added Azone. At
higher Azone concentrations, either there is increased water pene-
tration into the bilayers, or extraction of surfactant into the bnayer
interior: both of these would give an apparent decrease in the ob-
served d-spacing. The initial increase in the quadrupolar splitting
of the water would be expected if the degree of Azone penetration
was increasing, as it is the water molecules associated with the EO-
groups adjoining the ttlkyl chains that have the greatest contribution
to Avobs. Thus, increasing Azone penetration leads ta a decrease
in the area occupied per amphiphile at the EO-hydrophobe interface
and increasing order.
62 Ward and Tallon

C. incorporation of Mixed Enhancers


The differing influences that propyIene glycol and either Azone o r
oleyl alcohol have on the stability of the bilayers should be reflected
in the behador of the system in the presence of binary mixtures
with propylene glycol. Such mixtures have been studied (4,29,30)
as topical delivery vehicles, and syner@stic effects have been found
in the delivery of certain drugs. A preliminary study of the ef-
fects of incorporating binary mixtures of the type oleyl alcohol-
propylene glycol and Azone-propylene glycol into the lamellar phase
is summarized in Figures 10 and 11. There is an enhancement in
the meurimum uptake of the mixture by the bilayers that is greater
than for either of the individual components. The enhancement is
most pmnounced for oleyl alcohol-propylene glycol mixtures (about
55% w/w) which compares with about 26% for oleyl dtcohol or about
5%for propylene glycol alone. The increased water capacity exists

Figure 4.10 Partial phase diagrams of the system C lgEO5-water


with wlubilized oleyl alcohol-propylene glyool of different PG mole
fractions (numbers at the apex of the diagrams).
Penetration Enhancer Inco~porationin Bilayers

MOLE FRACTtON O F P O IN MIXTURE

Figure 4.11 Maximum uptake of Azone-PG (closed circles) and oleyI


acohol-PG (open circles) mixtures in lamellar 612E05-water (3: 2
wfw) at 298 K, showing the limit of compositions at which increased
water capacity is found; * - * - . - * - oIeyl alcohol-PG; - - - - Azone-
PG.

only for compositions with gropylene glycol contents below those in


the range of maximum synergism (i-e, , propylene glycol mole frac-
tions less than about 0.75). Therefore, a minimum amount of in-
corporated oleyl alcohol is required to obtain maximum synergism
and water capacity of the bilayers.
Preliminary ZH NMR measurements of the water in the systems
containing solubilized mhtures of Azone-pmpylene glycol or oleyl
alcohol-propylene glycol (Fig. 12) again show changes in slope of
Ward a n d Tailon

(a) O ~ L I S U R F A C T A N ~MOLE RATIO

Figure 4.12 Quadrupolar splittings of [ 2 ~ ~in~lamellar


2 0 C12E05/
water at 298 K in the presence of solubiIized Azone-PG e t u r e s
(a) or oleyl alcohoIlPG mhtures (b). Qleyl alcohol mole fraction:
o, 1.0; e, 0.85; A , 0.65; A , 0.52; o, 0.00.

the plots of bVobs against mlub~izate/surfactantmole ratio at con-


stant water content. Future small-angle x-ray investigations will
be necessary to determine if the compositions at which these changes
occur will correspond with d-spacings as found. for Azone (31).
The initial increase in Avobs is common to all of the binary compo-
sitions in the region showing synergism. There is obviously a bal-
ance established between the stabilizing (and ordering) effect of the
01eyX alcohol and the destabnizing effect brought about by replace-
ment of bound water molecules by progylene glycol. In these
terms, there is &most camplete balance at the camposition of maxi-
mum promotion of bmyer phese stabmty, as denoted by the dmoslt
complete independence of the water quadrupolar splitting from the
amount of the mixture 8olubMzed (see Pig. 12).
Penetration Enhancer Incorporation in Bilayers

(bj OILiSURFACTANT MOLE RATIO

D. fmpllcatlons for Vehicle Design and interaction


with the Stratum Corneum
A model system of this simplicity cannot, of course, 'be expected to
give a comprehensive description of a system as complex as the
stratum corneum. However, these results do give some indications
relevant ta the design of a topical vehicle and the potential interac-
tion and modification of the skin to which it is applied. It can be
seen that components often used in enhancement studies can poten-
t i d y lead to increased water capacity in the system for which head
group hydration is an important factor. Changes are also brought
about in the mleeular packing and dynamics of the biIayers leading
to increased enhancer uptake in the hydrophobic region of the sys-
tem. This descriptlion is in broad agreement with that recently in-
ferred from differential thermal analysis and permeation studies (321,
66 Ward and Tallon

Second, the model aIso shows synergistic aspects in both the water
and solubBzate retention of the system, both features being charac-
teristic of the r e d system* Extension of the model to include Salty
acid-soap components is being made t s heorporate other- aspects of
skin p H and, composition.

ACKNOWLEDGMENT

This work has been part funded by EOLAS, The Upjohn Company,
and thanks are due to P~ofessolsStig E. Friberg for making avaga-
ble time on his smdl-angle x-ray apparatus,

REFERENCES

M, A , Lampe, M, L , Wsfiams, and P. M. Efias, J . Lipid Res,,


24: 131, 1983,
H , J , Yardley , in Biochemistry and PhysioEagy of the S k i n ,
.
Chap. 16 (L. A. Goldsmith, ed,) Oxford University Press,
Oxford, Eng*, pp, 363-377, 1383.
D. J . Mitchell, C. J. '1". Tiddy, L. Warhg, T, Bostock, and
M , P, McDonald, J, Chem. Soe. Faraday Trans, 1, 79:975,
1983.
R , B , Stoughton, and W, Q . MeCTure, Drug Bev, I n d , Pharm.,
9:725, 1983.
J. Charvolin, and B , Mdy, in Magnetic Resonance in Cotloid
. .
and In terface Science (IZ A , Resing, and 6 C. Wade, eds ,) ,
ACS Symp , Ser. 3 4 , American Chernicd Society, Washfngton ,
D.C., 1976,
J, Seelig, and A. SeeXig, Q. Rev. Biophys,, 13: 19,1980.
E , Qldfield, in Techniques in the Life Sciences: Lipid and
Membrane Elioehemistv (J. 6. Metcdfe, and R, Hesketh, eds,),
B427: 1, 1982,
J. H, D s ~ s Biochirn,
, Biopkrys, Aeta, 737: 111, 1983,
M , Ranee, K . R . J e f f r e y , A . P . TuXlocE.1, R . W e Butler, and
1 . C. P. Smith, Bhehim. Biophys, Aeta, fiBO:245, 1980.
3. Seelig, &, Rev, Biophys,, 10: 353, 1971.
J, Gharvosn, in Ly~tropieLiquid. G;1~ystals ( S . E. Friberg,
ed.), Adv, Chern, Sere 152: 101, 1978,
I", Ahlnas, O Soderman, C , Hjelm, and B. Lindman, S, Plzys,
I

Chem, , 87: 822, 1983,


If. Mmfsch, H. Sdto, and 1 , C . P . Smith, Prog. Nuct. Magn,
Reson, Spectrasc, , 1 2 : 211, 1977.
d , Seefig, and W, Niederberger, J . A m , Ghem. Soc, , @@: 2069,
1974.
Penetration Enhancer Incorporafbn in Bilayers 67

3 , Gharvalin, P , Mannewille, and B . Deloche, Chem. Phys,


Lett. , 23: 345, 1873.
N. Baden, Y , K . L e ~ r z e ,and A . J , 1 , Ward, Biochin, Bio-
phys, Acta, 4 1 @ : 3 9 5 , 1976.
D , W, Gmen, Prag, CoE1, PoZym, Sci,, 70:6, 1985.
.
S Marceljlja, fEioehim. Biophys, Aeto, 467: 165, 1974.
2. J, Burnett, and B , Muller, 6, Chem, Phys,, 55:5829,
1971,
V , Luzatti, in Biological Membranes (El. Chapman, ed .), Aca-
demic Press, London, p, 71, 1368.
P , EkwaXX, A d v , Liq, Cryst,, 2: 1, 1915,
K . Renddl, and G , J, 34'. Tidrfy-, 17, Ghem, Sac. Faraday
Trans. 1, 80:3339, 1984.
I , 6 , Lyle, and G . S. T, Tiddy, Chem, Phys, Lett,, 124:432,
1986,
EM. Garvelle, D, C . Wdl, I , G , Lyle, md G. 8. T, Tiddy,
Faraday Discuss, Chem, SOG. 81: 1, 1986,
A . 3. I , Ward, C. Marie, L-A, S y l d a , and M , A , PhilXipi, 6,
Dispersion Sei, TeeZznol, , 9; 149- 170, 1988,
I , H. Blank, 6, Invest, Dermatal,, 28; 133, 1952,
R , J, Schmpleirz, and 1 , M, Blank, Plzystsl. Rev,, 51;702,
2971.
R , J. Scheuplein, J, Invest, Dermatol, , 67: 672, 1916.
.
A , Nioelgaard, and IEZ MoBgaard, J, Cantrolled Release, 2: 111,
1985.
.
P K Wattan, B , MaXlgaard , S , Hadgraft , and A , Haelgaard,
I

Znt. 6. Pkarm,, 24:19, 1985,


A , J. 1. Ward, and R , Tallan, Drug Dev. jfnd. P h a m , , 14:
1155, 1988.
B , W, Barry, in Skin Plzamokinetics. (B. Skzroot, and fiX,
Schaekr, eds,), Karger, Basel, pp. 121-137, 1987,
The Influence of Skin Surface Lipids
on Topical Formulations

DAVID W. OSBORNE and DOUGLAS A . HATZENBUHLER The Upfohn


Compcuzy , Ralamazoo , Michigan

I. INTRODUCTION

As both the largest and most visible organ of the human body, the
skin is of unequaled importance in portraying an individudts state
of being. Therefore, the biology and chemistry of the skin and its
appendages is important for both the cosmetic and pharmaceutical
industries. The heterogeneous nature of the skin provldes one of
the diffimlties in studying this tissue, Just as the skin itself com-
prises morpholoQEically different layers, the glands, follicles, and
micmvasculatu~teof the skin provide zones with different charaeter-
istic pmperties. One of these properties is that a thin, noncontin-
uous film of lipid is deposited on the skin surface from the seba-
ceous glands. Although often neglected in the design of a topical
formulation, the presence o f skin surface lipids can significanay in-
fluence the delivery of topical drugs. This infiuence may be either
beneficid to, or detriment& in, obtaining the desired drug delivery
characteristics and, thus, should be conddered , especially for top-
icals applied to sebum-rich areas of the body (e .g. , forehead,
chwks, and scalp). It is impo&-tant to realize that the interaction
between a drug and the skin surface lipid is not only dependent up-
on the physicochemical properties of the drug, but also upon the
physicmhemical nature of the vehicle. Topical agents can be formu-
Iated that will target sebum-rich zones of the skin (e,g., the hair
follicle) or, attematively, a topical agent can be formulated to mini-
mize deIlvery through the hair follicle. Backgmund information on
skin surface lipids and a brief description of the approach neces-
sary to formulate sebum-sdective vehicles will be the focus of this
70 Osborne and Watzenbuhler

chapter, The discussion is desimed. La remain as simpEstic as pos-


sible in an attempt to introduce concepts that can be studied in
more detdl by consulting the ofiginal, research efforts that w e ref-
erenced.

If. S K i N AND SURFACE LIPID


EXCRETION

The skin is traditionally divlded into three major re@ons: the stra-
tum corneum, the able epidermis, and the dermis, The outermost
of these layers, the stratum corneum, proddes a barfier against
the permeation of most substanees. This layer is composed af non-
viable, keratin-faed cells (squames) that are roughly pentagonal
plates 0.5 vm thick and. 30 to 40 urn across ( 2 ) . Faling the inter-
ceBuIar space between these ceUs are bsayer-stmctured lipids ( 2 ) .
The stmeture of these lipids has proved to be important to the
moisture-retaining absifity of the stratum comeum (3,4) . The viable
epidermis Ees bdow the stratum comeum and consists of stratified
keratinizing epitheliaf cells whose fin& function is to produce the
stratum comeum, This layer does not contdn blood vessels, rely-
ing on noulrlshrnent by cell Buid fmm the deeper dermis layer. The
deepest layer of skin is the dermis, which conslsts of dense, irreg-
ularly arrmged connective tissue, and it is nourished directly by
blood vessels (5). Combined, these layers form the skin and are
connected to the subcutaneous tissue by bundles of collagen fibers.
Embedded in the skin are ecesine sweat glands, apocrine glands,
h d r follicles, mef sebaceous glands, Ecerfne sweat glands are sim-
ple tubular glands distlrlbuted over dmost all of the human body,
Each gland has a secretory part located below the dermis in the
subcutaneous tissue and an excretory duct that ultimatc3liy opens di-
rectly on the skin surface. These glands produce perspiration and
are particularly numerous on the p&m of the hand. Apocrjlne
glands produce eharaeteristic body odors and are primarily located
in the axarz. The secretion of this gland contains chdesterd, ster-
oids, and proteinaceous substances ( 6) and contributes substantially
to the surface lipid on the a i E a , thus, b e h g primarily im.po&ant
to deodorant and antiperspirant hrmulations. When present, tho
apocrhe gland empties into the hElris follicle above the sebaceous
gland.
Two major elassidcations of h d r exist on humans: terminal h&rs
and vellus hizirs. Termin& hairs are the courser h d r s of the scalp
and male tmnk hair, and the root of termhal hairs may extend
more than 3 mrn belaw the skin surface linto the subcutaneous fatty
tissue, Vellus hair is the fi-ne, often unnoticed, body hair that
popujates re@ons such as the forehead, and extends less than 1 mm
Skin Surface L i p i d s and Topical Formulations 71

into the dermis. The hair fiallicle comprises the ha$r, combined with
its surrounding root sheath, The sebaceous gland empties directly
into the upper third of the hair fcirliele, and the cambination of se-
baceous @and and follicle is oAen referred to as the piEosebaeeaus
unit. The sebaceous gland is the major source of skin surface lipid
in the adult. A diagram representing both the skin layers and skin
appendages is shown in Pimre 1.
Because the sebaemus @and is the primary srsctrce of the skin
surface lipids covering large portions of the anatomy, understanding
the properties of this skin appendage becomes important, The se-
baceous glanc;2 is well defined in the embqo, remains relatively smdl

Figure 5, I Grass-sectional. representation of f'U1X-lthi~kn8~s


thin skin.
72 Osborne and Natzenbuhter
dudng chgdhmd, and becomes drasticdly enlarged dur;ing puberty
(7). n each gland, a common excretory. duet is supplied by smdler
X
ducts that originate in the acini of the gland. As the Battened or
cuboidd pedpherd sebaceous cells move toward the center of the
gland, lipid spthesis within the cell bcreases, The cell sweBs with
fat unts il 1100- to 25Q-fold &crease in the cell volume occurs, The
entire cell then ruptures, expelling its contents into the excretory
stream of the gland as sebum, Finfly, the sebum passes through
the pnosebaceous foaide and is deposited into the uppermost re@on
of the hair s h d t . Synthesis and discharge of the lipid contdned
in the sebaceous gland requires 1, to 3 weeks (?), The discharge
of this lipid has been demonstrated by Kligman to be constant, with-
out regard to either season or the amount of l@id &ready on the
skin surface f 8). However, the quantities of sebum that am nor-
m a y encountered on unprotected hedthy skin depend mare upon
the rates of flow (or refatting process) of sebum thtb upon the ab-
solute rate of production of sebum in the sebaceous gland, This
results because there is a large reserve of fufly synthesized sebum
contained in the filtfcular ~esemtotr(91, that is, the upper portions
of the hair foflicle and the osfice to the sebaceous gland, This
source of sebum is not depleted, even by repeated solvent extrac-
tions. Thus, after careful demsing of the sk-in, sebum contdned
in the follieular reservoir initimy appears to Bow aut onto the skin
surface at a canslant rate m e r a several-hour period until the
amount of sebum normmy present on the skin surface (defined by
Kfigman as the casual Zevel] is reached. After the skin has re-
gained the ori@nal casual level of sebum, it appears that the nor-
mal loss pmeess (i,e, , reabsoqtion into the skin, k r t h e r migration
to less sebum-rich areas of the skin, and loss from the skin sur-
face to the surroundings) reaches an equiliibP.lum with the refatting
processes, and no furtkrer increase k sebum concentration is ob-
served, For example, on a relatively sebum-rich skin me&, such
as the forehead, where the foI1ieular reservoir is estimated to con-
tain a few milligrams of sebum per cubic centimeter, it has been
shown ( 16) that the refatting of the skin surface occurs at a rate
of from about 0. L to 2.5. pg of sebum per square centimeter a min-
ute, and that the casud level of sebum is essentiany restored 3 or
4 hr after defatting. For areas of the skin that are less rich in
sebaceous glands, it can be expected that the rate of refatting will
be slower and that the ultimate casuaI level of s&um will dso be
lower.
Once on the surface of the skin, the sebum has already been.
chemicd1-y modified by micraorganisms in the p30sebaceous unit astd
becomes mked with lipids of epidermal ori*, Hence, the terrn
sebum is more precisely reserved to describe the lipid contdned in
the sebaceous gland, and the terrn skin surface lipid is used to de-
scribe the lipid mkture on the skin surface*
Skin Surfice L l p i d s and Topical Formulations

Ill, COMPOSITION OF SKIN SURFACE


LIPIDS

S k h surface lipid pxtimafiy c a n t ~ n sdiglycerides , cholesterol , fatty


acids, teglycerides , w a esters, cholesterol esters, and squdene
(I), The amounts of these components and their vadations d t h
anatomical location have &so been, investigated, TabXe 1 lists the
results obt&ned by Greene et ale (11). When the spedfic stmctures
of the lipids in these general categodes are examined, the skin sur-
face lipid becomes significantly more complex, Hot only do the
chdns of the lipids exhibit different lengths and degrees of satura-
tian, but they also contdn odd-numbered-carbon chdns, uncommon
in biolo@eaX systems, and significant branching ( 12)- The amount
that the lipid fractions of human seburn exhibit these charactedstics
is given in Table 2 (12).
The complexity of the chemicafty diverse mkture of lipid that
eventually reaches the skin surface is a result af equrrXly diverse
inRuences on the sources of this lipi4. The two sources already
mentioned me dways associ&ed with s k h surface lipids and warrant
fudher discussion .
The sebaceous gland is considered the prirnav,
if not exclusive, source of squdene, Classic work by Nicolaides
and Rothman (13) demonstrated that the sebaceous gland8 contdn an
incomplete enzyme system, compared with the epidermis, rendesling
the sebaceous gland heapable of finishing the synthesis of choles-
terol fmm squdene. For this same reason, the source of chdester-
ol is distinctly epidermd. Use of this fact explained the bulk dif-
ferences in the matomicd variation in human skin surface lipids,
As seen in 'Fable 1, the forehead, which contains 400 to 800 sebez-
ceous glands per square centimeter (71, produces skin surface lipid
rich in squalene and law in cholesteraf. This is because the skin
surface lipid of the hsehead is; primarily of sebaceous origin, Con-
versely, the surface lipid of leg skin is ~ g inh cholesterol (9.4%)
and low in squalene ( 6 . 2 % ) because of the greatly reduced sebaceous
coxrtrlbution ( 11).
The source of the fatty acids found on the skin surface is both
epidermal and hdireetly sebaceous. Elias et sf. (14) has character-
ized human epidermd lipids (Table 3) and found the free sterols/
free fatty acid ratio for the face to be 0.9, Assuming that &I of
the cholesterol charactedzed in TabXe 1 far the forehead is of epi-
dernnaf, ori@n, then only about 6% af the free fatty acid on the fore-
head skin surface is af epidermal o d e n , The remainder is the hy-
dmlysis product of triglyce~desthat ori@nate in the fatty acid-
.
free sebum ( 15) HXigman showed that the bactedurn Corynsbaete-
rium acne8 , which inhabits the follicular canal, is responsilble for
this hydrolysis (16). Variations In the &acted& &lor&among differ-
ent hdidduals is used to explain why comparisons between i n d i ~ d -
uafs exhibit dramatie differences in the amount of fatty acid in their
Table 5.1 Anatomicd Variation in the Amunt and Composition of Human Skin Surface Lipids in Adult
Male Subjectsa 3 h r after Washing with Soap and Water

Composition
(wt%) Forehead Cheek Chest I3 ack Side Arm Leg

Free fatty acids 24.1 22.4 28.1 24.9 23.4 36.4 37.8
Wax esters 27.0 24.3 21.7 21.9 17.7 15.8 12.9
Squalene 12.3 13.5 10.7 10.7 8.5 6.9 6.2
Cholesterol 1.2 2.0 1.6 2.1 3.9 4.1 9.4
Cholesterol esters 2.9 3.9 2.9 3.0 5.3 4.4 7.5 0
G;
Total lipidb 160 104 59 38 29 30 19 tr
D
(glcm2) 3
(D

aThe data are averages for three subjects, each examined on two occasions. 9a
b ~ r a p h i cextrapolation indicated that the epidermal lipid contributed an average of 4 g/ern2.
Sourn: From Ref. 11. 2
.v
h)
Skin Surface Lipids and Topical Formulations 75

Table 5.2 Fatty Acid Composition of Each Lipid Fraction Taken

Cholesterol ester Free fatty


Chaln type (%I Triglyceride and hydrocarbon acid

Branched chains 15.0 13.6 12.4

Odd-carbon-number
chains 22.5 26.2 21.8
Even-carbon-number
chains 77.4 73.8 78.1
Saturated chains 70.9 56.2 69.7
Unsaturated chains 29.9 43.8 30.2

Source: From Ref. 12,

Table 5.3 Composition of Human Epi-


dermal Lipid

Component Wt%

Polar lipids 4.9


Cholesterol sulfate
Nr?utr& lipids
free sterols 14.0
free fatty acids
triglycerides
sterol or wax esters 5.4

n-alkanes 6.1

Source: Pram Ref. 14.


76 Osborne and Natzenbuhler

skin surface lipid, although for a given individual, the fatty acid
component of the skin surface lipid remdns relaGvely constant with
time, Thus, each indiddud shows an inverse relationship between
the amount of free fatty acids and the triglycerides charactefistic of
their skin surf ace lipids,
A hydrolysis reaction is the source of one of the important
structurd components of the epidermaf lipids. The enzyme spkingo-
myelinase hydrolyzes sphhgomyelin to ceramide and phosg horylcho-
line 1(17). This enzyme degrades dl cellular gfiosphoBpids before
the cells move from the granular layer of the epidermis into the
stratum cornearn. The ceramides then become one of the mdn corn-
ponents of the layered interceaular lipids found between the termi-
n a y diffemntiated cells of stratum corneum. The other hydmlysis
product, phosphorylcholine , does not fiB the stratum corneurn spaces
but, rather, the phosphorus is thought to be reabsorbed and re-
utaized by the epidermis. This is an explanation for why the skin
surf ace lipids are void of phospholipids ( 18).
]in addition to the substances Ested in Table I, smdl quantities
af other compounds have been found in sebum. Detectable amounts
of mdragens are present in skin surface lipids ( 19) ; however, the
amounts of these steroids are considrtm*ed too smd1 to claim that the
s ~ ~ ~ C ~ glands
O U S play a notable role in the totd body excretion of
androgens, Parafsnic hydrocwbons , such as pdstane , &so have
been shown to be of sebaceous o ~ g i n(21. Their presence is spec-
ulated La be linked to diet.
The influence of diet on skin lipids is another factor that adds
to the chemicd eomple~tyof skin surface lipids. Fasting causes a
50% reduction in the synthesis of fatty acids, w a x esters, ~ n tA- d
glycerides , although not influendng squalene production ( 19) .
L&ewise, the absence o f dietary essential fatty a d d s (linolleic acid)
results in a stratum corneurn with a changed appearance and a sig-
nificantly reduced bar&er function ( 2 ) . Nicolajdes (21) showed that
feeding 1 - [ 1 4 ~ ] octadecane to the rat led to the prtjsence of the
labeled substance in the skin. Although a complete understanding
af skin metabolism is not yet avdable, it is important to remember
that some of the components found in the skin surface lipids are
present simply because the skin was the most efficient route of
""wasteffreclnnova.
When examining the components found h skin surface lipids,
the possibaity that the trace substmces are from exogenous sources
must be considered, The obdous sources are cosmetics, topicd
.
phtarmacmtic&s, toaet Aes , and environment& pollutants AltE-rough
most studies that use skin surface lipids collected from human da-
nors try ta elimbate these eontaminants, the reservoir function of
the stratum cornam m&es it part;icularly difacult to remove totdly
some substances from the skin surface. Beeauae chemicals can be
Skin Surface Lipids and Topical Formulations

readily retained in the stratum corneum, even after 16 days (22),


the total elimination of exogenous lipophilic m a t e ~ becomes
s diffi-
clt .
Methods used to obtain skin surface lipids for studies usually
involve application of a solvent to the anatomical site of interest.
This can be done by using cotton swabs (23), solvent cups (24),
and head soaks (25). Nonmlvent methods exist that take advantage
of the absorbent properties of bentonite clay (26), and the use of
absorbent paper (?,21) as well as ground-glass surfaces (28) has
been reported. The amounts and composition of the lipids are then
.
determined by chromato~aphy Other spectroscopic ( 29) and ther-
mal (3) techniques have also been used.

IV. FALLACIES CONCERNING S K I N


S U R F A C E LIPIDS

In 1963, Kligman, in the article The Uses of Sebum? (31), challenged


a number of previously held concepts concerning the human integu-
ment. Despite the vintage of this essay, some of these fallacies that
Kligman refuted still surface in the literature an&, thus, are worthy
of restating. Two of these fallacies are that sebum is antibacterial
and mtikngal and that sebum contains a vitamin D precursor. The
vitamin Z) precursor was found to be of epidermal origin, whereas
the a n t i b a c t e a and antffungal properties were considered to be
pfimarlly artifacts of in vitm testing methods, Kligman concluded
this article with the statement: "Human sebum seems to be useless.
Since hair has become vestigiaX over most of the surface of the hu-
man body, sebacwus gland are probably obsolescent appendages.''
Although this final statement seems extreme, it provides a balance
to the previously overemphasized benefits of skin surface lipids.
Another area of confusion fn the literature is the question of
whether or not there was a feedback mcrchanism (the concept that
the glands shut down sebum producfion when a certain Xevel of sur-
face fat is achieved) controlling the sebaceous glands. Kligman and
Shelly (32) demonstrated that, under careklly controlled conditions,
it was clearly possible to demonstrate a large buildup of skin sur-
face lipids, we11 above the Xevel that was speculated to eause shut-
down of the glands' production. This work, together with a better
understanding of the holocrine nature of the sebaceous glands (I),
has fully discredited the feedback hypothesis. Note, however, that
even though the rate of formation of seburn is constant, from the
practical point of view of, "how much sebum will a topical formula-
tion encounter on normal healthy skin?", the sebum Bow and refat-
ting pmesaes on the surface of the skin are a mom direct con-
cern, Uormver, it is also well established that the amount of sur-
78 asborne and Natzenbuhler

face lipid present on healthy, unprotected skin will plateau a few


hours after the skin has been thoroughly washed ( 3 3 ) .
The rate of sebum production does not show seasonal changes
(91, &though it is again possible to demonstrate seasond vadations
in perceived oniness of the skin, primarily because of the amount
of perspiration and other factors that afiFect the seburn refatting and
redistribution processes.

V, MODEL SKIN SURFACE LIPIDS

One appmach to determining the effect that skin surface lipids have
an topical formulations is to model either the prope&ies or composi-
tion of skin surface lipids by using readily available mater;irals, This
avoids the laborious task af pmling lipids from volunteers and pro-
d d e s the investigator with convenient quantities of lipid mate~a2.
Skin surface lipid models are of interest to not only cosmetic ~ n d
pharmaceutical hdustries but, dso, to the detergent industry, far
which they are used to compare the abnities of laundry formulations
to cXem sosed fabrics. Because of this interest from various fac-
tions of the industsid research community, a number of model skin
surface lipids are cited in the literature, However, most of these
modcls were designed for specific studies, thus, limiting their prac-
ticality in evdurr-ting cosmetic and pharmaceutical topical agents,
The cornpasitions of some of these model lipids will be @ven, and
the advantages and disadvmtages o f each model will be discussed,
in this section,
The F~berg-Osbome model skin surface lipid (34) given fn Ta-
ble 4 is based on the chemical composition elucidation studies that
have been carried out by various investigators, In this model, each
of the major chemical fwtions of human S ~ ~ Uare M represented by
a purified commerciaUg available chemicaf, For instance, the tri-
glycerides are represented by triole.in, whereas the wax: esters are
represented by oXeic acid pdmitic ester. This modrtl has two major
advantages: it chemicafly mimics naturd sebum, and it is a single-
phase liquid that is easy to work with at ambient temperatures.
This second advantage is also the modelrs strongest disadvantage,
because the physical properties of sebum (melting point &out 3S06)
are not obtained. This model was initially used to describe the
phase behador of a simplified cosmetic system when mixed with the
model lipid, l t is in this capacity that a single-phase Xiquid model
becomes essential for practical studies.
Another model, which is based on the chemical, components
found in natural s k h surface lipids, is given in Table 5, Agdn,
the major advantage to this model is that it chemically. mimics the
skin lipids. At room temperature, this mixture is EI waxy solid that
Skin Surface Lipids and Topicat Formulations

Table 5.4 Composition of the Friberg-


Osborne Model Skin Surface Lipid

Component %

Oleic acid 16.5


Myristie acid 1.9
Triolein 41.8
Ofeic acid palmitic ester 20.3

Squalene
Lecithin

Source: From Ref, 34.

Tabte 5.5 Model Skin Surface Lipid

Component

Palmitic acid
Myristie acid
Oleic acid
Tripalmitin
Triolein
Palmitic acid palmitic ester
OIeic acid palmitic ester
Cholesterol
Cholesteryl palmitate
Cholesteryl oleate
80 Osborne and Hatzenbuhler

Table 5 . 6 Composition of Spangerrs


Sebum Model

Component %

Olive oil
Coconut oil
Pdmitic acid
Stearic acid
OleSc acid
Paraffin wax
Squdene
Spermaceti
Cholesterol
Total

Tabte 5 . 7 Composition of the Gordon Sebum Model

Component %

Hydrocarbon oif (medium-viscosity lubsicaiing oil) 25


Tristearin 10
Arachis oil 20
Stearie acid 15
OIeic acid 15
Cholesterol 7
Octadecanol 8
Skin Surface Lipids and Topical Formulations

Table 5.8 Composi~onof the %orris


Skin Surface Lipid Model

Component %

Squalene
Tristearin
Triolein
Stearic acid
Oleic acSd
Cholesterol
Octadecanol

contains considerable crystalline material, even after repeated melt-


ing and cooling cycles. It is the presence of crystalline matedal,
even at 60°C, that limits the usefulness of this model bmause it is
difficult to manipulate samples of this nonhomogeneous mhture at
ambient temperatures.
Spanglex's sebum model (Table 6) dates back to 1964 and com-
prises a mixture of naturaf fats and oils (35). After melting, mix-
ing, and then cooling these components, a homogeneaus waxy mate-
rial results that mimics well certain physical properties of skin sur-
face lipids. This model has the chemical advantage of comprising
multiple components that have a wide range of branching and of
degrees of saturation. Unfortunately, the chemical properties of
vegetable oils are unlikely to be similar to those of human sebum,
and the rationale for the materials used appears to be based on ob-
taining a wax at ambient temperatures. This model has the added
problem of crystal formation after approximately 1 month on the
shelf.
The Gordon model (Table 7) resembles that of Spangler, but it
was designed for the double-label radiotracer studies that are used
to compare the detergency of a series of detergent systems (36).
The advantages of this doubly labeled system also apply to topical
formulations; namely, the selectivity of a formulation can be inves-
tigated, while the number of experiments is minimized. The com-
mercial avariIab%ty of the racliolabeled matedd dictated the composi-
tion of this model. The very similar Morris model (Table 8) was
oeinafly used to determine retention of an artificial oily soil on
rotton and polyester-&ton durable-press fabrics after laundemjlg
(37). This model has the slight advantage that all of the compo-
nents arc3 readily available.
82 Osborne and Hatzertbuhler

Vt. 1NTERACTIOMS BETWEEN S K I N S U R -


FACE LIPIDS AND TQPfCAL
FORMULATIONS

Research that is concerned with the interaction of topical agents


and skin surface lipids can be divided into two major areas, The
Arst is the research that is concerned with targethg seam-rich
zones of the skin (e. g. , hair follicles and sebaceous glands). This
is of o b ~ o u simport~nceto acne formulations. The second is re-
selllrch concerned with structural changes "tat ocmr in the topical
vehidr; after mixing with skin surface lipid, This area is most im-
portant when formulating topic& agents that will be used on lipid-
rich re@ons of the body, such as the forehead or scdp.
The initial observation. that it is possible to fttarrget" delivery
of d m g substances to the hair faIBcle appears to have been made
by investigators at Unilever ( 3 8 ) . Their work noted that if a ve-
hicle with low polarity (i,e. , matching the polarYity of sebum) and
low viscosity was used, greater amounts of drug could be detected
near the base of the hair follicle, Further work In our laborato~es
( Hat zenbuhler , unpublished results) has demonstrated that this ap -
pears true, even for drug molecules that are inherently too polar
to be readily soluble in sebum, Thus, for localized deliveq into
the re@on of the hair follicle, seburn appears to present an irn-
partant barAer, somewhat anallogous to that of the stratum corneum
barrier (although stmcturally totally different) across the remain-
der of the skin, fmgartantly , the muck greater surface area of
the stratum corneum makes this route of delivew generdly dominant
for the systemic delivery of topically applied drugs (39). However,
this does not negate the foregoing observations for local delivery
to seburn-fich re@ons such as the hair folliele,
The technique most widely used for the measurement of lacar
drmg di~tributionsin tissue i s autoradiography ; a method that uses
radiolabeled drug to measure the concentration of' drug throughout
thin sections (typicany only a few microns thick) of the tissue.
The sectionhg process has been described as highly technique-de-
pendent for any type of tissue, and it is particulriirZy difficult for
the study of skin tissue. Thus, one should carefulfy reproduee
results before malcing final, judgment on their vdidity. Classic
methodology in this technique is p r a ~ d e din Ref, 40,
An dternative method to autoradiography, which can &so be
used to speciacally measure the amount of drug deltlivered into se-
bum-dch re@ons, has been pmposed by Elias (41). This technique
uses different solvents to differentidly extract sebum , rather than
the lipids of skin surface ol..igin. B y comparing the relative amounts
of drug present in extracts of skin surfaee lipids with extracts of
Skin Surface Lipids and Topical Formulations 83

sebum , an estimation of the relative efficiency that two vehicles show


for delivering a drug into the sebum-rich areas may be made,
The basis for understanding a vehicle" effect on the delivery
into sebum-rich areas, such as the h d r follicle, appears to be fully
explained by conventional solubnity properties. Hildebrand solubill-
ity coefficients ( 42) appear adequate to predict this performance.
The HSdebrand coefficients for model sebum compositions demonstrate
that sebum is an averaB nonpolar, oily matefirif [with a Rsdebrand
coefficient of approximately 7.5- 8 ceilll2lcmZf2) , Many topha1 ve-
hicle components, such as water, propylene glycol, and ethanol,
with HiJdtsbrand coefficients of 23,4, 14.0, and 12.55, respectively,
are t w polar to be readBy soluble in sebum (typically .1. 2 units on
the Hllldebrand coefficient indicates miscible matedds) , Therefore,
for many rczfatively polar d m g s , if the vehicle is not spedfieally
designed to solubilize the sebum reservoir in the hair follicle, there
wBl be little chance to effectively deliver the drug into the deeper
portions of the hair folGelei. This may be txtnized to either avoid
delivedng a local excess of drug o r , perhaps mare significantly, to
attempt to XocaEze drug delivery h t o the paosc;baceous unit,
The determination of interactions between skin surface lipids
and topical agents bvolves utilizing various physicochemicd tech-
niques. These techniques include contact angles, solubalty parant -
.
eters , and phase b e h a ; ~ o rdeterminatbn Contact angles are used
to prodde information on the abnity of a formulation to wet the
skin. These measurements can be done in d v o and in d t r o . In
d t r o studies cim, use stratum corneum sheets that have been sepa-
rated from cadaver skin with a twpsin solution, Natural or made1
skin surface 1;ipids can then be added in. known amounts to the
stratum eorneum sheets, This technique has shown that eontact an-
gle measurements can differentiate between relatively polar vehicles
that show superior wetting on stratum comeurn sheets and less polar
vehicles that mare effectively wet a sebum film deposited on the
stratum corneum (Hatzenbuhler, unpubUshed results). This work
demonstrates that the difference in the polaAty of the protein sur-
face of the stratum corneum and the polarity of sebum is sufficient-
ly large ta show observably different vehicle performance. D e t a e d
elinicilil or cosmetic evslluation of this performanee difference appar-
ently has not been reported, However, use of solubnity parameters
in cosmetics formulations was thoroughly descAbed in a recent arti-
cle by Vaughm ( 4 3 ) , in which he lists the solubjility parameters
For over 150 cosmetic matedds.

Vll. CONCLUDING REMARKS

A s described, the skin and its appendages combine to form a com-


plex , heterogeneous network of epithelial( tissue and ghnds that
84 Osborne and Walzenbuhler

produces both the proteins and lipids that form the skin" barrier.
One aspect of this network, the skin surface lipids, can influence
the topied products that are *plied to the skin, This infuence
may be either beneficid or detrimental, depending upon the forma-
fator" awareness of the physicwfiemicd p m p e ~ i e sof skin surface
lipids and the desired drug delivery characteristics. By matching
the po1ar;ity of the formulation to the poladty of the skin surface
lipids, the hair follicle can Be targeted for d m g delivery, even if
the a m g is inherently too polar to be readay soluble in sebum or
skim surface lipid. The reader is further encouraged to consult
the review articles by Strauss et d. (7) and. WhernIlg (44) for more
detajlled consideration of skin surface lipids

REFERENCES

6. F, Bdjland, in Biochemistry and Physioto$y o f the Skin,


Vol. 1 (L. A , Goldsmith, ctd .). Oxford University Press, Hew
York, pp. 3-83, 1983.
P , M. Elias, Int, J. Llermatol. , 20: 1 , 1981,
6 . fmokawa, and M, HattorI, 6 . Invest, DermatoZ, , 84: 282,
1985,
D. W . Qsborne, and S. E. Friberg, J . Dispersion Sei, Tech-
not,, 8: 173, 1987,
A . W , H a m , liistogogy , 7th ed, J. X 3 , Lippincott , Bhifadelphla,
pp. 593-623, 1914.
J. T?T. Labows, R , J. McGinlsy, and A , NI, Rligman, In Prin-
eiiptes of Cosmetics for the Dematofogr'st ( P . Frost, and S. E\J,
Horwitz, eels,), C. V, Masby, St. Louis, pp. 89-97, 1982.
6 . S . Strauss, D. T. Downing, and X", J. Ebling, 3n Biochem-
istry and Physiology of the S k i n , Vol, 1 (ti. A . Goldsmith,
.
ed.) Oxflord. University Press, New York, pp. 569-595, 1983,
.
A . M. Kligman, and W, B Shelley, J. Invest, Dematot, , 30:
99, 1958.
A . M , Kligman, 2 , J, Leyden, and K . J. McGinley, in Prtn-
c;iples of Cosmetics for the Dematotogist , Chap. 3 , 6 , V ,
Mosby, St, Louis, 1982.
I>. Sdnt-Leger, and E, Cohen, B r , J , Dematol., 113: 551,
1985.
R. S, Graene, I), T. Downing, P, E. Pchl, and J * $3, Strauss,
J, Invest, Bermatol, , 54: 240, 18T0,
H, Kosugi, and N . tleta, J p n , J, Exp. Med,, 47:335, 1977.
N , NicdLfibes, and 5, Rothman , d , Invest, Dermatol. , 24: 125,
1955.
M. A. Lampe, A , L . Burlingme, d , Whitney, M. L. Walians,
B, E, Brown, E . Roitman, and P, M, EUas, J . Lipid Res,, 24:
120, 1IfS3,
Skin Surfinee Lipids and Topical Formulations 85

N. Eiiicolddes and C . C . Wells, J . Invest. Dermatol., 29:423,


1957,
R , R, MarpXes, D . T. Downing, and A . M. KUgman, J. Invest,
Dermatot, 58: 127, 1911,
P. A. Bawser, and G . M, Gray, J , Invest. DemaCoE. 70:331,
1978,
55. W. Ernara and H e S. El-Makaddem, Z. Matrtkr, $4:641, 1979.
M. E , Karunakaran, P. IE. Pochi, J e S . Strauh363, E, A. Valedo,
W, lif, Wotiz, and S . J, Clark, S, Xnvesl, DematoE, , 60: 121,
1973.
X-I. J, QfNr;Bl,L, L. Gershbein, and R. C. Scholz, Bfoclzem,
Biophys , Res . Commun . 35: 946, 1969.
N , NicoXaides , Lipids, 1: 8 7 , 1866,
R , El. Staughtan, Arch, Dermatol. , 91: 657, 1965,
K , I(. Jones, M, C . Spencer, m cX S , A. Sanchez, S. h v e s t .
Dermatoi,, 17:213, 1951.
N , Ni~~lajides, R. E. KeXfum, an& P . V , TNwlley X I X , Arch,
Bioehem, Biophys, , 105:634, 1964.
D. T, Downing, A , M , Stranier, J. S. Stsauss, J, Invest,
DermatoE. , 79;226, 1982,
S , J , Lewis, A . R. Shdita, and W , t e e , J. Invest, Demmatol,,
72;370, 1978.
J. S. Strauss and P. E . P s ~ h i ,IT. I n v e ~ t .DermataX., 36:293,
1961.
Lf, Sdnt-Leger, 6. Berrebi, C, Duboz, and P. Agache, Arch.
.
Dermatol Res ,, 2635; 79, 1979,
P. Bare, N, Goetz, and 6 , C, Camn, Int. J , Cosmet. Sei,,
2: 117, 1980,
.
W e J, CunXiffr;, J. N , Kearnq, and N , EJ Simpson, 9, Xn-
.
v s s t Dermalal, , 75: 394, 1980,
A , M 'Kligman, in Advances in BfoEogy o f Skin, Vof , 4 , The
V1,

Sebaceous Glands (We Mantapa, R e A. Ellis, m d A. F,


SBver, eCls ,) , Permagon Press, Oxford, pp. 110-124, 1963.
A. M. Kligman and W. B, Shelley, J . Invest. Dermatol., 30:
99, 1958,
D, Sdnt-Lsger and J. L , Leveqzle, B r , J . Dematot, , 106:
669, 1982,
S , E , Friberg and D . W. Osborne, J . A m , Oil Clzem. Soc,,
63: 123, X98Se
I#, €2, Spane;ler, J , Am. OZE Chem. Sac,, 4f:SOQ, 1864,
El. E , Gordon, 3 . Roddewig, and W, T. Shebs, J, Am, Qil
Chem. Soc,, 44:289, 1961.
&I A,, H~sntitn.and M. A , Morris, Text. Res. J . , 41:657, 1971.
T. Rutherford and J, 43. Black, Br, b. DermaEol., gl(supp1.
4):75, 1969.
8Ci Osbarne and Natzenbuhler

39, R, 3. Scheupfein, S , Invest, DermaloZ, , 48: 79, 1961.


40, W . E. Stumpf, and I&. J. Roth, Autoradiography of Diffusible
Substances , Academic Press, New ?fork, 1969,
41, P . M, Elias, Personal eommunieation.
$2. 3. H , Hadebrand, and R . L. Seatt, Tke Solubility o f NoneEec-
troly tes , Reinhold, Mew York , 1948.
$3, C , D. Vaughan, j.. Soe. Gosmel, C h e m , , 36: 319, 1985.
44, V. R , Wheakly, PFzysiol, Bathaphysiol, Skin, 9: 2829, 1986,
ogy of the Skin

DAVID W . LYNCH
fmmurrex Corporation, Seattle, Washington
LEE K . ROBERTS University o f Utah School of Medleint;, Salt Lake
City, Utah

The skin, by. weight, is the largest organ of the body, Human
skin serves to provide several important functions. These include
maintaining physic& proteetion (barr-ictr function) against extern&
agents and dessication; r e c e i ~ n gsensory stimuli from the endron-
ment; regulating body temperature and water balance; excreting a
.
variety of substances ; participating in metabolic pathways ( e g. ,
initiation of vitamin ZD synmhesis and subcutaneous fat metabolism) ;
and serving as a earnpartmentalized component of the immune system
to provide protection agdnst certain pathogens, toxins, and neo-
plasia, To protect the host from foreign mate~alsand organisms,
an extremely complex relationship has evolved between the skin an&
the immune system, Many o f the cell types in skin synthesize a
variety af bioactive compounds, many of which have profound ef-
fects, not only on lac& in'tlarnmcltory responses and skin-associated
immune responses but, &so, on systemic immune responses,
The development sf ~frate@esto transdermdly deliver phama-
cologically active compounds is an. exciting new area of research
with far-reaching implications. However, transdermal delivery of
phasmacolo@cdly. active compounds also presents a number of chd-
lenging pmblems because the skin is normillly fairly imperdous to
even relative1y small molecules , T hus , from the pharmacolo@c point
of view, one of the mdor requirements for effective transdermd
dalivem is to accentuate the rate of movement of compounds acm8s
the many different layers of cells and connective tissue that collec-
tively form skin, However, transdermd dellvery of a nuniber aE
88 Lynch and Roberts

different compounds has resulted in the generation of significant


hy-persensiti~tyor inflammatory- responses. Far example, &though
transdermal delivery of arrrchdene (a compound that is useful. in the
treatment of Alzheimers disease) results in the sustained delivery of
pharmacolio@caIlg beneficial eoncentrations of the d m g , it results in
the generatlion of symptomatic eont act hypersensitidty (CN) re-
sponses (Kmeger, Pershing , and Roberts, person& communication) ,
Furthermore, subsequent oral, administration of this compound elicits
a wmernoryftCH response. Thus, the induetion of an immune re-
sponse to a transdermally delivered compound may result in a condi-
tion in which it is no longer feasible to achieve delivery by conven-
tion& routes.
Historicdly, during the early development phasea of transport:
systems, little attention was paid to the potential immunalo@cal con-
sequences of t~msdernnrrldrug delivery, A further complication of
the issue is the reticence on the part of a number of companies and
research gmups to report the deletesous effects af transdermal
transpart sy sterns in the scientific literature, thus, making system -
atie evaluation of many of the gmblerns associated with transdermal
delivery systems extremely difficult, if not irnpossibfe, This chap -
ter will present an overdew of the immunobiolom of the skin and
its interactions with systemic immune responses as an aid in eon-
sideration of the potential immunolo@c& ramifications of transdermd
d m % delivery systems, It is hoped that this will lead to develop-
ment of rational approaches to obdate patentidy deleterious side
effects that may arise as a result of these deliveq systems,

f l. STRUCTURE AND HISTOLOGICAL QR-


GANIZATIQN OF THE SKIN

Like other organs, the skin is well organized into histolo@cally de-
fined tissues to faeBft(ste the performance af its vafious functions.
The stmctrxral and functiond relationships of the skfn for its im-
munolo@caj. role are discussed in the following:
The skin is di*ded into two mdn layers, the surface epitheli-
urn, termed epidermis, m d the underlying connective tissue layer,
termed the dermis, Beneath the dermis is the hygodemis, which
3s ai layer of loose connective tissue consisting mainly of subcutane-
ous adipose tissuee The hypodermis is lmsely connected to under-
l y h g deep fascia, aponeu~osis,or perhsteum. A schematic repre-
sentation of the skin is presented in Figure 1.
The epidermis is a keratinizing , stratified squamaus epithelium
composed of several distinct eel1 populations. The keratinocytes
represent mare than 90% of that cells d t h i n the epidermis, These
are the keratinizing squamous cells of the self-renewing epithelium
ImmunoloSy of the Skin

7-LA&GERWANS CELL.

STRATUM CORNEUM

DENDRITIC EPIDERMAL T-CELL


KERATINOCYTE

INFLAMMATORY CELLS

ADfPOSE TISSUE

Figure 6.1 Schematic repwsentation of a histofo@cal, cross-section


of the skin.

which terminay differentiate into the nonviable, flattened corneo-


cytes of the stratum corneum. The stratum comeum is the outer-
most layer of the epidermis. The interlocking platelike structure
of the highly keratinized corneocytes provides the major barrier
function for the skin,
Melanocytes , Merkel cells, Langerhans cells, dendritic epidermal
T cella, and epidermotropic lymphocytes represent the remainder of
the minor epidermal cell populations. Melanscytes are pigmented
dendritic epidermal cells that function to provide pmtection against
ultraviolet radiation (UVRI-induced skin damam through the mela-
ninization, or tanning, process. Merkel cells do not, as yet, have
a defined function, &though it has been suggested that they serve
as neumsensory receptors. Langerhans cells, dendritic epidermal
T cells, and epidermotropic lymphoey tes are distinct immunowmpe-
tent cell populations within the e p i d e d s . Their phenotypic char-
acteristics and functional properties are outlined in greater detail
in the fallowing section,
Lynch and Roberts

The dermis is a dense connective tissue layer that supports the


epidermis. The bulk of the dermis consists of bundles of collagen
and elastic fibers that form a reticular network to support the vas-
culature, nerves, lymphatics, and adnexd structures (hair, sweat
glands, and such). Fibroblasts, which secrete the reticular fibers,
represent the major cell population within the dermis. Mast cells,
which are in close proximity to the blood vessels, represent the
major inffammatory immunocompetent cell population of the dermis.
These cells synthesize and secrete a number of active inflammatory
mediators, Other immunocompetent and innammatory cells within
the dermis include: resident antigen-pmmnting cells and transient
inflammatory-lymphoid cells (e. g, , polymorphonucleocytes, mono-
cytes , and Xymphscytes).
Distinct immunocompetent cell populations do not reside within
the subcutaneaus tissue layer of the skin. However, this layer of
the skin does support blood vessels, lymphatics, and lymph nodes
that are important structural components of the immune system.
Transient inflammatory and lymph&d cells are also observed in the
hypodermis .
Ill, INFLAMMATION AND IMMUNE RE-
SPONSES IN THE SKIN

To understand the immunobiology of the skin, it is important to


distinffuish between inRammEltion and immunity, because the conse-
quences and approaches to circumventing the two are different.
Inflammation is a basic nonspecific response localized in various tis-
sue sites in all multicellular antmds. A variety of stimuli are capa-
ble of triggering an inflammatory response, which is characterized
by a series of interacting and interdependent physiolo@c reactions.
The initial phase of an inflammatory reaction is marked by edema
resulting from an increase in both the blood supply to the local
site and the permeability of the capillary walls. Phagocytic cells
( p ~ a r l l yneutrophns and monocytes) leave the circulation and ac-
cumulate at these sites. This is accompanied by the release of a
large number of soluble mediators produced by lymphoid cells, mac-
rophages, and neutrophils, as well as by the clotting, kinln, and
complement systems. The inflammatory response is a complicated
series of reactions that sets the stage for the repair of damaged
tissues by stimulating collagen biospthesis and the proliferation of
both connective tissue cells and small blood vessels,
The hallmarks of immunologically mediated responses, on the
other hand, are speciBcity and memory. The lymphoid cells, which
compose the immune system, are separated into two broad catego-
Immunology of' the, Skin 91

ries: B cells, which can differentiate into plasma cells that produce
antigen-specific antibodies; and 2' cells, which are further subdi-
vided into a number of subsets based upon their function, These
include cytotoxic I-' lymphocytes (CTL) , delayed-type hypersensitiv-
ity eells ( l i " r ) ~ ~T-helper
), fTh) cells that aid in the generation of
either cell-mediated immune responses or antibody-mediated immune
responses, and suppressor I\ (T8) eells that act as regulator37 cells
for both cell-mediated and antibody-mediated immune responses.
Functiondly associated with the lymphocytes are antigen-presenting
cells (APC), such as macrophages and dendritic cells, which act to
f'processft and "presentt' antigens to the various subpopulations of
B and T cells, The APG are required for the induction, generation,
and revlation of immune responses. Antigens are "presentedf7to
antigen-spedfic T lymphocytes in the context of major histacornpati-
bnitly complex-encoded eeU-surface molecules that are expressed by
APC (1). The induction and elicitation phases of the immune re-
sponse that are in;i"tiated by the APC causes both the activation and
the clonal expansion of the antigen-specific T cells.
Mature T cells are continudly trafficking between the circula-
tion and various different secondav lymphoid organs such as the
spleen, peripheral lymph nodes, mesenterie lymph nodes, and Pey-
erts patches ( 2 ) . The eontinud surveBlance of different organ sys-
tems in the body by- these recirculating lymphocytes thereby pro-
e d e s an enhanced protection against not only pathogens and toxins,
but also aga;inst neoplszstically transformed somatic cells, This con-
tinud immunosurveillanee is of pairticular importance for those organ
systems, such as the skin and the gastrointestinal tract, that me
in constant. contact with the numerous harmful agents encountered
in the external environment.
The immunosurveBlance capabnities of the immune system are
further enhanced by the compartment&ization of subsets of li" lym-
.
phacytes into defined cf rculatory circuits For example, recent
studies have demonstrated the prefemntial homing of a subset o f T
cells to the Peyerts patches, mesenteAc Igmph nodes, and lamina
propria of the htestine, a feat mediated by a distinct lymphocyte-
binding molecule expressed only within the microvasculature of these
tissues ( 3 ) . These data indicate the existence of a population of
lymphoid cells that &re restricted to gut -associated lymphoid tissue.
Although much work rem&ns to be done, data from a variety of
sources also hdicate the existence of an imrnunoloajical circuit, re-
stricted to the skin, that has been termed skin-associated lymphoid
tissue (SALT). Compartmentalized irnrnunofo@cal circuits thereby
increase the pr0babBit-y of an interaction between an antigen-spscif-
ic I-' eel1 and an antigen-bearing APG by directing effector cells to
the anatomical sitesf s) of antigen deposition ( 4,5) .
Lynch and Raberls

The immunologicd effects of antigenic exposure are mmifold,


with the simultaneous induction of both positive effector responses
(which may be cellular or antibody-mediated) and negaave r e e l a t o -
rg responses. In addition to the necessary cellular interactions, a
vadety of lymphokines- (soluble mediators) are required to elldt an
immune response. Thus, the ultimate response detected is the re-
sult of a complex sedes of interactions between these two arms of
the immune response. Furthermore, the amount and route of anti-
gen exposure also play a major role in determining which of these
two arms of the immune response is preferentially induced,

In an anatomic& sense, the s k h is uniquely situated and faces per-


haps the most diverse array of antigenic stimuli, as well as phys-
ic& and chemical insults, of any organ system in the body, For
example, the skin is bombarded ddlg ~ t ultradolet h (UV) radia-
tion, a physic& enPrironmentaX agent that is- known to be c a p ~ b l eof
causing neoplastic transformation of cells. Recognition of the un -
usual set of immunolsgical requirements placed upon the skin led
Streaein to hypothesize that a special relationship elnlsts between
the skin and the immune system ( 6 ) . Thus, it has been proposed,
and a substantid amount of e ~ d e n e esupports, the existence of
SALT that prodde skin with a unique and .vitaXly important sur-
ve2leznce mechanism.
The SALT is hypothesized as being an integrated irnnnunosur-
venlance system compdsed of ( 1) antigen -presenting cells that re -
side within the epidermis; (2) a population af redrculating T lym-
phocytes that are distinctly epidermotropk ; and C 3) a unique ana-
tomic& end~onmentthat can affect the induction, generation, and
regulation of immune responses both locally and systemicdly, Sup-
port for the concept of SALT includes (1) the identification of a
population of cells (Langerhans ec3Ils) that can aet as APC (I);
(2) identificatian sf functional. subsets of T cells that exhibit a def-
inite epidermotmpism (8) ; (3) the demonstration that antigen-spe-
eific T cells can recopize antigen within the skin ( 9 ) ; C4) charac-
terization af a unique population of d e n d ~ t i eepidermd T cells with-
in the skin (10-12); (5) the demonstration that epiderrnd I :cells
are imrnunocompetent in situ ( 13,141 ; ( 6) the elucidation of unique
i m m u n o r e p l a t a ~chcuits that operate within the skin (15-17);
(7) the demonstration that keratinocytes can be hdueed to express
majar histacomgatibBity complex -encoded elass X X molecules ( 18) ;
(8) the identification and efiaractedzation of a number of different
keratinocyte-derived cytokines that can have profound effects on
the rewlation of immune responses, T-cell differentiation, and hem-
Immunology of the Skin 93

atopoiesis C19-28); and (9) the observation that a number of phys-


ic& (e, g ,, ultradolet radiation, heat, sur@cal trauma) and chemical
..
(e g , steroids, arachidonie acid, transdermal delivery- systerns)
agents can have profound, effects on the generation of immune re-
sponses to antigens Brst encountered through the skin (29-32).
Thus, by affecting any of the various components of SALT the nor-
rnal inzmunolo&cal capadty of the skin can be chrznged, with the po-
tential fox a l t e h g not only local, but also systemic, immune re-
sponses.
The p e ~ p h e r a lsecondary lymphoid tissues that drain the skin
provlide the necessary irnmunolo@cal envlironment for the induction
of an immune response. Antigens first encountered through the
skin are transported to the dracining lymph nodes by epidermal
Langerhans cells and macrophages through the afferent lymphaaes
(33). f n cantrast , more than 90% of the lymphoid cells that m t e r
the lymph nodes do so from the blood stream, The entry a f T cells
from the circulation into pefipheral lymph nodes is mediated and
rewlated by specific interactions between the T cells and anatomf-
cslly distinct sites of the ~ulboidalendothelid cell-lined gostcapalaw
venules , the so-called high endothelid venrxXes (WEV ; 3 4 ) . After
binding to HEY, lymphocytes are able to extravasate from the eir-
culatioon and enter the parenchyma of the lymph node (35).
Events that affect skin can have pmfound effects on the SALT
drcuit, For example, Spaneade et al. found that the number of
radiolabeled mudne T cells that "homediqto pe&pher& lymph nodes
rose from 1 l . l B in control mice to 21.8% in mice exposed to 5000 J I
m 2 psr day of UVB radiation over a pedod of 6 days (36). Fur-
thermore, this alteratbn in the lymphocyte trafficking pattern was
observed to persist for up to 2 months. The increase in the num-
bers o f T cells homing to peApheral lymph nodes in UV-irradiated
mice appears to be due to many factors. For example, increases in
WEV stmcture and function have been observed in the pedpheird
lymph nodes of mice exposed to relatively modest doses of UVB ra-
diation (37). This may well be due ta an herease in the number of
activated APG in the peripheral lymph nodes as well as an increase
in the production of cytokines, such as interleukin-l (XL-I), by
UV-irradiated keratinoeytes ( 38) . The situation is ful.t;her accentu -
ated by the effects of increased production of prostaglandins during
the hflammatory response induced by UTT exposure. The increased
release of prostaglandin pxladuces an efferent lymphatic blockade
that, in turn, leads to increased retention times for rec_irculating
lymphocytes in the lymph nodes (39). It is important to note that
the changes in lymphocyte redrculation patterns produced by short-
term exposure to U V radiation are simjilar to those caused by cuta-
neous administration of antigen ($0). Whether similar changes are
94 Lynch and Roberts

induced by various transdermal delivew systems (either by them-


selves or in conjunction with the compounds being delivered) is now
unknown and deserves serious consideration,

V, IMMUNOCQMPETENT CELLS WlTW IN


THE SKIN

We will now focus on those skin cell populations that either have
been shown to play, or are suspected of pladng, a rale in skin-as-
socialed immune responses. The expedmentd model of aller@c con-
tact dermatitis, referred to as contact bypersensitiety ( C W ) , v\ml
be discussed in detail because this type of skin reaction represents
a major obstacle faced. by the pharmaceutical industry in developing
successful trmsdermal drug delivery systems, The roles played by
individustl cell types in the afferent, or induction, phase (which in-
volves T-cell mtigen-priming by APC) and the efferent, or elicita-
tion, phase (which involves the activation and migration of effector
T cells to sites of antigen deposition) of an immune response will be
outlined. The kinetics and mediators ascribed to CH responses w2l
be compared with those involved in antigen nonspeeidc inflammatory
reactions.

Langerhans cells (LC) account for 3% to 5% of the cells in the epi-


dermis (41). These cells constitutively express high levels of ATP-
ase actidty, receptors for the Fc portion of immunoglobulin and
the complement proteh C3b, as well as major histocompatib2ity gene
complex-encoded class If molecules (termed fa in mice and HLA-B in
man; 41). Human LC also express the GI24 (T4) and C f ) l (T6) T-
cell differentiation antigens (42,431, Birbeck granules, which are
detected by transmission eleetran microscopy, serve as the definitive
markers of LC (44). With use of ATPase or immunohistochemicaf
st dning techniques, LC are morpkalo&cally identified by light mi-
croscopy as evenly spaced denddtic cells, An example of mudne
LC stained for cell surface class EX molecules by an indirect immune
peroxidase method is presented in Figure 2 ,
Functionally, LC are very effective antigen-presenting cells ( 4 I),
In vlitro and in Grivo studies have shown that LC are capable of act-
ing as APC for the induction and elicitation of humor& and cell-me-
diat ed immune responses. Their location in the suprabasal. cell lay -
e r of the epidermis indicates that LC are the primary sentinels for
the processing and presentation of antigens that are introduced
through the s k h to the immune system of the host,
Contact hypersensitivity reactions are mediated by antigen-spe-
cific T cebfs in response to eplcutaneously encountered contact sen-
Immunology of the Skin

Figure 6.2 Murine Langerhans cells. Indirect immunoperioxidase


stdning was used to visudize Ia-expressing Langerhans cells in an
epidermal sheet removed from a BGLBlc mouse. Monoclonal &ntibod-
ies against I-A and 1-43 class 11 molecules were used as primary an-
tibodies for identifying Langerhans cells.

sitfzing agents (45). A variety of eontaet sens.itizers have been


identified.. They include metds, plant oils, organic dyes and chem-
icals, preservatives, cosmetics, medicaments, and petroehemie,eerl
products ( 4 6 ) . Although this appears to represent a highly diverse
group of compounds, eontaet sensitizers display a simaar chemical
property in that they finction as haptens. Waptens are classified
as smalf, ( < 2000 molecular weight), highly reatctive eornpounds that
form antigens by establishing chemical bonds with soluble proteins
or cell surface molecules,
B y employing the contact sensitizing agent dinitrofluorobenzene
(DNFB), it has been shown that LC carry DNFB from the epidermis,
through afferent lymphatic vessels to the draining pefipheral lymph
96 Lynch, and Roberts

.
nodes (- 9) Whlile in transit in the afferent lymphatics, the LC are
histolo@cillly identified as velled cells. Once in the drdning lymph
nodes, the LC initiate the afferent phases of a CH immune response
by presenting the cell surface-bound antigen to responsive clones
of T cdls (41). The antigen-stimulated T cells undierp clonal ex-
pansion and seed the peripherd lymph nodes with memory. cells that
are capable of mediating the lintense CH responses observed in al-
ler@c contact dermatitis after subsequent chaBenge with the immu-
nizing cant act sensitizer. Thus, under norm& conditions, any agent
that is introduced through the skin (assuming it has the pmperties
of a f o r e i p antigen) can induce or elicit an immune response through
the actions of the LC to stimulate an effector T-cell response,
Although immune responses (whether cell-mediated immunity or
antibody production) are generdy attdbrzted to the actidties of
the effector cells, it must be appreciated that the immune system is
highly rewfated. The induction af Ts-eel1 responses by most anti-
gens serves to regulate the intensity and duration of an immune re-
sponse. Xt has been shown that distinct populations of antigen-spe-
cific Ts cells are capable of inhibiting the afferent and efferent
phases of a CH response (17,48). Thus, the balance between the
actidties of the 1C"~~-ceXl and Ts-cell populations dic-t-inteswhether
.
or not an d e r @ c contact dermatitis reaction will ocmr It has
been suggested that a population of epidermtll APC that are distinct
"Erom LC may selectively elicit antigen-specific Ts-cell responses
( 152, Lariwise, a rsoluble antigen absorbed through the skin that
gains access into the blood may d s o selectively- stimulate a T"~-cell-
dominated immune response (50). This may account for the elicita-
tion of Ts-cell responses to the topied administration of high doses
of contact sensitizing agents C 51) . The immunolo@cal conditfons
and mechanisms responsible fbr the selective elicittation of Ts-eea-
medhted responses are outlined in detall in Section V X of this chap-
ter.

E3. Dermal Mast Cells


The dermal mast cell population plays a central role in the efferent
phases of CH responses, inflammatory reactions to skin i r d t ant s ,
and immediate-type hypersensitietg responses 6 52- 54) , Mast cells
reside in close approximation to the derrnEif capnlaries. These cells
synthesize and store in cytoplasmic granules a number of vasoactive,
.
compounds (e g ,, serotonin and histamine ; 55) , When activated,
the mast cells release the contents of these storage panules into
the Interstitiurn where the soluble mediators are able to activate en-
dothelid cells to cause increased vascular permeabBity, The ac-
tions of the mast cell pmducts thus account for the edema and e m -
Immunologry of the Skin 97

therna that accompany inflammatov and immune skin reactions. The


increased vascular permeability also allows the movement of in8am-
matow and immune eells into the dermis.
Loveren and eo-workers have reported that cert &n antigen -
primed 1: cells produce factors that activate rntsist cells during the
early phases of the elicitation of a CH response (56). As the re-
sponse progresses, the increase in vascular permeability at sites of
antigen. deposition allows the influx of antigen-speciac Tcfi cells,
The TCW eells secrete a vaAety of chemotactic, cell-activating , and
migration inhibitow factors that sequester inaarnmataw cells at the
antigen - reactive skin site. The reaction proceeds to effect clear -
ance of the antigen from the skin, The response subsides with the
appearanee of the Ts--cell-mediated regulatoq response,
Mast cells mediate antigen-nonspecific hBammatory reactions and
immediate-type hypersensiti~tyresponses in a simDar manner. The
appearance of inflammatory cells within "the skin is directly asssci-
ated with the increased vascular permeablility produced after mast
cell depmulation. In these two reactions, mast cells are activated
by either the direct or indirect effects of the skin irritant (inffam-
matouy), or direct activation by @ergen binding to IgE on the sur-
face of the mast eens (immediate-type hypersensitidty) ( 54) . A1-
though the appearmce of the skin reaction produced by either in-
Barnmatory or immune cells i s sirnsar, the mechanisms that are in-
volved are distinct and, thus, different apgsaaches should be t&en
to circumvent the elicitatjlon of a specific skin reaction pmgtu~edby
a @ven compound introduced through the skin (f . e . , does the com-
pound act as an irritant, contact sensitizer, or dlergen?).

C . Epidermotropic Blod-Borne Celts


Lymphocytes , m o n ~ c y t e ~and
, poFymorphonuclear cells are the medi-
ators of the various skin reactions. The inBux and concentration
of these cells ;In a skin site are responsible Tor the urticada ob-
served at inflammatory and immune reactive sf es. We h a w already
discussed how these cells gain entry into the skin. However, it
should be appreciated that certain populations of T cells display
epidermotrapic properties (57). When activated, these eells tend to
selectively recirculate to skin sites, Mycosis fungoides cells repre-
sent a neoplasia of T cells that have retained this prope&y (58),
Thus, it appears that in addition to possessing antigen-spedfic ac-
tivity through the expression, of the I"-cell antigen. receptor (Ti l
CD3 complex), epidermotropic T cells also express receptors for
skin endothelium-expressed cdX adhesion molecules which a ~ ~ o w s
them to selectively bind to, and exlravasate across, the dermd vas-
culature. These t m e ~ of adhesion molecules have been identified
98 Lynch and Roberts

on lymphocytes that redrculate between mesenterie lymph nodes and


the lamina propria of the srnafl intestine (59,603.
D. Dendritic Epidermaf T Cells
Several years ago a population af mudne dendritic epidermal cells
(DEC) was identified that expressed the T-cell differentiation anti-
gen Thy 1 (10- 12). These cells appear to possess natural killer
eell-like aetidty (61). Furthermore, it has been ~uggestedthat
these cells are responsible for the induction of Ts-cell responses to
skin-associated antigens ( 6 2 ) . However, their exact functional sig-
nificance is unknown. Recently, it has been reported that cloned
d e n d ~ t i cepidermd T-ceUi lines express the gammaldelta form of the
T-cell antigen receptor complex (63). These cloned cells do not ex-
press either 604 or CD8 cell-surface proteins, which are differen-
tial matarkers of mature functiond T-cell subsets (61). It is possible
that these cells represent epidermotropic population of a minor sub-
populatjlon of 3' cells found in the pedpheraf blood that have simaar
phenotypic properties (64) , Alternatively, it is &so possible that
dendritie epidermaf T eells may be immature cells that have under-
gone limited extrathymic T-cell maturation. Thfs is consistent with
the data that indicate that the epidermis possesses similar morpho-
lo@cal and limited hormonal charaetedsties of thymic epithelium (85).
Whether DEG play a role in the generatition a r regulation of effector
functions of skin-associated immune responses is now unknown.
What is clear is that further analysis is required to deternine the
immunobiolo@cd role of this particular cell population,

Under norm& conditions, the I;@ are the only cells within the epi-
dermis that express class X I cell surface determinants ( 4 1 ) . This
is consistent with their known ABC function, since class XI mole-
cules are important far antigen presentation and the ee'Ilular inter-
actions required for the induction of an immune response. Gu&aus-
ly, it has been shown that keratinocytes in the skin of patients
with certain skin diseases are induced to express NLA-DR antigens
(18,66-61)). Sim2ar results have been obt&ned expeAmenta2ly
where murine keratinocytcss can be induced to express Xa (Fig, 3 ) .
Interestingly, those conditions in which the ker~tinocylesare in-
duced to express class 11 antigens are genertblily as~ociatedwith the
infiltration of lymphoid eeUs into the skin. For example, y -inter-
feron (y-IFN) can induce keralinacytes to express class I1 antigens,
which suggests that the release of this lymphokine by activated I"
eells may be responsible for inducing expression of these eell-sur-
face molecules f '72). Efowever, experriments in mice suggest that
Immunology of the SkSn

Figure 6 . 3 Induced expression of fa molecules by mufine keratino-


cytes. X-A and 2-E: reactive monoeland antibodies were used in an
indirect immunoperoxidase-stdning procedure to .vjsudize Ia-ex-
pressing keratinocytes in the epidermis of BALBlc mice treated with
high doses of reeornbinml y -interferon ( 106 unitslday IP for 6 days).

stimuU aside fmm y-inlerferan may allso be cqabXe of indudng kes-


atinoeyte expresshn of cell -surface class 11: molecules .
That keratinocytesj are induced to express class If antigens
suggests that these cells may possess some immunolo@cd potential.
Numerous studies have been undertaken to determine whether or
not keratinocytes expressing class 11 molecules can function as APG
(75-71). The results of" these studies Indicate that these cells pos-
sess v e n limited, if any, APC a e t i ~ t y , For example, y -interferon-
treated human HLA-BR' keratinocytes fiunctlon as stimulators of al-
lageneie T cells only when interleukin - 2 (IT; - 2) is exogenously sup -
plied ( 7 5 ) . Furthermore, recent studies lave indkated that class
100 Lynch and Roberts

11" keratinocytes are incapable of processing soluble antigen (16-


77). Thus, it is not surpdaing to find that Xa" keratinocytes are
capable of presenting only tp"peprocessedffantigenic fragments to
certaln cloned antigen-specific T-cell lines, We have obtdned simi-
lar results using the T-cell mitogen MAS which is prepared from
supernatants of eultures of Mycoplasma arthrftidis (78--79). The
MAS mitogen does not require processing (79) and is reestdcted in.
its presentation to ;Ip cells by the I-I3 encoded class If mdecule (78).
fn our study the I-E-expressing keratinocytes were ewable of pre-
senting MAS to mufine T cells (Ghadess-i et af., mmuscdpt in prep-
aration). Thus, the data hdicate that class f 1" kemtinoeytes dis-
play extremely limited APC actidty.
The genera-f lack of APC a c t i ~ t ydisplayed by HLA-DIIa-ex-
presshg keratinocytes apparently is not due to their inabifity to
produce the appmpriate lymphokines , Keratinocytes smthesize and
secrete both the esduble and the membrane-bound forms of inter-
leukin- l (IL- 1) (22,381. Expression of d a s s I]: mtigens and pro-
duction of IL-l &ppear to be the primary requisites of a function&
APC (801, Furthermore, keratinoeytes have been shown to produce
a wide array of lympktokines, including IL- 3 , EL-6, granulocyte-
monocyte colony-stimulating factor (CM-(=SF) , tumor necrosis factor-
a (TNF- a ) , and prostaglandins (24,25,83-84). Several of these
keratinocyte-dedved lymphokines are produced in increased amounts
after apprap&ate stimulation (e ,g ,, exposure to UV radiation ; 22,
38). The fact that these lymphokines have potent irnmunomodulatory
actidties indicates that keratinocytes can actively- influence an ins-
mune response. Further investigation is regufred, however, to dif-
ferentiate between localized and systemic effects of keratinocyte-
derived cytokkes ,
Although class I1 ~ntigen-expressingkeratinocy-tes do not ap-
pear to function to any sipificant degree as effective APC, recent
studies suggest t h a these cells may play alternative immunolo@caf
roles. Robe*s et &, have reported that Xa expressbn by musne
keratinocytes is strongly correlated wi"t the intensfty and duration
of CH responses as well as the migration of LC into the epidermis
(85,86). This suggests that these cells may function ta direct (ei-
ther attracting or sequesteAng) lymphoid cell movement in the skin,
This concept is suppoHed experiimentciilly by the results of studies
h which it was observed that cloned epiderrnotropic 2' ~ e l I sdisplay
.
increased migrational actidty toward Ia" keratinocytes (87) Xt was
Ellso recently reported. that la+ keratinocytes mrtgr, in fact, function
to selectively elielt antigen - specific Ts -eel1 responses (88) , Thus,
these cells may have an important immunoregulatory function, Fur-
ther evduation of class XI mtigen -expressing keratinocytes is clear-
ly required to determine their exact imnrunoIo@cal role.
immunology of the Skin

VI, EFFECTS OF AGENTS THAT DEPLETE


SKfN OF LANCERHANS CELLS ON 1M-
MUPIE RESPONSES 70 ANTIGENS EN-
COUNTERED THROUGH THE SKIN

The best-eharactedzed system in which dterations of functions of


some of the cells in the SALT drcuit can profoandly a t e r systemic
immune response uses UVB ( 280--320 nm) radiation as the modulato-
ry agent, The results of s t u d i e ~by both Toews et d, (89) and
Lynch et af. (31) have shown that exposure of skin with even low
doses of UVB radiation causes a rapid, and dramatic reduction in the
density of LC in the exposed skin sites, For example, Toews et al.
found the density of ATPase' LC in mudne epidermis obt&ned from
shaved abdominal skin sites that were exposed on four consecutive
days to 100 ~ / m 2of UVB radiation had decreased from 800 cells/mm2
(in nonirradiated skin) to Xess than 50 celfslrnm2 (89). Contact hy-
persensitidty responses to DNFB applied to the UV-irradiated skin
sites were markedly depressed relative to responses elicited in ei-
ther nonirradiated mice or in UVB-irradiated mice that were sensi-
tized to DNPB through a nonexposed skin site. Depletion of ATP-
ase' LC by UTTB irradiation persisted for as long as UV treatments
were continued, but returned to normal densities within 2 weeks
after the cessation af treatments (31). Furthermore, a direct eor-
relation was observed between the density of "returningn A~l"l"ase"
LC and the capacity of" irradiated skin to promote sensitization to
DUIFB. The results of these studies have been extended to humans
by Coaper et alZ, and Slot et al., who found that UV exposure of
skin caused a marked reduction in the density of LC that correlated
with a depression Sxl the abaity to elicit CH response to eplcutane-
ouslly *plied contact -sensitizing agents ( 9Q,91) .
In addition to being unable to induce CHs responses through
UV-irradiated skin sites, application of contact sensitizers to UVB -
exposed mudne skin results in a condition in which the mice are re-
fractory to resensitization thmugh nonirradiated skin sites (31,89),
The refractory state is antigen-specific, and the results of later
studies by Elmets et d. demonstrate that the antigen-speeiec unre-
sponsiveness hduced by topic& application of contact sensitizers to
UV-exposed skin is mediated, at Xeaat in part, by Ts eelfs, which
function to downremlate the afferent, but not the efferent, phases
of the CH response (92).
Xnsight into the mechanisms by which the Ts cells are induced
is pmvided by the results of the much earlier studies by Macher
and Chase ( 9 , 9 3 ) . Their investigations revealed that if the site of
sensitization was sur@caIly excised 22 h r after application of a re-
active hapten that only 4%o f the sensitized animclls developed de-
tectable CW responses, as compared with 80% of the mlmda .in the
102 Lynch and Roberts

control group in which the sensitization sites were not exdsed. Ex-
cision of the sensitization sites 24 or 48 h r after sensitization re-
sulted in the induction of CH responses in 148 and 61% of the ani-
m d s , respectively, Animals that were unresponsive to the contact
sensitizer a&er excision of the sensitization site were also refractory
to subsequent immunization. 0%major importance was the observa-
tion that most of the contact sensitizer left the site of application
through the venous outflow within minutes of sensitization. As little
as 5% was found remaining 24 h r after application. Morwver, intra-
venous administration of the contact sensitizer resulted in the gen-
.
eration of specific C W S unresponsiveness These data suggest that
epicutaneou s application of reactive chemicals to normal skin results
in the simultaneous induction of bath positive and negative immune
responses, and that antigen must persist at the sensitization site
for an apprapeate pedod to activate an effector response,
This scenario has received considerable support from the re-
sults of number of different investigators who have demonstrarled
that positive and negative immune responses can be induced either
simultaneously, or in isolation, depending upan the immunization
prot;oeol used, For example, pretreatment of animds with cyelopkos-
pharnide (to eliminate suppressor cell precursors) before sensitiza-
tion by conventional skin-painting techniques results in enhanced
GH responses ( 94) . Subcutaneous immunization with haptens or in-
travenously injected haptenated, unfraetionated epidermal cells has
&so induced both positive and negative immune responses (95-97).
This should not be interpreted to mean that both types of immune
responses are Q Z W Q ~ S induced simultaneously. For example, skin-
p&nting with threshold levels of a contact sensitizer leads to induc-
tion of C H responses in the apparent absence of Ts-cell responses
(98). Conversely, skin-painting with supraoptimal doses of hapten
preferentiaEy induce Ts-cell generation (5%). Xt remdns to 'be de-
termined which of these situations will be mimicked with each indi-
Actual antigen that is used in a transdermal. delivery system.

Trornsdermd delivery systems potentidly offer both therapeutic and


economic advantages over many of the more canventiand drug deliv-
ery systems currently in use, However, it is isso becoming in-
creasing clear that there are stal major problems associated with
transdermal delivery systems that must be overcome for these de-
livery moddlties to gain widespread applicabnit y and use .
In addition to determining what types of immune responses may
Immunology of the Skin 103

be induced to trarrsdermdly delivered dmgs, it is also important to


evduate the consequences of immune responses to antigens that en-
ter the system thmugh a t'coat-tailrf effect. For example, induction
of a suppressor response to antigens expressed by- pathogens ( e . g l t
baetefia) that normally reside on the skin may render the i n d i ~ d u a l
unable to mount effective immune responses to that agent and may
result in an increased susceptibility to disease caused by that path-
o w n , Nor is it ~ I e a rwhat the effects of transdermaf. carrier sys-
tems will have on the expression of other immunolo@caUy and phys-
iolo@cdly active rnoleeules (such as TL-1, IL-3, IL-6, y-IPN, kxista-
mine, GM-CSP , TNF , prostaglandins) , on the expression of eel1
surface class Xi rnoleeules on either epidermal LC or keratinocytes,
on the abixity of LC to process and present antigens to T cells, or
on the function of the "Z" cells (DEC) which normally reside in the
skin. Thus, the effect of transdermarl delivem systems an the func-
tions and effects (both IoeaXly and, systemically) on all of these as-
pects of the immune response capab2ities of the skin also deserve
scmtiny and carefinl evduation ,
The major obstacle to be hurdled is not in the development of
effective transport systems but, instead, is in understanding the
interaction sf these systems on both local and systemic immune re-
sponses. Should either local. or systemic immune responses be in-
duced by a transdermally delivered compound, there are certainly
a number of approaches that can be taken to resolve the problem.
One approach would be to modify either the vehicles used in the
delivery system or the amounts of the compound being delivered, to
obdate the induction of undesirable immune response. A second
approach would be to chemica31y modify the drug being delivered in
such a way that the compound (or its metabolites) are no longer Im-
.
munogenie, yet still retdn phmmacoloEfical aetidty Finally, it
might be possible to tempomrBy modify localized SALT circuits in
such a way that compounds delivered through a p g r o p ~ a t eskin sites
no longer result in the induction of deletedous immune responses,
Clearly, none of the aforementioned potential approaches are
t r i e d in the amount of effort that will be expended in their impIe-
mentalian, ft also seems to be an inescapable conclusion that if ra-
tional strate@es are to be developed to evduate the exact nature of
irnrnunolo@c~~Iy mediated infiammatory re8ponses induced by trans-
dermdly delivered compounds, then we must first possess a s i p i f -
ieantly enhanced understanding of immune response in the skin,
This goal can only be achieved by devclting increased e f k r t s to-
ward understanding t he fmmunobiolow of the skin , its interaction
with systemic immune responses, and by improtring the dissemination
of the knowledge obtained by the different research groups,
104 Lynch and Roberts

ACKNOWLEDGMENT
The authors would like to thank Janet Swapp, Carol Peck, ~ n d
Patecia Tibola far their excellent secr-etariaf assislmce in the prep-
aration of this m a n u s c ~ p,t
This work was supported by EPlZ Contract 68-01-7288 and Na-
tiand Institutes of Health Grant AM 33543.

REFERENCES

D. Hmsburg, E, Neber-Katz, T . Fairwell, and E, Apella, J ,


Exp. B e d , , 158:25, 1983.
3, L , Gowans, and E , J, Knight, Proc, R , Soe, B i o l , , 5@:257,
1984.
P. R. Streeler, E , L , Berg, B . T. N. Rouse, R , P , Bargatze,
and E. C. Butcher, Nature 331:4f, 1988.
G , E. Walosckr&, and T , B e Tomasi, CRC Grit-, Rev, Immunol,,
4: 11, 1983,
J . W, Streifeh, J, Invest, Dematol, , 72; 167, 1378,
J. VV. Strelileh, J. Invest, Demalol, , 71: 167, 1978,
G , StingX, R, Tam&%, and S, I , Katz, fmmunol. R e v , , 53: 149,
11980.
R, L, Edelson, J, A m , Acad, Dematol., 2:89, 1980,
E . Wtacher, and M. W , Chase, J , E x p , M e d . , 129:103, 1969,
D. A , Chambers, B r . J , DermatoE, , 213: 24, 1985.
E l Tschachler, C, Schuler, J, Hutterer , H, Letbf, It. Wofff,
and 6 , Stingl, J , Invest, Dematol,, 811~282, 1983.
P, R . Bergstresser, R. E . Tigelaar, J . H. Dees, and J . W.
SitreUein, J , Invest. Bermatol, , 81: 286, 1983.
.
d , B Solomon, Transplanatation, 2 :3 2 1 , 1963.
C. F. Barker, and R , E. Biflingham, Transplantation, 14:525.
1972,
R X3, Granstein, 2'. Invest, Demalol. , 84: 206, 1385,
I

J, W, StreBein, J. Invest, Dermatol, , 80: 128, 1983.


W. Pat&, B, Rozycka, P . W. Askenase, and R. IC, Gershon,
J, Exp, Med, , 151: 362, 1986,
J , Smoffe, Acta Dermmato-Venereal, , 65: 9, 1985,
C. Lisby , C . Amslorp, and G, L. Wantzin, Int, J , DermatoZ. ,
263: 8, lif187.
. .
T A . Luger , m d J , J Oppenheirn , A d v , Infaamm, Res. , 5: 1,
1983,
D. N. Saucfer, T. V. ArsenauXt, D, Stetska, and C . B ,
Harley, J . Invest. DematoE, , 88: 515, 1987,
T, S . Kupper, A . 0. Cfrrua, P. Fload, 3 , McCuire, and U,
Gubler, J , Glin, Invest,, 80:430, 1987.
Immunolo*gy of the Skin 105

A . C. Chu, J, A , K, Patterson, 6. Goldstelin, C , L e Berger,


8, T&ezaki, and R , 3;. Edelt~san,9, Invest, Dermatol, , 181:
194, 11383,
M. Danner, and T. A . Luger, b, Invest. Dermatol,, 88:353,
1987.
T I S, Kupper, M , Harowitz, F. Lee, B, Coleman, and P ,
Rood, J. Invest, Bermalal, , 88: $01, 1987.
3 . F, Hicolas, D, Kdserlian, M, Dttrdenne, 84, Faure, and 5 ,
Thiv'rrdet , 2 , Inztes t , Dermalol, , 88: 161, 1987,
T. Schwarz, A. Urbanska, F. Gscfinait, and T'. A . Luger, J,
Immunol, , 138: 1451, 1981,
L, 6. Eahdving, A , Bucklq, m d R. A , Daynes, J, Cline In-
vest, , 76: 1585, 1885,
f3. Berman, fl. S. France, G , P, MartineEi, and A. Hass, J.
Invest. DematoZ, , 80: 168, 1983.
W , L, Morison, and M. L, Kdpke, C e l l , Immunol,, 85:270,
1984,
D. H e Lynch, M, F. Gurish, and It. A , Baynes, cl, Immunol, ,
126: 1892, 1981.
I;., A, Rfneins, and 5 . J, NoretXumd, J, Immunal, , 1315":867,
1986,
I, Silberberg-Sinein, 6. Thorbecke, R , L. Baer, S. A ,
Rosenthaf, and V , Berezowsky , Cell, Immunol, , 25: 137, 1976,
Eil , D. Anderson, A. 0,Anderson, and, R, 6. Wyltle, Immu-
nolow, 3.2: 455, 1976.
6. J. Spangrude, B. A , Braaten, &nd R. A . Daynes, J, Im-
munal. , 132: 354, 1984,
G, J'. Spangmde, E. J , Bemhard, R. S. Ajiotra, and R, A .
Daynes , d , Xmmunol. , 130: 2974, 1983.
W , E, Samlowski, W . T. Chung, f), K. Burnham, and R, A .
Baynes, Regional fmmunol ,, 1:41, 1988.
E , Gahrbg, M, Baltz, M, B , Pepys, and R. Dames, Proe,
Natl, Acad, S c l . U S A , 81; 1198, 1984,
H. T, Ghung, W. l3, S&miowski, I), K . Kelsey, and R . A ,
Dames, CeEE, Irmmuno!, , 103: 531, 1986,
W. C , Kimptort, A. Walsh, T), C. Poskitt, &nd W. K , Muller,
Xnt, Arch, A E l e ~ wA p p ! , Immunot. , 74: 4 0 , 1984.
C. Stingl, K. Tarn&I, and S. I , Katz, Xmmunol, R e v , , 53:
149, 1980,
V, Grokr, M. T a d , A . Harper, K. WoXff, and 6. S t h g l , 3 ,
Invest. Dermatol, , 86: 115, 1986,
F. Fithian, P. Rung, G . Goldstein, M. ftubenfeld, C.
Fenogfio, and R, Edelaom, Pme. Natl, Acad, Sci. USA, 78:
2541, 1981.
M. 8. Birbeck, A , S. Breathnach, and J, D, EveraE, J, In-
vest, Ctermalol, , 37:51, 1961.
Lynch and Roberts

R . 1;. Baer, and J. t, Turk, in Biochemistry and Physiology


of the S k i n , (L, A. Coldsm-it;h, ed .I. Oxford University
Press, New York, pp, 921-937, 1983,
3, D, Wskinson, and It, 2. C. Ryeroft, in Textbook of Der-
matolow ( A , R w k , 21. $3. Wikinson, F , 9, G , Ebfing, R , H ,
61-tampian, and J. L. Burton, eds ,) . BlaekweU Scientific,
Oxford, pp. 435-532, 1986,
G , L, Ashershon, and M , Zembafa, Proe, R . Soe. Lond, B , ,
187:329, 1974.
P, Phanuphak W . Moorehead, and I3t. N , Claman, J, Immunol, ,
123: 1230, 1974,
R. D, Granstein, J. Invest, Dermatoll, , 84:20ti, 1985,
fr". N&ano, and K . Nakano, Immunology, 42: 111, 1981,
M-S. S y , S. I). Mnler, and fl. BI, Claman, J . Immunol, , 139:
240, 1977,
R , K . Gershon, P. W. Askenase, and M. W, Cershon, J , Exp.
Med, , f 42: 732, 1675.
A , Schwartz, P, W. Askenase, and R. I(. Cershon, J , Immu-
nol, , 118: 159, 1877,
6 , 8 , SolIey, G , J . Gleich, R , E , Jardon, and A. L ,
Schraeter, d , Cline Invest, , 58: 4 0 8 , 1976.
5 . 1 , Wasserman, in B t o c h e m t s t ~and Physiology o f t h e Skin.
(L, A. Goldsmith, ed .), Oxford University Press, New York ,
pp. 878-898, 1983.
H , Loveren, R , E l Ratzlaff, IC. Kato, R . Meade, 'I'. A . Fer-
guson, 6 . Xverson, 6. A. Janeway, and P. W. Askenase,
Eur. J , ImmunoZ,, _16:1203, 1986,
M, L , Rose, ID. M. V. Parrot, and R , C , farnee, Cell, Immu-
noE,, 27":36, 1976,
R, L, Edelson, 6. Am, Acad, Demcxtol., 2 ~ 8 9 ,1980.
S . Jalkansen, R, F , Bagatze, d . d e l o s Toyos, and E. C.
Butcher, J , C e l l , BiaZ., 105: 983, 1987.
P, R , S"treeler,E. L. Berg, B e T. N. Rouse, R , F ,
Efmgatze, and E . C. Butcher, Nature, 331: 41, 19138.
N , Romani, 6 . Stingl, E , Tschachler, M. D. Wftmer, R , M,
.
Steinman, E M. Shevaekx , and G Schuler , J , E x p . Med, ,
I

161:1368, 1985,
M, Bigby, T. XCwan, and M. S. Sy, J , I n v e s t , Dermatol,, 89:
495, 1987.
F . Kaning, C , StingE, W . M , Kokoyama, HJ. Yomada, W. L.
Mdoy, E l Tschachler, E l M. Shevach, and S , E. Coligan,
Science, 236: 834, 1881,
L , L, Lanier, N. A . FederspfeX, J. J. Ruitenberg, J , H ,
Phillips, J, P. AUison, D. LSttman, and A . Weiss, J . Exp,
Med, , 165: 1018, 1987.
%2, I?, Haynes, CZln, Res., 34:422, 1986.
Immunology of the S k t n 107

P. G I Natdi, C . D. Martino, V . Quaranta, M , 12. Nieotra, F.


Frezza, M , A. Pellegdno, ~ n d S , Perrone , Transplantation,
31:75, 1981,
U, M, Tjernlund, Arch, Bermatol, Res, , 261:81, 1(378,
A, Seheynius, and fJ , TjernXund , Scand. J . Xnzmunol. , 2 9: 141,
1984,
$3, Aiba, and H , Tagami, B r . J, Dermatol,, 3 ~ 2 8 5 , 3984,
I . A , Lampert, A , J. Suitters, and P . M. Chisholm, Nature,
293: 149, 1981.
A. J , Suitters, and I. A . Lampert, B P , J , E x p , PatZzoZ., 63:
207, 1982,
lt. Y . Basham, B. J. Nickoloffy ?I" C , Merigan, and TI, B e
Morhenn , J, itnvest , DermatoE. , 83: 88, 3984.
B , D, Jun, G , G , Krtleger, and L. K , Roberts, J , Invest.
Dernzatol, , (in press).
t. K . Roberts, L , G. Gahring, S. E , Wfedmeier, 42, C.
Krueger,md R. A . Daynes, J , Invest. Dermtttot, , 86: 503,
1886,
V. B. Nlarhenn, and B . J. Nickofoff, d. Invest, Demalol,, 89:
464, 1987.
8 . M. Breathnach, 5 - Shirnada, Z, Ro-vac, and $3, I . Ratz, J ,
Invest. Dermalol, , 86: 226, 1986,
A. A . Gasjparini, and 5 , 1. Katz, 6, Immunsl,, 140:2956, 1988,
D. M. Lynch, M. F. Gudsh, Z Z . 6. Cole, and R. A. Daynes,
J , Immuno2. , 131: 1702, 1983,
M , C. Bc;koff, B. C , Cole, and H. M. Grey, J , Immunol,,
f 39: 3189, 1987,
E . A , Kurt-Jones, and W, Virgin I V , J , Immunol., 135:36352,
1985.
TI. A. De Leo, H . Borliek, II], Hanson, M, Eisingsr, and L. C.
Harber , J , Invest. Ifematole, 83: 323, 1984.
T, Ruzicka, J , F , Wmlter, and M , P , Printz, J. Invest. Ber-
matol, , 81 r 308, 1883.
J. A , Chodilkewitz, T. S. Kupper, and B. L , Coleman, J. Im-
munot, , 140: 832, 1988.
D. M. Sauder, D , Wong, R, McKenz-ie, L). Stetsko, D. Har-
nish, 77. Tran, B. Niekoloff, T, ArsenauTt, and C, 1 3 , Harley,
J'. Invest, Dematol, , 90: 605, 1988.
L. K. Roberts, G. J. Spangmde, R , A , Daynes, and C. G ,
Kmeger , J . InzmunoE, , 135: 2829, 1985.
L , K . Roberts, Q . G . Krueger, and R. A. Daynes, 2 , Immu-
nol, , 134: 3781, 1985.
T. Shiohesra, and M. Nagashima, 6. Invest, DermatoE, , 88: 518,
1987,
A . Gaspari, M. Jenkins, and S, L, Ka-tz, J. Invest, DematoE,,
90: 562, 1988,
f 08 Lynch and Roberts

6. B . Toews, P , R. Bergstresser, and J, W. Strelilefn, J ,


ImmunoE, , 224: 445, 1980.
It;, I). e m p e r , P. Fox, 6. Nelses, and S, I , Kalz, J , Immu-
no$,, 134: 129, 1985.
W , B, Slat, J. R. Taylor, and J. W , Streaein, J. Invest,
D e ~ m a t a l , , 88:519, 1987.
6 , A, ETmets, P. IE. Bergstresser, R . E , Tigelaar, P, J.
Wood, and J, W , StreBein, J , E x p . Med. , 3158:181, 1983,
E l Maeher, and N. W, Chase, J , Exp. Med, , f29:8L, 1969,
fi, Pol&, and J, L , Turk, Nature, 2411: 654, 1914,
M - S . S y , S, D..N2ler, 3, W. Moorhead, a n d H , 1V, Clanan,
J , Exp. Med, , 149; 1191, 1919,
M. E. Sunday, EZ, Benacerraf, and M. F, Dorf, J , E x p , M e d , ,
153: 811, 1981.
K , Tamalri, £3. Fujfwara, and S , I. Katz, 6. Invest, Dermatol, ,
76: 215, 1978.
$3. SuXlivan, J, W. StreiXein, P . R. Bergstresser, and R. E.
Tigelaar , J , Invest, DematoZ, , 82: 446, 1984,
DRUG FORMULATION SELECTION
AND TESTING FOR TOPICAL USE
Computational Methods for Prodtug
or Drug Analogue Selection
Optimized for Percutaneous Delivery
DAVID W , OSBORNE The Upjohn Company, Kalamazoo, Mdchigan

The introduction of transdermd patch systems cap&le of delivering


therapeutic systemic levels of clonidine , nitrogycerin , est radio1 ,
and scopolamine ( 1 , 2 ) has caused the realization that the skin is an
dternative route for dmg administration, For the topical formula-
tor, an important result of the work conducted on patch systems is
that models are avdlable that can predict the rate at which a d m g
can cross the stratum eorneum barrier, Although topicatl formula-
tions will seldom t q to mdntdn systemic drug levels, IocaIxlized de-
livery of a drug will require that the stratum corneum b a r ~ e rbe
erossed, Thus, it iis impo&ant to know the Bux value of the dmg
across the stratum corneum, By hadng this information early bur-
ing development of" the formulation, timely discussions of drug con-
centration within the formulation can be completed. The need for
addition of absorption enhancers to the formu1at;ion can also be de-
termined,
Recently four empirical models were evaluated to determine their
ability to predict percutaneous absorption of dmgs (3). Two of
the predictive models evaluated were developed by Berner and Coo-
per ( 4 1 , the third was taken from a publication, by Michaels et al,
(5), and the fourth was proposed by Albery and Hadgraft (6,7).
In this p r e ~ o u sstudy, ten drmgs, hadng a ~ d rangee of physical
properties, were evaluated and their resulting predicted transder-
rnaX flux values compared with experimental vdues, These permea-
tion equations are u s e h l for predicting the flux of a drug through
the skin barder when vdues for the water soluhBity, partition eo-
efdeient , and molecular weight of the drug are known, However,
all of the models examhetl occasionally predicted transdermal fluxes
that were unacceptably high or low when compared with expedmen-
t d l y determined data. It was hoped that insight into these discrep-
ancies could be gdned by obtdning addition& expednentezl trans-
dermd flux rrdues from the literature for cornpadson with predicted
vdue6. Thus, sn appropdate combination of the predicted values
fmm each of the models might g r o ~ d ea more reliable overax pre-
dieted flux.
Another apI-zffcation for these computation& methods becomes ap-
parent by reaBzing that a reasonable transdermal flux value for a
d m g can be predicted with knowledge o f the molecular weight and
two read3y cdculable physicocktemicaX d m g properties (8,E)). Be-
cause group contdbution methods &low the cdculaeon of both par-
tition coefficient and water solubafiy, it becomes possible ta predict
a trmsdermaf flux vdue for m y liquid drmg molecule whose stme-
Lure c m be drawn, regardless of whether or not the drug has &c--
tudly been synthesized, For c ~ s t a l l i n edmgs, knowledge of the
melting point is required in addition to the stmcture of the male-
cule. This could be of particular vdue when desiwing analogves
to existing drugs or prodrugs in hopes of obtaining increased sys-
temic breathrough aRer topical administration, Indeed , by calcu -
lating flux vdues over a range of water solubilities and parti-
tion coefficients, par2icularly useful insight into how to alter d m g
structure can be obtI3lixted.

II, DESCRlPTfON OF THE MODELS

Because the models have been descdbedi in detdl in both the a ~ d -


nal publications and in the predous technjical update, only a brief
descdptbn will be presented here, Model 1, the two-parallel-path-
wscy mode1 for s k h permeation pmposed by Bemer and Cooper, (4)
separates permeation through the skints barrier function into two
paths, a polar (aqueous) path and a nonpolar (lipophnic) path, The
diffusion eonstmls for the polar, Dp, m d lipalphMc, I)L, pathways
are calculated u s h g the equations

where WX is the rndecular we4ght of the d m g , and the resulting dif-


fusion constants are given in cm2/hr. Because the fluxes of the
polar m d Hpophilic pathways are considered additive, the tat& flux
Computational Me thotls for Prodrug Selection I lf

( 6 ) of the drug through the stratum corneum can be stated in terms


of the diffusion coefficients from Eqs. I and 2 , as foaows:

where Ap and AE, are the area fractians of the polar and fipophilic
pathways, respectively; P is the partition coeftfbient of the nongo-
lar phase; Gw is the drug" water solubixty (in mglml); and L is
the thickness of the stratum cornearn (in cm). Bemer and Cooper
suggest that Ap = 0.1 and AL = 0.9. The octanol /water partition
coefficient is used for P , and a vdue of O.Q015 cm is suggested for
the thickness of the stratum corneurn, Using these vafues for the
parameters, J will have the units mg/cm2 * hr.
Berner and Goapelr have &so proposed a three-pardlel-pathway
madel ( 4 1 , which combines not only the polar and lipophiaic pathways,
but &so an oil-water multiXaminate pathway (2). Use of the same
vdues far I , DL, Dp, and Ap as for the two-parallel-path model,
and decreasing the value for AL from 0.9 to 0.5, the equations to
caIculate the upper ( J U ~ ~and ) lower ( J L ~ bounds
~ ) of the max-
imum fiux thmugh the stratum csrneum are e v e n in E q s , 4 and 5*
The values for $Urn, and JL,, have been averaged in this evalu-
ation to [ p r s ~ d ea shgle v&ue for the three-parallel-path model,

Xn the heteragenmus strueturd model (model 3) described by


Michaels and associates (51, the barrier Eunction is treated as a dis-
persion of hycfrophsie protein in a continuous lipid matrh through
which the drug migrates by disscilution and Fickian diffusion, Giv-
en this reasonable assumption about the s"t.cture of the stratum
comeurn, the shplified flux equation is where P is the lipidlpratein
partitian coefficient, which in the ori@naf paper, is set equill to
the mheral a&/water pastition coefficient. The oetmdlwater parti-
tion coefficient has; &so been successfully used,
The last model ((4) was adapted from a theoretical description of per-
cutaneous absorption published by A l b e e and Hadraft in 1979 ( 6,7) .
Xn steady-state applications, the total fluxes can be predicted by
using the following equation :

Ill. COMPARfSON QF PREDICTED AND EX-


PERIMENTAL FLUX VAI-UES

Table 1 lists the molecular weight , partition coefficient , water solu -


bBity, expefimentd Qux, and predicted fiuxes for 50 compounds
.
(5, f 0-20) When compgifing a list of experimental values, the most
str-iking result is the huge value range that is found on the few
occasions in which values are determined by more than one investi-
gator, Partition coefficients and wa_ler solub3ities can vary by 2
orders of mamit-lude, whereas experimentally determined transdermal
flux vaues can vary by 3 to 5 orders of magnitude. With trans-
dermaE flux measurements, such variabnity is usually the result of
using substantid& different expeAmentaf techniques (see Chizpt .
12). Xn many respects, the ""bal.fparktT"predictive nature of these
empidczll models is desirable over ""quick-and-dirtyf' in ~ t r trans-
o
dermal experiments, especially if the purity of radidabel is ques-
tionable. The average predicted fiux vdue is prodded for compar-
ison purposes.

IV, USE OF T H E PREDICTIVE MODELS

A f'ew examples should demonstrate the usefulness o f these predie-


tive models. First, if the drug or dmgs to be evaluated have been
physicochemically elharacterized , then paHition coefficient and water
so1ubBity data will be available, For 13 substituted melamines and
related compounds (Table 21, these properties have been deter-
.
mhed ( 2 1) This information, eombined with the molecular weights,
directly p r o ~ d e spredicted transdermlll flux values. If either max-
imum or minimum systemic breakthrough was desired, then these
predf~tedvdues would aid in the decision of which compound to
test in ~ t r o . However, it is important to remember that for ana-
lowes that are not expected to be metabolized to the same parent
cornpound (i.e, analowes that are not psodmgs) the actidty of
each species must be considered, If an a n d o p e has a tenfald high-
er flux rate, but is 100-fdd less patent, then it obfiously wfll not
Computational Methods for Prodrug Selection 113

be the drug of choice. Once again, the balance between localized


effect and systemic breakthrough is important.
The physicochemical characterization of a drug may not be com-
pleted early enough during drug development efforts to have both
the partition mfficient and water solubility data avdable when
evaluation by computational methods is desired. If pitrtition coeffi-
cients are not available, then they can be calculated using various
group contribution methods (8). Because these computational tech-
niques are becoming more readily available commercidly, they will
not be fu&her discussed here, High-performance liquid chmmatog-
raphy (HPLC) methods are also cited in the literature that correlate
.
retention time to partition coefficients (22,23) These methods have
the advantage that small amounts of the drug matedal are required
to determine the partition coefficient. Thus, partition coefficients
can be aperimentally determined by HPLC when only a small quan-
tity of the d m g is available, or partiton codlfficients can be calcu-
lated based upon drug structure alone.
KnowIedge of the partition coefficient and melting point of crys-
talline compounds can be utilized to estimate the water soIubilily of
the compound by use of the method of Valvani and Yalkowsky (9).
Specific equations for various classes of organic compounds are @v-
en in Table 3. This water-solubility value combined with the cal-
culated or experimental partition coefficient can then be utilized to
predict ffux values as shown previously. Table 4 gives predicted
transdermal flux vdues for a series of benzodiazegines, calculated
using partition coefficients and melting points (241, This is an ex-
ample of how predicted flux values for a series of analowes can be
combined with pbarmacokinetic information (25) to select the d m g
candidate most likely to provide the most efficacious topical treat-
ment. The use of these models for a series of timolol prodrugs (26)
is given in Table 5. Because these compounds axe expected to be
metabolized to the same active species, the Rux values are the pri-
mary concern, However, the rate of hydrolysis may also be impor-
tant for localized delivery, since the prodrug could be nswept'9y
the targeted site of delivery before conversion to the desired spe-
cies, if the rate of hydrolysis was sufficiently sIow ,

V, CONTOUR PLOTS

AIthough the models descdbed in this chapter may prove useful in


evaluating existing drugs for use as topicals, a more fundamental
utility exists for these computational methods; namely, a set of
gualltative guidelines can be established that will aid the synthetic
chemist in designing drug analowes and prodrugs. In the past,
vague guidelines existed indicating that increases in lipophilicity re-
k
Table 7.1 Comparison of Predicted a n d Experiment& TransdermaI Flux Values (mg/cm2 0 fir)

Cw Model Model Model Model


Compound (Ref. MW PC (mtyld) Exp Flux 1 2 3 4 Avg

A-taminiphen ( 10) 151 2.95 0.012a 4.6 0.033 0.032 0.064 0.032 0.038

Atropine (5) 289 0.006 2.4 0.02 0,061 0.081 0,028 0.017 0.05
Benzoic Acid (10) 122 89 3.4(11) 720 439 286 280 38 270
Benzyl Alcohol ( 10) 108 10.5 40(11) 1,060 779 589 727 251 590
Caffeine ( 10) 194 0.955 22(fl) 1.5 12 14 40 23 20

D i d t o x i n (5) 765 0.014 0.01 0.00013 1.3 x 1.7 x lom7 0.00027 0.00017 0.00009
Dextromethorphan (10) 271 13,489 0. 017a 10 30 19 2.8 0.21 14
Diazepam ( 10) 285 631 0.05(14) 4,7 3.4 2.2 7.0 0.6 3.0

Fentanyl (5) 337 200 0.2 2.0 1.9 1.2 22 2.3 5.7
Flurbiprofen
Nydralazhs! HCI (10) 161
N y d m r t i s o n e ( 10) 363
Ibuprofen ( 10) 206

Idolyl-3-amtic acid 175


(10)
Indomethacin ( 10) 358

Methotrexate ( 17) 454


Methyl salieylate (10) 152
Minoxidil (10) 209
Mitomycin G (MMC) (13) 334
Benlryl MMC (13) 434
Benzoyl MMC (13) 448
Benzylcarbonyl MMC 462
Table 7.1 (Continued)

Cw Model Model Model Mdel


Compound ( R e f . ) MW PC (mgiml) Exp lux 1 2 3 4 Avg .
B enzyloxycarbonyl 478
M M C (13)
Pmpyloxycarbonyl 430
MMC (13)
Pentyloxycarbonyl 458
WNC (13)
N onyloxyca~bonyl 514
MMC (13)
Morphine sulfate (10) 285

Nicotinic acid (10) 123


Nitroglycerin (5) 227 10.0 1.3 13 3.6 2.8

Salicylamide ( 10) 137 7.76 0.02(11) 53


Salicylic acid ( l o ) 138 174 0.002 1,900
(11)
Scopolamine ( 5) 303 0.026 75 3.8
Testosterone ( 10) 288 2040 0.039 34
( 15)
Theophylline (18) 180 0.66 7.5 1.08

Trbmcinolone 434 339 0.04(20) 0.18


aeetonide ( 103
Correlation f o r plot of In(theoretica1) vs. In(experimental)

aCalculated using equations in Table 3.


b ~ e r s o n a lwmmunications - various sources.
128 Osborne

Tsbte 7 . 2 Predicted Tranadermal Flux Values for a Series of Substi-


tuted Me-amines Calculated from Experimentally Determined Octanolf
Water Partition Coefficients and Water Solubilities (mglml)
Drug Structure

Calc .
DruglRl R2=Rg MW PC CW fluxa

NMe2 NMe2 210 350 0,091 8

N HMe NMe2 196 67 2.16 65

aAverage of the four models (uglcm2 hr).


Source: From Ref, 21.
Computational Methods for Fmdrug Selection 119

Table 7.3 Equations for Prediction of Water SolubFUty

Fundamental equations
liquid solutes
Iog Sw = -1.072logPC + 0.672

crystalline solutes : rigid nonelectmlyte drug molecules

log Sw
C
= -logPC - O.OlMP + 1.05

rigid polycyelic aromatic hydrocarbons, e 0 . ., nap ht &&en@

mono- and multifunctional halobenzenes , e. g. , 1,2-diehlorobeneee


c
log SW = -0.99lagPC - O.01MP + 0.72

steroid hormones, e g ., progeatemne


e
log Sw = -0.88logPC - O.OlNP + 0.08

hexamethylmelamines e. g I ., hexmethylneEamine
log S'
W
- 0.007MP + 0.07
= -0.904IogPC

alkyl para-substituted benzoates , e. g ., hexyl-p-aminobenzoate

log sC = -1.14logPC - 0.005MP + 0.633


W

alkyphenols
C
log SW = -0.997 log PC - 0.008 MP + 1.43
-.--- --

Source: Prom Ref. 9.


Table 7.4 Predicted Transdermal Flux Values for a Series of Ben-
zodiazepines Calculated Using ExperimentalIy Determined Partition
Coefficients and Melting Points

Melting Calc. Calc.


Drug IwW PC point (OC) CWa fluxb

Diazepam
Oxazepam
Lorazepam
Temazepam
Desmethyldiazepam
Triazolam
Midazolam
Bromazepam
Chlordiazepoxide
Alprazolam
Clobazam

aUsing equation for rig%&polycyclic aromatic hydrocarbons in Table


3,
bAverag-e of the four models (vg/cm2 = h r ) .
Source: PC from Ref, 24; MP from Ref, 11.

sulted in increased permeability and that major changes in molecular


weight were required to produce significant differences in diffusivi-
t y ( 2 7 ) . Although these generalities were very useful, more de-
tailed insight into the relationship between a drug" physical pmp-
erties and percutanwus permeation can be obtained by use of con-
tour plots of the predicted flux that are plotted as a function of
the partition coefficient versus water solubility at a fixed molecular
weight.
Contour plots for mleeular weights of 50, 100, 300, and 500 are
given in Figures 1 and 2. For molecular weights of 500 to 1000, the
contours were indistinguishable on this scale from the plot for mo-
lecular weight 500. A s seen, for each nolecular weight, the pre-
dicted flux values smoothly increase as both water solubflity and
partition cwfficient increase. Although drugs, such as the pm-
staglandins and some antibiotics, have both relatively high water
wlubilities and partition coefficients, generally spetlking, the upper
Figure 7.1 Plot for dmgs of molecular weight (MW) 50 and 100.
Contour lines are values of predicted transdermal flux in units of
vg/cm2 hr. The ordinate and abscissa are natural IogaPithms of
water solubility (mglml) and octanol /water partition coefficients , re-
spectively.

Figure 7.2 Plot for dmgs of molecular weight 300 and 500. Con-
tour lines are values of predicted transdermal flux in units of pg/
em2 * hr. The oranate and abscissa are natural logarithms of water
solubility (mglml) and oetanollwater partition coefficients, respec-
tively.
h.,
hu
h3

Table 7 . 5 Physic& Data and Predicted Transdermd Flux Values of Timolol

Melting Cale
Ester R MW PC point (OC) CWa Csle. flux

0 -Aeetyl -COCH3 394 13 204 3.1 18


&usingequations for rigid nonelectrolyte dfug molecules given in Table 3 ,
Source: From Ref. 26.
right corner of t h e contour plot is seldom attainable. Most notable
on these plots is the shape of the contours and changes in this
shape as molecular weight is increased, Chan&ng the physical
p r o p e ~ i e soP a drug in a manner that results in movement parauel
to a flux contour line will result in neither an increase nor decrease
in the predicted transdermd delivery of the dmg. Conversely,
chan@ng the physical properties of a drug in a manner that results
j;n movement perpendicular to a flux contour line will provride the
v e a t e s t change in predicted delivery, For example, increasing the
pa&ition coefficient fmrn 0.05 to 0.37 (InPC = - 3 to - 1) , while
maintdning the water solubility at 2.72 mglml (Ins = 2) results in
a 5,s-fold herease in predicted flux for a d m g of molecular weight
300, but only (ft 2.4-fold increase in the predicted flux for a drug
of molecular weight 100.
A good example of modng along a contour line is seen in the
experimental transdermd flux results for the mitomydn C (MMC)
analopes listed in Table 1. Note that the paTt;ition coefficient of
pentyloxyearbonyl MMC is twofold greater than the partition c ~ e f f i -
c-ient for benzfloxycarbonyl MMG , whereas the water solubBities and
molecular weights are approximately the same. The experirnen-tirl
results s b w that the in ~ t r pereulaneous
a absorption of these sub-
stances are essentially the same, just as anticipated based upon the
contour plot for molecular weight 500, Alternatively, benzyl MMC
m d propyloxycarbonyl MMC can be compared, Both compounds have
essentially the same molecular weights and l@ogkfiiefty , but benzyl
M M C has about fourfold greater water s01ubBitg than prspyloxycar-
bony1 MMC, Here, the increased water solubSity of the analope
increases dmg flux by greater than an order of mamitude, Agdn,
this increase in flux of the drug would have been anticipated from
noting the perpendicular crossing of the cantour line in Figure 2.

VI. ADDITIONAL CONS tDERAT tONS

Although these contour plots are of obdous utility, it must be re-


membered that these computational methods provide predicted Bux
values that are g e n e r a y within an order o f magnitude sf the ex-
perimental values determined using in v;itro transdermd methods.
It is fully expected that such computation& methods may not be ap-
propriate for topical drugs, such as steroids, that uniquely interact
vuith the baayer-stmctured epidermal lipids (see Gkap , 3) (28-30) ,
or for d m g s that alter the barrier properties of the skin. This
second considerhll.ian can be seen in Table 1 when c o n s i d e ~ n gdata
for the Iresatslytic salicylic acid, Thus, expeAmenta1 values
ic,

10,000-fold greater than the predicted vdues result, Xt is also


noteworthy that each o f these models asBarnes that diffusion is stra-
tum eorneum-controlled , rather than dermzllly controlled,
Computational Methods for Prodpug. Selection 125

Also it is important that the compounds initially investigated to es-


tablish the empirical parameters were nonelectrolytes , Thus, drug
molecules that wBl dissociate should be evduated with reservation,
As seen. in Table 1, the cornputatland methods ean be predictive
.
for dmgs that dismciate However, predictions for compounds that
fall outside of the assumptions of the models must be considered
suspect, L&ewise, use af these techniques to predict the pereula-
neous delivev of polypeptides is probably unfounded. Octanoll
water partition cmfficienls, for polypeptides are often larger than
would be axpeeted compared with non-hydrogen-bondhg solvent I
water partition coefficients. The use of mineral oillwater, or af-
kanelwater partitian coefficients may be more predictive c-sf the ac-
tu8t percutaneous flux vdues. Finally, remember that axzly the two
models by Bemer and Cooper utBize the mdecular weight of the
compound, Far compounds of molecular weight h o v e 1000, the re-
sulting flux values fmm these two models might be more predictive
of experiment& results,

REFERENCES

.
A . F, Kydanieus, and B Bemer, eds, , Transdermal DeTivev
of Dmgs, Vols, 3 - 3 . GIRC Press, Boea Raton, Pla., 1987.
B, 6 , Monkhouse, and A . S . Wuq, Drug Dev, I n d . Pharm,,
24: 183, 1988.
D , W, Osbome, Pharm. Manufact., 3(4):4J, 1986,
. .
B Besner, and E , R Cooper, in T r a n s d e m a l Belivery of
Drugs, V d . 2 ( A , F', Kydonielxs, and B , Bemer, eds,).
CRC Press, Boca Raton, Ra,, pp. 41-62, 1987,
A , S e Mi~haeX~, S 1 K, Chandrasekaran, and 3 , E . Slrzaw,
AlCEfE J , , 22: 985, 1975,
W e 2 , Ailbery, m d 3 , Wadgraft, S, Pharm, Pharmacol, , 31: 229,
1919,
W, J, A l b e q , and J , Hadgraft, J , Pharm, P k a m a c a l , , 31:140,
1879,
J, T, Chou, and P , C. Jurs, in Physic& Ghemicacl Properties
af Dmgs ( 2 3 , W e VdkowsXcy, A , A , Sinkula, and S . C . Val-
vani, eds-). Mared Dekker, New Vork, pp. 1.63-199, 1980.
S, 6 , Valvani, and S N, Vdkowsky , in Physical Chemical
I

Prope&ies o f Drrxgs (S. H. Yalkowsky, A . A , Sinlrula, and


S , C , Valvani, eds ,) .
Marcel Deklrer , New Uork , pp 201-
229, 1980,
.
G , B e Kasting, L. Smith, and E. R Gmper, in Pharmacology
and the Skin. Val. 1, Skin Pharmaeoktnetks .
(E3 Shroot arrd
Eli. Schaefar, eds.). Karger, Basel, pp. 138-153, 1987.
Osborne

The Nemk Index, 10th ed, Merck Ilr Co, , Rahwsy , N ,J., 1983.
S. H e Udkowsky, $3, el Valvarai, W-Y Kuu, and. R-N Bannen-
felser, Arr'zona Database of Aqueous Solubility, 2nd e d , , copy-
dght 1987 by S . flalkswsky*
E, Mukai, K, Arase, M, Hashida, m d W . Sezaki, Int, J.
Pharm. , 25: 95, 1985.
A , MacQondd, A. F, Michaelis, and B. 2;. Senhowski, in Ana-
.
Eyticat Profties of Drug Substances, VoX, 1. ( K Flarey , ed ,> ,
Aeadernjlc Press, Hew Uork , pp, 79-99, 1972.
$3. H . Udkawsky, S. 6 . Vafvani, and T . J , Rosemaur, J,
Pharm. Sci, 72: 866, 1983.
3, A , Akhter, and B. W . Barry, J, Pharm, PharmaeaE, 37:27,
1985.
S, 8 4 , Wallace, J. O, Rura&.is, and W D, Stewart, Can, J,
Pfiarm, Set, 23: 66, 1978.
K. E3. Sfoan, S. A , M, Koeh, K , GI Siver, and F. D. Flowers,
J, Invest, Dematol, , 87: 244, 1886.
R, GarNl, K. Engle, G. Rork, and L, 3. Calidwell, Pharm,
R e s . , 3:225, 1986.
.
K F l o r q , hi Analy tical PpnoffEes o f Drug Substances, Vole 1.
( K , Florey, e d , ) , Academic Press, New York, pp. 397-421,
1872.
A . J. Cumber, and W. 6. J, Ross, Ghem, Biol, Interacttons,
17: 349, 1917.
R , Kdfszan, Quantitative Stmcture-Chrornato~aphicRatention
Relationships, John WSey 8s Sons, New Uork, pp. 232--278,
1987.
W, J , Lambe-t, and L , A . Wright, J . Chromatag., 464: 400, 1989,
B. If, Greenblalt, R, IVI. Arexzdt, D, R. Abemethy, H. G ,
Gjiles, E. M, Sellers, and R 1 . S h a a r , B r , J . Anaesth, $5:
I

985, 1983.
D. J, GreenblaPt, M. DivoXI, D. R. Abernetfiy, H. R , Ochs,
and R. I. Shader, Drug Metab, Rev., 1 4 : 2 5 1 , 1983.
I-I, Bundgaard, A, Buur, S,-6. Chang, and V. W . L. Lee,
Xnl, J . P h a r z , (Amst,), 46:77, 1988,
J, L, Zlatz, CTFA Scientific Manograph Series No. 2 , , CTFA,
Washh@on, D,CI,1983, pp, 29--45,
I). W . Osborne, COLL, Abstract 072, l(318.t ACS Nation& Meet-
ing, New Uork C i t y , Mew Usrk , Aprif 13- 18, 1986,
C. Ackernnann, and C . L. Rynn, Int, J , Pharm, (Amst ,) , 36:
61, 1987.
C , Ackermann, 6 , L, Flynn, and W. M, Smith, Int. J.
fharm. (Amst,), 36:67, 1981,
Kinetic Considerations in the
Design of Surfaclant-Based
Topical Formulations
ANTHONY J . I . WARD* University College Dublin,
Dublin, Irelattd

I, INTRODUCTION

The development of formulations using surfactants as vehicles for


transdermal drug delivery is currently of great interest. Much of
the attraction of this type of system lies in the acceptable rheologi-
cal and wide-ran&g physicoehemical pmperties that can be ob-
tained. The objects of system design are to achieve a suitable res-
ervoir dosage and controlled release by percutaneous absorption.
Contact betwmn such vehicles and the skin inevitably entails modi-
fication of the skin's barrier function, either by hydration changes
or structural changes in the molecuIar arrangements in the stratum
corneum. This is a result of absorption of components either from
a vehicle into the skin or vice versa. The thermodynamic equilib-
rium properties of the system, as always, will control the ultimate
state of the system; however, the time scales of the various kinetic
processes can be of more impo&ance POP consideration in practical
applications because controlled or sustained release over a period is
usually a major requirement.
Recently (1-lo), attention has been foeused on the kinetics of
the transfer of components across oil-water interfaces in systems
contdnhg surfactant aggregation structures. The amount of avail-
able data is small, however, when compared with the corresponding
literature relating to equilibrium properties. An understanding of
the mechmisms found for mass transfer across interfaces in such
systems is becoming clearer as a result of such studies. The pur-
pose of this paper is to review the available information in the con-
text of what it tells us about processes, such as solubilization, in-
*Current affiliation: Clarkson University, Potsdam, New York

127
128 Ward

terfacid liquid crystal formation, and diffusion, that control the


overall kinetic behavior of systems containing surfactant aggregates.

I I. EXPERIMENTAL BACKGROUND
A. Passive Drop-on-Fiber
One of the problems of quantitatively examining systems, such as
emulsions, to obtain the kinetic and contained mechanistic informa-
tion has been that of simultaneously defining the volume and inter-
facial area of the system as a function of time. Studies of formula-
tion stability to determine the factors involved in their time evolu-
tion are usually somewhat qualitative and system-specific. An at-
tempt to overcome some of the inherent diffimIties in such measure-
ments has been the development of the passive drop-on-fiber ( 3 , 5 )
technique. This defines the system by fixing in space one droplet
of emdsion size using an inert cylindrical fiber. Provided the dis-
tortions from gravity are negIi@ble, the shape of the axis-symmet.tric
dmp is purely determined by capnlary forces. An analysis of this
type of system ( 3 ) shows that the rate of solubBization into solution
is related in a simple fashion to the relative dimensions of the drop-
let and fiber. Both the volume (V) of the drop and the interfacial
area (A) can be obtained and used to calculate the rate from the
general relationship :

1 dV
Rate = - -
A dt

The way in which the experiment has been performed has mainly
been concerned with systems in which the amount of oil in the dmp
is extremely small compared with the amount required to saturate
the micelles. In this respect, the data obtajned are concerned main-
ly with the initial kinetic p m e s s so that contributions from micelles
containing oil can be neglected.

8. Rotating Disk
Cases in which the solubilizate under investigation is in the form of
a solid that can be fashioned into a shape allows the use of methods
in which the materm is suspended in the solubilizing medium. Dis-
solution of the solubilizate is then folIowed analytically a s a function
of time, with the particular technique being dependent upon the
nature of the system. Radioisotopes, for example, have been used
(1,2) with fatty acid solubilizates. An advantage of this technique
is that it potentiany allows the effects of rheology on the kinetics
to be determined by spinning the solubilizate at different rates.
Surfactant-Based Topical Formulations

118. RESULTS A N D DlSCUSSlON


A. l nsoluble Oil-Aqueous Micelkar Surfactant
Application of the drop-on-fiber technique (3,5,7,8) to the study of
initial solubilization kinetics of highly water-insoluble oils into aque-
ous surfactant solutions has yielded consistent mechanistic descrip-
tions. Pseudo-zero-order kinetics (Fig. 1) were observed (3,5-8)
for single-component oils solubilizing into different surfactants
above the critical micellization concentration (CWC) , whereas the
rates were not observable below the CMC (Fig. 2). This behavior
was considered ( 3 ) in terms of the following possible diffusion mech-
anisms:

Figure 13.1 Solubilization stage of n-nonane droplets solubilizing in-


to surfactant solutions ( 1%w /w) as a function of time [An is the
drop radius/fiber radius compared at zero time and time (t) ] ; solid
triangles, C 12E06; solid square, DDAPS ; open circle, dodecyltri-
methylammonium bromide, and dosed circle, SDS .
Ward

Surfactant Cancentration

Figure 8.2 Solub3ization rate of oils as a function of C 12EOt; con-


cent ration ; open circf es, tetradecane ; sdid circles, n - nonane .

1. Diffusion o f oil into water is =rated to the water solubBity of


the ofl Go,, as

Do,wCo,w
Rate =
6

where D,,, is the diffusion coefficient of the oil and 6 is the


diffusion Iayer thickness (Fig. 3 ) . This mechanism was shown
to be significant (3) anfy for oils with soXubO'jitiea in water
greater than about 10-6 wt%,
2. A process that i s limited by the rate of micellar diffusion to the
oB--water interface with mass transfer of 03 to miceIle,

The rates o f solubBfzatbn predicted on this mechanism arcs &so


too high compared with the expedmentally observed values for wa-
Surfic tan t-Based Topical Farmufations

OIL

Figure 8 . 3 Solubaization mechanism involdng micellar assodation-


dissociation within a diffusion layer of thickness, A ,

ter-insoluble 02s. It was a r p e d that this wag because the dissod-


athn of the mjicelle , being a precursor to adsorption, with consequent
desorption of the rnirsed micelle-cantdning oil, did not occur for
every excursion of rnicelles into the hterfacld reeon ( 3 ) . A factor
describing the probabnit y of the miceltlle dlsssdating mrit'hh the v-l-
cinity of the 0%--water htsrface, leadhg to solubgization, has to be
included (3) @ d n g the relation

A
Rate = - TI [bla) (C
21. 0
- CMG) [ 31

where A is the thickness of the interfacial re@on (see Fig. 31,


which has dimensians of the order of the miceEe diameter; r is the
t i m e hterva2 bet ween micellar dissoeiatlons that lead to adsowtion
and concerted r;olubBization steps; 710 is the molar volume of the
solubSizate, and bla is micellar capacity for the solub3izate (i.e. ,
moles oil per mole surfact~nt).
Ward

Equation 3 has been used to d e s c ~ b ethe b e h a ~ o rof alkanes


solubBizing into aqueous micellar solutions of various surfactant
types ( 3 , s - 8 ) . The term (bla) is found to be proportional to the
equilibrium solubilization capacity of the micelle for the homologous
serYies af n-hilkanes (1). Comparisons among oils of diffeAng archi-
tecture (5) indicated that the constant of proporti~xlalitybetween
the equsibrium and kinetic values af the solubaizftt.ion capacity was
different, possibly reflecting differences in the ail packing in the
.
micelle A further cievelopment upon the packing requirements of
the micelle-solub3izate aggregate has been made (8,11) in the con-
sideration of solubfiizati~nfrom binary oil mixtures, It was shown
( 21) that, within the restriction af na preference in the solubsiza-
tion of the 02 components (i.e., the composition of the solub2izate
mixture rerndning the same as the bulk contacting oil phase), rela-
tionships of the following form should apply.

where b! are the solubilization capacities of the pure oils and the
I
represents the oil mole fractions in the binaly mixtures. The
vslue of the factor P can vafy between zero and (1 + bp/bq), the
former value representing ideal mixing of the solubnizate in the mi-
celle interior, Comparison of data determined from kinetic experi-
ments with values determined at equilibrium show apeement in the
vdue of I" required to fit the data in some eases (&), whereas, in
others, different values are needed in each case. The solubilization
of oil from dodeeane-tetradecane mixtures into micelles of the non-
ionic surfiretant n -dodecylhexaoxyethy1ene glycol ether, f: 12EO6 ,
( 12) shows gmd agreement (Fig. 4a) with a I? value of 1,s f i,e, ,
v e q close to ideal mixing). Xn contrast, solubBization from simBar
mktures into a micellar zwitleAonic surfactant, n -dodecyldirnethylam-
moniurn pmpane sulfonate, shows (Fig. 4b) that, although the equi-
librium data require P = Q, the kinetic date i s fitted only if the
mmimum possible vdue of FZ is used. This discrepancy may be a
result of micelle shape changes in the first stages of the solubniza-
tion process that are not considered in the defivation of E q . 4, A
furf;her manifestation of micellar shape changes that oceurs in the
initial stages of solubilization is the nonlinear temperature depend-
encies observed for the rates in the re@on of the phase-inversion
temperature sf some nonionic surfaelants ( 5,8).
One consequence of the proposed mechanism is that the rate i s
essentidly independent of stirring in the system, unlike the passive
Surfactant-Based Topical Fornula tions

8, Y Composition dependence of solubilization far (a) C ~zEOC;-


F Egu re
C 12H~g-6lf;t.I34 and (b] DDAPS-C 12W26-C X ~ 34
W at 298 IC,

diffuskn process described by Eq. 2, Were, the value of S 18 de-


creased and the rate increased by increased s t i r ~ n gf 1 , 2 ) . A sim-
Bar conclusion has been reached more recently from the dcsscfiption.
of interfmiizl kinetics in experiments involdng contacting phases
w i t h different states of surfactant aggregation (9,10). A linear in-
crease in volume of the aqueous micellar phase with time in contact
with a lamellar phase was found, If the rate of mass transfer from
the miceXXar phase had been diffusion-limited, a square root of time-
dependence of the layer thickness would have been expected. The
restmcturing to groduccz surhctant monomers f r o m the micellar dis-
134 Ward

sociation in passage into the XameUar phase was regarded as being


the rate-limiting step.

Oils that have sufficient water so1ub;llity are more likely to praduce
kinetics that are dependent upon mechanisms in which a diffu~ion
step is rate-determining . Preliminaq studies ( 7,13) of systems
containing benzene, as either a single component or a component of"
a b i n a v mhtur'e, have indicated that this is tme. Benzene, for
example, is soluble to the extent of about 1500 pprn in water, and
any oil that has a water solubility greater than about 1 0 ' ~w t % may
be regarded as soluble in this context. Oil transport across the
liquid-liquid interface is essentidy that of solubility and is gov-
erned 'by the passive diffusion E q , 2 , Rates determined by the
dmp-on-fiber experiment are for systems under conditions of nini-
ma1 stirring and, therefore, c2o not represent rates that are at the
.
diffusion lirnlit. (i e ,, the diffusion layer thickness , 6 , being on the
order of the dimensions of the ail molecule). Some rates typically
found for soluble oils d i s s o j ~ n gin wafer are presented in Table 1,
fnterestingly, the diffusion layer tltzieknesses d e ~ v e din these
experiments are simnar to those found in membrane pmcesses in
v h o , Furthermore, the ratio of the rates is the same as the ratio
of the oU sdubsitieies in water', as required by E q . 2.
Equation 2 shows that the rate is independent of the surfactant
concentration, inasmuch as it does not affect the act-idty of the oil
isl the aqueous phase. A surfactant concentration scan of the rate
of benzene in contact with aqueaus solutions of sodium dodecyl sul-
fate (Fig. 5) s h ~ w sno i-ncrease untit the total sudactant concentra-
tion is above the CMG. This rntty be understoad in terms of the
relative sizes of the oil "sinksw "prodded by the bufk water and. the
micelles; thus, only when the volume of micdles avdable for solu-
baization is comparable with, or greater than, that from the water
salubaty will there be an observable effect from the surfaatant on
the observed rate, The concentration at which this occurs a 1 oh-
dously depend on the size of the water solubility, GMC, and micelle
volume. The larger the micellar volume and lower the CMG in a
system with oil of a @ven water solubnirty, the Xower will be the to-
tal surlFslctat concentration at which micellar solubaization will have
an important contribution. Factors such as the size of the S U F ~ ~ C -
tant hy drophobe , therefore, will be impor^fiant, as will any factor
that influences the CMC of the surfactant, Sim-ifarly, because non-
ionic surfactants tend to have larger sotubBiz&ion capadties for
oils than do ionic ieurfactmts of simnar rnoleeular volume, this ef-
fect should aceur at lower surfaet ant: concentrations in formulations
based on nonionics.
Surfactant-Based Topical Formulations 135

Table 8.1 Typical Ratee for Soluble Oils Dissolving in Water

Rate Diffusion layer thickness


(n .min-l) (wm)

Benzene

Toluene

p-Xylene

Figure 8.5 Solubilization rate of benzene as a function of cetyltri-


methylammo~umbromide concentration (normdized to the CMC) at
288 K.
Ward

C, I nterfaclal Liquid G rystaf Formation


Early work (14,151 on systems with a fatty acid or soap constituent
indicated the importance of interfacial liquid c y s t al formation in
solubilization and associated detergenq processes. This was further
emphasized in studies of soltzbifization kinetics in such systems using
a rotatbg disk apparatus ( 2 , 2 ) , The meehilnisms found in these
situations involved the relative temperature of the system compared
with that required to produce interfacial liquid crystal formation,
diffusion of components in and out of the interfacial layer, and the
degree of stirring. Experiments to determine the pathways of phase
formation in surfactant-og-water systems have shown the impostant
role played by the formation of lamellar phases in interfacial mass
transfer processes (9,10). Furthermore, the rate at which compo-
nents diffuse in and out of this interfacial layer, when considered
in tho context of the phase behador of the system, determines the
overall observed kinetics, An example has been @ven in which the
formation of a layer of interfacial water effectively stopped the
transport of the 09 component in the system ( 9 ) . Here, although
the bulk composition of the total components indicated. that a micellar
phase should be formed, this state was, in. fact, never reached. be-
cause the kinetic pathway led to the formation of an ef"Ec3tive bar-
rier, to transport in the unstirred system; namely, the water layer
Tormation. These considerathns must be %&en into account in sit-
uations in which multieompanent systems are involved and interfaces
hrmed, An important interface to consider, in this context, is that
of the vehicle and skin cxrntaining unequal water wntents on either
side of the interface, The occlusive effect of the vehicle w21 lead
to an increase in the water content of the stratum corneurn, Slmi-
larly, there wnl be diffusion of the water component of the vehicle
into the interfacid re@on, The tendency of transport of water
from one side to the other will depend upon the relative thermody-
namic a c t i ~ t i e sthat, in turn, w2X
i X vary relative to each other, de-
pending on the rates at which their concentrations are being estab-
lished, This will have an influence upon the transport of any hy-
drophaie or polar components acmss the interface. The same kind
of eonsideratians apply to the nonpolar components of both the ve-
hide and the stratum carneurn.

D , Differential Solubltization Effects


An example of the differential extraction of oil components from the
skin has been demonstrated (16). Molecules interacting with the
stratum eorneum lipids, p Amafiy thmuglr hydrophobic forces and
located withh the lipid bBayers, were extracted the most rapidly.
Those molecules that were more amphiphBic and interacted more in-
Surfactant-Based Topical Formulations

Figure 8.6 Solubilization stage of a benzene-cyclohexane mixture


(7:3 w/w) into aqueous SDS (0.032 M) at 298 K.

timately with the bilayer took much longer to be removed. This im-
plies that the nature of the oil component location within the sur-
factant association structure i s also of importance in the solubiliza-
tion and related processes. Demonstration of such preference in
the solubilization process in simple micellar solutions is found in ex-
traction from binary mixtures of benzene and eydohexane into either
water o r surfactant solutions (6). The kinetic profile derived from
a drop-on-fiber experiment (Fig. 6) is no longer linear, implying
that the composition of the oil drsp is also time-dependent. A buIk-
scale expesiment, in which the composition of the oil in the micellar
phase was monitored by gas chromatography (131, showed the com-
position to increase to a ratio of about 9: 1 from an inital I: 1 bulk-
oil composition within the first hour of contact between the bulk oil
and micellar phases. Here, the process is determined by the diffu-
sion rates into water, which are in a ratio of approximately 30: 1 in
favor of benzene. Similar effects have recently been found in the
138 Ward

initial. solubiaizatian kinetics from binamy mktures af n-aTkanes C 131,


for which the mechanism is that discussed in Section I .
Such observations for the formulation of topical agents are im-
portant because they raise the question of the relative tendencies of
molecules to leave (or enter) the vehicle. This is determined by the
nature of the molecules and the molecular properties of the surfac-
tants used to prodde the vehicle. For solute species for which the
rates are determined pdmarily by diffusional pmcesses, this, in a
first approximation, reduces to a consideration of their relative sol-
ubilities. However, the loeation of the sofubSizate within an aggre-
gated structure must &so be considered because it will have a bear-
ing on the rate at which it is released or solub2ized. Molecules
that htercdate between the surfactant molecules, with an orientation
toward the agpegate-solvent interface, that form the aggregated
stmcture, whether simple micellar sr lystropic liquid cryst&, re-
quire a greater energy to overcome the constraints of their e n ~ r o n -
ment, A consequence of this is a slower rate of rdease than that
of nonpolar components held in the essentially liquid hgdroearbon
interior of such aggregated sur&ctant structures. The rate of ac-
crxmulation at, or transport across, the vehicle-stratum csrneum in-
terface of the different components in a typic& mrnufti~omponentsys-
tem wB1 be determined, to a certain extent, differentidly as result
of their locations within the system,

The time evolution of the vehicle----stratumcorneum hterface will un-


doubtedly be such that steady-state penetration of a d m g from the
vehicle will not be instantaneous, It is usual to express these non-
.
equsibrium effects in terms of a lag time (Fig, 1) Conventionally,
this has been developed in terms of the passive diffusion equation
w ~ t t e n as
,

where Qt is the total amount absorbed in time t , D the solute diffu-


sion coefficient (dC,) is the difference in the solute concentration
between the vehicle and tissue, P is the solute paflition coefficient
between vehicle and skin, and 6 is the thickness of the stratum
comeurn-vehicle interfaciczl re@on. The lag time has been shown to
be equzil to 4 2 1 6 ~ . Generdty the vdue of 6 has been taken as the
thickness of the stratum cornam using the measured lag limes to
obtdn vaIues for the diffusisn eoefficients. The considerations pre-
sented here, however, indicate that such a; simple passive descrip-
Lag T irne
TIME

Figure 8.7 Ideazed dmg--penetration- time profile for diffusion


through skin,

tion of the diffusion barriier is limited, because the active interac-


tion of components at the vehicle-stratum corneum Interface will ef-
fectively alter the vdue of 6. Exact description, of the time evolu-
tion of 6 depends on a knowledge of the effects discussed earlier.
This is important in e e w af the large range of lag times that have
been observed {minutes to sever& days) in studies of drug pene-
tration from topical vehicles.

The work outlhed shows some of the insights to be obtained fmm


a cantrolled and well-defined errperlnrental appmach ta the study of
interfacia3 kinetics. The importance of interfacial surfactant asso-
ciation stmclures and their rearrangements has been demonstrahd.
Such considerations are also important in many processes, such as
liquid chrornatogralpl.1y ( I T ) , potential low-enerp cost oil separa-
tions, as we11 as multiva~ousphy siolo@eal processes.
Ward

The importance of the phase b e h a ~ o rin determining kinetic be-


h a e o r f I, 2,9,1Q), or vice versa, has also been demonstrated. It
should be noted that a combinatian aE local adsorption phenomena
and development of interfacial. concentrations dependent an the rela-
tive rates of ingress or diffusion away can lead to an interfacial
phase formation that is unexpected from the total eoneentratbns of
the components in the system. Thus, for example, the observation
of interfaciizl liquid crystal Earmatian at the 02-water interface in
systems in which the total 03 is less than 1%of the amount required
to saturate the surfactant micelles with which the oil is in contact
in initial kinetics expedments ,
Both of these considera~ans(i.e., time scde and phase evolu-
tion) w i l l affect the performance of a topicd vehicle formulation,
Humectant aspects of an applied cream, for example, will rarise the
water concentration at the vehicle-skin interface, thereby altedng
the local concentrations of the system, The effect of this w i l l be
determined by the phase behador of the system in response to in-
creased water and the rates at which the components diffuse h t o
and out of the interfacial region. Qne possible scenario would be
that for which the rate of ingress of components from the vehicle
was much slower than the rate of water Buildup. Here, an essen-
t s l y impenetrable harder to the passage of a hydrophobic &drug
could be established.
In generd terms, for the rate of steady-state penetration of a
drug to be att&ned (see Fig. 71, the effects discussed will be man-
ifested h terms of the lag time ($, The range of lag times en-
cauntered is from minutes to severaf days; thus, they are of clinical
significance, m d an understanding of the factors underlying them
is of major imporlcamce in the d e s i a of topied systems,

REFERENCES

I, .
3 , A , Shaeiwitz, A . F, -C Chan, E. L. Gussler, and D. F,
Evans, J. CatEoicl Interfizee Sci, , 89: 47, 1981.
2, 6 , Nuang, f), X;". Evans, and E l L. Cussler, J . Colloid Inter-
face Sci, , 82: 499, 1981.
.
3. B J , Carmll, J , Colloid Interface Set, , 79: 126, 1981.
4. Y . C. Chiu, V. 6. Wan, a d W. M , Cheng, rln Structure/
Performance Relationships In Surfactants, ( M e J . Rosen , ed .I.
ACS Symposium Series, p. 23, 1984,
5. B , 9, Carmll, B . 6, Q'Rourke, and A , J. I. Ward, J. Pharm,
PhamacoZ, , 34:281, 1982,
6. A . J, I , Ward, M . C , Carr, and J , Cmdden, J. Colloid Inter-
face Sci. , 106: 558, 1985.
Surfactant-Based Topieat Fomulatians 141

A , C . Donegan, and A , J. 1. Ward, J . Pharm, Pharmacol,, 39:


4 5 , 1987.
B , G , 6. QfRourke, A , J, I , Ward, and B. J. Carroll, J.
Pkarm, PharmacoZ. , 39: 865, 1987.
S. E . Friberg, M, Mortensen, and P. Nea@, Separation. Sci,
Technol, , 20: 285, 1985,
P. Mead, M , Kim, and S, E l Friberg, Separation, Set, Tech-
noE,, 20:613, 1985,
B , J, Carroll, J , Chem. Soe, Faraday Trans, I , 82:3205, 1986,
P, Faulkner, M. S G , Thesis, NatX, Universli-ty of Ireland, 1985,
A. J, 1 , Ward, unpublished results,
A. S, C. Lawrence, Disc, Faradory SOC,, 25: 51, 1958.
A. S, C . Lawrence, A . Bingham, 6. B. Capper, and XC, Wume,
J , Phys, Chem., 68: 3470, 1964.
6 . Imokawa, and M , HattoA, J , Invest, Dermalol, , 84: 282,
1985.
D. W , Armstrong, li". S . Ward, and A. flerthod, Anal, Cfiem,,
58: 579, 1986,
The Use oi Phase Behavior and Laboratory Robotics for the
Optimization of Pharmaceutiea Topical Formulations

DAVID MI, OSBQRNE The upjohn Company, Kalamazoo , Michigan

Formulators seldom have the time required to optimize topical phar-


maceutical formulatians over complete ranges of both active and in-
ert ingredients. Traditional cream, lotion, and ointment; formula-
tions are usudly taken from cosmetic fitrmulatories. The d m g is
added at the desired level to the heated ointment base or emulsion
oil phase (wahr phase far hydrophilic dmgs) and the resulting
formulation is observed for physical st ab;ility m d appropdately aas -
.
sayed f'or potency, preservative challenge, etc Since we assume
the vehicle was optimized prior to being included in the farmulatom,
no changes are made unless the formulation shows indications of?
phase separation, a r chemical incompatibD3ty with the dmg. While
this process should work well for the formulator, anyone who has
followed this procedure knows that drugs tend to either destabixize
the vehicle or have inadequate solubgity in the vehicle. Even
worse physical sstabi2ity problems ean also occur, such as slow crys-
tilnization from a super saturated solution, or e~stallizatisnof a
more stable polymorgh sf the drug, These latter occurrences are
particularly troublesome because the resulting changes in product
proper-t-ies may not became mpparent untiI months after manufacture.
Optimization of a pharmaceutical topical formulation can help to min-
imize the likelihood of encountering these problems,
Typicdly, a formula is considered acceptable when cfitfcal prod-
, content uni-
uct characteristics f appearance, potency, ~ s c o s i t y
formity, ele,) have been identifled and shown to be maintained for
the shelf life of the product. The acceptable formulations are then
further scmtinized by consideAng raw mfsterial costs and processing
restraints until a final formulation is selected. It is important that
the criliclll. product characteristics be obtainable using raw materials
whose properties span the supplier" specification range, This is of
particular impor-f;anceif one or more of the ingredients is a naturd
product that may tend to have extensive lot-to-lot vaAab2ity. 16
the fornuXation Is not tested using at least three dififerent lots of
each ingredient, then costly problems may arise during manufacture
of the product. L&ewise, the ranges for both active and i n e d in-
gredient amounts that are tolerable to the formulation should be de-
termined. If more than one dmg concentration is antieiptzt-ed, then
the ranges of inert ingredients should be cheeked at each d r u g con-
centration. It becomes quickly apparent that the optimization. of
even a five-component formulation is a substantial, undertaing. Fi-
ntltly, the criticd and noncritical process vadables must be defined
once an optimal formulation is selected, Although this last step is
usudly associated with scaleup, laboratory scale expe&ments can
help define processing steps such as holding times, holding ternper-
atures, and the effect of heating and cooling cycles.
O b ~ o u s l y ,the formulator is faced with a tremendous challenge
to actually optimize a topical formulation. To meet this ehaXXenge ,
an efficient , we11- planned series of labor atom investigations must
be performed. These experiments can be divided into three steps:

1. A preliminam f'ormullation is selected and component ranges are


determined by conducting phase behavior studies.
2, A statistically valid experiment is designed utilizing these com-
ponent rmges md any othhri process variables considered i m -
.
port ant
3. Responses from the experimental tridils are empirically modeled
and optimized.

The last two steps in this process have been the foeus af much at-
tention for solid-dosage forms ( 1 , 2 ) and, more recently, for dis-
perse systems ( 3 ) . However, these optimization techniques require
that the responses be smooth and continuous throughout the vari-
able ranges, For topicals, smoothness and continuity must be es-
tablished for the composition ranges before optimizatkn. Thorough.
characterization of the formulation by determining the phase behav-
ior of the system is a way of establishing smmthness and continuity
throughout the composition range. Thus, selection of component
ranges by phase b e h a ~ o rdetermination will be the main focus of
this chapter,
Optimization of Topical Farmulatisns

11. DETERMINATION Of= COMPONENT RANGES

The fkst step in the optimization of a topied d m g delivery system


is selecting a preliminary formulation and determining the component
ranges to be optirnlzed by conducting phase behador studies. Be-
cause these studies require preparation of a large number of sam-
ples, laboratory robotic techniques can be used to automate the
.
quantitative dispensing and mixing of the components Although
the data that result from evduating each of the samples can be rep-
resented by vadous techniques, plotting the data to compose a bi-
nary or ternary phase diagram is the method used in this chapter,
Representing the data with b i n a v or ternary diagrams is the most
efficient way of presenting all of the equilibria for systems c o n t ~ n -
h g only pure components. For complex systems typical of indus-
trial applications, binary and ternary diagrams prodde a method of
mapping (characterizing) the '"phases" whose component: composition
range must be known. Thus, the general usefulness of binary and
ternam phase diagrams warrants the following d e t d e d descdption
of how to assimilate the informatian provided by these diaparns.
This tutorid on phase diagrams will be followed by a general, yet
dataaed description of the laboratory robotic techniques that are
useful in prepadng the samples required to complete the phase dia-
grams. Remember, adequate characterization of any disperse sys-
tern is required to e a r a n t e e smoatkrness and continuity throughout
the composition range to be optimized,

A, Phase Behavior
Disperse systems a m unique among pharmaeeutic& dosage forms be-
cmse the physicochemied intctrilctions between the components dorn-
inate the stabBity of the system, Small changes in the component
composition can completely destab2ize or dramsllieally improve the
sti3lbQity of the dispersion, A well-characterized example would be
the stable emulsion re@an for the wat~r-p-xylene-decaaxyethylene
glycol nonyphenol ether system, In this systern the stabnfty of the
emulsion dramiltjielnlly increased when the surfactant concentration
was changed, from 3 to 4 wt % ( $1, This dramatic change in physical
stablility was related to the formation of lamellar liquid crystdline
baayers around the oil droplets. The formation of a third phase
(e. g. , a liquid eqstdline phase) is now a common method of h-
creasing emulsion stabiEty (5). Other examples of small composi-
tional changes affecting product properties would include the effect
of pH on, clay suspensions or the effect of solids e o n t e n o n the
melthg point of a suppository,
To understand dispersion stabgity, it must be realized that
changes in stabaity are the result of chmges in the phase behador
of the system, For surfact ant systems , sufficient scientific inves -
tigations have been completed to be@n to correlate emulsion (4,5),
foam (61, and gel (7) stabifity with the equilibrium that exists be-
tween the thermodynamically s a b l e micellar and liquid crystalline
phases. Although the techniques for representing phase b e h a ~ o r
described in this chapter are not limited to surfactant systems, they
have been pdmafly applied to this area of study, Nontraditional
formulations may genertllly be less dependent upon emulsifiers than
current topical creams; however, binary. and ternary phase mpre-
sentations are useful in characterizing any of a vadety of systems,
Most topical f"armu1ations are composed of matedals that fa11 into one
of three groupings: polar components, nonpolar components, and
..
other components (e g , amphiphilie, polymeric, therapeutic, bio-
technolow products) . For any formulation whose components can
be d i ~ d a dinto two or three such groupings, b i n a q or ternary
plots of the phase behavior can be extremdy useful in ehzzracteriz-
ing the physicoehemical interactions between the components of the
system. These physicschernical interactions between the indiv_idual
components ultim~telylead to the stabnity or 1acR of stzibility char-
acteristic of pharmaceuticll.l. dispersions.

1, Binary Systems
Bixlary diagrams are typiedly plotted on Cartesian coordinate grjids
in which the composition is plotted along the abscissa and tempera-
ture is plotted along the ordinate. Phase boundarYies are plotted on
this grid to separate single-phase re@ons fram multiple-phase re-
@ens. An idedized example of a two-component liquid system hav-
ing m upper criticd solution paint is shown in. Figure 1, The left
verticical axis represents the phase behador of pure A over the ten-
perature range of interest, whereas the right ve&ical axis is far
pure B. Both A, and Ea are liquids for this temperature range at
the given pressure. A s expeeted, the so1ubiXity of B in A in-
creases as the temperature is increased, This change in solubgity
is shown by the left phase boundary. Analogously, the changes Jn
the solubility of A in B with increasing temperature is represented
by the right phase boundary, The two-phase liquid-liquid re@on
f&ls between these boundazllies. The phase diagram p r o ~ d e seon-
siderable information about any sample that is mixed within the con-
ditions represented by that diagram, First, for any binary mixture
of components A and B , it can be discerned whether the sample
will be one or f wo liquid phases. Second, if the sample is two
phases, it can be discerned what the composition of each phase w21
be and how much each phase will weigh. For example, consider the
point circled in F i ~ r e1. Overall, ibis sample contains 40- and
60% B. We ertn immediately see that this sample will separate ento
Optimization of Topical Formulations 147

two liquid phases I

CONIPOSlTlON
Figure 9.7 Idealized binary diagram for a two-component Iiquid sys-
tem having an upper critical point.

two phases at 100°C, with one of the liquid layers having the eom-
position 85%A and 15%B , and the other liquid layer having the
composition 20%A and 80%B . The constant-temperature line that
conneds these three colinear composition points is called a tie line.
For binary systems, all of the tie lines are understood to be parallel
( i .e. , constant temperature) and need not be drawn. Thus, for
m y composition at 100% within the two-phase region, the eomposi-
tion of one liquid layer will be 85%A and 15%B , whereas the other
Iiquid layer wiIl have the composition 20% A and 80% B. The only
diffemnce will be the relative amount of each of these phases as
one moves along the tie line.
To calculate the amount of each layer, remember that the weights
of liquid layer 1 (ml) and liquid layer 2 (m2) must add up to the
total weight of the sample, thus
The sample Ellso has a gross weight percentage of component A that
is designated by y in Pimre 1. However, as stated before this
sample breaks into two phases that have weight percentages of corn-
ponent A equal to y~ and y ~ . Thus, the conservation of' camponent
A &lows us to wAte

Substitution of the expression for mtot from E q . 1 into E q , 2 @ves

which rearranges to

As seen, the weights of the two phases are in the proportian of the
len@lrs of the tie line segments extending from each side of the
sample point. Another way to write this relation is

where a and 'b are the lengths, as shown in F i p r e I [i.e,, (y -


y1) and ( y 2 - y)], respectively, To continue with the example for
the point circled in Fimre 1, if the total weight of the sample was
5 g , then liquid layer 1 f 856 A, 15% B) would weigh 1.54 g , where-
as liquid layer 2 (20% PI, 80% B ) would weigh 3.46 g.
For the diagram shown in Figure 1, as the temperature is
raised, the tie lines became shorter because the solubility of each
component in the other is increasing. UlMrnately, the tie line be-
comes a point as it crosses the upper portion of the phase bound-
ary. This point could also be considered the point at which the A -
in-P3 solubility boundary joins the B-in-A solubs-ity boundav. A
sample kavling this composition will consist of two roughly equal-
volume phases below the upper consult boundary. When heated suf-
dciently to reach the wc&ticalltpoint, the interface between the two
liquid phases will disappear, and a single-phase liquid will result.
Upper erjitical points, lower critic& points, or both upper and lower
erjitical points can occur for binary liquid systems, In systems in
Optimization of Topical Formulations 149

Figure 9.2 Freezing-point diagram for the binary system benzene-


naphthalene at 1 atm pressure (After Ref. 9, p. 324).

which both upper and lower critical points exist, the phase bound-
ary (also called the coexistence curve) forms a closed loop. Thus,
as the temperature of a miscible mixture is lowered, the mixture
first separates into two phases and, then, reforms as a miscible
mixture. A very readable article describing these reappearing
phases recently appeared in Scientific Amel-ican (8).
If the temperature range of interest for the two-component sys-
tem is such that one or more solid phases now exist, the binary
diagram has a considerably different appearance. A s seen in Figure
2, two components that are miscible when liquids, and in which only
pure crystalline forms of the two components form as solids, pro-
duce distinctive phase behavior when combined. Again, it is un-
derstood that the pressure for the system is held constant. Above
the curved lines AE and BE, the components are mutua11y soluble,
forming a single liquid phase. Below the horizontal straight line,
the temperature is sufficiently low that no liquid phase exits. Be-
tween the curved lines AE and Be ~s the two-phase re@on in which
the pure components are in equnibrium with the single-phase liquid.
Just as b e k r e , any point in either of these two-phase re@ons will
have a horizontal tie line passing through it. The location at which
this tie line intersects the phase boundary will give the composition
of the liquid phase, whereas the relative amounts of the liquid phase
and solid pure component can be calculated by Eqs, 4 through 6.
A s outlined by Barrow (91, it is instructive to consider what
happens when solutions of two particular concentrations are cooled,
Consider Srst the composition labeled b (60% naphtkdene in benzene)
at 70°C1 As the sample is cooled, it wig remain a liquid ant3 cmss-
ing the phase boundaw labeled as curve BE, Upon cmssing this
boundav the first solid naphthalene crystals will become detectable
in the sample. Note that for this diagram the length of the tie line
segment to the &ght of this composition will not change, whereas
the length. of the left tie line segment will increase with decreasing
.
temperllbure Thus, the amount of solid naphthdene wilt continue
to increase in the sample until the hodzontal line at approximately
-4OC is reached. As stated p r e ~ o u s l y ,below this temperature no
liquid exists; thus, the sample that contained mughly equal amounts
of primaw naphthalene crystds and liquid (2Q% naphthdene and 80%
benzene) at -3OC becomes completely solid at - - P C , consisting of
40% by weight prjlmav naphthalene crystals and 60% by weight of a
fine crystat-gr&ned mixture of 20% naphthalene and 80%benzene
Ci.e, , the errteetic mixture). This behador can be compared with
the phase changes encountered by the composition labeled e (20%
naphthdene) at 5QQC. As the sample is cooled it remains a single
liquid phase until reaching the --.4aG line, st which time the solution
becomes in egusibrium with pure solid benzene and pure solid naph-
thdene. The sample remains at this temperature untiil completely
converted into a bne-grained crystal mixture of the two solids, hav-
ing the composition 208 naphthalene and 80%benzene, The point
labeled E is cdled the eulectie point. Mote the eutectic mixture has
a melting point lower than either of the pure, components.
Although the benzene-naphthalene system is useful in under-
standing eutectics m d identifying eutectics on a binary diagram,
most systems of interest Gontain components that interact with one
another in the solid state, Such interactions result in solids or
semisolids that are mixtures of the two components. Interactions of
this type can also result in the formation of solid-state compounds
consisting of simple mole ratios af the two components. The effect
that formation of campounds and mutual. component solubility has on
the appearance of the binary phase diagram is shown in F i w r e 3 .
The mutual. solid-state solubnity of A with B can. be seen along the
vertical axis of the plot, When B be@ns to solidify at temperature
Optimization af Topical Formulations

csolld E3; with


dir~sokedA

Figure 9 . 3 Ideaazed freezing point diagram showhg multiple com-


pound formation in the solid state,

y , a certain amaunt of A is accommodated within the salid to form a


solid-state mixture. Analopusly, as A is cmled below temperature
.
x , the resulting solid is a mixture of A. and B The vertical lines
at 8 0 ~ 2 0AIB and 5Q:5Q A I B correspond to the formation of com-
pounds, Although this phase behavior may seem eaml?;lex at first,
the formation of two compounds merely divlides the diaeam into
three simple eutectic diagrams, Eaeh of these euteetfes can be evd-
uated in the same manner as F i e r e 2 was.
The phase behavi;sr eharacte~stieof miscible solids is consider-
ably diffemnt from the eutectic diagrams that result when the solid
phases are only partially soluble in each other. Ffmre 4 shows
idealized phase behavior for two miscible solids. The uppermost of
the two curves is the f ~ e e d n gpaint curve and represents the tem-
perature at which the vadous compositions bedn to freeze, The
lower curve e v e s the eompositian of the solid that separates out at
.
that freezing paint ( 9) Obdously , the solid will alw~lyseant%in
more of the higher-meltkg component than the solution from which
it separates. The practical implication of this phase behador far
formuXation of semisolids is that adequate mking must be mdntained
as the temperature is lowered through the re@on in which solid is
in equilib~umwith liquid,
The binary diagrams descfibed thus far have been limited to two
eomponclnts that are chemically pure, and that hawe sharp melting
points , For most ""red-lifev "rmdation efforts, especidly for semi-
Osborne

Figure 9.4 Idealized binary diagram for a two-component system


in which the components are miscible in both the liquid and solid
phases but have different melting points.

solids, the components will be mixtures characterized by melting


ranges, rather than exact melting points. Although the phase equi-
libria of these practical systems are less elegant than the p r e ~ o u d y
described academic systems, the phase diagram representation is
still very useful. The primary difference between the systems is
that both the single- and multiple-phase regions must be labeled
descriptively. Phase equilibrium for complex multicomponent sys-
tems often becomes a mapping of characteristics on a composition/
temperature grid, For example, the phase equ2ibria for the white
petmlatum-anhydrous lanolin mixture is shown in Figure 5. Where-
as the mutual semisolid solubility regions are definitive, the pres-
ence of a eutectic is not certain. The phase behavior at high pet-
rolatum concentrations seems to have the appearance of a eutectic;
Optimization of Topical Formulations 155

I I I r I I r I I I
100% 60160 10076
Whlte Anhydrous
Petrolatum Lanolin
Figure 9.5 Freezing point diagram for the semisolid system white
petrolatum-anhydrous lanolin, Component impurity and wide melting
ranges result in an appearance si@f"rsantly different than for ide-
alized diapams. The phase regions were observed to have the fol-
lowing appearance: (a) single-phase , clear, colorIess liquid ; (b) two
clear Iiquid phases, top colorless, bottom yellow ; (c) single-phase
clear yellow liquid ; (d) white petrolatum with dissolved lanolin ;
( e l cloudy semisolid top phase, colorless clear liquid bottom phase;
(f) homogeneous semisolid ; ( g) two semisolid phases ; top, white,
bottom, yellow ; and (h) two semisolid phases,

however, the single phase becomes a semisolid at temperatures above


the temperature at which the lanolin is completely liquefied. This
raises an important consideration when working with semisolids whose
components are mixtures. SolidiiFication is a very nebulous term.
It could mean uniform clouding with an increase in viscosity, or
solidification could mean hardening into an amorphous solid. For
these systems, it is usually sufficient to characterize the tempera-
ture effects upon phase equilibria and utilize this information as
necessary. For example, the diagram in Figure 5 indicates that any
combination fmm 20% through 45%lanolin in petrolatum will result in
a homo@neous cloudy, slightly yellow semisolid upon cooling below
50QC. To obtain this semisolid, no mixing is required, the liquefied,
miscible system above 60OC will cool to form a homogeneous semi-
solid. If greater than 50% lanolin is used, then the system will sep-
arate upon cooling. Although vigorous mixing of the system until
reaching temperatures below 40PC may result in an acceptably ho-
mogeneous system, this composition range should be avoided. Mix-
tures containing less than 20% lanolin should be avoided for the
same reasons, unless 5% or less lanolin is to be mlubilized into the
white petrolatum.
154 Osborne

t I I 1 I I 1 I
100% 60150 10096
Whlte Anhydrous
Petrofatum Lanolin

Figure 9.6 Freezing point diagram for the semisolid system white
petrolatum-anhydrous lanolin. Comparison with Figure 4 emphasizes
the phase behavior changes that can result from lot-to-lot variabif-
ity of a natural product. The phase regions were observed to have
the following appearance: (a) two clear liquid phases ; top, color-
less, bottom, yellow: (b) multiple phases both liquid and semisolid.

When considering the phase behavior of natural products, such


as lanolin, the lot-ta-lot chemical variabflity of the material must be
wnside~ed. Phase behavior can often be very sensitive to changes
in the chemical composition of the component, especially if the im-
purities are surface active. To illustrate the dramatic changes that
can occur, compare the phase behavior of Figure 5 with the behav-
ior shown in Fig-ure 6. Both diagrams are for the same white pet-
rolatum-anhydrous lanolin systems, using the same lot of white pet-
rolatum. The only difference in the systems is the lot from which
the lanolin was used. Although both lots of lanolin were from the
same supplier, meeting all of the specifications set for the material,
the phase behavior is in no manner similar. The region in Figure
5 that provided the homogeneous semisolid is completely absent in
Figure 6. To further emphasize the need for consistent raw mate-
rial quality when formulating components that are mixtures of com-
ponents having wide melting ranges, Figure 7 has been included.
This diagram shows the phase behavior for a third different lot of
lanolin from the same supplier, and white petrolatum from the same
lot as for Figures 5 and 6. For this system, the phase behavior is
chwacteristic of two components miscible in both the liquid and sol-
id states, but having different melting temperatures. Needless to
say, this third lot of lanolin provided a superior ointment after rnan-
ufacturing .
Optimization o f Topical Formulations 155

si@e phase semisolid

%
White Anhydrws
Petrolatum Lanotin

Figure 9.7 Freezing point diagram for the semisolid system white
petrolatum-anhydrous lanolin. For this particular lot of lanolin,
the phase behavior is similar to the idealized phase behavior for two
miscible solids as shown in Figure 4.

2. Ternary, Pseudoternary , and Quaternary Systems


The use of ternary plots is a convenient representation of phase
behavior when three or more components are involved. Because
ternary plots are frequently encountered in the pharmaceutical. lit-
eratum (10), a description of how to fully utilize this type of dia-
gram will follow.
The two-dimensional representation of a three-component system
is possr%le only if the temperature and pressure &re fixed. The
phase behavior, as a function of composition, can then be repre-
sented on a triangular plot. The triangular plot is merely a eompo-
sition grid in which any point of the triangular plot represents the
relative amount of each of the three components. When the amount
of each of these components is added up for any single point on the
plot, the total composition will always be 100%. The geometric re-
sult is that the sum of the three perpendicular distances from any
point to the three sides of the triangle is equal to the height of the
trian- (i.e,, 100%).
First consider the three corners of the ternary plot. These
corners represent the pure component, that is, 100%of either A , B ,
or C. The lines connecting the 100%corners constitute the three
Figure 9.8 The 1: I eonstant weight ratio line af I3 / C , Circle is
the composition before addition of A ; X represents addition of 10%
A , and the square is positioned at 20% A , 40% B , and 40% C ,

different b-inary systems possible, AB, B e , and CA at a @ven tern-


perature and pressure. Therefore, although the corners define the
components used in the study, the perimeter of the t d a n p l a r plot
shows the mutual solub2ities of the binary systems. This is impor-
tant when constructing a diagram, because any solubiUty extrapolcat-
eeX to the diagram perimeter must correspond to the solubsity limits
of the binary system,
Every point contained within the intedos of the triangle corre-
sponds to a mixture of three components, Consider a 50: 50 mix of
component I3 and component C: denoted by the circle on F i p r e 8.
If enough of component A is added so that it is 10%by weight of
the now three-component mixture ABC, then the mixture contains
I O U A 45%
, B , and 45% C , and the position an the diagram is marked
by x in Figure 8. If more of component A is added to the sample,
bringing the composition of A to 208 (wtlwt), then the sample would
contdn 40% B and 4 0 W , and be denoted by the square in Figure
8. Note that, although the percentage of B and C change with the
Optimization of Topical Formulations 157

addition of A , the ratio of these two materials remains the same,


This, of course, must be true as long as neither component B nor
eomponent (2 is allowed to enter or leave the sample. Fmm these
few examples, it is quickly realized that the three colinear points on
Figure 8 fall on a line that passes dirwtly through the 1008 A cor-
ner, and that this line is actually the 50:50 of B / C constant ratio
line.
Now that it is appwent how the addition of a third component
to a binary system moves a point from the perimeter of the diagram
to the interior of the diagram, it is time to desedbe the most relia-
ble way to read a composition from a ternary plot. Figure 9 has
five compositions denoted by circles of different shading patterns.
Consider the open circle. What percentage of A , perrentage of 3 ,
and percentage of C does this sample contain? First, find the per-
centage of A by moving to the side of the triangle opposite corner
A (side B C ) . Move perpendicular to the plat lines, parallel to side
B C , until the point in question is reached and record the percent-
age of A , For the sample plotted as the open circle, 20% of the
sample is component A . Second, move to the side opposite of cor-
ner B (side AC) and follow the same procedure to match the point
with the plot percentage line of the diagram. The open-circle sam-
.
ple contains 30%B Finally, and very important, repeat the proce-
dure for component C as a check, rather than cdculating the per-
centage of C, The summation of the perrentages of the components
must always equal 100%. Xf they do not, then an error In reading
the diagram has occurred and the researcher should start from the
beginning to read the composition from the plot. Remember that
the point will be nearest the corner of the eomponent with the larg-
est percentage. One of the more common mistakes in reading ter-
nary plots is to interchange two of the components. This ermr
can be minimized by always beginning at the side opposite the cor-
ner of the eomponent whose percentage is being determined. The
compositions of the samples denoted by the other shaded circles are
listed with Figure 9 for use as examples.
To gain greater dexterity in reading ternary diagrams, consider
the following scenario. After -Wining and mixing numerous two-
and three-component samples, it is found that upon standing, some
of the samples maintain a homogeneous appearance, whereas other
samples separate with either two or three distinct layers. In these
two- or three-phase samples, each homogenmus layer is separated
by a distinct, unchaneng interface, indicating an equilibrium be-
tween the phases. The samples can be grouped into three different
categories based on their unchanging homogeneous appearance at
equilibrium. Likewise, it is noticed that the samples in each of
these categories fall within a distinct re@on on the ternary plot.
Thus, boundaries can be drawn on the ternary plot separating the
Figure 9.11 Each point of the tfianglzz2ar grid denotes a unique m i x -
ture of components A , B , and C . Circles denote 20% A , 30%f3 ,
50% C ( 0 ) ; 15%A , 55"2,, 30% C f e); 5% A , 90% B , 5% C ( 0 ) ; 45%
A , 258 B , 30% C f a ) ; and 45% A , 45% B , 10% C! ( 0 1 .

composithn forming the single-phase compositions fmm the eomposi-


tions formlirrg two- or three-phase re@ons. Tie lines can &so be
d r a m that will indicate which phases are in equsibdtrm when the
composition of the samples falls outside of the homogeneous, single-
phase re@on, The phase boundaries for this hypothetical system
am shown in F i p r e 10 and the unique eategades or diagram re-
@ens are f &beled a , @ , and y.
From the foregoing discussion, it. iis eddent that we can answer
a variety of questions using the phase diagram in P i p r e 10, For
example :

1, Row much C can be added to A, whne maint&ning the physic&


gropeTties characteristic aE phase a3
2, f f 40%A , 20%EZ , and 40% G are mixed, will the sample exhibit
only those properties charactedslic of the B phase?
3. What is the maximurn amount sf B that can be stecornmoclated
within the f! phase?
Optimization of Topical Formulations

Figure 9 , f 0 Ternary. representation of the phase behador resulting


.
from mixing components A , E3, m d @ The bold curved lines are
the phase boundaries. All compositions within the phase boundary
labeled a will form a single phase system (it will not separate afSter
mixing) that is uniquely different fram either compositions mixed
withi;n the @-phaseboundam or the y-phase boundam, LiErewise,
single-phase B sampfes are different from single-phase y samples.
Cornpositions that are mixed that fall outside of the phase bounda-
ries will be either two-phase systems (separate in two layers) or
three-phase sys"tms (separate into three layers), The numbered
points or re@ons on this diagram refer to the questions asked in
the text,

4. To reach the maximum amount of 3 in phase B , to what AIC


ratio should B be added?
5. Xs it possible to solubilize 25"k;of I3 into phase a?
6. Can B and C be mked to form a single phase?
7. Xs A more soluble in 3 , or is B more soluble in A ?
8, If 1. g of A is mixed with 9 g of a 20%A , 60% 61, 20% C nth-
ture, will the sample still be single-phase y ?
9, Each of the phases a, 6, and y e m be encountered at some
point by adding one of the eomponenk to a mixture of the oth-
e r two. What is the ratio of the two components, and which
single component must be added?
10, What B f C ratio will neither pass through phase a nor phase @ ,
nor phase y , even up to the addition of 952% component A ?

The answers are

I, Approximately 32% of G can be added to A before crossing the


phase boundam.
2. Ha , the sample will also exhibll-l characteAstics of component C .
3. Slightly more than 40% B can be accommodated within the f3
phase.
4, The 42: 58 ratio of A f G will accommodate the maimum amount of
B possible whne remdnhg within the @ phase.
5, No, the 25% f3 line does not pass through phase a.
6, No, the B e side of the trianwlar plot shows no mutual solubil-
ity areas.
1. Less than 1%of B is soluble in A , whereas slightly over 2%of
A is soluble in E3.
8, Yes, the final eompasition 28% A , 54% B , 18%C falls within the
r phase boundary.
9. Addaion of 13 to a 67:33 through 59:41 r a t h of AIC will Grst
pass through the a phase, then the phase, and, firrally, the
y phase.
10, The 13/C ratio of 60: 40 will not pass through ac single phase
re@on up to 99+8 of A .

Understanding how to read the compositions from the sin@@-


phase re@ons of the diagram is obdously an important tool, Be-
cause a complete ternary diapam represents the complete charac-
terization of a three-pure-component system, a l l questions of mutu-
aX solubgity- are mstvered merely by determining whether the corn-
position is inside or outside of the phase boundam, However, even
more inEormix.tion can be retdeved from a completed diawam, Not
only can questions of mutud solubaity be answered, but also gues-
tions of egu3ibdum B y effectively utiEzing the experimentdly de-
termined tie lines, multiple-phase equaibrium problems can &so be
solved.
Just as with b i n a v diapams, a tie line drawn between two
phases in equiEbdum can be used to determine the comp~sitionof
each phase and the amount of each phase, In this manner, they
tie the phases together. When two lines intersect at the same loca-
tion an the single-phase boundaw, a three-phase tdangle will be
formed, mapping out the composition. redon of the diagram at which
three phases are in equgibriurn with one another. Consider the
diagram in Fimre 11, I t is seen that the tie lines denoting equi-
librium between cx and 6 move dong the a-phase boundary until in-
tersecting with the tie line denoting equilibrium between a and y.
Optimization o f Topical Formulations

Figure 9.11 Example of the utility of tie lines for determining the
amount and composition of each phase in a two-phase system. The
composition marked X will split into two phases upon reaching equi-
librium. The composition of each phase wiIl be the same as the
composition at the phase boundary-tie line intersection (circled
points), whereas the amount of each phase is proportional to the
geometric distance of the phase boundary-tie Iine intersection
points (circles) and the sample point (X) ,

Analogous events occur dong the 6- and y-phase boundaries, re-


sufting in the tie lines forming the triangular tie line boundary of
the three-phase region.
The tie lines can be used to read both the composition of each
phase and the amount of each phase. For samples that split into
two phases, the composition of the separated phases will be the same
as the compositions at which the tie line intersects the phase bound-
aries. For example, in Figure 1I , the sample containing 65% A , 22%
I3 , and 13%C is mixed. Upon standing, the sample splits into an
ct phase on bottom and a d phase on top, with a distinct interface
between the two phases, The totd composition of the sample falls
on the tie tine shown in Mgzlre 11. The tie line that passes through
the sample camposition intersects the a phase at 79% A , 10%B , 11%
.
C , and htersects the B phaee at 60% A , 26% B , 14% C Thus, a
162 Osborne

sample containing 65%A, 22% B , and 13% C wiIl split into a lower
phase with the composition 79% A, 10% B, and 11% C and an upper
phase containing 65% A, 22% B , and 13% C . The amount of each
phase is propo&ional to the geometric distance of each phase bound-
ary from the sample point. By measuring the distances denoted on
the tie line as a and b and knowing the total amount (mass) of the
sample (mtot), the equation for the amount (mass) of the B phase
(i, e. , top phase) would be

whereas the amount of a can be obtained by either having cdculated


the amount of phase 8 and subtracting it from the total amount of
sample, or by use of the equation

m a = m tot [ b / ( a + b ) J

As seen, Eqs. 7 and 8 are identical with Eqs. 6 and 7 and can be
derived andogously once it is realized that the sum of the three
perpendicular distances from any p i n t to the three sides of the tri-
angle i s equal to the height of the triangle.
For the samples that are contained within the three-phase tri-
angle, the composition of each of the phases is given by the point
of intersection between the phase boundaries and the three-phase
triangle. Figure 12 shows an expanded three-phase triangle of the
system shown in Figure 11, Any sample that is Iocated within this
triangle of tie lines will separate into three phases with the a-phase
component having the composition 71% A, 12% B , and 17% C, where-
as the 8 and y phases will have the compositions of 63% A, 19% B,
18% C, and 61% A, 7% B , and 32% 6 , respectively. Thus, once
within the trisulgular three-phase region, only the amwnts of each
phase will change, whereas the compositions of the phases will re-
main the same. The amounts are determined in a fashion analogous
to the method used for two-phase samples. Equations for the
amounts of a phase, B phase, and y phase for the sample denoted
by the open circle in Figure 12 am

m a = m tot [a/(a + b)l

m = m Ec/(c + dl1
$ tot
m y = mtot [e/(e + f)5

Thus, a 10 g sample containing 66% A , 12% B , and 22% C will sep-


arate into three phases: 4.45 g of phase a with a composition of
Optimization of Topical Formulations

Figure 91.12 Det&l of the three-phase re@on shown in Pimre 4.


Composition of each phase is @ven by the phase boundav-tie line
intersection, and the amount of each phase is pmporl;iond to the
distances shown, See text for explanation.

7'2% A , 12%B , and 17% C ; 2.23 g of phase 6 with a composition of


63%A , 19%B , and 18%C ; and 3.32 g of phase y with a compssition
of 61% A , 7% B , and 32% C ,
Ternary diagrams e m &so be used to identilfy e d t i e d solution
behavior. Critical states of mixtures are considered to be the eon-
dition. at which the prope&ies of coexisting liquid phases become in-
distinwishable, that is, the point at which the interface between
the phases disappears, In simplest terms consider the phase be-
havior in F i p r e 13. The fnitid camgosition. of 50% A and 50%B
will result in a sample with roughly equivdent volumes (depending
on density) of the immiscible liquids A and B . As the third liquid
component, 6 , is added, the location of the tie lines show that C:
w%tll be equally distdbuted between the immiscible liquids A and 13.
The volumes of each phase wBX, increase with the addition of C until
the point at which the single-phase boundary is crossed, At this
point, the critical point, the interface disappears between the two
phases of roughly equivderzt vafume .
With this understanding of how to read trimgular plots, a few
examples of how these diagrams have been utiIlzed for pharm&ceu.li-
e d formulations is warranted, The biolo@calIy compatible system,
Osborne

Figure 9.13 Idealized phase behavior representing how addition of


component C to an equal mixture of A to B (point x) will cause an
equal increase in the volumes of phase A and phase B until a criti-
cal point is reached (*), and the interface between the two approx-
imately equal-volume phases disappears, resulting in a single-phase
sample.

water-sodium cholate-lecithin is shown in Figure 14 (11). Although


the structure and characteristics of surfactant liquid crystalline
phases are outside the scope of this chapter, it is sufficient to
state that Iiquid crystals are thermodynamically stable regions, sin-
gle-phase, and reversible in their formation relative to temperature.
Various texts describe the surfactant systems in detail (7,IQ). For
the water-sodium cholate-leeithin system, the mixed micellar region
has been used to sollubilize vitamin E to form an injectable sterile
.
solution ( 12) The liquid crystalline regions, espc?cially the cubic
phase, are stiff phases that are characterized by stow diffusion of
.
incorporated materids Similar cubic liquid crystals have been pat -
ented for use as sustained-release vehicles (13).
When c o n s i d e ~ gthree-component systems not consisting of pure
materids, tie lines are no longer reliable, m d considerably less in-
SODIUM CWOLATE

Figure 9 , I Y Phase diagram f i r the ternary system egg yolk lecithin-


sodium eholate-water at 22OC. (Modified from Ref, E l , with permis-
sion fmm Plenum Press).

formation can be obttzined from the multiple-phase re@ons of the


diagram. Trimmlar plots of systems ha&ng more than three pure
components are eaEXed pseudoternary diagrams. The pmpylene gly-
col-petml&tttum-emulsifier system is an example of pseudoternam-
phase behavior being used to salve a manrrfaeturing problem, A s
shown in F i v r e 15, the base Izaflng the composition 7% propylene
glycol, 88%white petrolatum, 5% glyceml monosterate separates into
a petrolatum-deh phase and a propyXene glycol-dch phase at 70°6.
The physic& stabaity of this base Is completely dependenupon ki-
netic stabBization caused by the increase in dseosity of the vehicle
upon solidification of the white petrolatum -glycerd monasterate
(melting point, 60aC) mixture. Although the semisolid state appears
White Petrolatum

Figure 9.15 Pseudoternav diagram f'or the pmpylene glycol-white


petrolatum-surfactant systems at 7QQC. The mutually miscible solu-
b a t y ama is denoted by L for the surfactants - glycerol mono-
stearate, - - - - - sodium stearoyf Xaetylate, and
---.,.--sarbltan monolaurate.

to be able to accommodate prspylene glycol at ambient conditions for


an indefinite peAod, the vehjde will spontaneously separate inis two
phases if not egorously mixed when heated. Any sep~rationof a
propylene glycol-rich phase will c o n t ~ nhigh concentrations of the
d m g , resulting In content unifclmity problems for the product.
The replacement of glyeeml monostearate with a less lipophilie sur-
factant, sodium stearoyl lactylate (SSL) , @ves phase behador more
suitable for the stabitization of a propylene glycol-white petrolatum
ointment base, Although incorporation of 7%propylene glycol is not
possible at higher than 65% white petrsfaturn, SSE is miscible with
both propylene glycol and white petrolatum at 70QC, 'This implies
that the surfactant is appropdately balanced between the polar
propylene gl_yeol and the nonpolar petrolatum. Although the desired
compositian still falls out side the single-phase boundary , composi-
tions above 80% white petmlatum form emulsions that require a few
hours to separate at 60°C, Samples .in. this two-phase re@on have
an addition& stabsiang effect because of the ability of SSL to gel
propylene glycol. Thus, for these two-phase systems, both the
white petrolatum and the pmpylene glycol solidilFIy upon emling. Use
of the nonionic emulsifier sorbitan monolaurate results in decreasing
the single-phase re@on of the system,
Just as with the binary systems discussed earlier, cornpXex sgs-
tems can be represented using pseudoternary plots. Each of the
three corners should be single phase to mdntain ease in sample
preparation, and descriptive terms must be used to desedbe each re-
@on within the diagram. Because each of the corners does not rep-
resent pure components, tie lines are no longer strdght m d eannot
be assumed, An example of a pseudoternary diagram for an 1%-esm-
ponent system is @ven in Fimre 16 ( 1 4 ) . Each of the three cor-
ners of this diapam were single-phase liquids. The left corner was
distilled w&er in this example, but could have been water--s&t or
water-preservative combinations. The top comer consists of oleic
acid ( 16.5%) , mydstic mid ( 2.9%), friolein ( 4 1.8%) , pdmitic oleate
(20.3%), cholesteryl oleate ( 3 . 0 $ ) , p&stme C2,8%), squalene ( 12,2%),
and lecithin (1,5%). The right earner is a mixture of oleic acid
neutralized with excess tdethanolamine (TEAIQL , 1: 2 weight ratio),
Although the phase behaAor is complex, a large emulsion re@on
(labeled d) that cannot be broken by centrjifugation (3600 rpm , 3 3 O
Gxed-angle rotor, 24 kr) i s present from 8%to 84% water. Likewise,
if sufficient water is added to re@on k to bre& the emulsion, a
clear colorlerss liquid layer w i l l separate f m m the emulsion. If water
were a2lowed to evaporate from region d , then, an anisotropic liquid
cvstalline phase will be present in the sample, that is, ei.ther re-
@on e or f wjll be crossed provided TEA-OLISSL ratio is greater
than unity.
From the foregoing discussions, the utitity of representing phase
equgibrium by use of either b i n a q or t e r n a v representations 18
clear, Because the information contained an these two types of
plots is often interrelated, the transfer of information from binary
to t e r n a n pltats should be briefiy descAbed before considexling
quaternam phase diagrams. Xn Fiwre 27, the blnary diagram fir
the surfactant system water-sodium oleate is shown. Because two-
dimensional representation of three component s is only possibl e at
fixed temperature, the tsmperaturelconeentration plot provides
phase data for the base of the tdanmlar gfid for any t e m a v plot
within the determined temperature range. A s seen in F i p r e 17,
the phase b e h a ~ o sshown for the base of the water-oil--sodium ole-
LIPID MIXTURE

Ab isotropic solution

solution

( 1; f wt, ratio)
tiquid cr ystal
Eigura 9.16 Dia-am showing the complex phase behador for the
multiple-component water-lipid mixture-TEA :QL system. The mul-
tiple-phase re@ona were observed to have the following appearance
after centrifugation: f a ) cloudy white and elear gray isotropic lay-
ers; (b) cloudy white, clear gray, and anisotropic layers ; (e) cloudy
white and clear cdorless isotropic layers ; ( d ) cloudy white-stable
to centdfrx@ng; ( e ) cloudy white, elear colorless isotropic and
anisotmpic layers ; (f) lamellar and hexagonal Xiquid c q s t a l mixture ;
Ig) isotropic solution and lamellar liquid c q s t a l mixture.

ate ternary system at 50°C must correspond to the phase encoun-


tered by the 50°C line as it extends across the binary composition/
temperature plot. Laewise, the 100°C line must correspond with
the phase b e h a ~ o rfor the higher temperature ternary plots, be-
cause these are identical systems under identical conditions (ambient
pressure). If the phases do not correspond, then differences in
component purity or investigator interpretation may have occurred.
ORen information plotted in one represent ation is difficult to
interpret, but becomes more apparent when plotted in the comple-
Op timizathn af TopieaZ Formulations

Figure 9.17 Relationship between a tenperature/concentration dia-


gram and a ternary diapam. The water-surfactat base of the
ternary diagram can be taken from the temperaturelcamzcentrat~on
diagram for a given temperature.

mentary form, For example a special, type of blInary temperature/


concentration plot is the cloud point or cloud temperature diagram..
These diapams are plotted for water-nonionic surfactant mixtures
containing less than 261% surfactant agdnst a temperature range not
.
exceeding 1OOQC A s a fuid , isotmpic nonionic surfact ant solution
is heated, the water hydrogen bonded to the ether oxygens de-
creases. Because this hydration of the ether oxygen@i s the pdma-
ry interaction keeping the nonionfc surfactant in solution, the mieel-
Tar wefght of the palyoxyethylene-type? nonionic surfactant solution
.
increases with increasing temperature (2. e , decreasing hydration) ,
This increase in size ultimately causes; the solution to become turbid
(cloud point temperature) , f~lXornretSb y separation into two phases.
For an ideafized system in which cloud point temperature is plotted
Osborne

Cloud Temperature
Of agram

Ternary Phase Behavf or


at Temperature a.

5% 10% 15%
SURFAETANT

clear

WATER 5% 10% 15% 20% 30%


el oudy S@RFACTANT CONCENTRAT1ON

Figure 9.18 Relationship between a cloud temperature diagram for


a nonionic surfactant system and the corresponding idealized terna-
ry diagram can be taken from the temperature/concentration diagram
for a given temperature.

against surfactant concentration (Fig. 18), the corresponding water-


hydrocarbon-nonionic surfactant ternary phase behavior can be es-
timated. As seen, the minimum in the cloud point temperature ver-
sus surfactant concentration curve corresponds to a two-phase re-
gion within the normal micelle area on the ternary diagram. As tem-
perature is increased, the ternary phase behavior will change a s out-
lined by Friberg (15). For added nonelectroXytes that are structurally
simifar to hydrocarbons, the cloud point temperature elevating or de-
pressing capabitities of the added nonelectrolyte can be understod
by examining these ternary diagrams. Cloudiness results when the
addition of 5% nonelectrolyte causes the final composition to fall out-
side the micellar single-phase region. If the added nonelectrolyte
is more soluble in the micellar re@on, or if it alters the phase be-
havior by changing the phase transition temperatures of the surfw-
tant , then the final composition will be a clear solution, If a series
of binary diagrams contdning progressively larger amounts of non-
electrolyte had been plotted, rather than a ternary plot constructed,
then the pseudobinary set of curves would be difficult to interpret.
However, the ternary representation shows that a characteristic
prog.ression of phase equitibdum exists for these systems.
Figure 9.19 Quatemaw pyramid with the 50% I) plane erngfiasized,

To represent a system that contains more than three pure corn-


ponents, psettdoternaq diagrams can be used, or, if only four pure
components are involved, quaternary f four-camponent , three dirnen-
shnal) pyramidal volume diagrams can be constmeted or projected
into two dimensions. Consider the four-component diagram of Fig-
ure 19, A s d t h the ternary diagrams, the addition of one of the
four eompanents to any mixture of the three other components will
c w s e the totd composition to move in a strdght line toward the
added components corner. Also notice that any three-component
mkture will fall om one of the four surface t ~ a n g l e sthat form the
boundaries of the pyramid. Et is immediately obGous that such a
four -component system becomes very complicated, very quickly. To
simplify this representation, slices through the pyramid& diagram
can be %&en, For instance, rather than determining the phase be-
r the system A-B-@; if A and D were mixed I : l by
h a ~ o for
weight, 33 and D were mked 1: l by weight, and C and I) were
mhed I: 1 by weight, then a pseudoternary diaeam for this system
could be determined using the same techniques as used for a true
three-component system. Determination af the mutual solubility ar-
eas would be completed exactly the same way; however, what were
tie lines defined by two points for a ternary system are now tie
surfaces that may be curved. Thus, an a r b i t r a v slice tkrcrugt?
this surface vv3X produee tie lines that are no longer necessarily lin-
ear.
Osborne

Figure 9.20 Quaternary pyramides in which the ratio of two of the


components is varied to form pseudoternary planes. Varying ratio
of AID (a), B I D (b), and CID (c). RepresentaNons resulting from
the varying the ratio of A/B , A /C , and B IC for the A, B , C , D
system are not shown.

Seldom will eamponent D be soluble up to 50%in A , B , and C.


Thus, often the fourth eompnent will be mixed with one of the oth-
e r three in v a m n g ratios, p d u e i n g a series of pseudoternary dia-
pams that will characterize the system. Figure 20a shows how var-
ying the AID ratio will alter the slices taken through the pyramid.
Variations of the B I D and C/D ratios are shown fn Figures 2Ob and
20c, respectively.
Another way to represent a four-variable system is to have the
base formed from a triangular grid, while increases in the magnitude
of the fourth variable move perpendicular from the base, thus, Eorm-
Op timization of Topical Formulations

Temperature

Figure 9 - 2 1 Example of" a three-component phase diapam with the


temperature varjiable as the verticd direction,

ing a trianwlar pfism, This representation is useful when the


f o u ~ hva~zlbleis temperature, or the fourth component is added to
only a few percent (Pig. 21). For these trianmlar prism, repre-
sentations for which temperature is the long axis, slices through the
psiism pardXe1 to the base result in true ternar~ydiagrams, rather
than pseudot e r n a n diagrams, whereas parillXel slices thraugh a four -
component systems result in pseudoternary diapams.
As seen, quaternary diagrams are actrxdly ff"builtfv by construct -
ing a series of pseudoternam diagrams. Thus, all of the concepts
described earlier for t e r n a q systems are applicable, with the minor
changes that now one or mars of the corners is a miscible mixture,
the tie lines are na longer necessargy strltight, and critical points
are found on a ~ u r f a c erather than a curve (Pig. 22; 16).
In summary, this tutorrlal on binary , telmtav, pseudoternary ,
and quaternary phase diagrams has emphasized some very important
facts concerning pharmaeeuticd fluids and semisolid ksmulations.
First, that homogeneous t i .el , stable) preparations exist only for a
discrete set of compositional ranges, whieh are enclosed by a phase
boundary. Preparations whose composition falls outside this phase
boundary will be unstable and separate into layers of diffedng com-
positions. Within the phase boundary the pmduct eharactedstics of
the preparations (vjiscosity, drug rdease , appearance, consumer ae -
ceptance) should vax.y in a continuous and smooth manner through-
Osborne

Figure 9.22 Idealized two-phase region in a three-component dia-


gram with an intensive variable as the vertical direction. The up-
per critical solution point (UCSP) and the lower criticd solution
point (LCSP) are connected by a plait-point loop IPPL).

out the range of the component composition, Second, once the suf-
ficient number of samples have been mixed and examined to pmvide
the information required to construct the diagram, detemining the
composition ranges over which to optimize is very straightforward,
Values for the ranges can be read directly from the plots based
upon the phase boundaries. Thus, the determination of the phase
behavior of the fluid or semisolid system is required before the op-
timization of a topical formulation.

8. Automation of Phase Behavior Deternination Using


Laboratory Robotics
The introduction of relatively inexpensive, commercially avaable
laboratory robotic aystems provides the formulator with a technique
capable of pmparing many samples without the tedium traditiondy
associated with phase behavior determixlation. The laboratory ro-
botic procedure can be mnceptually didded &to a six-step process:
(I) selection of a preliminary formulation to be studied and deter-
mination of the cornpasitions of the components to be mixed; (2) dis-
pense components and check compositions; (3) cap and label sample
tubes; (4) mix samples well; (5) observe the samples for an appro-
priate length of time to determine when critical product characteris-
tics are established; and ( 6 ) record results and repeat steps X
through 6, as necessapy, to complete phase behavior evaluation. In
Optimization of Topical Formulations 175

the following sections, the robotics techniques required to complete


each of these steps will be considered in d e t a , The discussion will
be generaX enough to be applicable to any of the cornmereidly avail-
able systems, but speciac enozxgh to be used as a p i d e for automa-
tion of the sample preparation required for estabEshing compositiontiX
ranges before formulation optimization.

Successful preliminaq formulation selection is dependent upon vari-


ous factors: (1) the robotics system must be able to adequately
dispense, process (mix), and evalfuate the stabnity of each sample
prepared ; ( 2) the formtxlation should be as simple (i ,e ,, contdn as
few components) as possible ; (3) the formulation should consist of
readsy avaaable raw materials that are as chemically pure as possi-
ble; (4) the attributes of the formulation that are important for con-
sumer acceptance of the eventud product must be cleaznXy defined
and accommodated; and (5) manufacturability of the scaledup formu-
lation must be considered. Qnce a preliminary formulation is select-
ed, the components can be separated into two gmups. The first
grouping is far mateAds that wBl be used at a preset concentration,
regardless of other formulation, changes that may occur because of
optimization, Concentrations af the active component, preservative,
and fragrance mag be examples of matedals that wilX be added in
smclll amounts, or over narmw ranges and, thus, are included in
this 6rst gmuping, The second matedd group is for those compo-
nents that mag be used over a wide ran@ of concentrations. Ex-
amples of this group of materids for topical8 would usually include
.
water and other excstlpients The concentration ranges forming
homogeneous phase re@ons for this second s o u p of m a t e ~ a l swnl
prczvide the Iframework of the phase behavior studies. It is hoped
that no more than three to five of the components will fall into this
second material group.
Next, each of the components fafling in this second grouping
are divided into polar materials, nonpolar mateAals, and other mate-
rids. This allows pseudoternaq diaeams to be constmcted. If
only two of the foregoing divisions of polarity exists, then a b i n a v
diagram may be more appmpsiiate. Five weight percent increments
of the component compositions is usually sufficient to obtain a rough
diagram that can then be refined along the phase boundaries.
An example of this pmcedure for a hypothetical system should
clarify the steps in determining the composjil;ions to be mked. The
preliminalvy formulation selected is a cream ~onsistingof 52 to 15%
active (on soluble), 6% emulsifier, 11% eoemiulsiEer, 15%oil, 8%glyc-
erin, and q.s. with water eantdning the preservative. Glycedn
and water are the polar mateAaXs, the oil is nonpolar, and the emul-
Osborne

Figure 9.23 Points on the ternary grid at 5% intervals that are -


>
45% A, and -
< 30% C, and between 5% and 30% B inclusive,

sifier and coemulsifier are grouped as "otherf' materials. Preserva-


tive challenge testing of the preliminary formulation indicates that
0.1 % preservative concentrations adequately pratects the formulation.
Thus, this level will be added to each sample prepared by the ro-
botic system. Three discrete levels of the active agent will be
evaluated (i-e., 53, lo%, and 15% by weight). Because this is a
complex seven-component system (five components unconstrained,
two components discrete), it is useful to limit the ranges of some of
the materials on the basis of previous formulating experience or raw
material costs. For this preliminary fomulation we will not evaluate
final formulations containing (1) less than 45% water-glycerin, (2)
peater than 30%oil., or (3) greater than 30%emulsifier-coemulsifier.
With these constraints, the region on the ternary diagram has been
limited as shown in Figure 23, Note, that the polar materials will
be plotted at the left corner of the triangle, the oil will be plotted
at the top of the triangle, and the emulsifier-coemuisifier mixture
plotted at the right corner, in keeping with convention for surfac-
tant systems (71. For each concentration level of active agent, the
40: 60, 35: 65, and 30: 70 ratios of ernulsifier/coemulsifier will be
evaluated. Likewise, two water /glycerin ratios will be evaluated.
All combined, 18 pseudoternary diagrams, consisting of 41 sample
points each, will be constructed as outlined in Table 1. This trans-
Optimization of Topical Formulations 177

Table 9.1 A Listing of the Phase Diagrams to be Completed to Es-


tablish Smoothness and Continuity of the Hypothetical Cream Sys-
tem Example

Dkgram
number A B C

5% Active, 0,1%preservative
1 92% Water/8%glycerin 40%Emulsifier/60%cwmulsifier on
2 92% Water/@%
glycerin 35%Emulsifier /65%caemuIsifier on
3 92%Water/8%glycerin 30% Emulsifier /TO%coemulsifier Oil
4 86%Water/ 14% glycerin 40%Emuldfier / 60%coemulsifier OiT
5 86%Water / 14% glycerin 35%Emulsifier/ 65%cwmulsifier on
6 86%Water/14%glycerin 30%Emulsifier/7O%coctmulsifier OiT
10%A c ~ v,e 0.1% preservative
7 92%Water/8%glycerin 40%Emulsifier / 60%coemulsifier oil
8 92%Water/8%glycerin 35%Emulsifierf65%eoamulslfier oil
9 92% Water/8%glycerin 30%EmulsiEer/?O%eoemulsifier Oil
10 86%Water114% glycerin 40%EmulsiEer/60%eoemulsifier oil
11 86%Water/ 14% glycerin 35%Emulsiaer/ 65%coemulsifier Oil
12 86%Water/ 14% glycerin 30%Emulsifier/?O%coemuIsifier Oil
15%Active, 0.1% preservative
13 92%Water/8%glycerin 40%Emlsifier/60%coemulsifier on
14 92% Water/8%glycerin 35%Emulsifier /65%coemulsifier OiI
15 92%Water/8%glycerin 30%Emulsifier/70%coemulsifler on
16 86%Water/ 14% glycerin 40%Emulrsifier /60%coemulsifier on
17 86%Water/ 14%glycerin 35%Emulsifier/65%coemdsifier Oil
18 86%Water/ 14% glycerin 30%Emulsifier/?O%coemulsifit?r O i l
lates into the preparation of 738 samples, Dependent on the corn-
glexity of the dispensing and mixing steps and whether or not the
samples will be centrifuged, each sample will require approximately
1Q min to complete, which leads to an estimate of 123 h r (5-118
days) of robotic time, Approximately 2 h r of technician tirne is re-
quired for each day (24 h r of continuous, unattended operation) of
robot time. Thus, the initial samples for this evduation will require
approximately 1 technician-day and 1 robot -week to complete, Ad-
ditional samples mag be required to pinpoint cfiticaX re@ons of the
phase boundary , as described in the following discussion.

2. Weighing Qut Components


Pharmaceutical topicals may c o n t ~ nfluid liquids, semisolids, amor -
phous solids, cryf~t alline solids, or any combination thereof. Be-
cause the raw materials for topicals v a v great& in their physical
properties, a number of diserent dispensing techniques must be em-
ployed to evaluate topical formulations. Solids are generally di8-
pensed by a ~ b r a t i n grobotic h m d , whereas liquids and semisolids
are dispensed, from automatic syfinges. Materials that are not read-
ily dispensed by either o f these methods can be added by hand,
Conceptual detdls of each of these dispensing methods are de-
scribed below.

Solids. To deliver powders using robotics, a Lube filled with


solid is t-iilited and dbrated in discrete pulses. The dbratiornt
causes the solid to flow from the tube into an appropriate, tared
container before being weighed. Although, in theory, this may
seem a simple task, in practice, it can be very demanding. Aeeu -
rate delivery. of soZids depends on a number of factors, including
brati ion intensity, g d p location, dbration duration, tube diameter,
and tat angle. The dbration intensity is usually varied dudng the
course of powder addition. At the be@nning af powder addition,
the intensity is at a preset vdue near, but not necessaray at, the
mmimurn possible. However, as the target weight is approached
the intensity is reduced, allowing more control over the amount of
solid added. For best results, the tube containing the powder
should be gripped near the base, resulting in maximum transfer of
e n e r w from the hand to the powder in the tube. This ensures that
the solid is added continuously. Vibration duration is also varYled
over the course of addition, decreasing as one reaches the target
weight. The amaunt of variation is dependent on the nature of the
powder being added, Far crystauine solids, tfie duration is re-
duced, whereas for amorphous solids, a longer vibration time is re-
quired. The diameter of tfie tube being used to dispense the pow-
der should be smdler than the opening of the container receiMng
Op timizalion of Topical Formulations 1 79

the powder to avoid spillage. A lip on the mouth of the tube con-
tdning the powder has proved useful, but requires that the dis-
penshg tube is returned to an exact orientation, This can be ac-
complished by notching both tube and rtack sufficiently to warantee
return to a constant orientation, The tilt an@e at which the tube
is held &so determines the rate at which the solid is dispensed*
Accurate addition of solid depends on an appropriate combination of
tilt mgle , *bration intensity, dbration duration, and a knowledge
of the physical pmperties of the solid,
To pour powder, the robot grips the powder tube and tilts it
to an initial angle over the contrainer. A dbration pulse foIlows,
after whkh the tube is weighed, A lag time usually occurs, repre-
senting the time required for the solid to reach the tip of the tube,
If only a s m d l fraction of the target: weight has been. reached, the
tilt angle is increased, The hand ~ b r a t e sand powder is added.
A. compaAson between the target weight and the amount added is
made. Ass the target weight is approached, both the brati ion in-
tensity and duration are decreased and the pouring routine contin-
ued, When the target weight is obtdned, the pouxlling mutine is
stopped, and the net weight of solid cdeulated, By using this
technique, we have obtdned g w d reproducibsity with target weights
as low as O,O5 g.

Liquids, Liquids can be dispensed using automatic, gas-tight


syrbges that are eommercidly avililable and capable of dispensing
accurate volumes, Typicdly , these s y ~ n g e sme computer -contmlled
to position a valve so that the desired volume of liquid can be
pulled from a reservoir into the barrel of the syr;ings. The valve
is then switched and the liquid dispensed through Teflon tubing in-
to a mbot hmd-held sample tube, Bectause the syringe dfive is
subdi~ded,h t o approximately 1000 positions, the repraducibB.lPy.yof
the volume delivered will be 0.1% of the total sydnge volume. Thus,
if a 10-ml sydnge is used, the volume degvered will be 2 0.01 ml;
5.0-, 2.5-, and 1.0-ml. sydnges can readily be substituted to pro-
vide greater accuracy. Although volumes can be accurately dis-
pensed, it is best to weigh each sample tube aRer each liquid addi-
tian. This &lows an exact accounting of what is added into (or
adhered onto) the tube by diminating volume ermrs that may result
fmm air in the sydnge lines or unpurged syringes, Because dl of
the h p u t and output compositions are recorded in terms of weight
percents, a n u r n e ~ e deonversion from weight to volume is necessary-
before each liquid addition. This requires knowledge of the density
of the liquid being dispensed. Appmximate density values, such
as those $iven in Table 2, are u s u a y sufflicisnt fir this canversion,
The dispensing of nondscaus liquids ( ~ s e o s i t yless than I, cp) is
Table 9.2 Approximate Densities for Liquids

Liquid Density (chain l e n e h )

n- Alkanols 0.79 (n = 2) 0.84 (n = 17)


n-Fatty acids (saturated) 1.05 (n = 2) 0.91 (n = 8)
n-Fatty acids (monaunsaturated) 1.03 (n = 4) 0.85 (n = 22)
n-Alkanes 0.66 (n = 6) 0.79 (n = 20)
Alkanes
pristane
squalane 0,810
squalene 0,858
petrolatum 0-82-0.88
Glyerides
glycerol trioleate 0.91
glycerol monooleate 0.94
Esters

cetyl palmitate 0.83 (at 50°C)


propyl stearate
ethyl oleate
Surfactants approx. 1.0

readily accomplished using virtually any syringe configuration.


However, more vlscous matedals present a greater challenge. To
accurately dispense more viscous materfals (viscosity between 1 and
8000 cp), varfaus changes can be made to the dispensing system.
The diameter of the tubing, dispensing probe (if present), and
bore dimeter of the valve can be increased. Tubing lengths can
be minimized, and a smaller syringe volume can be used when pos-
sible. The time required for the syringe to travel full stroke can
also be increased. Table 3 gives a series of syrlnge configurations
for the Hamilton Microlab 900 series dispenser that accurately dis-
pensed liquids of known viscosity without stalling the syringe.
Optimization of Topical Formulatians f 82

Table 9 . 3 Filling and Dispensing Confiprations for the HamBton


Microlab 900 Series Dispenser

Tube diameter 5yringe speed


Syringe (in ,) (0- 15)a
Viscosity volume
(cP B 25OC) Id> Inlet Outlet Filling Dispensing

"Sydnge speed of 4 is equivdent to 1.4-see full stroke, whereas a


s y r h g e speed of 25 is equivalent to 15-see full stroke.

Semtsoltds, Viscous materials can be dispensed by use of corn-


nzercidly avdable liquid application systems that are designed for
dispensing adhesives, solder pastes, and other sirnBar matedals
(EFD, Xne. 917 Waterman Avenue, East Provzidenee, Rf 029143). The
dispenser is connected to mtered compressed plant air, Output air
pressures up to 100 psi p m d d e cantrolled application of even thick
matefials that will not pour. An adjustable vacuum system prevents
ddpping between dispensing cycles for lower- viscosity liquids,
These dispensers can readity be controlled by interfacing to the
robotic ieantrol computer. To dispense the desired amount of mate-
rial, the s y ~ n g ebarrel, with appmpriate dispensing tip, is placed
above the sample tube pXacled on the ba-iance. The ~ s c o u smaterial
is dispensed into the sample tube for a preset time (e.g., 50 cycles
through a timing loop), Dispensing is halted, and the amount of
added materid weighed, Based upon the amount added for the
first 50 q c l e s of the timing loop, the number of timing-loop cycles
required to complete the ~ s c o u smate&& addition is calculated, and
materid is then dispensed into the sample tube for that pedod.
Greater accuracy can be gdned by increasing the number of addi-
tion iterations.

Hand additions. Some materials, such as stiff amomhous semi-


solids, are unreasonably difficult to dispense with any of the fore-
going methods. Hand addition of these tmublesorne materials is the
most practiczil method. The robot is pmgrammed to tare each sam-
ple tube, after which the researcher adds a quantity of the materid
to the tmed tube. The mbot then weighs the sample tube contain-
ing the hand addithn, calculates the weight of the added materielil,
and cerIculates the amounts of the remaining liquid or senndsoIid corn-
ponents that must be aefded to obtdn the orl@ndIy desired conpo-
sition. In this way, hand addition can be compieted rapidly, pro-
d d e d the total volume of the sample is not required to be constant.
The researeher must be carefin1 that the amount of matedal added
by band is of an approgriate mapjtude so that neither overfBling
nor underfilling of the sample tube results upon subsequent com-
ponent additions, Qbeously, this pmcedure &lows for only the
first matedal to be added by hand,

3, C o n t a i n e r s , C a p p i n g , and Labeling
Ilirtudfy any contdner is suitable for a laboratory robotic system,
proeded the weight of the f2Eed sample cont&ner does not exceed
the lifthg capacity of the robot* Otherwise, contdners should be
selected based upon the ease of completing pmper mixing within
the container and the washabnity or disposabBity of the contdner,
We have found 16 x 125-mrn disposrzble culture tubes ~ t caps
h to
be ideal for our phase behavior investiigations.
Capphg stations function by either having a deAce that will
grip and rotate the cap whife the tube is held stationary by the
robotic hand, or by having a efedce that tiltill g ~ and
p rotate the
tube, whae the cap is held stationary by the robotic hand, The
former system holds the cap unt2 the tube is repositioned to the
capplistg station, and the cap is torqued onto the tube, Thus, only
one tube is typically uncapped at any @ven time, The latter sys-
tem sows more than one tube to be uncapped during any single
command cycle.
Optimization of Topical Fomulations 183

The labeling of the numerous samples can be readay ac~ornplished


by printing the sample name, compositbn, and preparation data on
peelable adhesive labels that are applied to the sample tube by the
researcher, A dot matrix pdnter can be set up in a manner anal-
ogous to that required fbr pdnting address labels and interftlced
to accept output data fmm the robotic control computer,

4, Mixing Samples
After component addition and weighhg, the next stage is to effee-
tively mix each sample, For Buid samples, an automated vortex
mker can be instdled with the robotics system to operate in e&her
a single-action or pump action manner, The length that the sample
is vortexed can eassy be adjusted to meet the requirements of the
system studied, To facaitate mking of more dscous samples, sr
samples that contain components that melt at sBght1-y h i e e r than
ambient temperatures, a heating block can be positioned d t h i n the
work envelope of the robot, and samples may be heated for a pre-
set Xeneh of time, The method of centz.ifu@ng ~ s c o u ssamples
through: a constricted glass tube can also be useful Tar mixing sam-
ples by use of laboratoq robotics. A soft glass tube 5 to 6 in,
Xong is sealed at one end, and a constriction, is placed roughly in
the middle of the tube. After addition of the components, the
mouth of the tube is then flame-sealed, and the mat^ is repeated-
ly centrifuged firam end to end of the tube through the matdx,
Slight annealing may be required after mking by this process (17).
The canstruetian and seding of the constdeted tubes w3l need to
be completed manuafly, although use of robotics is ideally suited for
f XI@ repeafed inverting and centrlfu@ng of the eonstdcted tube,

5, Observation O f Samples
After a we& of robotic effort, the reserzrcher is fsced w.i"eh evaluat-
ing 738 samples and plotting the phase behador of each sample on
one of 18 pseudoternary plots, Xt quickly becomes apparent why it
is necessary that all of the critic& processing steps be completed
using the laboratory- mbotics configuration. If each sample re-
quired a manu& processing step (e.g., mking vvlith a spatula) the
amount of human efforpt; would be unacceptable. Sample evduatiran
should consist of a simple observation to establish the presence or
absence of a homogeneous phase, Thus, each point on each dia-
gram can be quickly platted as a drcle if one phase, and as an x:
if two or more phases are present. The best smmth curve passing
between the xs and os is considered the phase boundary, If a
best mrve cannot be drawn because of lack af data, then add-i(ional
Figure 9.24 Comparison of a rough diagram (a) drawn from an in-
sufficient number of data points and, a finetlized d i a p a n (b) that
has a well-defined phase boundav.

samples can be prepared to better charactedze that re@on of the


diavarn. F i p r e 24a shows a rough diagram drawn from a linti_lsd
number at" avdable points, and F i e r e 24b shows the refined dla-
gram.
Thus far, we have discussed sh@e-phase re@ons and homoge-
neous phases. It may be necessary to Ezlarlfy what is meant by a
phase, A, general definition quoted, fmrn Ref. 8 is: A sample of
matter is safd to be in a certain phase when It has a certain well
defined set of"macroscopically observable properties. The phase of
a sample Is ~"eallyan indication of the degree of order ar disorder
inherent in the molecules of which the sampEe is composed, Be-
eause the researcher can estabBsh the set of macroscopie&ly ob-
rservetble properties, a phase can be defined in various ways. For
our cream example from the Erst paragraphs of this section, the
phase could "be defined as any mixture that rern&ns hornogerzmas
after b e h g heated to 40QC for 10 m h and then immediately centd-
fuged at 4OOO rprn for 1 h r , Note, that this was not referred to
as a single phase, Usudy the term single phase i s reserved for a
mkture of molecules .forming a phase that i s thermodynamicany srta-
bf e, Such systems include simple solutions at equnibdum, lyotrapie
.
liquid crystda , and other surfactant association stzructures Emul-
sions are kineticdly stabi~zedby the presence of the surfactant
and, thus, cannot be considered wsjslgle-phase't "systems, Distinct
phase bounda&es w i l l enclose the stable emulsion re@ons on the di-
Optimization of Topical Formulations

Table 9.4 Component Ranges of the


Formulation that Provide a Smooth and
Continuous Phase Suitable for Optimization

Range (%I Component

5 - 15 Active
0.1 Preservative
Emulsifier
Goemulsifier
Glycerin
Water

agram as seen in Figure 14 and in Ref. 18. Thus, for most com-
mercial applications, the researcher defines the phase according to
the stability requirements for the product.
Before the experimental design, modeling, alid optimization steps
can be established, the researcher must guarantee that a phase
boundary is not crossed as the composition of each component is
varied. Pseudoternary diagrams have now been constructed to
bracket the constrained ranges of each of the component variables.
If each of the 4 1 samples on each of the 18 diagrams was within the
desired phase boundaw, then the compositional ranges to be op-
timized would be identical with the values of the constraints we
placed on each component. However, if the phase behavior was as
shown in Figure 25, then the compositional ranges would be more
limited. Note that for 5% lo%, and 15%active compound, the 30%:
70%emulsifier/coemulsifier provide a stable emufsion only at higher
than 20%to 25% by weight of the emulsifier-coemulsifier mixture.
Thus, the emulsifier/coemulsifier ratio amst be greater than 30: 70.
For the other two emulsifier/coemulsifier ratios, the mount of oil
can range from 5 to 20 wt%, whereas the amount of emulsifier-co-
emulsifier can range from 10%to 25%. Within the above constrslints,
the water-glycerin mixture can range from 45%to 85% (remember,
the formulation contains a minimum of 5%oil. and 10% emulsifier-eo-
emulsifier). The component ranges based upon the phase behavior
results shown in Figure 25 are given in Table 4.
Qsborne

Figure 9.25 Phase behavior results for the ternary systems listed
in Table 1 for the hypothetical cream system example. X s denote
either clear single-phase regions, two liquid phases (i.e., an un-
stable emulsion) or stable emulsion plus another phase.
Optimization of Topical Formulations

Figure 9.25 (Continued)


Figure 9.25 (Canthued)

Before discussing how to design an experiment to op~mizea formu-


lation, some terminolom should be cladfkd, First, the difference
between trial and expelcliment, as used in this sedon, should be de-
scribed. When cansidering formulations, the result of each tdaZ is
a finished pmduct po~sessing8 set af properties ( e .g, , potency,
dseosity , density, appearance). After completing a suffident num-
ber of tdals, arx optimized formulation can be selected to satis&
the edticd pmduct ekaractedstics. A set of trlals resulting in an
optimized formulation is called an expevz'ment, Thus, to design an
experiment is to select a sedes of unique or replieate %ridsthat
will ultimately pm.mide sufficient informat-lon to select an optimized
formulation. Each trial d 1 differ in the mnditiorrs , that is, in the
varfabtes used to formulate the product (varjiables might include
Optimization of Topical Formulations 189

composition, mixing times, shear rates, pH). The variables that


axe chosen for inclusion in the study should be those that have the
most dramatic effects upon the critical pmduct characteristics,
Likewise, variables that are unlikely to affect final product charac-
teristics, or vahbles that are constrained because of equipment or
procedural limitations, should be held constant throughout the tri-
als. Finally, the terms response and level should be discussed. As
stated earlier, each trial wiIl result in a finbhed formulation pos-
sessing a set of properties. The resultant characteristic propemties
of this formulation are called the responses. The variable levels
and eor~ispondingresponses are combined by use of an appropriate
regression analysis to empirically derive a function. For two vari-
ables and one response, an easily visudized three-dimensional re-
sponse surface results when this function is plotted. As the num-
ber of variables or responses increase, a multidimensional response
surface w i l l result that cannot be easily visualized. However, any
three-dimensional "slJceWof the multidimensional surface can be plot-
ted. For each variable, a range of values exist that could be test-
ed. In thL section, the level of the va&ble will refer to the value
of a variable selected for a particulm trial. If the variable range
is viewd in terms of highest, middle, and lowest values, then the
phrases level of the varhble and variables at three levels can be
understood. Simirarly, one might be required to evaluate 10 levels
of an excipient between 10 and 20 w t % in a formulation. In summa-
ry, an experiment is a carefully designed set of trials in which var-
iable levels have been selected. Each response can then be em-
pirically modeled to provide a response surface. The optimum value
of the response surface is located, and the corresponding variable
values are obtained, thus providhg an optfmZzed formulaHon (I.e,,
optimized critical product characteristics).
The quality of an optimized fornulation is only as good as the
design of the experiment used for optimization. The techniques to
be described mlnimize the number of trials in a manner to mmimize
the amount of information obtained by setting up a statistically val-
id experimental design. These techniques require that the response
surface is smooth and continuous and, thus, for dhperse systems
in which a phase boundary might be crassed, it is necessary to
have established the phase behavior as det&led earlier. Because
the application of statistical methods for optimization of disperse
d o s q e forms (e.g., suspension, emulsions) has recently been re-
viewed by Franz and co-workers (31, these topics will be only
briefly described to serve as an introducaon to the reader who is
unfamiliar with experimental design, modeling, and optimization
strate@es.
A, Experimental Design
An ernpiricali response model surface can be created by conducting a
sedes af exgedmentd trWs hadng selected vadable levels. Each
varfablo level for the tdafs is carefully chosen such that a maimurn
arnmnt of information on the relationship between the variables m d
responses is obtdned with a minimum number of trials, A number
of assumptions are made about the data to be obt&ned from the
statigticay desimed experiment, It is assumed that observations
are representative of the population, that replicate observations for
vaf.ious combbations of variables have the same variance, and that
the rmdom errors are bdependent m d nomdilfy distfibuted, In
adctftion, it is assumed that outliers m d missing values will not be
.
obtdned Each of these assumptions can be statistiefly vdidated ,
either by random sampling of a proper size from the a p p r s p ~ a t e
populati-on rn by use of standard statistied methods of testing (29).
The most eornmanly used analysis within the pharmaceuticaf industry.
is the full f s l e t o a design in which all the varialbles are at two
levels. For this design the number of trials will equal 2k, where k
is the number of vadables. Because of the exponentid increase h
the number of t d d s for this design, it iis generdly used when ffve
or fewer variables will be investigated. The advantages of this de-
s i p are that (1) a low number of Z;ri&s per independent variable
are required; (2) both quantitative and quditative variables can be
examined; (3) the results are eers2y iinterpreted; and (4) this de-
8ign forms the basia for several other designs such as the fractiand
fsrctoAal, face centered, and centrd composite expedmenfd designs.
When there are five or more va&ables, a fraction of the full factor-
i d can be used. All essential information on varhbEe ma-En effects
and two-way interactions are stBX obtdned , but with less exped-
mentajt ef"fort.
The design layout or matrir for a 2k factorial design for k = 2
thraugh. 5 is shown in Table 5. This table pmddes infirmation on
the number of trials required far a statisticdfy vdid experimental
design (2k) and the value or level that each variable must be evalu-
ated for a @ven trial, Remember that we have limited ourselves to
a two-level design; thus, the maximum and minimum vdue of the
variable range are the only two levels of the variables to be evaluat-
ed, The (+) notation means thait: the maximum value within the
vadable range will be tested, whereas the (-1 notation represents
the mbimum value of the vaAable? range, Therefore, fmm the table
it is apparent that experiment& tdal number 25 d 1 use the mhirnurn
vdues for vwableco XI, X 2 , ~ndiX 3 , whereas the! maximum values
of vadables X4 and X5 are implemented. Similar information can be
readay obfhecf, fmm the table for any trial. Note, these t d d s
should not be pedommed in the order listed in Table 5, rather the
Qptimization of Topical Formulations l i)_Z

order should. be randomized, This d l help to ensure that errone-


ous conclusions wnl not be drawn from the effects of any systemic
or eneronmental. vadables not included in the studiy,
Once the elrperirnentd design has been selected, dedeians must
be made concerning the proper response a t t ~ b u t e sfar the formula-
tion. Remember, when the topic ""determination of compos~tfonsta
be mhedff wvvas considered earlier, it was staled that sueees~fulpre-
liminary farmutallon seteetian is dependent upon ... ( 4 ) the attri-
butes af the finnutation that are important for consumer acceptance
of the eventual pmduel must be clearly defined and weornmodated.
Because the attdbutes of the formulation were clearly defined be-
fore phase behador characterization, the attdbutes (or responses)
tested for each formulation have &ready been thought thmugh.
For a topical agent, the attfibute ~ s c o s i t ywBl likely be cansidered.
Cost of the fclrmulation and purchase btent , as judged by a con-
sumer panel, may &so be possible responses selected for each trial,
Potentid attfibute responses to be tested may hclude bioavaability
(as estimated by in vitm percutaneous absorption studies) or irri-
tation potential (as shown by anima3 testing). An excellent example
of optimizing a topied pmduet for consumer accqtance is @yen by
Moskowitz (20). Because generation of a reliable response surface
is the goal of this entlire process, the researcher must discern front
the v e be@nning
~ what pmduct charactedstics or response attri-
butes are impo~antto the cammer~ia3.success of the pmduct. Xf
the researcher b bterested in optimizing only one response attri-
bute, then the complete ch~racteAzationof the phase b e h a ~ o ro f
the system and subsequent optimizkllian techniques descdbed in this
chapter are probably not worth the effor/t: required for completion,

B e Modeling and Optimization


W e now know the concentration ranges over which the formulation
proddies a smwth and continuous phase re@on. The experimental
d e s i p has been selected to include both compositional and pmeess
variables, If the full EBctorjid expedmentacI d e ~ i mat two levels
was selected for five vadables totd, then the 32 formulations or
t m s (25) have been manufactured, Six tests have been selected
to evduate the gmduct attdbutes, and results from these have
been obtdned for each of the 32 formulations. These expelrfm.entaX
observations p m ~ d ethe response attributes we consider most im-
portant for the commercial, success of the pmduct, All of this ob-
served data are then used to eonstmct a model of the response
surface, Optimurns (either mmimums or minimums) can be deter-
mhed by u s h g standard techniques (3) or by using statistfed soft-
ware packrages that offer multiple experimental deslips and perform
192 Osborne

Table 9.5 Two-Level Factorial Design Patterns

Trial Variable
Optimization of Topical Formulations

Table 9.5 (Continued)

Trial Variable

both modeling (by use of multivariant curve-fitting routines) and


optimization (21). Predicted results can then be confirmed with ad-
ditional experiments. Although these last steps of modeling and op-
timization are truly the "guts" of optimization techniques, the thor-
ough reviews and articles dealing with this subject ( 1-3,20,21)
should be consulted for further information.

I\(. CLOSING REMARKS

The task of designing an experiment, preforming the trials, devel-


oping a model based upon observed data, and conarming the pre-
dicted results with additional experiments seems Sn itsdf a hermean
effort that may appear justifiable only for the most important of
projects. Now, added to these efforts, is the burden of proving
that the phase behavior of the disperse system is both smooth and
continuous before even starting the forewing procedures. Al-
though this chain of studies may seem too involved to be practical,
remember that (I) the final product will be of the highest quality
possible with the greatest lfkeljhood of commercial success, and (2)
that the steps are well defined and readily predictable in terms of
time and cost required for completion. It should also be noted that
compared with solid-dosage forms, disperse systems can be f o r m -
Iated in small quantities (2-200 ml) to check the stability and re-
sponse attributes, thereby minimizing the drug and raw material
costs during optimization,
AKNOWLEDGMENTS

Specid thanks to Roger Klassen for rev;iew af the section on optirniza-


tion and to Carolyn Pesheek and Kllim O%eill for redew of the
seetion an laboratory mboties. Prof. Stig Fribergk method of
t e m h h g surfactant ternary phase behador was thoroughly utjEXized
in writing this chapter. H i s mentorship during my graduate studies
was, and still is, greatly apprec%ted.

REFERENCES

f), E. Fonner, d r , , J , R. Buck, and 6 . $3. Bander, S, Pharm,


Sci, , 59: 1581, 1970,
N, R. Harris, 3. B , Schwartz, and 3, W. McGinity, Bmtg,
Dev. Znd, P h a m , , 1 2 : 1589, 1985,
Ft , M, Franz, 3 , E. Browne, and A , R e Leuris, in Pharmaceu-
ticaZ Dosage Forms: Disperse Systems, Val, 1. (H. A .
Lieberman, M, M, Rieger, and G , S Banker, eds,). Marcel
I

Dekker, Mew York, pp. 427-514, 1988,


5, E , Fdberg, and 6 . Solans, Langmufr, 2: 121, 1986.
G , M. Ecefeston, Cosmet. ToiEetrz"es, 102(11):13, 1986,
S, E . Fdberg, X, Bfute, H, Kunieda, and P , Stenius, Lang-.
mur'r, 2: 659, 1986,
P , Ekwiiifl, in Advances i n Liquid C r y s t a l s , Vof, 4, ( G , H.
Brown, ed. 1, Academic Press, New York, 1975,
J , S, Wdker, and C I A. Vause, Sci, A m , , 258(5):138--105,
1987.
G. M. Barmw , Physical Chemistry, 4th e d . , McGsaw-Hal, New
York, gp. 32-2-329, 1979.
&. Attwood, an@A , T I Florence, Surfactant Systems: Their
Chemistry, Pharmacy and Biology. Chapman and Hall , London,
1983.
D, M. Smdl, in The Bile Acids, Vol. 4, (P, P. N d r , and B.
.
Kritehevsky , eds ) , Plenum Press, New York , 1971,
European Patent 133-258-A, VitaminlE mked micelle h j e c t h n
. . .
solns contg dtamin/E, phospholipid, e g , lecithin and b2s
acid, e ,g ,, glyco :cholic acid, 1983 ,
World Patent 84 / 02016, method of prepadng controlled release
preparations for biolo@cdly active materials and resulting eorn-
positions, 1984,
S, E . FAberg, and D , W, Qsborne, J. Am. Oil, Chem, See.,
63: 123, 1986.
IC, Shinoda, and 23. El F ~ b e r g Emulsions
, and Soilubz'Ztzation,
John W2ey & Sons, New York, 1986,
Op liimization of Topical, Formulations 195

16, B . M, Knickerboeker, 6 , V . Pesheck, H. '1'. Dafis, and I;. E.


S e ~ v e n ,J. Phys, Chem,, 86: 393, 1982.
17. F, f).Blurn, E, I , Franses, E;, D, Rose, R. G I Bryant, and
W. G , Nsfer, Langmuir, 3: 448, 1987.
18, S , N, Ng, and S. G , Prank, J, Dfsperston Sci, TeehnoZ., 3:
217, 1982.
19. €2, E l P , Bax, and lif , R , Draper, Emplricat Mode2 Building and
Response Surfaces. John Waey ga Sons, New Vork, 1987,
20, H , a. Moskodtz, in Cosmetic Pmduet Testing: A Modern
Psyetaophysieal Appmaeh , Chap, 10. Marcel Dekfrer , Hew
Vork, 1984,
21. d , Baykroff, J, Eng, Cornput. AppE, , fall, pp. 101, 1986,
General Considerations for Stability
Pharmaceutica

GARY R , DUKES The Upjohn Company, Kalamazaa, Michigan

Stabflity testing is a necessary and vftd means to help assure that


.
a formulation fl mdntain its integdty (i.e , strenelrz, q u a t y , and
pudty) during its assigned shelf life. It would, however, be futife
to attempt ta propose a simple, universd stabzty-testing protocol,
which could reasonably be expected to assure the h t e g d t y of every
topical formulation of e-vev pharmaceuticd manufacturer, because
there are necessaq differences in spl~?"cific stabnity pmtwals be-
tween difsrent formufatians of the same manuf'aeturer and between
simaifar firmulations of different manuflacturers,
Mthough a number of factors may hRuenee the determination of
an expiry date f"os a given product, this discussion will be confined
to t e s t h g designed to determine the chemnied and physicd proper-
ties cfitic& to the stabisty of the product, with the assumption
that one, or a combination, of these is the limiting factor. Although
much of the discussion may be applicable to the testing and dating
of nonpreaer;lptbn p r o d u ~ t s ,it wal be confined to g r e s c ~ p t i o n
products that progress from investigational new d m g (END) status,
through the new d m g application (NDA) rstage, and on to market-
ing.
The theme of this chapter is based on the premise that despite
the differences in specific stabnitg-testing protocols among products
and manufacturers, there are actidties that are common ta any sta-
baty-testing program, The first set of stabaty-related actieties
in the development of a new active ingredient into a marketed
product generany invalves the profiling of its physical and chemical
198 Dukes

pmperties, This is followed by a period in which the stabnity of


the bulk active ingredient and its assodated forrnuktions is assessed
and an expiry pe&od established, Subsequently, stability studies
are hitfated and continued ta oeonff~mand expand the results of the
preceding studies, F h d y , the stabBity of the marketed product
is pedodicably reassessed,
Thus, the development of an understandkg of the stabBity per-
formance of a &ven formulation is basicay an evolutionaq process,
For convenience of dismseian, this evolutionam progression may be
divided into six stages f 1) :

Preformulaaon, stage
Formulation development stage
Proposed product stage
New product stage,
Established pmduet stage
Re.rriseEI pmduct stage

Within this prctgression from prome to assessment, through con-


firmation and expansion, and f i n a y to periodic reassessment of
product stabif;jfy, a range of testing protomls, methods, and math-
ematical models may be rxtnized, with. each successive stage desiped
to augment the data base of the product and, thereby, strengthen
and expand the eonelusions reached during each preceding s t a p .
At each stage diffe*g types and amounts of informatkn are sought
that are important to the assignment of an expiry pedod, thus,
@ ~ n g &ss to widely differing concerns and objectives for each
phase of the overall stablJity-testing program.
It fs also important to understand the relationship between.
these stages and tkre pmgression of the product through clinic& de-
velopment to marketing (Table 1). The evolution of the product
stabsit-ly performance praf2e be@ns with the preformulation stage,
which is associated with the pre-END phase, then proceeds to for-
mulation development which is generally associated wfth Phase f and
11 clinical tri&s, Prom there it psoceeds to the proposed product
stage, which is tempordly associated with Phase ZIS dinical tgale.
The new product stage is assodated wfth scdeup, pmcess vaida-
tion, m d appmva2, and is foillowed by the established and r e ~ s e d .
pmduct stages, which are associated with the marketing of the pmd-
uct. Note that tkre timing relationships @ven in Table 1 represent
a general approach and are not intended to be d&dly bterpreted.
A s the product pmpesses through the v a ~ o u sstages, different
kinds of studies are carried out to meet a vadety of objectives.
There are a number of elements thiat. are common to a% of these
stability studies and that must be included in the study protocoI.
Stability Testing 199

Table 10.1 Evolution of a Product Stability Performance Profile

Study stage Timing

Preformulatbn Pre-IND
Formula*n development Phase 1-11
Proposed product Phase 111
New product Scaleup and approval
Established product Postapproval
Revised product

The design of each study must begin with a clear objective. With
the objective in mind, the attmutes to Be measured are defined,
and appropriate test methods are defined or developed; the storage
canditions, assay schedules, and types of samples are selected ; and
the timing and method for data evduetion are detemined,
The importance of a clearly stated objective cannot be overem-
phasized. For example, a study to determine the effects of metal
ions on the stability of a new active ingredient in aqueous formula-
tions will be quite diffentnt from a study on an established product
to confirm its stabnity performance profile.

I I. PREEORMULATlON STAGE

The first assessment of the stability of a potential new product be-


gins in the pmformulation stage. During this phase the first re-
quirement i s the development of analytical methods that are speciflc
for the drug substance in the presence of process impurities and
degradation products. Two types of stability studies are conducted
during this stage,

A. Profile Studies
The first type is to determine the profile of physical and chemical
properties of the new drug substance. To identie probabIe routes
of demadation, the active ingredient should be subjected to short-
term accelerated studies to produce obvious changes (Table 2). The
storage conditions should be geeared to possible degradative condi-
tions, such as heat, light, maisture, and oxidants. Testing sched-
ules are chosen to allow estimates of stability parameters. The sam-
Tabfe 10.2 Profile Studies

Protocol elements Activities or results

Objectives Determine any reactivities of the new


drug substance
Establish storage conditions and packaging
requirements for the new drug substance
Study design Short-term studies under accelerated con-
ditions, loakhg for gross changes
Tests Potency, appearance, physical properties,
degradation products, and the like
Storage conditions Geared to possible degradative conditions
of heat, light, moisture, oxidants,
catalysts, pH, and others
Schedules As appropriate to atlow estimates of
stability parameters
Samples Representative of the batch
Evaluation Review of data to estimate relevant
st ability parameters

ples should be repmsentative of the batch of materid used, and the


evduation will consist of a review of the data to estimate relevant
stability parameters. The results of these studies will provide im-
portant information about the molecular reactivity of the new drug
substance, an indication of the types of formulations that may be
successful, and information on the storage requirements of the bulk
drug.
An example of the type of study and the results that can be
obtained are shown in Figure I. This study involved the effect of
metal ions on the stability of a drug substance in solution at 70°C.
It is apparent that zinc (II), even at micromolar concentration, as
well as nickel (111, and chromium (VI), to a lesser extent, exert a
very deleterious effect on the solution stability of the drug. (The
mlution with no added metal ions retained 93%of its initial potency
after 2 months at 70°C.) Because these metals may well be present
in packawg materials, excipients, or even the synthetic process
for the bulk drug, this knowledge was of great value in the devel-
opment of formulations, the selection of packa@ng materials, and
the monitoring of the bulk-drug synthetic process.
Stability Testing

Figure 10.1 Effect of metal ions in solution at 70°C. Squares, 3.0


D M Zn (EX); triangles, 1.0 mM Ni (11); diamonds, 1.0 mNI Cr (VI).

B . Toxicalagy Supplies
A second useful sauree of stability information during the preformu-
lation stage comes from the monitoring of toxicology supplies. These
represent the first rudimentary formulations of the new drug sub-
stance. The objectives are to assure that the potency and level of
significant degradation products are documented during the use of
the supplies. An important added benefit is the development of
supporting data for use during the formulaaon development stage.

1 11. FORMULATION DEVELOPMENT STAGE

The emphasis in the formulation development stage shifts from pro-


filing the physical and chemical properties of the new drug substance
to the assessment of the stability of both the bulk drug and its m-
sociated dosage forms as they are developed. During this stage,
preliminary formulations are manufactured in small-scale lots for use
202 Dukes

in clinical studies, and their stability is monitored. Early in this


stage, additiond studies specific to the type of formulation chosen
are conducted on the bulk drug to complete the profile of its phys-
ical and chemical properties. The effects of pH upon solution sta-
bility (Z), for example, would be criticat if the active ingredient
were to be formulated as an aqueous solution, cream, or lotion. If
the product is to be formulated as a topical suspension, studies to
determine changes in particle size with changes in temperature or
over time would be very useful. A knowledge of the various crys-
tal forms in which the active ingredient can exist, their intercon-
version pathways, and the solubilities of each could prove important
to the manufacturing process for the formulations and for their re-
sultant chemical and physical stability ( 3 ) . Finitlly, studies should
be conducted to determine the compatibility of the new active in-
gredient with probable formulation excipients. These are short-
term studies under accelerated conditions of binary mixtures of the
drug substance with potential excipients. Comparison of results
provides for the early elimination of potentially deleterious excipi-
ents. If the drug substance is sensitive to the presence of mois-
ture, hydrated excipients should be avoided. Sometimes, the stud-
ies may indicate the types of controls necessary to ensure the suit-
ability of the excipient for use. A n example is shown in F i p r e 2.
A cream that was formulated with two different grades of the same
excipient was stored at 40°C. The results indicated the necessity
to control the pH of the excipient to avoid the basic pH range.
Later in this stage, studies are conducted to determine the op-
timum formulation. A number of different formulations should be
prepamd and compared with each other.

A. Formulation Comparison
A convenient means for conducting formulation comparison studies is
to first screen the trial formulations for undedrable physical
changes under accelerated conditions. For example, ointments and
creams may be subjected to temperature cycling to determine the
relative tendency to separate, Topical suspensions should be
screened for changes resulting from particle growth by Ostwdd rip-
ening, for changes in polymorphic forms, or for a tendency to cake
(4). Next, the formulations should be screened for chemical stabil-
ity. By using the assay for the drug substance Pram the previous
stage as a starting point, a stability-indicating assay can be devel-
oped for the most promising formulations. Specificity may be deter-
mined by spiking the assay preparation with like compounds or au-
thentic samples of known or suspected degradation products.
Extensive use can be made of short-term studies at high-tem-
perature and high-humidity conditions (5). For example, the ex-
Stability Testing

Months
A

Months

Figure 10,2 Effect of exdpient pH at 40OC. Diamonds, pH 5;


circles, pH 8.

perimental formulations may be stored at 40°, 47O, 56O, and 70°C


and assayed for potency at several intervals. If there does not ap-
pear to be a change in the mechanism of the degradaCion reaction,
Arrhenius plots may be utilized to calculate energies of activation,
and the shelf life of the proposed formulations at other temperatures
may he estimated. Regression analyses may also provide a crude
estimate of the sheff life of each formulation. The shelf lives esti-
mated from the use of the Arrhenius plots and the mgressfon anal-
yses can be used to compare the formulations with each other to
determine the one that is most stable, The data for the selected
formulation are also useful in pmviding the first indications of a
suitable storage condition for the finished product (Table 3).

B . Container and Closure System Selection


The next component of this stage involves studying the selected
formulation in several different container and closure systems.
Again, short-term testing under conditions of elevated temperature
and humidity ape used to determine the stability characteristics of
204 Dukes

Table f 0.3 Formulation Comparison

Protocol elements A c t i ~ t i e sor results

Objectives Determine any formulation reactivities and


significant drug substance degradation
products
Compare stability characteristics of the
experimentat formulations
Collect preliminary information on possible
stability -limiting factors
Study design Short-term studies under accelerated con-
ditions, looking for gross chmges
Tests Potency, significant degradation products,
physical properties, plus indiddud
tests relevant to specific dosage forms
Storage conditions Geared to possible deeadative conditions
of heat, light, moisture, and the like
Schedules A s appmpriate to allow estimation of
stability parameters
Samples Representative of the batch
Evduation Direct comparison of raw data, comparison
of rates, and so forth, to estimate
relative stabaity performance

the formulation in the various contdner-closure systems. Test and


assay data are obtained and regression analyses are used where ap-
plicable to obtain crude estimates of shelf life for comparison pur-
poses. Any packages that are found to adversely affect product
stability are eliminated.

C. Cllnical Supplies
An additional component of this stage involves the monitoring of
cEnicd supplies (Table 4).
Were, the objectives are to determine the values of the critical
quality parameters of the formulations used to establish safety and
efficacy as a function of time and to ensure that only satisfactory
material is used in the clinic.
Stability Testing

Table 10.4 Clhical Supplies

Protocol elements Activities or results

Objectives Determine the values of the critical quality


parameters of formulations utilized to
establish safety and efficacy
Assure that only satisfactory material is
used in the clinic
Study design Medium-term studies looking for changes
in critical quality parameters
Tests Basic attributes, such as potency, signi-
ficant degradation products, physical
properties, plus individual tests rele-
vant to specific dosage forms
Storage conditions Compatibh with label storage conditions
Schedules More frequent for less stable formulations
Samples Representative of each formulation and
container-closure system in use in the
clinic
Evaluation Periodic evauation to assure continued
suitability of supplies by direct com-
parison to estabfished specifications or
statistical evaluation

Generally, the study design for clinical supplies is a medium-


term study (covering at least the period for which the supplies are
in urn) looking for changes in critical quality parameters. The
tests should include potencuy, sipifieant dewadation products,
.
physical properties (e g . , appearance, odor, consistency, and
weight loss), plus individuai tests relevant to the specific dosage
form [e,g., particle size (suspensions), pH (aqueous formulations),
drug release (transdermal patches) , resuspendabillty (lotions), and
.
(water content (nonaqueous formulations) 1 In addition to measur-
ing the potency of bulk ointments and crems, the potency of sam-
ples from several locatjons inside the contdner should also be meas-
ured to determine the tendency of the active ingredient to eoncen-
trate. The storage conditions are those that correspond to the la-
bel storage conditions. The samples are representative of at least
each formulation and container elosure system in use in the clinic.
206 Dukes

Sometimes, every lot of cliniezll supplies may be on stabaity. There


is a pedodic evaluation of data by direct comparison of the data to
.
established specifications or by statistic& evaluation (e g ,, regres -
sion analysis), to assure the continued suitabiaity of the dinicaf
supplies in the field.

IV, PROPOSED PRODUCT

The proposed product stage covers the pedod fmm selection of the
find formulation to NDA fsing ; it roughly corresponds to the pedod
during which Phase Ill clinical studies are conducted, At this
point in the development pmcess, a substantial amount of fundarnen-
tal stabgity informiation will be available. Any inherent sensitftivlities
of the active ingredient have been determined, sipificant degrada-
tion products identified, and major stabiXfty-limiting charactefistics
of the various expedmental formulations have been d e t d e d .
The objectives of this stage are (1) to assess the stabi2ity per-
hrmance of the selected formulation for the purpose of defining
storage conditions and stabiIity -limiting factors ; (2) to eonfirm deg-
radation pathways in the selected formulation; and (3) to establish
the b i t i d expiration dating periad for the contdner-closure system
jU1 which the product is to be marketed.
A greater variety of test data can be collected du&ng this phase
to establish the stabfii2-y-limiting factor(s). Both short;-'term accel-
erated studies and long-term studies at proposed label storage con-
ditions are conducted. The tests and assays to be conducted
skould be essentially the same as those given in Section 1If.G.
Many sf the tests and, assays ineluded in the stabnily protocols uti-
lized during this stage will not be included in the protocols tn later
stages because they will be deemed redundant or unnecessary, In
generd, at least three lots of the proposed formulations are placed
on test in the proposed contaker-closure systems, These lots are
usually produced in production-scale equipment f Table 5) ,
There are eases when scale t s s t h g may not be sclentific&ly
necessary. X f a stable, noncornplex flormulation. (formulation defined
as er f h e d ratio of active ingredient(s1/excipients) is being devel-
oped in several strengths or coxlcentratioians, full testhg should be
conducted on the highest and lowest strength or coneentratbn, The
data from this testing regimen wiU prodde information to fuHy char-
acterize the stabnity performance profs@of formulations at all inter-
mediate strengths. For example, the lowest concentratha would test
the case for which the package surface area per maligram of active
ingredient is at a maximum, whereas the highest concentration would
test the case for which the probabglty of nucleation and crystal
growth I mlutions) or agglomeration (suspensions) is maximized,
Stability Testing
Table 10.5 Proposed Product

Protocol elements Activities or results

Assess stability performance to define


stability-limiting factors and storage
conditions ; confirm degradation path-
ways in the market formulation; estab-
lish initial expiration-datkg period
Study design Both short-term studies at accelerated
conditions and long-term studies at
proposed labeli storage conditions
Tests Potency, sipLfEcant degradation products,
physical properties, plus hdividual
tests relevant to specific dosage forms
Storage conditions Accelerated : geared to known reactivities
More frequent testing for less stable
formulations
Samples Representative of the formulation and the
contetiner-closure systems ; the least
protective container-closure should be
included
Evaluation Direct comparison of raw data with pro-
posed standards or statistical analysis
with a determined confidence level

Another instance in which full-scale testing may not be scien-


tifically necessary involves bracketing of package sizes within a
container-closure system. (A container-closure system fs defined
as a family of packaps made with the same materials.) Again, for
a stable, noncomplex formulation packaged in several sizes, within
a given container-closure system, full testing should be conducted
on pmduct packapd in the smallest and largest sizes. (The small-
est package size provides the greatest possibnity for product-
package interaction .) Therefom, comparison of data for product
packaged in the smaUest and largest sizes (as well as that from
other container-closure systems) will provide an additional assess-
ment of the effect of the package upon the stability of the product
208 Dukes
m d will support paekag;lng af the produet in any intermediate sizes
of the container-closure system,

V. NEW PRODUCT STAGE

At the tirne of NDA, approvd, development enters the new product


stage dux"mg which the objectives shiR fmm assessment of stability
performance and establishment of the initial expiration-dating period
to confirmtian of the stabBity performance and expansion of the
stabnity data base. The new product stage is a time of transition
to the established product stage and usually lasts fmrn 1 to 2 years,
The studies initiated dufing the proposed product stage are contin-
ued and, in general, at least three marketed lots w i l l be placed on
sstabnity. These lots are placed on long-term study at label storage
conditions. Testing schedules should be selected aecordhg to the
stabaity performance of the product,
A s the data base for the product maturns, the protocol w2l
evolve such that the studies in this stage, &though generdly less
intensive than those s f the preceding stage, will be more jlntensive
than tktose of the succeeding stage. When warranted, any redsions
of the expiration dating peAad for the pmduct are supported by a
mature data base.

V I , IESTABLBSHED PRODUCT STACE

The last stage i s the established product stage, whkh continues f"or
as long as the firm manufezetures and markets the gmduct. Dudng
this stage, concern again shifts. At this point, a comprehensive
data base is in place, and the objective beeones one of pedodicany
reassessing the stabntg of' the pmduct .
Because much stabgity data has been generated on the pmduct
during the preceding stages, generally fewer data per product are
required in any @ven tirne span dufing thiss continuation of the
stability testing pmpam.
In general, one lot per year should be placed on test, t o t
selection should be designed ta ro"tte among the various marketed
cantminer-closure systems. The protocol should include tests and
assays far significant q u d t y attributes, as estabashed in predous
stages, The study desim is for full-term studies at label storage
conditions. The schedules t&e into account the relative stability
of the product as well aa the relative sensitidties of the vadous
stability parameters determined during the g r e ~ o u sstages (Table
6) *
Stability Testing 209

Table 10.6 Established Product

Protocol elements Activities or results

Objectives Confirmation of stability performance de-


termined in preceding stages
Study design Long-term studies
Tests Significant quality attributes as determined
in preceding stages
Storage conditions Consistent with Iabel storage conditions
Schedules Based on relative stabMty of the product
and the relative sensitivities of the
various stabaity parameters
Samples Representative of the product; lot selec-
tion rotated among the various mmketed
container-closure systems
Evaluation Periodic evaluations to allow timely deci-
sion -making ; evaluation by comparison
of results with the established stability
profile

Vll. PRODUCT REVISION STAGE

The expiPation dating periods of established products are assigned


or revised based upon a great deal of experience and upon a mature
data base. When pmducts are revised, however, there are many
instances in which the number of months of stabaity data are sig-
nificantly fewer than the number of months of the proposed dating,
an event that experience indicates would be entirely reasonable.
Lf the programs previously discussed were wnscientiously imple-
mented, treating revised products as entirely new products requir-
ing full, arectly suppoHive data bases would be neither advisable
nor necessary. Changes in pmducts occur in several ways and for
various reasons. In general, they may be categorized as formula-
tion changes, such as the addition or deletion of excipients; proc-
essing changes; changes in the source of active i n p d i e n t s ; or
changes in packa&g material that comes into direct contact with
the dosage form, When such changes am antidpat&, a scientific
judgment should be made c o n e e d g their probable effects on
product stability. Such judgments would be based on a sound,
21 0 Dukes

technically based understanding of the product's decomposition


pathways and physical stability properties that were developed dur-
ing the new active ingredient, formulation development, proposed
product, new product, and established product stages,
If it is determined that the changes will have no effect on sta-
bility, then no additional stability testing is necessary. Many
changes, such as a change in the supplier of a raw materid or de-
letion of &or excipients, would fall into this category. Here, sta-
bility monitoring would be the same as that for estabgshed products.
If it is judged that the change is likely to affect the stability
performance profile of the product, short-term studies under accel-
erated conditions should be carried out to compare performance of
the revised product with the original product's performance. Com-
parisons of the data obtained from these accelerated studies are the
key to rapidly determining the effect of a change on the product's
stability. The conclusions drawn should be confirmed with full-
tern studies under label storage conditions.

VIII. CQNCLUSION

The foregoing discussion reflects the current stability-testing phi-


losophy of the Upjohn Company. It has been generd in nature be-
cause the finer details of stability testing propams of different
manufacturers will almost certainly be different. Indeed, the finer
details of the stabiBty programs at the Upjohn Compmy are dynamic
(6.71. However, it is hoped that it has stimulated some thoughts
on the basic underlying concepts that should be common to all sta-
bility-testing programs.
In particular, the eancept of a progression of stability testing
with product maturation-a system maranteeing a certain continuity
of data aumentation-should be applicable to any stability pmgram
(8).

REFERENCES

1. Stability Concepts, PMA Joint QC-PDS Stability Committee,


Pharrn. Tech, , June 1984.
2, J , J , Winheuser, and T. Hfguchi, J. Pharm. Sci., 51:354, 1962.
3. A. J. Aguair, J , Krc, Jr., A . Wr. Kinket, and J. Samyn, J.
Pham. Sci,, 56:847, 1967,
4, C. T. Rhodes, Modem Pharmaceutics, (G, S. Banker, and C .
T. Rhodes, eds.), Marcel Dekker, New Uork, p . 329, 1979.
Stability Testing. 21 1

5. J, T , Carstensen : Sotid Pharmaeeutics : Mechanical Brapertr"es


and Rate Phenomena. Academic Press, New York, pp, 243-2144?
1981.
6. C. 1R. Dukes, Pharm, T e c h , , January 1982,
7. G , R . Dukes, Drug Dev. Ind. Pharm,, f0:1413, 1984.
8, 6 , H, Bfiart, Drug Dev. I n d , P h a ~ m . , 5:349, j1979.
In Vitro Testing of Topical
Pharmaceutical Formulations

RONALD C. WESTER and HOWARD I. MAIBACH University of


California School of Medfcine, San F~uncisco,California

I. INTRODUCTION

The development of dermatological and transdermd drugs requlres


knowledge of the pemutaneous permeation (local skin) and absorp-
tion (systemic) of the drug. The uItimate in both relevance and
cost is in vivo human testing. To retain relevance, but compromise
on cost, alternate test procedures are desired. One of these pro-
cedures is in vitro skin absorption using diffusion cells. A recent
report (I) attempted to define principles and practices related to
in vitro percutaneous penetration studies. Thia chapter vviU discuss
and elaborate on those procedures and principles as they apply to
drug testing in our laboratory.
We hold only one principle that relates to in vitro skin absosp-
tion studies: the system is artificial and subject to all variabilities
of its parts. Relevance must be reserved until in vivo (preferably
human) testing is done. On this principle, we build our in vitro
studies.
We also offer an altemaave study design for in vitro skin ab-
sorption drug screening that has proved valuable, especiatly for
hydrophobic compounds that will not partition into reservoir Ruid.
The information that follows is a small part of in vitm percu-
taneous absorption. The reader is referred to the fine wo~kof
Robert Bmnaugh on in vitro percutaneous absorption (2).

I!. BASIC STUDY DESIGN

Table 1 gives the basic study design that we approach ~ 5 t hin


vitro absorption studies. We believe that
21 4 Wester and Maibach

Table 11.1 In Vitm Study Design

Number of repficates
Human skin
source Formula~on1 Formulation 2

A 4 4

Total 12 12

Assay: Reservoir fluid


Skin
Skin surface wash
Apparatus wash
Material balance

I. All formulations should he compared within the same skin sam-


ples.
2, Variability exists between skin sources, and thus a minimum of
three skin sources is desired,
3. Reservoir fluid, skin, skin surface wash, and apparatus wash
should be assayed for material baknce.

Obviously, the abbreviated study design depicted in Table 1


wilI not meet all needs. However, it will avoid pitfalls of human
skin variabBty, and the material balance is a check on scientific
quality. In addition, as will be discussed later in this chapter, as-
say of chemicals within skin may be a viable alternative when salu-
bility in reservoir fluid is a problem.

Ifl. FACTORS
A, Membranes
The literature is full of reports showing that skin abmrption is dif-
ferent in most animals than in humans (Table 2; 3). Human skin
is available, so it should be the membrane of choice.
Human skin should be used as soon as possible. Obviously,
freshness is best. Reports suggest that the freezing of skin
changes fts permeability characterhtics, If skin metabolism data
are warranted, then appropriate procedures and verifications are
warranted.
In Vitro Testing of Topical FormuIations 215

Table 11.2 Absorption of Paraquat and Water Through Human


and Laboratory Skin

Paraquat
permeability rate (vglcm2)

Human (in viva)


Ruman (in vitm)

Permeabaty constant (cmlhr x 105)

Water Paraquat

Human
Rat
Hairless rat
Nude rat
Mouse
Hairless mouse
Rabbit
Guinea pig

B. Cell Design
The diffusion cells, connecting lines, and collection chambers should
be made from inert, nonreactive mateAa1. The drug can disappear
thraugh volatifity and apparatus adsorption. Material balance in
the study design is a check on this. Table 3 shows this dvferclnce
( 4 ) . If solubility in reservoir fluid is a problem, then the larger
volume from a Row system may result in relevant data (Table 4; 5).

C , Temperature
It is fortunate that cirrmlating bath water of 37OC results in skin
surface temperature of 3Z4C, just as in humans in vivo. Obviously,
temperabres should be veflied .
D, Dose
This is easily determined by taking a known volume of formulation,
~preadingjt over human skin, and measuring the skin area covered
218 Wester and Maibach

Table 11.3 In Vitro Percutaneous Abso~ptionand Skin Distribution


of Chemicals in Water Solution

Chemical (%)

Parameter Benzene 54%PCBs Nitroaniline

Percutaneous absorption
(systemic) 0.045 t 0.037 0.03 t 0.00 5.2 t 1.6
Surface boundlstratum
comeurn 0.036 f 0.005 6.8 2 1.0 0.2 + 0.17
Epidermis and dermis 0.065 2 0.057 5.5 t 0.7 0.61 t 0.25
Total (skin/systemic) 0.15 12.3 6,l
Skin wash /residua2 2.51 2 0.94 71.2 k 5.5 92.4 2 0.8
Apparatus wash 0,006 f 0.005 0.25 t 0.2 0.47 2 0.03
Total (accountabBity) 2.67 83.3 99.0

Table 11.4 In Vitro Perrmtaneous Absorption of Tridocarban in


Adult and Newborn Abdominal and Foreskin Epidermis

Doae absorbed
Type (% +- SD)

Static system, 37OC


human adult abdominal (n = 14) 0.23 f 0.15
human newborn abdominal In = 6) 0.26 -t 0.28
human infant abdominal (n = 4) 0.29 t 0.09
monkey adult abdominal (n = 8) 0.25 f 0.09
human adult foreskin (n = 4) 0.60 f 0.25
human newborn fareskin (n = 7) 2.5 t 1.6
Static system, 23OC
human adult abdomhal (n = 8) 0.13 t 0.05
Continuous Bow system, 23OC
human adult abdominal (n = 12) 6.0 2 2.0
Man in viva (n = 5 ) 7.0 2 2.8
In Vitro Testing. of Topical ForrnuEatians 21 7

by the f"ormulal;ion CspreadabIIfty). Generafly, I to 5 v l of formula-


tion will cover 1-cm2 skin s u r h c e area. The drug mass within that
amount will be determhed by the concentration of d w g within the
formulation.

E, Open Versus Occlubd


Ths use of the product under normd ~ i r c u r n s t ~ n cwill
e ~ determine
its application.

F". Finite Versus Infinite Dose


For dernatolo@e& products, a dose relevant to actuat use will usu-
&By suggest a finite dose. A delF?livery system that erngfays constant
delivery may justify infinite dose, Common sense says to follow
haw the produet will be used.

C , Equilibration
Phy siolo@cam (except for. some psoriasis treatment) a person does
not sit submerged in a tub before appXHng a dermatofo@cal or
transdermcsl d m g , Therefore, it seems irrational to use eguaibra-
tion beyond 30 min to determine if the system is working (bubbles).
Because human skin tends to faXl apart after 24 to 48 Izr in an in
d t m rmn, adding more water time generally seems irrelevant.

The re1evmt in viva d m g use should determhe in vttm run time.


Most absorption is determined over a 24 ta 48 h r pedod because of
the tendency of human s k b to fall apart afier this,

I. Receptor Fluid
The pmdous published recammendation is as folXaws ( I ) : Far most
studies, an isotonic mlution buffered to pH 1.4 is a suitabXe and
preferred receptor f"luid; a different pH can be used if it can be
justgIed, fn all Instances, the thermodpamic actidty of dmg in
the receptor Buid should not exceed 10boE its thermodynamrie ae-
t i d t y in the donor medium to maintzlin, a favorable drigr.ing k r c e for
permeation and to assure reasonable and efficient colfection of per-
meant. The receptor medhm may need alteration h r n a str;Ictly
aqueous medium to attdn this endpoint for hydrophobic compounds ,
This factor supercedes coneem for rnahtdning a minim& receiver
volume,
Wester and Maibach

Hydrophobic compounds may be defined as compounds that have


uniquely low solubilities in water (less than 10 mgfL), and solubili-
ties In both water-miscibIe (alcohol, propylene glycol) and water-
immiscible (ether, octanol , chloroform, hexane) solvents that are
orders of m a ~ i t u d elarger. These compounds have large octanoll
water partition caefficients that are about 103 or higher. Such
compounds have a low tendency to partition into the receiver medium
beneath the skin. It may be neeessaw to use a nonphysiolo@caf
medium in which the drug is more soluble to efficiently elute such
substances. When this is a concern, studies should be performed
either by using a lipophilic receptor fluid that is without effect on
the skin membrane, as demonstrated experimentally, or by using an
isotonic solution containing an appropriate concentration of solubil-
izer for the hydmphobic compound. Flow-through cell designs also
tend to minimize collection problems.
The practical problem is with hydrophobic compounds that will
not partition into water receptor fluid. The afternative is to go to
a nonwater medium. This creates a nonphysialogical situation. The
data can then reflect an extraction of compaund from skin, possibly
with damage to the skin membrane. A pseudoextraction system
without in viva human verification does not appear to be a strong
position.

J. Alternative Approach for Drug Research


Historic& precedent states that with the in vitro system, drug con-
tent is assayed (only) in reservoir fIuid, and these data are sub-
jected to vigorous mathematical treatment for lag time and permea-
bility determinations. Drug development may not be so interested
in vigorous lag time as they might be interested in which formula-
tion allows a drug to penetrate skin. We offer this alternative ap-
proach,
Table 5 gives the in vitro absorption screening in human skin
for spironolactone topical formulations. Spironolactone will not par-
tition into the water-based reservoir fluid. Therefore, formulations
were screened for spimnolactone content in skin, and it was the
skin content that was used to determine differences in formulations.
With appmpdate control and total recovery of material balance, an
effective screening system can be instituted.
We believe in in vivo veridcation, and the data in Table 6 show
the in vitro absorption in human skin and in vivo verification in
rhesus monkey and humans for the same formulation. Comparison
of human skin eontent in the in vitm system (with material balance)
with in vivo percutaneous absorption in rhesus monkey and humans
(by the standard urinary 1% excretion method) shows excellent
correlation. The in vitm system is verified.
In Vitro Testing o f Topical Formulations 21 9

Table 11.5 Spironolactone Formulation Screening

Spironolactone content ( %)

Reservoir Surface
Formulatbn fluid Skin wash Total rwovery

3 0.0 3.5 96.1 99.6


4 0.0 3.1 93.0 96.1

%ontrol = 0-hr application time (system veriacation); 1-4 = 48-hr


application time.
WLPC assay for spironolactone only,

Table 11.6 Percutaneous Absorption of Spironolactone

Total percent dose (n = 11)

En Vitm Receptor Surface


(human skin) Skin Ruid wash Total

Rhesus monkey 2.9 + 3.2 (24-hr skin application time)


Man 0.5 t 0.4 (8-hr skin application time)
220 Wester and Nnibach

Xn dtm, percutaneous absorption is best done with human skin, The


skin should be used as soan as possible, and stared in the re-
ffigeratar no longer than 7 to 10 clays.
In. ~ t r penetration
a into skin @ves results suitble to distinmish,
among d m g f"ormulations, especially when the dmxg will not par-
tition into reservoir f u i d ,
Natedd bbalance in an in v-itro study design adds to the overall
data present ation.
In viva vefidcation of skin absorption, preferably in humans, adds
relevance to the in vitro data.

REFERENCES

1. J , P . Skelly, V. P. Shah, If. 1 . Maibaeh, R. H. Guy, R . C .


Wester, G , FXynn, and A . J18cabill Pharm, Res., 4 ~ 2 6 5 , 1987.
2, R. L , Bronerugh, in Percutaneous Absorption, (Re Bronaugh,
and H. Mdbaeh, eds,) , Marcel Dekker, New York, p 267, I

1985.
.
3, R , C Wester, and H , I , Rlaibaeh , in PereuEaneous Absarp tfan
( R I Bronazxgh, and W. Maibach, eds.). Marcel Dekker, Hew
York, p. 251, 1985.
4 , R , C 1 Wester, M, Mobayen, and H. 1 , Mdbzich, J. Toxicol,
Envlron, Health, 2 1 ; 3 8 7 , 1987.
5, R. G , Wester, W . I. Mdbaeh, J. Surinchak, and X f , A , W .
Bucks, in Pewutanclous Absorption. t R . Branaugh, and H .
Maibacfi, eds,). Marcel Dekker, Mew Uork, p. 223, 1985.
Drug Delivery from Topical Formulations:
heo ore tical Prediction-and Experimental Assessment
WILLIAM J , ADDICKS* and NORMAN D, WEIEaER University of'
Michigan College o f Pharmacy, A n n Arbor, Miclcli~n
RANE L. CURL University of Michigan College of Engineering, Ann
Alnbor, Michigan
GORDON L, FLY FSN C ygnus Research Corporation, Redwood City,
Calr'firnia

I. INTRODUCTION

D m g delivery f r o m topical. formu1ations for local or systemic e;Efects


essentiafly InvoIves passive diffusion of the drug through the skin.
The release of a drug molecule from a vehicle into the skin and dif-
fusion across the skin are controlled by physieochemicd factors
sensitilre to the molecular groperlCies of the permeant, the vehicle,
and the skin (1). Katz and Poulsen C 2) have identified four cate-
gories of interactions with bearing on the delivery pmcess, R~tte-
infiueneing interactions are possible between, the (1) drug and skin,
( 2 ) vehicle and skin, (3) drug and vehicle, and (4) d m g , vehfcle,
-
and skin Examples af dm g--~ k interactions
h inelude dteration of
the surface structure o f the skin by the drug components of formufa-
..
tions ( e g , hydration. of the stratum corneum by sodium pyrrsli-
dine carboxylate) as well as the binding of drugs to canstituents of
the skin as they diffuse throueTfi the tissue field. A vekride-skin
interaction 86cur8 when one or another of the v&iele% main corn-
ponents, here m e a ~ n gthose components that psn,eip&ly determine
the deliveq matrix: of the formula, effects a change in the physic&
state of the s k h , in turn, affecting the skin" pe~meab%ity, Ve-
hide-skin interactions can be dirridecf into three categories : ( 1)
penetration enhancement effects brou @kt about by direct solvent exc -
tion on the skin, (21 the vehicle" general hfluenca on the state of
"

"Current affiliation: E , I , du Pont de Nemours and Company,


Wlilmingtan , Delaware
the skin's hydration, and (3) vehicle effects on the locat vaseula-
lure, which alter d m g clearance and skin temper&ure, in either
case, subtly hfluencing permeabSity, Drug-vehide interactions
are those in which physieochemical ;interactions between the drug
and the vehicle kineticalfy or thermodynarnicalXy govern the release
of the d m g into the skin. Such interations can become the rate-
controlzng factors and be clinically highly important when the stra-
turn comeurn is impaired as the consequence of disease or i n j u v ,
for, when the skin is damaged, solution and diffusion in the vehicle
may be relatively slow relative to skin permeation. Here, one is
concerned about the solubgity of the drug and its diffixsive mobility
within the vehiele, as each is a factor infiuendng the rate of pre-
sentation of the drug to the vehicle-skin interface* FOP the more
usual case, h which the permeability of the intact stratum comeurn
is tow, partitioning of the d m g into the skin can still exert a p m -
found, if not dominant, influence on the rrate of delivesy.,
As one c ~ t i c d l yredews the literature dealing with vehide ef-
fects in transdermral d m g delfvexuy, a confused state o f affairs
about the theom of topicd d m g delivery is apparent. fn a redew
of the subject, Xclsan (3) has cited sever& reasons for this eonfu-
sion: (1) Exper;Iments have been performed on a large variety of
anirnitfs; (2) cfiffermt methods have been used to estimate skin pen-
etration; ( 3 ) there has been a lack of awareness of possible dmg-
vehicle hteractioms; and (4) there has been a lack of eonsider8tion
of thermadynamic factors in the interpretation of results, Although
aU. the possible types of interaedions between d m g , vehicle, and
skin mentioned here are important to topical bioarraSab2ily, this
pager mainly focuses on dmg-vehicle and dmg-vehicle-skin (mern-
brane) interactions.

I!, IN VITRQ RELEASE FROM VEHICLES


A. Theoretfcat Background
A s part of the assessment of a .vehicllefs potentid for deBvering a
d m g to the skin far topic& absorption, researchers have long
sought method8 to determine a drmg's release rate f'rom its vehicle,
f t is generally assumed that the results obtained in such experi-
rnents at leaat quagtatively relate to the release of the d m g to the
skin in a clinic& situatisn, FaDure to properly interpret and in-
tegrate results from such studies could lead to a substandard d m g
pmduct .Thus, by necessity knowledge of the thermodynamics and;
kinetics involved fn the in vitro situation is imporPt.ant if one wishes
to suecrtssfully pmject the results to the in vfvo circumstance (4).
There are sever& identifiable pmcesaes that, theoretically, can
establsh the rate at which a dmxg is deavemd from a film into an
absorbhg membrane, Minimally, these include diffusion thraugh
Delivery from Topical Fornzulations 223

the vehicle to its interface with the skin and diffusion across the
outer skin. Even in the absence of specific dmg-vehicle-skin in-
teractions, there are special circumstances in which dissolution of
drug suspended in the vehicle o r diffusion through the inner living
strata of the skin may became rate-determining.
A theoretical basis for the study of the release kinetics of
drugs from both suspension and solution ointments for the case in
which release from the vehicle is rate-limiting was established by
Wiguchi (5). Wiguchi first depicted the situation in which the oint-
ment vehicle is initidly saturated with solute, with excess solute
uniformly suspended a s tiny particles, The exact assumptions for
the derivation of the time dependency of release are a s follows:

1, The particles are present in a fine enough state so that dissolu-


tion of the particles is not rate-limiting.
2. Q (the total concentration (mass/volume) of dissolved and undis-
solved drug) is much v e a t e r than C s (the solubility (mass/volume)
of the drug in the ointment).
3. A sink condition prevails at the ointment-receiver phase inter-
face.
4. Release occurs through a planar surface.
5. There is no significant boundary layer adjacent to the ointment
(assumed implicitly).
6. Quasi-steady-state diffusion exists between the dissolution in-
terface at the edge of the particle field and the interface with
the sink.
7. Although not explicitly stated, the model is semi-infinite, as in
the ori@nal derivatian no limit was placed on how far the
boundary could recede.

Assumption (2) is fn effect a directive that dissolution takes place


rapidly at the edge of the receding particle field in the course of
drug release, Thus, unit activity (activity of the crystalline state)
is continuously experienced along this front, and, with the passage
of time, a well demarcated zone lying between the releasing surfaee
and the suspension phase develops that is spanned by a linear
concentration gradient given by the solubility divided by the thick-
ness af the layer, that i s , (Cs/h). The release of drug through
the surface is controlled by the momentary steepness of this gra-
dient. A s time progresses, the distance between the edge of the
saturated suspension phase within the ointment and the ointment's
sink-side surface widens, resulting in a decrease in the rate of
drug delivery. The following equation describing the release of
solute was derived (51,
where D is the diffusidty (cmzlsec) of the drug in the ointment.
After differentiation for time, an expression for the instantaneous
rate of release is obtdned,

When Q > > C g , the amount of drug released into a sink bears the
following relationship to time :

and the rate becomes

Equation 3 predicts that a plot of the amount of"drug released (per


unit area) versus the square of time should be linear, whereas
E q , 4 predicts that the rate of d m g release fs groportiond to the
reciproed of the square root of time.
Hiwehi dso deduced a relationship characterizing the release af
d m g fmm an ointment with i t s d r u g totally in soXut5on (""soluti.on
ointment"") ,so &om a planar surface directly into a diffusion&
sink (fi), This equation i s a sczlution to F.ickts second law, and can.
be found in severd standard texts coneemfr-lg dlffasion pmcesses
( 1 . 8 ) - A s in the predous case, uptake into a sink is assumed,
with diffusion to the relea~inginterface being the rate-limiting step
in the overall process, The folXowing mathematied descsiption of
the pmeess was presented :

In this expression, h i s the thickness of the ointment phase and 60


is the initid b m g eoncentration in the ointment, The following
simpXiBect equation closely describes diffusion for the first 30% of
release (9):
Drug DeEiveq from Topical Formulations 225

The solutions to the 8olution release ease embodied in Eqs, 5 and 6


assume a semi-infinite geometry for the ointment phase, A s in the
suspension case, it is once again! predicted that the amount of drug
released wilf be proportland to the square root of time, but in this
instance, as a practical matter, this is so only up to 30% of the
total release,
Paul and McSpaciden (102 have pointed out that although the
Hiwchi approximation works well in Instances in which the totd
concentration of d m g in the oi~tmentis much larger than the soXu-
bBity of the d m g (Q > > CB), appreciable departures from actual,
theoreticd behador occur in the limit Q -+ CB. In this case, Eq, 1
reduces to

In the. limit of $ + G s , Eq. 7 predicts 11,3% less d m $ will be re-


leased than does Eq. 6. Paul and McSpadden have eliminated this
discrepancy by dedving a solution, for this ease that can be applied
over the complete spectmm of drug concentrations, Sneluding the
suspension sit~s1tions, f n this dedvation , boundav eond&ions have
been applied that remove the pseudo-steady - state appmximation of
.
Eq 2 , In additjon, a semi-innnite geometry is assumed. The re-
sulting relatknship for the mass relea~edat a @ven tfme i s

In t h b equation, erf(@)i s defined by the fdlowfrrg transcendent&


expression :

A s in Eqs, 1 and 6, E q . 8 indicates that release should follow a


square r m t of time dependence. However, the dependence on Q in
Eq, 8 is different t h m in the p r e ~ o u streatments. To BXustrate
the p r o p e ~ i e sof this exact wlution, Paul and Mc8padden have ex-
anhed some limiting cases. When Q -+ C s , Eq. 9 indicates that @
becomes very large md, thus, erfC &) approaches unity. A s a re-
sult, Eq, 8 reduces to
This equation is the same as E q , 6, for which aX1 of the & m g was
assumed to be in solution. For the case of Q > > Gs, Eq. 9 pre-
dicts that f3 is smdl. A sefies expansion of" Eq. 9 permits an ex-
pB&t c&cula~onof erf( @) as follows:

Upon combining this with Eq. 8, the following result is obtdned:

Equation 12 i s identicat in form e t h E q , 1 except that Cs appears


.
in lieu of Cs12, which i s a r e ~ u l tof the limit pmcess Paul and
MeSgadden point out that no relevant difference e d s t ~in the the-
oretied treatments when Q > > C8. In effect, they eanclude that
the pseudo-s"tacZy-st~te assumption that resulted in Eq. 1 is ade-
quate when Q > > C , , To test the u t ~ i t yof IEq, 8, Paul and
McSpadden (10) performed a series of c3xpedments for which the re-
lease rates of an organic dye, diffusing out of a snicone pdyner
and into an acetone sink, were measured over a wide range of' s d -
ute loadings ( Q I C s .(: I to $ ICs > 1). The theoretical predktions
were experimentdly borne out over the entire range of solute load-
ings.
Yet another appmach ta the problem of dmg release f m m a
nonerodible solution-suapensio~ matdx has been developed by- Lee
(11). This approach involves the application of a refined heat bal-
ance htegral method to the moving boundaq pmblem first solved
by Hiwehi and then more rrigorouslsly addressed by Paul and McSpad-
den. A s before, Lee assumed that diffusion occurs from a planar
surface, and that a sink condition edsted at the mat&-receiver
interface. The result of Gee's deAvation is

where H Is defined as

Like the Himchi treatment, E q . 13 is an approdmate solution to the


problem, Lee compared the predictions of Eq, 13 and the Hiwehi
Drug Delivery fk.om Topical Formulations 22'9

solution f E q , 1) ag&nst the exact mlution of Paul and McSpadden


(Eq, 8) over a wide range of Q/Cs vafues (11). The dedaaons of
Lee% solution from the exact solution were condstently one order
of mapitude smder than the devriations obttlined upon using
Hiwchl" eequatian, Thus, it was concluded that the Lee equation
could be applied safely over a wider range of $/Cs ratios.
The models discussed thus far have descsbed d m g release h n n
topic& formulations directly into a sink. Although these models are
useful in terms of their abnity to characterize the diffusional be-
h a ~ o of
r a drug within a formulation, they do not d e s c ~ b ethe re-
alistic situation in whkh 8 d m g diffuses from the vehicle, partitians
into, and subsequently diffuses across a resistant membrane f 2 .e .,
the skin). To treat situations such as these, Paul and McSpaddm
&so derived 8n exact solution for dmg release from a suspension
matAx when a finite mass transfer resistance is encountered be-
tween the edge of the releasing matfix and the sink (10). The fol-
lowing equations to describe this case were put forth:

and

Here, w i s the mass transfer cmfEefent, and, erfC 8) t&es the iden-
tical form as in E q . 9. Notably, E q s , 15 and 16 are asymptotic re-
lationships that are only applicable st long time. The term & is
the finite intercept on the tff- axis.
Rosernm and Ripchi (12) also developed equations for the sit-
uation in whick finite mass transfer resistance is present; here, in
the form of 8 hydrodynamic bboundam layer, Xn this model, release
fmm a suspension m a t ~ xdepends upon two distinct processes.
The first of these invoIves an initid zero-order phase of release in
which the d m g partitions f r o m the matm";uinto the botxndaq layer.
During this period, diffusion through the boundary layer deter-
nines the o v e r d release rate, The duration of this p e ~ o ddepends
mostly upon the drug's external pbaselmatrix partition coefficient
as diffusion in the liquid fam at the surface af the mat* is pre-
dictably fast and should be reasonably simgar from liquid to liquid,
Xn a more generd ease in which the film is any type of m a t e ~ d ,
the duration would be comparably dependent on the perxneantfs dif-
fusidty through it. Regardless and eventudly, a trmsithn pedod
is reached at which the matrix g r a d u a y assumes control of the
process, ultimately with tho re-fease rate being proportiond to the
square r m t of time, Xn this model, it is assumed that $ > > Gs,
where & is once agdn the total drug eoncentration and Cs is the
drug's solubQity in the matrix. The expression for the quantity of
drug released within an area A is

where krm is the thickness of the zone of depletion within the matdx.
This quantity is related to time through. the fallowing expression:

In this equation, Dm is the drug" diffusion coefficient in the matrix,


wis the mass transfer eoef6cient, and K is the external phaselmatrix
partition coeffident, (In the original paper, Roseman and Higuehi
attributed the mass transfer resistance to an aqueous boundary- layer
and thus assumed w to be Dblfhbl. Clearly, axlather value could be
@ven to this resistance, justifying the generalizations made in the
text .) Two limiting cases were presented, During the initial, usually
aeeting, period the receding boundary- within the matrix, character-
ized by hm, is sufficientfy thin that

In this drerrrmstance the amount of d m g released is

This equation disclose& that the release of the dmg is djrectly pro-
portiond to the drug" ssdubsty in. the matfix and the mass trans-
fer coefficient, Zero-order release is predicted, as the excess o f
solid d m g ensures the mdntenance of a saturated solution at the
releasing surface, When differentiated for time, an expression for
the instantanwus rate of release is obtdnerl,
Drug D e Z i v e ~fmrn Topical Formulations

However, as tirne elapses, the receding boundav thickens, with a


commensurate increase in its resistance, Eventudly, the f o l l o ~ n g
condition is met:

and, as a result, Eq. 19 simplifies to

which will be recopized as the Hipchi equation presented earlier


(Eq, 3). The shiR to square root of time b e h a ~ o rindicates the
release kinetics have come under matdx contml, Althcnzgh Roseman
and Hipchi applied these equations to a situation in which a d m g
was suspended in a sllicone rubber m a t h and released into an
aqueous sink, it has hem pointed out (4) that they could eonceiv-
ably be applied to other situations, such as d m g release from an
ointment across a didysis membrane or weeping wound, the latter
as might occur when a tclpied drug is spread over badly darnaged
skin.
An eltlborate solution. to the problem of release of a d m g from
a suspension ointment thmugh ii resistant membrme has been prei-
sented by Ayres and Zindstmm. C 132. In their model, terms are
even induded to account for a finite sate of dissolution of any sus-
pended d m g pmtides, NlatfiernatietaX solutions far the possible situ-
ations were derived with the aid of Laplace transformation teeh-
niqzles. The arrived at equations desedbe the loss of a drug from
an ointment into the blaods-tream as a function of tirne under many
scenados. A s such, the Ayres m d Lindstrom equations represent
the most d p r o u ~theoretical characterization of d m g delivery from
thin -mm applications into the skin yet attempted, Unfortunately
and unavoidably, the resulting equations are unwieldy. When in -
terpretetrt by example, one does get an intuitive feel for the release
and permeation processes. However, nurnedcaf analysis is required
to implement this method when actually analyzing data and, at least
up to the present tirne, there me na real data precise enough to take
advantage of the power of the theory,
Guy and Hadgraft (1.4) have also evduated the theoretical re-
lease from ointments, gaHicularly focusing on delivery o f the drugs
as a function of the thickness of the applied film. They provlide
theoretied curves in which optimum dosage redmens nay be
8chieved. Another treatment, admittedly a much more approximate
soluthn to the problem of d m g permeation of the skin fmm a sus-
pension ointment, was presented by Rynn et al. (15). This aolu-
tion is an adaptation of a standard, solution to Fick's second law for
diffusion through. a resistant membrane under conditions such that
drug is mdntdned at a high, steady concentration on one side of
the membrane and diffuses h t o a sink that is maintained at the
other side, These seem to be reasonable asserclcions covedng many
r e d instances, as the skin's stratum comeum is a thin, but higMy
resisttnnt, membrane and the body of the patient to which the topi-
cal system is apgied is overvuhelmi~glylarge relative to the volume
of the applied fdm. The equation in terms af the fraction af tot&
drug initidly present from a system in whkh the d m g is finely
suspended is

where ZC is the stratum corneumlvehicle partfiion coefficient and &


i s the inititif loading of d m g in the vehicle. Other terms are as
defined in predous derivations, with the s u b s c ~ p t ssc and v re-
ferdng to the stratum corneum and vehicle, respectively. An im-
pliei"tassum1ption is that the d m g remajms diffu~tonallywell mixed
within the vehicle phase, even as it is permeating through the skin,
because the later process is slow. Equation 23 correctly indicates
that. the fraction& d m g delivery through the skin is a function of
the reciprocaf of the product of Q and h,, which &ves the total
d m g content in the application (XM-) , It also suggests that as long
as one has a suspen~ion,and pmdding the vehicle has no influence
on the skin membrane, the release rates from different vehicles wifl
be equall, because the product of K: and C s will be constant (1 15).
The situations discussed to this point have mostly dealt with
d m g release fmm topical fisrmulatians in which the dmg is present
as a suspension, For such situations, as long as the skin acts as
a highly resistant membrane and there is residual suspended drug,
an invariant d m g concentration (thermodynamic aetidtg) wfthh the
formulation is expected, D m g diffusion through. the skin from a
topical solution can be expected to be more complex, as the eoncen-
trfllion (actidty) af the drug decreases continuously irrespective of
whether the solution vehicle is funetiondly well mixed or is marked
with steep gradients, It must be redized that the SmicaX applica-
tion thickness of a topic& product is on the order of 20 um (13).
Diffusion of even a smdl amount of d m g from such 8 thin applica-
tion resrxlts in a rapidly deereasing drug concentration in the donor
phase, resulting in non-steady-state kinetics. There is scant in-
formation in the literature &out topical. dmng delivery under these
circumstances, Flymn et al, (115) have deduced a solution to the
problem ~ " f i f f u s i a nfrom a sdution ointment through a r e s i s t ~ n t
membrane into a sink. The boundary conditions are as follows:

The initial conditions to the prablem are

where x = O is defined as the membrane-sink interface, hw and hsc


are the thicknesses of the vehicze and stratum corneum (membrane),
respectively, and DSG is the dmg" diffusivlity within the stratum
corneum. It is assumed that the d m g is dfffusioxsdy weu mixed
within the vehicle. The following concentration-distance-time re-
lationship was deduced:

where 6, are the r m t s of Bnhsctan(FSnhscI = IChsc/hv. This equa-


tion is an ad~ptatianof the solution to an analogous heat transfer
pmblem (16). Computer simulations have been presented that show
the affects of waving the parameters in the equation [ 15). The
ratio K/hsG was shown to affect fraction& drug release with time,
232 Addicks et a!.

Increasing K , with d l other parameters kept constant, increases


the absolute rate of delivery, whereas increasing the thickness of
application slows the retative clearance of the d m g from the fam.
(Relative clearance is in terms of the fraction of the total d m g
present. This should not be confused with the absolute rate of re-
lease, which is always equal. to, if not greater than, that of a thin-
ner f ~ m , ) Experimental verification of some of the basic dependen-
cies has been proaded by Addicks and eo-workers (17).

f 11. APPLICATiONS OF THEORY: IN V1TRO


RELEASE STUDIES
A, Experimental Designs
.
According to Gemmen and Morrison ( 2 8 ) , ,.in vitro methods may
be of limited predictive value but they are the means o f assessing
the ability of a vehicle sr base to liberate medicament under the
.
conditions o f the test This statement vem accurately descfibes
the usefulness of in vitro release studies, In 6 t r o studies have in-
vaAably involved release from applications that are much thicker
than those seen clinically, Additionally, the chemicd nature of the
receptor phase used in such studies oRen bears m resemblance to
the skin, and is usually confipred as a thick slab. QveralX, the
conditions are such that the release kinetics reflect f"c3atures of dif-
fusion from a semi-infinite medium into a semi-infinite medium. If
the receptor phase is rate-controlling, and if this phase bears little
chemical resentblanee to the skin, then it is 1Foolishly unproductive
to m&e extrapolations to in viva situations (4).
Among the first experiments perfomed on the release of d m g s
from ointments were those developed 60 years ago by Reddish (19,
20). The prwedure involved the measurement af the antiseptic
properties of ointments contdning various drugs. The experiments
were carAed out by agp-lying ointment to agar that had been inocu-
.
lated, with Stapfxylococcus aureus After incubation, the agar plates
were assessed ta determine the presence of a zone of inhfiitian.
Thus, a cmde measure of the ointment" abBity ta liberate a drug
was achieved, A variation of this method was intraduced by Clark
and Dades (21) in which the minimum time required for an ointment
to cause inkribition. of bactedal growth was used as an index af
ointment efficacy.
The practice aP gau@ng the bioavdabnity of a drug fmm an
ointment from the dmgts ability to diffuse from the ointment into
agar was extended by Lwkie and Sgrowls (22). The d m g s used
in their experiments were sulfathiazote and iodine, both ctf whkh
can be detected c a l o ~ m e t ~ c ~ lZ"yhe
. procedure consisted of placing
a tube f2led with agar (cont dning an appropriate colo~metficindi-
Drzlg Delivery ft-om Topical Farmulations 233

cator) in contact with a tube fiXled with a drug-contdning ointment,


The diffusion of the drug (as eddenced by a color change in the
agar) was monitored, and the results were plotted as diffusion dis-
tance versus time, This method was later used by Bnlups and
Sager (23) to measure the antiseptic prope&ies of various ointments,
In these experiments, diffusion of drugs from ointments into bacte-
ridly inoculated agar was assessed by the measurement of the length
of a resulting zone of inhibition.
Ths method that has been used most *frequently to assess the
release of a drzag from a vehicle involves the release of the drug
from the vehicle dirwtly into a stirred solvent. Most oRen the re-
ceptor solvent is present in a volume that allows sink conditions to
be apprccrrimated throughout the course of an experiment. In this
confiwration, the vehicle slab is in direct contact with the receptor
phase and, therefore, a solvent is chosen that does not dissolve
the vehicle. This system was introduced by Poulsen et al. (24) to
study the release of Buocinolone acetorride and its acetate ester from
various prwylene glycol -water gels into isopropyl mgll.istate, With
use of simsar expefirnental designs, this group of researchers have
systematicagy investigated the release of other topical steroids from
vasous gelled solvent sy sterns ( 25,26) , Rank -order correlation be-
tween in viva physiola@cal responses and in d t r o release for these
dmgs was shown. The choice of isopropyl myristate as a receptor
phase has been @ven additional cred.ibility through experiments by
Seheuplein (273, who showed that topical co&icosteroids are trans-
ported through the skin by way of a lipoidal pathway, Other in-
vestigators have utaized systems In which the vehicle is in dire&
contact with an aqueous receptor phase ( 2 8 , 2 9 ) , In an interesting
modification of Potrlsenrs method, Busse et a1, (30) deTrised a three-
layer system that consisted sf a vehicle phase, an alcohol-water
phase, and a chloroform phase. The afcahol-water phase was in-
tended to represent the skin, and the eIzlaroform phase acted as a
sink. The release of b&amethasone 17- valerate from two different
ointment systems was studied with this experimentd c o n f i ~ r a t i o n ,
When performing studies i n v o l ~ n grelease from a topical -formu-
lation into a sdvent receptor phase, it is preferable to reduce the
probabaity that cross-eontarnination. of the two phases will occur.
Jurist (31) introduced a system for the study of the release of wa-
ter-soluble materials from ointment bases. We d e s e ~ b e da dialysis
cell method for measuring the Ion-exchange capacity of a resin in-
corporated into an ointment base. The method invdves the meas-
urement of Ion exchange through a semipermeable membrane, Muti-
mer et all, utilized this system for measufing the dm$-release char-
actedstics of severat ointment bases ( 3 2 ) . In their study, an ion-
exchange r e s h was incorporated into etleh base, whereupon the
formulation% aabiirity to release dmgs was judged by its abillity to
deliver hydrogen ions ta an dkdine solution,
On occasion, a dirslysis membrane has been used to separate the
donor from the receiver phase. Wood and associates ( 33) introduced
a diffusion cell for the study of the transport of a drug within a
vehicle. The cell consisted of dmor and receptor compartment sep-
arated by a cellophane membrane. The donor phase contdned d m g
incorporated into the vehicle components, whereas the receptor
phase contained only the components of the vehicle. Both phases
were stirred, and the cell was kept at a constant temperature, The
major drawback inherent in this expe~mentaleonfipration was that
it dlowed for a one-point analysis only, which did not allow the ex-
tractian of a diffusion coefficient from the data. A refined version
of this method was used by Rowze and Bnlups (34). As in the
g r e ~ o u s method,
, a cellophane membrane was used to separate the
donor and receiver phases. However, in this method, the receiver
compartment was faled with water, and samples were withdrawn pit
specifled time intervals. Thus, plots could be generated showing
the amount of drug released versus time. This procedure has been
used frequently by others as a means of assessing the relative ~ b 2 -
ities of ointment bases to release d m g s (15,32,35-- 38). A common
feature of the dialysis systems that have been used is that the
aqueous phase is usually stirred. The stirdng effectively sets a
lim-it;on the resistance of the watery re@ons wit-thin and adjacent to
the membrane, A s pointed out by Poulsen and n y n n ( 4 ) , the ef-
feet of this hydrodynamic resistance is the creEition of an initid
zero-order refease situation, the duration of which is, a finction of
the pfiysicochenticitf properties of the membrane and ointment, the
thickness of the membrane, and the rate of" stirring. The initial
zera-order release phase 3s governed by the dmgts (vehiclelwater)
partition cwffjieient , because the dmg's release rate is dependent
upon the steepness of the coneent ration gradient within the aque -
ous layers associated with the membrane during this time. A s dif-
fusion proceeds, a paint is reached at which release of the d m g
becomes praportional with the square root of time.
Alt hou glz dialysis membranes composed of cel?lruloseesters have
generrrlly been chosen for release studies, some investigators have
used other membranes, Elottari et Ztl. (39,401 and Di Colo et aX,
( 41) used silicone rubber membranes in release studjies involving
suspension gels and solution ointments. Sintsar studies have been
desedbed by others (42- 44).
A. novel technique war; employed by Tanaka et al, (45) in which
silicone rubber sheeting was used as the sole receptor phase r"or
the release of hydrocortisone butyrale propionate from v a ~ o u s
creams and gels, In this study, the effect of the evaporation af
vehicle components was investigated. When using a nonporous,
nonpolar membrane of thia type, one must be aware sf the relative
Drug Delivery fi"om Topical Formulations 235

difhsional resistance offered by the membrme and be prepared to


quantitatively account for it, This points to a fundamental Raw in
many of the studies that have employed membranes, namely that the
resistance offered by the membrane is ignored when analyzing the
results. Although the assumption that this resistance is negli@ble
may generdly be safe when the formulation is applied as a very
thick layer, nonetheless, without specifying vdues for K , Dm, and
hm, it d w s not seem possible that the experimental circumstances
under which the assumption is vdid can be assured.
Although the aforementfoned expedrnentaf designs have been
used frequently, there are sever& innate drawbacks in the desims
that should cause one to exereise caution when attempting to inter-
pret results (4). Although release of a d m g from a vehicle occurs
from thin vehicle fBms clinically, the vehicle is presented to the
receiver as a relatively thick slab in most in vitm expel*11ments, Re-
lease from such thick slabs is expected to assume an overall square
root of time dependency, which might not occur if the vehicle were
pmsent as a thin film. This problem may be present in any type
of release stultg invddng a thick ointment application, regardless
of whether or not a membrane is used. Another problem inherent
in these studies is that expedments are usudly conducted under
closed conditions, which prevents the wcurrence of compositi.ana1
changes in the vehicle that might otherwise be manifest if the for-
mutations were to be exposed to the atmosphere. In addition, the
absence of a membrane could lead to salub2izatian of vehicle eom-
ponents by the solvent present in the receptor phase, The choice
of a receptor phase itself creates problems, particularly if the phase
is purpotYted to be representative of the skin. For instance if one
is studying the refease characteristics of a steroid, isopropyl myris-
tate might, in fact, be an appropriate choice for a receptor phase.
However, far a small polar molecule, a phase with some capacity to
dissolve the drug, perhaps wilrrer Itself, might be a better choice.

B. Measurement of Diffusion Coefficients


Per"haps the first demonstration of the applicabaity of the theory
concernhg release of drugs from solution ointments was reported 'by
Hiwehi et al. ('7). Using data that had been pre~arxslygenerated
by Patel and ca-workers (481, W, Hfpehi was able to successfully
apply a simplified version of the theoretic& equation d e ~ v e dby T,
Wiguchi (Eqs, 5 and 6) to show that the kinetics of d w g release
from ointments da fdlow a square root of time pattern, The release
experiments involved the liberatiion of radialabeled iodine from hy-
dmphliXic ointments into a stirred aqueous reservoir. Visking mem-
branes were used to separate the two phases, and it was assumed
that the membrane contributed negli@bly to the reletlse resistance,
Hipchi selected a diffusion coeffident CD) that yielded. a best-fit
curve for each set of data using the following modification of E q , 6:

In this equation, 1R, is the percentage of drug released, and h is


the thickness of the ointment slab, The application of these t h o -
reticd equations to experimental data as a means of extracting dif-
fusion coefficients from the data was significantly advanced by
Spang-Brunner and Speiser (37). With the use of a diatysis cell
method employ-ing a cellulose ester membrane, the diffusion coeffi-
cients of resorcinol in various hydrogels were found to be on the
order of 2 . 3 to 4.1 x 10-6 cnn2s-1. Diffusion from ointments was
found to be 50 to 300 timers slower than diffusion from hydrogels.
Xn yet another study, Koizumi and Hipchi (47) applied E q . 25 to
the measurement of the diffusion coefficient of py ddine from water--
oil emulsions.
Subsets of the theoretical equations have been employed by Bot-
.
t a d et d , ( 4O) and fli Cdo et al ( 39,4 1) in studies invol.\ting the
release of drugs from gels and ointments across siricone rubber
membranes, Whife investigating the release of sdieylic acid a s a
functhn of cancentration from emulsions ( 40) , Eq , 6 was used to
cdeulate diffusion coefficients in the emulsions, B y plotting re-
lease rates against drug concentration for each ointment, the au-
thors attempted to determine d m g solubility by noting the concen-
tration at which a deTfiation from linearity occurred. Reasonable
agreement was claimed between these vdues and the independently
obtained solubBities as determined visudly. Through the use of
Eq. 6, it was assumed that the membrane offered neglidble resist-
ance to diffusion of the saficylic acid. Rowever, an appredable lag
time was observed in the mass versus square-
thus indicating the possibiIity of membrme involvement in the re-
lease process dudng early times, Similar studies in which benzo-
c a b e was suspended in aqueous gels were performed by the same
group (39). However, here, the membrane was assumed to o f h r
resistance to diffusion, and Roseman and Hipchi's vehicle---bound*
ry diffusion layer model was employed. In this instance, excellent
apeement was found between the experimentaX data and theow.
WahXgren et al, ($81, in an interesting application of a standard.
solution ta E"ickrs ssecnd law (71, cdculated thet diffusion coefficient
of hydmcortisone in a lecithin- water lyotropie Xiquid crystafBne
solution. In this met hod , the diffusion coefficient was extracted
Drug Delivery from Topical Pormulatlons 237

from concentration ldist ance pmfges. A glass cylinder containing a


hydrocortisone-liquid cryst& solution was brought into contact with
a cylinder of plain liquid crystal, Diffusion was &lowed to proceed
for several days, and the cylinde~swere then separated. The liq-
uid crystal was pressed from the cylinders, scrszped off, and
weighed, Each. liquid crystd segment was assayed for hydracorti-
sone, and the data obtained was used to calculate the diffusion co-
efficient. The faxlowing equation was employed in the calculations:

where x is the distance along the diffusion =is, t is time, C i s the


concentration of drug at x , and Cg is the initid concentration of
drug in the gel.

IV, DIFFUSION STUDIES lNVQLVlNC SKIN


A, In Vitro Studies
A frequently employed technique to test the relative permeabjifity of
topied dmgs involves the in d t r o use of excised skin mounted in
diffusion chambers. Historicafly, the method used most often is one
in which skin is mounted as a barrier between two stirred, fluid-
filled chambers. The donor chamber is filled with a d m g solutbn
with samples withdrawn fmm the receptor as a function of time,
TypicdX-y, a steady -state or quasi-steady-state esndition arises in
this diffusional situation, because generally there is an linappredable
alteration of the concentration differential existing between donor
and receiver chambers o m r the course of the expePlirnent. This
method has been used frequently to systematically study mechanisms
of skin permeabiility (49-52). PouXsen e t ale utnized inf'mite dosing
.
in the development of severd topical formulations (25,263,531 One
must continually *member, however, that this type of diffusion cell
expeAment has tittle in common with the delivery of drugs from
topic& appjieations and i s used to get mass transfer coefficients for
mernbrmes.
To more adequately depict true clinical. situations, so-ca2led A-
nile-dose diffusion eetls have been used ( 54,55- 57) . A membrane,
typicdly skin, i s mounted ve&ically in the diffusion cell, a small
amount of formulation is applied to the membrane, and samples are
withdrawn from a stirred receiver compartment beneath the mem-
brane (skin) as a function of time. An advantage of this method is
that, in principle, the amounts of formulated d m g applled can rea-
sonably reflect the usage situation. In addition, the cell can be
left open to the ambient conditions of the laboratosry for the formu-
lation to undergo the same compasitional changes as occur dinically.
Another advantage to the finite dose method is that phenomena of
topicd d m g delivery, such as skin penetration enhancement, can
be redistically evauated .
To test the usefuliness of the finite-dose difhsion cell, Franz
C54,55) studied the in vitm diffusion of v a ~ o u scompounds from
thin solid deposits through human skin, Data obtained were eom-
pared with previously obtained in vivo data, Good aveement was
reported between the two sets of data. MoXlgaard and Hoelgaard ,
with the use of finite-dose diffusion cells, bvestigated the diffusion
of estradiol h r n 21 different vehicles through excised human ah-
dominal skin (58). It was cmcluded that, for the vehicles chosen,
drug permeability was mast influenced by the direct vehicle effect on
the barrier. The degree to whieh the thermodynamic a c t i ~ t yof the
drug in the vehicle influenced. the course of events was relatively
minimal. One questionable aspect of this work was that the baseline
diffusion rate af estradiof through the skin was measured fram a
sogd drug deposit, a p e a t departure From the situation in which
the drug is in solution in the vehicle. This cert&nly affects the
quantitative interpretations of the studies.
Finite-dose diffusion cells have often been used ineor~ectfyto
assess drug delivery, in that formulations have been applied so
thickly that diffusion occurs from an infinite slab, chan@ng the
whole kinetics of the drug transport process, This basic flaw is
evjldent in several studies (59,tiO-63).

B. Rafe af Thermodynamic Activity in Topical Drug Delivery


important conceptud work concerning the role af topical vehicles in
the delivery af dmgs thmugh the skin was pioneered by Poulsen,
Bstrenga, and others ( 2 5 , 2 6 , 5 3 , 6 4 ) . These studies fwused on the
role of the drug's thermodynamic a c t i ~ t yin the overall diffusional
flux: through the skin, The first sf these studies dealt with the
effect that the dwgb thermodynamic .ectidty has on its release rate
from the vehidt: ( 2 4 ) . The work was based on T, HiguehiFs theo-
retical supposition that the steady-state diffusion of a d m g dis-
solved In a vehicle through the skin can be represented by
Drug Delivery from Topical Formulations

Where dMt/dt is the rate of penetration, K is the stratum comeurn/


vehicle partition coefficient, Cv is the d m g f s concentration in the
vehicle, A is the area of the application site, hsc is the effective
thickness of the skin, DS6 is the effective stratum eorneum diffusiv-
ity , av is the thermodynamic aetivity of the d m g in the vehicle,
and 6, is the effective activity coefficient of the drug in the s l i h
barrier. The latter equation indicates that only certain terms are
avdable for manipulation by the formulator to increase the rate of
diffusion of a drug. These are K , the partition coefficient, and
Cv, the drug" concentration in the vehide, K is altered when the
composition of the vehicfe is d t e r e d , and Gv can be adjusted up to
the Ievef of solubnity in the vehicle. Note that, for saturated solu-
tions in different vehicles, K Cv remains constant, and this product
is constant irrespective of the vehicle differences;, If one assumes
that no vehicle-skin interactions take place, and if the dmg itself
does not alter the skin. in any way, then can be regarded as a
constant. The d m g t s effective themodynamic activity is progor-
tionalt to the product of K: and Gv, which, in fact, y-i;elds the drugts
concentration at the surfme of the skin, Any change in this prod-
uct leads to a change in the tissue concentration that establishes
the gradient across the skin,
To test the firegoing concept, two topic& stemids were incor-
porated into a series of pmpylene glycol-water gels, and d m g re-
lease into isopmpyl myristate was studied (24). D m g sslubBi_ty de-
terminations were made for each system, and a receptor-vehgcle
phase partition coefficient was calculated. It was found that the
milximum d m g release occurred from gels contdning the minim&
amount of propylene glycol required to exactly dissolve all the dmg.
Excess propylene glycol required, to exactly dissalve all the drug.
Thus decreasing the thermodynamic aetivity of the drug and hence
its release rate from the vehicle, is fully consistent with the fore-
going argument. At the other extreme, when insufficient propglene
glycol was a v ~ l a b l eto completely dissolve the steroid, dissolution
and intragel diffusion became p ~ r tof the rate-limiting factors in the
overall diffusion process over the multihour collection, periods. The
concept of using the thermodynamic aetivity of a drug in its vehicle
as an indic&or of its bioavaifabififity from that vehicle was solidified
in a subsequent series of studies by the p o u p ( 2 5 , 2 6 , 5 4 ) . For both
fluodnonide and fluocinolone in propylene glycol-water @Is, positive
wrrelatians were seen when data were compared from in vitro release,
in ~ t r permeation,
o and in vivo vasoconstriction studies ( 2 6 ) . The
latter directly relate to concentrations of the steroids that build up
in the local tissues beneath the applications. It was found that
maximum bioavaitabiGty was obtdned f'rorn the gel system in which
just enough propylene glycol was present to completely solubilize the
drug, a result totdly consistent with the in ~ t r release
o studies.
Similar results were obtdned when Buocinonide was incorporated into
various creams an8 ointments ( 5 4 ) .
Along these same lines, in a study investigating the diffusion of
tritiated water from a wide r a g e of polar solvents, the absoqtion
rate of water through human epidermis was proportional to the
thermodynamic a e t i ~ t yof water in a particular solvent (65). B a r v
et al. 118) correlated the thermodynamic a c t i ~ t yof benzyl alcohol
vapor with the diffusional flux of the vapor through human abdorn-
inal skin. "These experiments were done in such a way that the
vehicle was not in contact with the skin, thus mling out the possi-
blXity of any direct vehicle-skin interactions* In an interesting
practical application of the finite-dose method, Bronaugh et al. (68)
studied the absorption of n-nitrosodiethanolamine thmugh excised
human skin, This campound is an ixnpuAty, believed to be a car-
cinogen, found in. many cosmetic products, It was found that the
permeability of n-nitrosodiethanoIamine increased with an increase in
the stratum eorneum /vehicle partition coefffcfent of the agent, In
another study, Turi s t al. studied the permeation of dinorasone di-
acetate through halrless mouse skin from vadouk vehicle forrnul&thns
(59). It was observed that the permeation rate increased with the
degree of solrxb;ilizatiaian up to a poht , and then decreased, This
obsepvation is because, at maximum permeability, the vehicle exactly
dissolves all of the d m g in the formuXation, Any further increase
in, d m g solubsization results in a decrease in the pastition. coeffi-
cient (but no further increase in the dmg's vehicle eoncentration).
This decrease in the thermodynamic aetirrity of the d m g results in
a deerease in its permeability. Sloan et al. ( 6 2 , 6 3 ; see Chap. 13)
have compiuced the experimentd flux of drugs through skin with
theoretical predictions obtrri.ned through the use of solubility parame-
ters. Studies of this type &resignificant in that they illustrate the
ways in which theoretical concepts can be applied to the development
of topical drug products, However, expefimentd conditions used in
these studies generally fail to approximate the clinical situation.

6. In V ~ V OStudies
When all is considered, mast topical products on the market have
been developed using more clinical research strategies than bench-
top ones, Iln viva assay methods have played an important POI@in
the development of many important topical products, Some of these
studies have invalved the use of animal models; however, the pres-
Drug Delivetly mnz Topical Formulatiuns 241

ent discussion is limited to studies i n v o l ~ n ghuman subjects, The


in vkvo study of the penetration kinetics of d m g s has often in-
volved the detection of some local biolo@cal response that the d m g
elicits upan penetration into the skin. Examples of responses that
have been utilized include vasodBation, vasoconst dction , kedtiniza-
tion, epidermaf proliferation, and changes in blood pressure. In an
early study that recognized the Importance of standardizing the
amount of drug plaeed onto the skin, Barrett et aX. (52) studied
the relative erythema pmduced by methyl, nieotinate when income-
rated into vadous vehicles, What is interesting about this study is
that it involved the application of a discrete disk of vehicle a f known
thickness to the forearms of human volunteers, perhaps the first
time that the importance of applying a standard amount (and thick-
ness) of vehicle was recognized, Qther in viva techniques involve
the analysis of blood or serum levels of drugs absorbed through the
skin or the mounts of drugs excreted in the urine, Urinary analy-
sis, in particular, often has been used by Peldman and M ~ b a c hto
study percutaneous absorption in. human subjects (6Ci--7Q).
Perhtlps the most welf-refined of all biol,o@cal assays used to
develop and assess derntatolo@cal formulatjions are topical cortico-
.
steraid bioassays f n some inst anees , an inflammatov stimulus is
applied to the skin, which is immediately followed by the adminis-
tration of the steroid. The anti-innammatory response is then grad-
ed. Using a method involving the application of croton on and
kerosene to induce pustules or blisters in humans, Kddby and
Kliman f 71) obtdned a rank order of corticostemid a e t i ~ t ythat
correlated well with the relative clinical efficacies of the dmgs.
Simaar studies using i r ~ t a n t ssuch as mustard dl and nitric wid
(722 , tetrahydrofurfuryl aleohoil (73) , and histamine (74) have been
used to test co&icosteroid actidty, The fact that certslin steroids
have the ab3;i-L~to c o n s t ~ c tlocal vasculature and turn the site of
application pale has often been utilized to study both the relative
potencies of stemids, as well as the abnity of vehicles to deliver
steroids topically. McKenzie and Stoughton (75) discovered that
treating psoriasis with stemids under plastk wrap blanched the Ie-
sion as well as normal skin. This observation was used to develop
a test to rra& commereially avaaable steroids in order of potency
(762. This method was adopted by Sarang et al. (77) to measure
the efficacy of sever& topical, steroids from vadous vehicles* A
method of application that had been predously employed by this
group (78) was used in this study. Subsequent@ , Woodford and
Barry ( 79) , among others ( 16,26,53), have used the blanching test
to compare topic& stemid fcQrmulalions,
242 Addicks et at,

REFERENCES

.
B Barry, in Dematological Formulations, Marcel Decker, New
York, p. 145, 1983.
.
M. Katz, and B Paulsen , in Handbook of Experimental Phar-
.
macology, New Series, Vol 28, Springer-Verlag, Heidelberg,
1971.
B. Xdsan, J. Pharm. Sci., 64:901, 1975,
G. Flynn, and B . Paulsen, in Percutaneous Absorption: Mech-
anism-Methodology-Drug Delivery, Marcel Dekker , Mew York ,
pp. 431-459, 1985.
T. Niwchi, J . Pharm. Sci., 50:874, 1961.
T. Niguchi, 3. Soc. Cosmet, Chem., 11:85, 1960.
J. Crank, in The Mathematics of Diffusion, 2nd ed, Clarendon
Press, Oxford, p. 48, 1975.
.
M Jaoobs , in Ddffusion Processes, Springer-Verlag, New
York, p. 44, 1935.
W. Higuchi, J . Pharm. Sci,, 51:802, 1962.
D. Paul, and S, McSpadden, J . Membr. Sci., 1:33, 1976.
P, Lee, 3. Membr, Sei, , 7:255, 1980.
T . Roseman, and W. Higuchi, J. Pharm. Sci,, 51:353, 1970.
J. Ayres, and T. Lindstrom, 3. Pharm. Sci,, 66:654, 1977,
R . Guy, and J. Hadgmft, Xnt. J , Pharm,, 6:321, 1980,
6. Flynn, E. Topp, and 6. Amidon, in Topics in Phamaceuti-
cal Science XBBS. Elsevier Science, New York, pp. 313-328,
1985.
J , Crank, in The Mathematics of Diffusion, 2nd ed. Clarendon
Press, Oxford, pp. 44-68, 1975.
W. 3. Addicks, 6 . L. Flynn, and N . Weiner, Pharm. Res., 4:
337, 1987.
D. GemmeE, and J. Mordsan, J , Pharm. Pharmacol. 9:641,
1957.
G. FXynn, and S. Yalkowsky, 3. Pharm. Sci., 6f:846, 1%?2.
G. Reddish, and N . Waes, 3. Am. Pharm. Assoe., 18:576,
1929.
6. Clark, and G . Davies, Br. 3. Dermatol., 00:521, 1949.
L, Loekie, and J. Sprowls, 3 , Am. Pharm. Assoc., 38:222,
1948.
N . BiUups, and R, Sager, Am. 3, Pharm., 137:57, 1965,
B. Poulsen, E. Young, V. Coquilla, and M. Katz, J , Pharm,
Sci. , 57: 928, 1968.
J . Helblian, B. Poulsen, and K. Burdick, Curr, Ther. Res.,
22:713, 1977.
J. Ostrenga, C. Steinnetz, and B. PouIsen, 3. Pharm. Sci.,
60: 1175, 1971.
Drttg Delivery from Topical Formulations 243

R. Schueplein, I , Blank, G , Brauner, and E), MaeFarlane, J.


Invest, Dermatol. , 52; 63, 1969.
R . Dempski, 2, Portnoff, and A . Wase, J . Pham. Sci,, $8:
579, 31969.
2;. Chowkrm , and R , Pritchard, J , P h a m , Sci, , 64:154, 1975,
M, Busae, P, Hunt, R e Lees, md N, Maggs, B r . J , Dermatol,,
81 :103, 1969,
A . Jurist, J , Invest. DematoZ., 20:331, 1953.
M, Multirner, G , Rif&in, J, Hal, M. GXiclrman, and Pa. GiIman,
J , A n , Pham, Assoe,, 45:212, 1955.
3, Wood, L , Rising, and W, E d I , J , Pharm, Sei., 51:668,
1962.
.
6 , Hawze , and N Bmups , Am, J , P h a m , , 138: 1143, 1966,
G , Bramanti, G I Bisatum, C. Mazzi, P, Mrxra, and P. PapillJi,
Boll, Ghim. Farm., 129: 738, 1980.
J . Ayres , and P , Lasker , J. Pharm, Set, , 63: 1402, 1973,
.
IZ Spang-Bmnner, and P. Spefser, J, P h a m , Pharmaesl, , 28:
23, 1916.
P . Yark, m d A. Suleh, S. Pharm, Sei., 65:493, 1976,
F. Bottari, C. Di Gola, E. filampic3&, M , Saettone, and M. Ser-
Bni, J e P h a ~ m ,Sci. , 66: 926, 1977,
P, BottaA, G . D i Cols, E . Nannipieri, 5111, Saettone, and M.
Serafini, J , Pharm, Sci,, 63:2779, 1974,
C. Bi Cdo, V. Garelli, B , Gietnnaecini, M I Seraffni, and F,
Bottari, 2. Pham. Sei, , 69: 387, 1980.
F , Emberg, and A , Brabin, Acta Pkarm, Suee, , 2s: 229, 1982.
A , Bmden, and A , PJygvliat-Mayer, Aeta Pharm, Suee, , 19:
261, 1982,
A , Kahn, S , D a ~ s and , J, Hadgraft, Pharm, R e s , , 3~214,
1886,
S, Tanaka, Y e Teashima, M. Nlurayaner, and S. Tstlchiya,
Int, J , Pharm, 87:29, 1985,
K . Patel, 6 , Banker, and El. DeKay, J. Pharm, Sei, , 50: 500,
29631,
T, Koizurni, and W, Himehi, J , Pharm, S c i , , 57:87, 1988,
8. Wsthlgren, A . Lindetrorn, and S. Friberg, J", Pharm, Scl,,
73: 1484, 1984,
C , Behl, G . mynn, T, R u f i a r a , N , Harper, W, Smith, W,
Higlmchi, N, Ho, and C. Pierson, J , Invest. Dermatol., 75:
346, 1980.
H e Durrheim, C , Flynrr, W. Hiwehi, and G . BefiX, J. Pharm,
Sci. , 69:782, 1980,
6 , Ffynn, H . Durrheirn , and W. H i p c h i , J . Pharn, Sei, , 70:
52, 1981.
R. Sckrapleh, and I. Blank, J. Invest. Demalol, , 6Q:286,
1973,
Addicks el a!.

J . Osfrenga, 5, Haleblah, B. Poursen, B . Ferrell, N . MueIIer,


and S. Shastr;l, J , Invest, Bernatob. , 56: 392, 1971,
1". Franz, Curr, Probl. Dermatol, , 7: 513, 1E178.
T, Franz, J, Invest, Bermato?, , 64: 190, 1975,
R . Bmnatlgh, and R. Stewart, J , Pharm, Sci, , 74: 64, 1985.
G , Hawkins, and W , Reifenrath , J , Bharm, Sei. , 75: 378, 1986,
.
B MoZlgaard, and A , Hoelgaard, Xnt, J , Pharm., 15: 285, 11983,
3 , T a d , D, DanieIeon, and 4, Woltersom, J , Pharm, Sei,, 68:
275, 1979.
R. Bronaugi.5, E . Congdan, and R, SchueyiEein, J , Invest.
Demalol, , 76: 94, 198 1,
E l Cooper, J , Plzarm. Scz", 73: 1153, 1983,
K . SToan, K, Siver, and $3. Kocfr, 3, Pharm. Sci., 75:?44,
2986,
K , Slaan, S. Koch, Ir;, Siver, 0 . Rowers, and P. Flowers, J ,
Invest, Berm,, 87:244, 1988,
1". Mdone, J. HaXebXian, B , Podsen, and R. Burdick, B r , J.
BematoZ, , 90: 187, 2974,
P . Dugard, and R . Scott, Int, J, Pharm,, 28:219, 1986,
R , PeXdman, and H, Mdbaeb, Arch, Dematol. , 91: 662, 1965.
R . Pelaman, and H , Mdbach, Arch. Bematot. , 134: 6 4 9 , 1S66,
.
R , Feldman , and H MGbackt, J . Invest, Demaio?. , 48: 281,
1967.
.
R FeXdman, and W Mdbach , J . Invest, De~maloE,, 50;351,
1968.
R, FeIdman, and H , Maibach, J , Invest, Dermatoit., 52:891,
1969.
K, Kaldby, and A , ICXigman, J , Invest, Dematol., E53:2132,
1974,
A , Scott, and F, K d z , J , Invest, Dermatol., 26:381, 1956.
W , Gray, md R . Wdfe, Arch, Bermatole 84~18,1961,
B , Reddy, and G , Singkt, B r , 3 , BermatoE, , 94: 191, 1976.
A , MclKenzie , and R , Stoughton, Arch, Ilermatol, , 86: 6 0 8 ,
1962,
A , MeRenzie, A w h , Dematol, , 86:611, 1962.
I, Sarkmy, J, I-fadgraft, G , Caran, and C , Barmtt, Br. J .
Dematol, , 77:569, 1965.
J, Barrett, J, Hadgraft, J. Sarkay, J , Pharm, Pharmacol. ,
IG(suppX, ) :1Q4T, 1964.
B , B a r w , in Dermatolop*caI Formulattoions : Pereu taneous A b -
sorption, Marcel tlekker, New Y o ~ k ,p , 242, 1983.
Skin Transport

KEN NETW E3. SLQAN University sf FZadda , GainesviZZe , FEoPz'da

The use of mlubnity parameters of dxllgs and vehicles to descebe


the transport of dmgs thmugh skin is based on the ef"?Fortsof Had@-
brand (I) and others (2--4) to account, in a ration& way, for sol-
vent -solvent and solute- solvent interactions in the prwess of dis-
solution. The results of those efforts to develop a theoretical b ~ i s
for solubility is represented in simplified form by Eq, l where X i is
the mole fraction sslubitity of dm$ (Ii) in the solvent or vehicle (v) ,
ANt: is the heat of fusion o f the d m g at its melting point, Tm is the
melting point af the drug, T is the temperature at which the solu-
bility is being measured and yr is the activity coefficient of the drug
in the vehicle, which can be calculated from Eq. 2, In Eq. 2 , 65
and cib, are the soXubflity parameters of the d m g and vehicle, re-
spectively, 4, is the fraction of the solution volume occupied by the
vehicle, and V i is the molar volume of the d m g , The calculated
mole fraction solubility of the dmg, then, is composed o f two corn-
ponents; one is a e ~ n t s b u t i o nfrom Its ideal solubnity, the other is
a eontdbution from its actidty coefficient to take into account devi-
ations of the actual solubility from the idea1 solublilfty. The evoIu-
tion of this approach to predicting soluts3ity stAclly from thwreticical
cansiderationa of the physicochernical properties of solute and so1-
vent ( d m g and vehicle) has recently been redewed by Martin and
Mauger (51,

v
- log Xi = ( A H f / 2 . 3 RT)[(T, - T)/T,l + lag Y
V
Because trmsport through skin is a diffusion process that de-
pends on a concentration gradient, the application of calculated sol-
ubiXities, befived in part from 6 , to predict the concentration gra-
dients and, hence, diffushn is a l,a@cal progres~ian. Pick's first
law Tor diffusion (Eq, 3) contains terms for the concentrations of
sl s2
the drug %nthe membrane (Ci and C i 1 that constitute the concen-
tra-t_iongradient ( F i g . I). They are difficult to determine. Hence,
Fickfs first law is txsudly expanded into a form CEq. 4) that con-
S,V
t a h s contributions from mare readny measured terms, where Ji
is the flux of drug (i) delivered h r n vehicle f v ) through a mem-
brane that, here, is skin ( s) , D? is the diffusion coefficient of the
1 a1
--
drug in the skin, h, is the thickness of the skin membrane, Ci
s2
and Ci are the concentrations of the drug in that layer of the skin
next to the vehicle and receptor phase or plasma (p) , respectively,
V S,V
Gi is the concentration of the d m g in the vehicle, and Ki is the
distribution or partition coefficient of the drug- between the vehicle
sl v
and skin equal, to G i IGi. En proceedhg from E q . 3 to E q , 4 ,
s,v v sl s2
K i Ci has been substituted for G i m d Ci has been assumed to
approach zero, because sink conditions are mahtahed on the plasma
side of the membrane, that is, ! C = O (see Fig. 1) +
1

The partition, coefficient ;far the diistdbution of the d m g between


two phases (here, vehicle rand skin) to form saturated, nonided 80-
lutions is @ven by E q , 5, Because the a c t i ~ t ycoefficients can be
calculated from Eq. 2 , the partition coe8fficient can be calculated by
substituting the identities for yyand vi from E q . 2 into E q . 5 to
give E q . 6, If it is assumed. that the d m g is not very soluble in
either the vehicle or the skin, and 4: approach a value of 1 and
E q . 63 reduces to Eq. I ,

V S
log K : ' ~ = log y i / y i E 59
Use of Solubility Parameters in Skin Transport 24 7

11. GENERAL APPLICATIONS OF REGULAR


SOLUTION THEORY TO PARTITIONING
PROCESSES

Various forms of E q . 6 and of relationships between 6 values and


bioactivity have been used successfully to describe a number of bi-
ological processes besides percutaneous absorption. Fctrmson (6)
showed that the toxic concentration of a chemical was a funcgon of
the intrinsic toxicity of the chemical and its distribution into a tar-
get phase. Here, the distribution is given by the difference in the
partial molal free energies of the chemical (phase 1) in its standard
.
state in two phases (phases 2 and 3; Eq 81, where, for example,
the mold free energy of the chemical in phase 2 is given by 2 . 3 RT
log YiP2 (TI.

log K = (PO3 - i?:)12.3 RT

Figure 73.1 Concentration gradient of a solute (i) across a membrane


of thickness hs where C r is the concentrmtion of the solute in the ve-
D
hicle in the donor phase and C; is the cancentratian of the solute in
sl s2
the plasma in the receptor phase. Ci and Ci are the concentrations
of the solute in the first and last layers of the membrane, respectively.
Khalil and Martin (8) used solubility parameters in their analysis
of the transfer of a solute from one phase to another in partitioning
expesiments. They measured the rate of transfer of sdicylic aeid
from a pH 2,0 aqueous solution (representing the stomach), through
nonpolar immiscible liquids exhibiting solubi2ity parameter values of
from 7.1 to 10.7 (cal /cm3) 1j2, and into a pH '7.4 aqueous solution
(representing plasma). The model was set up to analyze absorption
from the gastrointestinal tract, but a singar system could have been
used to model transfer through other membranes, such as skin.
The results for sdicylic aeid in this model showed that the closer
the 6 value of the nonpoXar liquid was to that of salicylic a d d (6i =
10.8), the faster the rate of disappearance of sdicylic acid was from
the pH 2.0 sdution and the slower its rate of appearance in the pH
7 , 4 phase. It was also observed that the optimum transfer of s d -
icglic add to the pH 7.4 solution occurred at a 6 value of the non-
polar liquid of about 9.7 (cal/cm3)1/2 to balance the rate of transfer
fmm the pH 2 solution into the nonpolar phase with its rate of re-
lease into the pH 7 . 4 solution,
Ostrenga (9) suggested that molar attraction constants (F vritl-

-
ues) for chemical structures, derjlved from solub%ity parameters ( 6 )
where 6 = P/V and where TJ molar volume, could be used to cor-
relate b i o a c t i ~ t ywith chemical stmcture. Because the Xi" vdues
were readiry avasable far most organic functional groups and were
additive on a constitutive basis, they could be calculated for most
ehernicd stmetures without resa~tingto additbnal experimentation .
This was in contrast with the Wansch approach ( 10) which used n
vafues dedved from the expedmentaf difference in logarithms of the
partition coefficients of a model chemical (RE) and its substituted
derivative ( R X ; E q , 9).

n
x
= log K
(RX)
- log EI
(R-EI)

Subsequently, Da?l-is (11) deduced E q . 6 from the Hndebrand-


Scatchard equations for the excess free energy of a remlar solution
and then used this solubfiity parameter-based partition coefficient to
show that .rr and F were related to each other thmugh Eqs, 6 and $3,
Thus, n vdues could. be cdculated for vadous functional groups
(74) fmrn their respective Fx and Vx values and the solubBity pa-
rameter values for the aqueous f 6), and organic phases ( 6 o) (Eq,
lo). Dads found that the agreement: between the expedmentallg
determined rr and the n cdculated from E q , 10 was not par"ecularly
good in many cases and armed against using calculated n or F val-
ues to correlate chemical stmcture with bioactiety,
Use of SoEubitity Parameters in Skin Transport 24s

Cohen and eo-workers ( 12,f3) have used regular solution t k e o v


and a more exact form of E q . 6 (14) to explain how stmcturdly
simgar molecules could be either depressors (anesthetics) or stimu-
lants (conwlsants) sf the nervous system. Their explanaBon of
this difference was based on an extension af the concept of Mullins
( 15) that linked the bialo@cal properties of a chemicd with its bio-
phase of action. Thus, drug specificity is a consequence of its
prefemntial solubBity in a subre@on of the membrane exhibiting a
bulk solubsity parameter simlJar to that of the dmg. In, this ease,
low 6 drugs cause excitation, whereas high 6 drugs cause depression,
Martin and co-workers (16) examined the effect of solvent polar-
ity and its abjility to solvate theophyfline on the in vivo absorption
of theophylline fmm mixtures of gslyethylene glycol 400 and water
into the rat intestin& membrane. Their results were quditatively
similar to those observed for sdicylic acid in et model system (8).
The closer the 6 value of the polyethyIene @ywl 400-water mixture
was to the 6 vdue of theophylline the slower the rate of disappear-
ance of theophylline from the solvent mixture into the rat intestine.
The authors afso used Eq. 6 to caleulats a 8 value for the intestinal
membrane of 12.6 (cd/cm3) 112 fmm their data,
Findly, Bustarnante and Sefles (17) have shown that the binding
of dmgs to plasma protein can be analyzed by E q . 11 in which 6~
and 8 ~ are
) the solIub2ity parameters l"or al'tsumb (or the drug-bind-
ing site in albumin) and drug, respectively, where the actidty co-
efficient ( a) is the ratio of maximum binding (gM 1 , to the real bind-
D
ing (BD) by andom to the ideiif and sctua3, salubifity fiom the Ha-
debrand equation, Data far the bhding of sulhrramides , penicallns ,
phenothiazines, and barbitude aeids to human serum albumin were
found to fit Eql. 11, which descdbes a saturation-type prmess, bet-
ter than an E q , like 6, which descsbes a distdhution-type process,
It was &so observed that mmimum binding to sermrn albumin oe-
carred when the solubifity parameters of the dmgs were similar to
that of the plasma protein, For the suXfonanides, 6~ was edmlated
to be 12.33 (cal/cm3]1/2 which closely corresponds to the average
solubility parameter of the seven-amino aeids sequence in human se-
rum albumh thmght to be the pdmary binding site for sulfonamides.

M 2
log o = log(BD IDD) = V
D E
12.3 RT(6E - 6D) 2 l 111

.
Iff APPLICATION OF REGULAR SOLUTION THEORY
70 PERCUTANEOUS ABSORPTION PROCESSES

Two p o u p s initiated the application o f remlar soIutian theom to


the partitioning grwess in percutaneous absorption, En 14184 Gohen
and co-workers descdbed the absorption of dkanoic acids through
skin in terms of the so1ubjtStity parameters of the acids (18). Short-
ly a&er the publication of the 1984 perper, Sloan, Sherertz, and co-
workers pubfished a sedes of eight a ~ i c l e s(19-26), starting in
1986, that used remlar solution theory to expldn the effects of ve-
hicles and the effect of changes in the physicmhemical properties
af dmgs cszused by their transient chemical modification Cprodmgs)
on the delivery of dmgs through skin. The articles by Slaan,
Sherertz, and co-workers w331 be discussed first, then the initid ar-
ticle by Gohen and co-workers, dong with two v e v recent articles
that have appeared from that gmup, will be discussed,
The expedmentd designs used in the work reported by Sloan
et af. f 19-26) were somewhat different fmrn the predous d e s i p s
used in sirnjtStar experiments ( 18). First, only saturated solutions of
dmg--vehicle were applied to the membrmes in the diffusion cells
as the initid part of each expe~ment, Second, after the injltial part
of each, experiment a '%econdwapplication of a standard sdute-ve-
hicle was made onto the membranes.
This second application part of each expe~mentenabled the in-
vestigators to determine whether or not the membranes were damaged
by the initial treatment of dmg-vehicle by comparing the rate of
delivev (fXux) of the standard sdute through skin after an initial
application (pretreatment) of dmg-vehicle , with the rate of d e l b e r ~ t
of the standard solute without pretreatment. The higher the ratio
between the t w o fluxes, the greater the degree of damage to the
skin and the greater the degree af devliations of the experjirnental
flux: of drug-vehicle from the theoretical flux predicted by rewlar
solution theom, Xn this wsy , it was possible to separate the ther-
modynamic effects (1) of the drug-vehicle interaction or (2) of the
pf-tysicachernicd prapert;ies of the prodrug on the partitioning proc-
ess from the chemicd or physical effects of the vehicles themselves
on the i n t e g ~ t yof the membranes.
Saturated solutions of drug-vehicle were used to ensure that
the dmgs were b e h g tested at theis maimurn activity. In addition,
only at saturation, on a mole fraction concentration basis, does a: =
,'
a where a: and as are the activities of the drug (if in vehicle
i x x 6 v
(v) and skin ( s ) , respeictively, Hence, K i from E q , 5 can be
calculated u s h g E q , 6, but only if saturated solutions of the solute
in the two phases of interest are b e b g evauated,
The effect af vehicles an flux of a drug through skin can be
correlated with regular solution theory through the relationship be-
s ,y
tween permeability coefficient (Pi 1 and partition coefficient
S, v S,V s,v v
(Ki ). Flux (Si ) is equal to the product of Pi and C: i , so
from E q . 4, ~f
pY is equal to the product of the partition coefficient
5'3
and the diffusion coefficient (Di) divided by the thickness of the
Use of Solubility Parameters in S k i n Trmspart 251

membrane (hs) ( E q . 12). If D; is independent of the vehicle used


and hs does not change much. withh the limits of biologic& vvariabili-
.
ity , then D; /h, is a constant and E q . 12 reduces to Eq 13 ( 19).
s,v
It has already been shown that K i can be calculated from E q , 6 or
?. Thus, to determine whether or not variations in the rates of de-
livery of a drug from a series af vehicles can be explahed by r e w f a r
solution t h e o q , ;it is merely necessav to plot the log e&culated
(theoretical) R y * v against the solubaity parameters of the vehides
1
s,v
( 6 v), and determine if the plot of log e x p e d e n t a l Pi versus 6 v
approximates the shape of the log thwretical lCTPv versus 6 v curve
and is separated from it by a constant amount (D?/hs),

constant

There are two reamns why it is useful to be able to calculate


these theoreticsil partition coefficients based an remlar solution the-
ory to understand the dermal delivesy pmeess. First, from the
faregohg discussion, It is obvious that if it is possible to predict
K Q ? ~, it is possible to predict permeability coefficients and, hence,
flux:. The ability to predict the effect of changes in vehicles on the
solubnity af a drug ;in a formulation and, hence, its Rux is essentid
for the rational d e s i p of topical formulations, Second, it is erfuBIX-
Xy impassfile to o b t ~ nan acmrate and appropfiate experirnentd
partition coefficient for a solute d i s t ~ b u t i n gbetween a vehicle and
skin. The concentration of a d m g in a vehgcle is essay detemined,
but the concentration fn the skin Is difficult to quantitate because
of the different demees of changes in hydration and of changes in
other physicocfiemied p r o p e ~ i e sof the skin that would result if
samples of skin were dowed to equnibrate with different whicles
fir extended periods (27). Thus, a method Tor determining theoret-
ical K Q p v would be advantageous because there are serious difficul-
53 I v
ties in obtainbg experimental K i .
Conversely, it is possible to calculate theoretied partition coef-
ficients u s h g other methods. The mast popular method uses frag-
ment constants far the vadous functional: gmups in a d m g to cdcu-
late the iSmgts distribution between water and a water-immiscible
solvent that serves as a model for the biolo@e& membrane (%9,30).
UsuaUy , the water-immiscible solvent is octanol , but many other wa-
ter-immiscible solvents have also been. used. The fragment const~nts
for a two-phase partitioning system are deAved from expedmexrtally
determined. partition coefficients for model elzemicdtls in the same two-
phase system, However, by using the avdlable data, this approach
is useful only if water is the vehicle being considered. In addition,
many, if not most, of the vehicle components for which it would be
useful to calcuXate theoretical partition coefficients are either miscible
with, or at least partially soluble in, both octanol and water. These
include, for example, solvents such as pmpanol, dimethyl sulfoxide ,
dimethylformamide , pmpylene glycol and formamide. Other solvents,
such as isopropyl myristate and oleic acid, are soluble in octanoi or
other solvents that serve as models for biological membranes. Thus,
it is virtudly impossible to experimentally measure partition eoeffi-
cients for model chemicals in a two-phase system that would give
fragment constants that could be used to calculate thmreticd parti-
tion coefficients for a drug distributing between the solvents (ve-
hicles) of interest and skin,
For the dmgs that were investigated by SIoan and eo-workers
( 5-fiuorouracil , 6-mercaptopurhe , theophylline, and salicylic acid),
s,v
the theoreticd parti~oncoefficients, Ki , were edculated assuming
that the solubiXity parameter of the skin, 6s, was about 10 (tali
cm3)1/2 (18). The eolubility parameters of the vehicles, 6v, were
literature values (28) except for isopropgl myristate (19) and are
listed in Table 1, whereas the wlubility parameters of the drugs,
61, and the molar volumes of the dmgs, Vi, were calculated accord-
ing to Fedors (29). These theoretical partition coefficients are giv-
S,V
en in Table 1. The experimentat permeability coefficients, Pi , for
the delivery of these four relatively polar molecule8 from vehicles
exhibithg a broad spectrum of polarity, from water to olefc acid,
through hairless mouse skin are also given in Table 1 along with the
corresponding fluxes, 58 * v , and solubilities, Cy.
1 1
S,V s,v
The values for log theoretical K i and log experimental Pi
for each drug are plotted against 6v in Figures 2 through - 5. In
s,v
each case, the plot of the log experimental Pi versus 6v approxi-
mates a parabola that is simtlar to that of the plot of log theoreticd
ICfSV versus 6,, except for the data obtained when actanol and
isopropyl myristate (IPM) were used as vehicles. However, both of
those vehicles caused a much higher flux of the standard solute-
vehicle in the second application part of the experhent than did the
other vehicles; therefore, it has been assumed that the inordinately
9.v
high flux of drug, hence, Pi , caused by those vehicles is due to
damage to the skin. If the data for these obvious outliers are dis-
regarded, a plot of log theoretical PSpVversus 6 v can be construct-
S,V i
ed from log theoreticd Pi values, each of which can be obtained
from the sum of the cofTesponding log theoreticd K ? t v and the
1 s,v
average of the differences between the log thwretical K i and log
C
01
"O
Table 13.1 T h e Effect of Single Component Vehicles on t h e D e l i v e v of D m g s t h r o u g h Hairless Mouse
Q
V s,v a s,v -b
Skin : Solubilities (C i ) , Fluxes (Ji ) , Permeability Coefficients (Pi ) a n d Iog Theoreticaf Partition Co-
efficients (log K7*V)b 0,
E.
C.
Solubility S,V
p;yv 4
v Rux (Ji 1
SolventC (Ci), lmglemZh x [cmlh x lo3 s,v 2
'
[ 6v; (eal/cm3) 1/23 (mglcm3) 103 ( 2 ST))] ( 2 SD)I log K i
1
~heo~h~lline~ 23
IPM (8.5) 0.062 41.2 (3.7) 660.0 (61.0) 1.12
3'
OCT (10.3) 1.70 547.0 (146.0) 320.0 (86.0) - 0.18 ~3
X
DMF (12.1) 34.6 17.0 (4.0) 0.49 (0.12) - 0.98 5.
Y
PG (14.8) 8.07 1.54 (0.31) 0.19 (0.04) - 1.22 3
EG (16.1)
FOR (17.9)

Salicylic acide
0A (7.6)
IPM (8.5)
UCT (10.3)
PRa (12.0)
PG (14.8)
fable 13.1 (Continued)

Solubility
V
Flux (al~
,1 v pisv
Solventc (Ci)s [mglemzh x [cmlh x 103 S?V
1 CSV; (callcm3f 1/21 (mg/cmfo 103 (5 SD)] ( 2 SD)] log Ki

FOR (17.9) 145 1150 (46) 7.9 (0.32) -0.64 (2.55)

OCT (10.3)
MEGB (12.1)
DMF (12.1)
nlwso (13.0)

FOR (17.9) 9.1 1.5 (0.10) 0.16 (0.011) - 0.42


-
OCT (10.3) 0.60 440 (14) 730 (23) - 0.13 "13

-.I
FOR (17.9) 13.7 15.Q (1.4) 1.1 (0.10) - 0.75
2
-.
aSteady-state fluxes, number of cdf run = 3. 3'
bealculated from Eq. 6; R =: 1.98 calldegree, T = 305 K , 6s = 10 ( c ~ / c m ~ ) lfor
/ ~ ;thwphylline 6i = Y
14.0 (cat/cm3)1f2, Vi = 110 cm3lmol; for salicylic acid 6 i = 14.4 (c&/cm3)1/2 or 61 = 11.0 (cal/cm3)1/2 2
for values In parentheses, Vi = 93.9 cm31mole; for 6-mercaptopuhe 61 = 14.4 (callcm3) l12, V i = 81.2 2
cm3lmol; far 5-fluomuracil 61 = 15 (cal/cm3)112, V i = 70.4 cm3/mol. %
COA, oleic acid; IPM, isopropyl myristate; DET, diethyltoluamide; OCT, 1-octanol; PRO, 1-propanol; 2
MEG , methoxyethmol; DNF, dimethylformamide; DMSO , dimethyl sulfoxide; PG , propylene grlycol; EG ,
ethylene glycol; FOR, formamide.
Sources: Data from d Ref. 19; e Ref. 20; Ref. 21; f3 unpublished resulte ; h Ref. 22.
Figure 13.2

Figures 13.2-13.5 The effect of the solubility parameters of vehicles


s8v
(tiv) on the log experimental permeability coefficimts (Pi ) of theo-
phylline (Fig. 21, salicylic acid (Fig, 3) , 6-mercaptopurine (Fig. 4)
s,v
and 5-fluorouraeil (Fig. 5 ) . In each figure the log theoretical Pi
versus 6, relationship is represented by (---I and the log theoretied
s,v
partition coefficient (Ki ) versus 6, relationship is represented by
I-), In each figure the solvents are 1, oleic acid; 2, isoprapyl my-
ristate ; 3, diethyltoluamide; 4, 1-octanol ; 5, 1-pmpanol; 6, Z-methox-
yethanol; 7, dimethylformamide; 8, dimethyl sulfoxide; 9, propylene
glycol; 10, ethlyene glycol; 11, formamide; 12, water. In ~ i ~ u 3r e
s,v s,v
there are two sets of log Pi versus 6v and Ki versus tiv curves
given. The curves labeled as (01 are for 6i = 14.4 (cal/cm31lf2
whereas those labeled as ( A are for 6i = 11.0 (eatlcm3)
Figure 13.3

6v
Figure 13.4
Figure t 3 . 5

s,v
experimental Pi . This average difference corresponds to the
constant in Eq. 13. If the thickness of the stratum corneum is as-
s
surned to be about 10 m , then Di is about 10-10 cm2/s for most of
the drugs studied, except for D: for salicylic acid, which Is one to
two orders of magnitude greater. The calculated D!, then, is of
about the right magnitude for such polar molecules.
There are five conclusions that can be reached on the basis of
the results given in Table 1 and F f p r e s 2 through 5. First, there
s,v
is generay a minimum in the log experiment& Pi versus 6v curve
that corresponds to those vehicles exhibiting 6v similar to that of
6i. The drug also usually exhibits its greatest mole fraction solu-
s,v
bility in those sane vehicles so that, generally, Pi are inversely
related to solubility ( 30). Second, for single-component vehicles,
those vehicles that exhibit 6v similar to that of & B also cause the most
damage to the skin as sssessed by second application studies. Be-
cause those 6v are similar to tis, they may be solubilizing a portion of
the membrane that is responsible for providing the diffusional resist-
ance to drug permeation and, especinlly, to polar noleeules exhibiting
a 6 i quite different from 68, Third, from the knowledge of the soh-
Use of SolubiEiZly Parameters in Skin Transport 259
s,v
bgities of a. d m g in various vehicles, the relathnship between Pi
and 6v, m d the flux of that drug from one specific vehicle it should
be passib2e ta determhe the ffux of that d m g from other vehicles of
bterest , assuming that the vehicles do noManrage the skin. Fourth,
s?v
the parabdic nature of the relationship between 6v and Pi expected
from remlar solution t h e o ~ yhas been confirmed for a number of dmgs.
Fgtfi, for vcharneleonicic"rsolutes (dmgs) such as saficry-lie acid (see
Fig, 3) that apparently can exhibit two differmt S i values, two differ-
s,v
ent log theoretical K i and, henee, two different log experiment&
P Q 9 Vcurves are generated.
In addition to the results obtained far single-component vehicles,
the rates of delivery. of 5-Buoroumcirl, (5-FU) and 6-mercagtopurhe
( 6-MP 1 from a b b a v component vehicle (oleic acid-pwpflene hTfycol)
.
were ailso determined (Table 2) Here, dthough the log experiment&
pQYV values for the delivery of 5-FU and 6-MP from oleic acid and
s,v
propylene glycol fell an the log theoretical Pi versus 1 5 curve, ~
8rv
none of the log expe~mentaf.Pi vdues for the mktures 4icX. In
each case, startling w i t h mhtures that were rich, in oleie a d d (21,221,
a s the st>lubi^lityof the dmg in each mkture hcreased, its experimen-
s,v
tal Pi v d u e decreased, Also, in each case, them was considerably
more damage to the s k h observed after treatment with the b k a q -
component soluthns than aAer treatment with either aS the 1~iTXgle
components (21,231, regardless of the cSv of the mixture. Thus, en-
hmced delivev of a drug by a mkture of vehicles may be primarny
the result of enhanced damage to the htegrity of the skin, rather
thm to some thermodynamic advantage,
In the investigations af the abnity of prodrugs to deliver their
respective parent dmgs h r n various vehicles through skin, there
was a good correlation between the calculated solubgity parameter of
a series of homologous pmdmgs and their abnity to deliver the par-
ent dmgs through hdrXess mouse skin (24-261, In addition, a ve-
hide effect on rates of delivew was observed that was directly at-
tebutable to solubsity phenomena. Generally, the results from
these articles were qualitatively simBas to the results of Ostrenga
( 9 ) that showed that there was a good correlation between mdar at-
traction constants (F values) and bloaetidty,
The most extensive studies of the effect of reI.llative solubBitles
of a homologous seAes of prodmgs on their abaities to deIiver a
parent d m g through s k h are the studies of ~6,s-bis-aeyloxymelhyl-
6-NIP (6,9-bis-6-MP) and ~ G - a e y l ~ x y r n e t h y l -(6-mono-6-MP)
~-~ pro-
drugs , The acety;loxymethylt through oetanoyloxymethyl abng w i t h
the pivdoyloxymethyl derivatives of both the bis- and mono-series
were synthesized, characte~zed,and the rates at which they deliv-
ered 6-MP through hairless mouse skins were measured. The solu-
Tabte 13.2 The Effect of Binary-Component Vehicles on the Delivery of Drugs Through Hairless Mouse
v s ,V 8 ,-v
Skin: Solubilities ( C i ) , Fluxes (Ji ), and Permeability Coefficients (Pi N
01
0

b
SoXventa 6
,
s
(mole ratios) (callcrn3)112

OA:PG (1:14.5) 13.5 11.1 88 (5) 7.9 (0.45)


PG 14.8 16.5 1.6 (1.0) 0.097 (0.061)

aOA, oleic acid; P G , propylene glycol. %


0
bcalculated on a volume fraction basis,
CSteady -state fluxes, number of cells run = 3.
9
Sources: Data for 6 = NIP from Ref. 21; that for 5 = PU from Ref', 22,
Use of Solubility Parameters in Skin Transport 261

Table 13.3 Solubilities of s 6 , 9-Bis-acyloxymethyl-6-memaptopurine


(6,9-bis-6-MP) Derivatives, the Fluxes (Jy v, and Log Permeabaity
I
S,V
Coefficients (log Pi ) for the Delivery of 6-Mercaptopurine (6-MP)
by the Derivatives fmm Isapropyl Mypistate (IPM), Pmpylene Glycol
(PG) and Water

Ji 9
5,v
Derivative, mg/cm2h x lo3 log pi
R = /vehicle Solubilitya ( 2 SD) em /k
Table 13.3 (Continued)
-

a;?vpb
s,v
Derivative, mglcm2h x 103 log Pi ,
R = /vehicle Solubilitya ( 5 SD) cmlh

&The solubnities are given in mg equivalents of 6-MP per ml of so-


lution.
bsteady-state fluxes, number of cells per run = 3.
cSource: Fluxes of 6-MP from the vehicles from Ref. 21.

biiities of the bis-derivatives in various solvents are given in Table


SYV s ,v
3 along with their respective Ji and Pi from the same vehicles,
and the comparable data for the mono-series are given in Table 4,
Plots of the calculated 6i of the prodrugs versus their corresponding
s,v
log experimental Pi for the delivery of 6-MP by both series of
prodrugs f m water, isopropyl myristate (IPM) , and pmpylene gly-
col (PG) are given in Figure 6.
Xt is clear from Figure 6 that there is a fairly good
- correlation
s,v
between log experimental Pi and 61 for each series for the deliv-
ery of 6-MP from each solvent. However, the form of the relatlon-
ship depends on the vehicle. For the delivery of 6-MP from IPM,
as the prodmxg becomes more lipophilic and more like the vehicle
StV
(i.e., the value of 6i decreases), the log experimental Pi de-
creases. Whereas for the delivery of 6-NIP by the prodrugs from
water, as the prodrug becomes more lipophtlic and less like the ve-
8,V
hicle, the log experimentdl Pi increams. Conversely, there i s
S,V
not much change in the log experimental Pi values for the deliv-
ery of 6-MP by the prodrugs from propylene glycol. This last re-
sult may be because the 6i am not that different from that of 6v.
The other article that has been published in this area describes
the effect of the structure of N-Mannich base prodrugs of 5-FU and
theophylline on their ability to deliver their respective parent drugs
Use a? Solubility Parameters in Skin Transport

Figure 13.6 A plot of the log experiment& permeabnity coefficient


(P) versus the calculated solubility prtrameter ( & f )for the ~6-aeyl-
oxymethyl-6-mercaptopu&e prodrugs delfverlng 6-mereaptopurine
from isopropyl myristate (closed circles) and from propylene glycol
(closed triangles) and for ~~,9-bis-a~1ox~rnethyl-6-mercaptopurine
prodmgs delivering 6-mereaptopurine from isopmpyl myristate (open
circles), from propylene glyeol (open triangles), and from water
(open squares). The numbers refer to the compounds in Tables 3
and 4.

through hairless mouse skin (26). The results are very similar to
those found for the acyloxymethyl-type prodrugs of 6-NIP, Because
of stability consideration, these N-Mannick base-type pmdmgs were
tested in only one aprotic vehicle-isopmpyl myristate. A direct
s,v
correlation between log experimental Pi and calculated 6i was ob-
served. Again, the more lipophilie the prodrug (lower 61 value and
Parameters ( ) of
the Fluxes IAn .. rAT.,.r of 6-~vlei'CaJptc•puri:rte
( 6-MP) (PG)

Calculated

IPM, PG 1/2 (± SD) cm/h

1, 6-Mpa 14.4 0.60 (0.30) 0.73 0.093 (0.006) -4.82 Cf)

9, 0.16, 14.2 14.4 (8.8) 0.55 (0.22)


s
30.8 -0.72 4.41 l':l
;:I
d
"The solubaities are given in mg equivalents of 6-MP per mf of solution. 3'
bsteady-state fluxes, number of cells per run = 3. tn
X
S o u ~ c e : Data for 6-MP from Ref. 21. 3'
266 Sloan
s,v
more like the vehicle), the lower the vdue for Pi for the delivery
of the parent drug. s,v
Thus, once the log experimental Pi has been determined for
the delivery of a drug by a particular type of prodrug from a specifie
s,v
vehicle, it is possible to predict Pi for the delivery of drug by other
prodmrgs, in the same homologous series, from that vehicle using the
8tV
slope of a previous plot of log experimental Pi versus 6j. To work
s,v s,v
backward from Pi to determine 31 , it is only neeesmsary to deter-
mine experimentdly the solubility of the prodrug in the vehicle of in-
v
terest. In view of the difficulty in predicting Cf , it is almost easier
S, v
StV
V
.
v
to predict Ji once C i is known than it is to predict Ci This may
s,v
be beeause Ji is predicted from calculated Ki which represent the
ratio of the solubilities of the prodrug in two phases at equilibrium.
Therefore, any contribution to the irregularity of the wlubility of the
solute peculkr to its structure would affect the solubilities in both
phases, and the results would tend to cancel each other (31).
The initial article by Cohen and co-workers (18) studied the ef-
fect of the structure of aIkanoic acids on their abaty to diffuse
through porcine skin and on the effect of the physicachemical prop-
erties of various solvents on the delivery of pmplonic acid from the
solvents thmugh porcine skin. For the first, it was assumed that
when a maximum flux of the dkanoic acids was reached the 6 of the
skin matched the 6 of the acid. Because the maximum flux was ob-
tained for butyric acid ( 6 = 10.0), the apparent 6 of skin was as-
sumed to be about 10.0 ( ~ a l f c r n ~ ~ There
~ f ~ was
. an excellent car-
relation between the permeability coefficients far the delivery of
various straight- and branched-chain dkanoic acids from pentane
with the square root of the volume cohesive energy product. How-
ever, the expected pwabolic mlationship between the permeability
coefficient for the delivery of propionic acid from various solvents
and the 6 of the solvents with a minimum in rate for the solvent
exhibiting a 6 closest to that of propionic acid was not observed.
Instead, there was essentMly a direct relationship - with a steady d e
s,v
crease in Pi corresponding to a steady increase in ti of the sol-
vent. This result was attributed to a breakdown in regular solution
behavior beyond a certain polarity. However, this result &so may
have been because saturated solutions could not be used in these
experiments.
The focus of the two most recent articles by Cohen and co-
workers (32,33) has been an attempt to separate and to quantify
the two variables responsible for the flux of a solute across the
skin barrier. These two variables are the thermodynamic activity
of the solute in the vehicle, on one hand, and the extent of the aX-
teration of the barrier function of the skin caused by the vehicle,
Use of" Sotubitity Parameters in Skin Transport 267

on the other. The first paper describes the effect of afkmoic acids
on the delivery of theophylline through human skin, To separate
the purely thermodynamjc effect of the dkanoic aeid vehicles ("'push
effectw resulting from excess free e n e r m ) from the abiXitg of the
Efilkanaic acids to solubaize the penetrant in, the skin ("pull e f f s t T t )
s,v( 1) s,v(2)
a ratio of pmtition coefficients ICi /Ki cztfculated fmm
r e p l a r mlution thwry ( E q . 6) was determined for the delivev af
theophyllhe from the two dkanoic acids b e h g compared. Because
s,v s,v
Pi = Ki .
constant (from Eq 13) , if the experimentally determhed
s , v ( l ) s,v(2)
ratio Pi /Pi was equal to the theoretically determined ratio
s , v ( l ) s,vC2)
Ki IICi then only thermodynamlcaXf.ygenerated push effect8
were responsible for the flux of thwphyllhe from the two solvents,
On the other hand, for example, if the vehicle v( 1) affmted 6& by
m&hg it more polar, then the experbentd ratio would be larger than
.
the theoretical ratio Conversely, if vehicle v( 22 is the one that
causes gs to become more polar for that diffusion exper;irnent, then the
experimental ratio would be smdler than the theoretie81 ratio.
sI
The experimentila results showed that the P i Tor the delivery
of theophyllhe from propionic , hexanoic , octanoic , and a mhture of
wtanoiie and hexmoic acids was inversely related to the 6 , for the
8 l v'
acicE, that is, Pi decreases a s cSv hcreased and bmane more like 6i,
8 v
However, the Pi for the delkery of theaphylline from mhtures of
a m o i c acids c o n t a h h g propionic acid was much higher than expect-
ed based on considerations of the mkturefs solubility paritmeter, At-
s , v ( l ) s,v(2)
so, the experirnentdly determined ratio of Pj /Pi in any
eomparimn of the performance of twa dkanoie aeidi was alwlzgrs smaller
s , v ( l ) s,v(22
than the theoreticdly determined ratio of Ki /Kg if propionic
acid was v(2) and always larger if propionic acid was v( 1). There-
fore, it was concluded that the flux of thwphylline from shgle- or
binary-component mktures of alkanaic acid vehicles was caused by the
thermodynamic push effect, if pr~pionicacid was not included but, if'
pmpiionic aeid was isleluded, then an additiond pull effect by. the ve-
hicle was observed,
The second of the tvvo most recent articles by Cohen and co-
workers describes the effect of alkanoic a d d s on the de1iver;y of
a d e n o ~ b ethrough human s k h , For mktures of hexanaic and pro-
s,v
piionic acids a bell-shaped dependence of observed In Pi on vol-
ume fraction of hexanoic acid was observed. This dependence was
attributed to the same two opposing uadables just identified, fn-
creases in the volume fractbn of hexanoic acid in a vehicle caused
an bcrease in the excess Tree e n e r m of adenosine 31that vehicle
~llnc3,hence, an increased push effect, Qn the other hand, in-
creases in the volume fraction of propionic acid in a vehicle caused
an increase in the flux of propionic acid from that vehicle and,
hence, an increase in the pull effect. A similar result was obtained
for mixtures of propionic acid in isopropyl myristate,
The enhancement caused by the pull effect of propionic acid in
the hexanoic-propionic acid mixtures was determined by first cdeu-
S,V(X) s , v ( ~ )
lating the corresponding ratio of theoretical Ki /Ki and
s , v ( l ) s,v(2)
equating it to a ratio of theoretical Pi /PI ( E q . 13) derived
s,v(l)
from regdar solution theory. The ratios of theoretical Pi 1
Pis*v'2) values were normalized by assuming that the flux of adenosine
from pure hexanoic acid was caused only by a push effect, therefore,
s,v(2) s,v(2)
Ki and, hence, Pi in each calculation is for hexanoie acid.
8,V
The product of the experimental Pi for hexanoic acid and the theo-
ssv( 1) s,v(2)
retical PI /Pi was assumed to give a value for the contribu-
S,V
tion to the total or experimental permeability coefficient Pi,total only
s,v S,V s,v
because of the push effect, or Pi, push. The ratio of Pi,total/Pi,push
was taken as the enhancement factor from the pull effect. The largest
enhancement factor observed was about 6.4 for a mixture containing
0.4 volume fraction of hexanoic acid.

IV. CONCLUSION

Regular solution theory can be applied to numerous partitioning


prmesses of biologic& importance to allow investigators to predict
the effect of vehicles or solvents and the effect of systematic
changes in the physicochemical properties of drugs on the partition-
ing process. In percutaneous absorption, the effect of vehicles on
the themodynamic dridng force for diffusion has been separated
Prom vehicle (solvent) effects on the integdty of the skin itself by
using second application experiments or by comparing experimental
permeability coefficients with theoretical permeabsity coefficients
calculated from regular solution theory. Analyses of results of the
thermodynamic effect of vehicles on percutaneous absorption sug-
gests that such experimental systems can be explained by remlar
solution theory and that such analyses have predictive value. Sim-
ilarly, the experimentdl permeabitity coefficients of homologous series
of prodrugs for the delivery of a parent drug from different vehi-
cles can be correlated with the calculated solubility parameters of
the prodrugs. Whether the relationship is direct or inverse depends
on the solubility parameter exhibited by the vehicle.
Use sf Solubility f a r m e t e r s in S k i n Transport

REFERENCES

5, W. Wildebrand, J , A m , C h e m , Soc, , 52: 66, 1929,


G. Scatchard, Chem, R e v , , 83321, 1931.
M. J, Chertkoff, and A . Martb, J. Am, P h a m . Assoe, , Set,
E d . , 49: 444, 1960,
A. Madin, P. L , Wu, 25. Gron, and S. Cohen, J, Pharm. S e i ,
74: 638, 1985.
A. Martin, and J. Mauger , Am, J , P h a m , Ed. , 52: 68, 1988,
J , Pergusan, Proe, R. Soe. (Land,), B12T: 387, 1939,
J. H , Hadebrand, and R, L , Scott, The Solubility of Nonelee-
t r o t y t e s , 3rd ed, Chap, 3, Dover, New Uork, 1964,
S , A. Kha12, and A , N , Martin, J . Pharm, S e t , , 56:1225,
1967,
J . A. Ostrsnga, J, Med. Chern., 12:349, 1969.
T. Fujita, J, Xwasa, and C. Han~cff,J , AM. Ghem, Sac,, 86:
5175, 1-964,
S . S, Dads, Experz'nentfa, 26: 671, 1910,
S, Cohen, A , Goldsckmid, G , Shtacher, S, Srebren&, and S,
GitZ-er, Mot, PhamaeoE, , 11:379, l975,
E . M e Landau, 3, Richter, and S e Gohen, S. Med. C h e m , , 22:
325, 1979,
S . Srebrenik, m d 3, Gohen, J , P h y s , C h e m , , 80:998, 1976,
L. J. MuEins, Ghem, R e v , , 54:289, 1954,
A. Adjei, J. Newburger, $3. Stavchansky, and A . Madin, 6.
Pharm. Sci. , 73: 742, 1984.
P. Bustamante, and E , SeBes, J. P h a m . Sei., 75: 639, 1986,
Z , Limn, and 6. Cohen, J, Pharm, Sci, , 7'3: 538, 1984.
K. 2 3 , Sloan, S . A , M, Koch, K , C. Siver, and F, B , Rower,
J. I n v e s t . BermatoE, , 87: 244, 1986,
K , B . Sloarr, K , G . Sivar, and S . A. M. Koch, J, P h a m ,
Sei., T5: 744, 29863.
R e P, Waranis, K . G , Siver, and K . B. SXoan, Int, J . P h a m , ,
36: 211, 1987,
E . F, Sherertz, I(;, B . STaarr, and R , G , McTiernan, S, I n -
vest, Dermatol, , 89: 147, 1981,
.
E , F, Sherertz, K , B Sloan, and R. G MeTiernan, J, In-
v e s t . Dermatol, , 89: 249, 1987.
R , P . Waranis, and K , B , Slaan, J , Pharm, Scf,, 763587,
1987.
.
It. P , Waranis, and EC. lE3r Slom, J, Pharm, Set. , T7: 210,
2988,
.
K , B Sloan, E , F. Skerertz, and R. G. McTlernan, Int, J ,
Pharm., 44:87, 1988.
R. J. Scheuplein, b. I n v e s t , B e m a t o t , , 45: 334, 1965,
2 70 Sloan

28, A , F. M. Barton, Chern. R e v . , 75: 731, 1975.


29. R. F. Fedora, Polyrn. Eng, S c i , , 14:147, ,974-
30. P. H. Dugard, and R. C. Scott, Xnt. S. Pharm., 28: 219,
1986.
31. S. H . PaIkowsky, in Design of Biopharmaceutical Properties
Thmugh Pmdmgs and Analogs, Chap. 13. ( E . B. Raehe,
..
ed .), American Pharmaceutical Association, Washington, D C
1977.
32. R, Kadir, D. Stempler, Z. Limn, and S. Cohen, J . Pham.
Sci., 76:774, 1987.
33. R. Kadir, D, Stempler, Z . Liron, and S. Cohen, J . Pharm.
Sci., 77:409, 1988.
Use of an Epiderma
Cutaneous Toxicity Studies

F. L. VAUGHAN University of Nichigan S c b o l of Public


Health, Ann Arbor, Michigan

t , INTRODUCTION

The skin is one of the organs most frequently exposed to gaseous,


liquid, and solid eneronmental substances that can result in a
breadown of Ezameostatic mechanisms, Contact with chexnieds used
in hy&ime, medication, cosmetic application, as well as chemicals
that result from the development of industrid products, is a con-
stant public health concern. Thus, substances that may cause skin
irfitation are among those that are constantly under surve-illance,
Animd assay systems usually involve the application of these sub-
stances onto the shaven backs of rabbits and other animds, followed
by inwbation pedods that result in lesions of vadous severity as
quantified by subjective pathoXogica2 examinations. The Draize skin
test ( 1 ) fs used most frequently by industry and is recommended by
replatory agencies in aissaying and classifying ckremieds potentially
irdtating to the human skin. The significant vaAab3it-y of this and
simaar in viva animal tests in accurately predicting irritancy So hu-
man skin has been observed and reported ( 2 ) . Studies have d s a
shown that results obtdned from such animal experiments are often
equivocal and sometimes do not refleet responses that occur when
humans are simaarly exposed ( 3 ) - It is also partieular'ly diffimlt ta
study mechanisms sf toxicity in the whole ianimd, in part, because
of the high levels of exposure necessary to elidt a measuretble re-
sponse and &so because of the undescdbed interactions sf vadsus
physiological systems, However, rewlatory apncies o&r little in
the way of alternatives for preevaluating pmdiucts bef"are they can
be released for public use (4). What is needed is an in TI-itro sys-
tem that will allow the investigator to examine the etioloa, at the
molecular level, of the toxicity of certain classes of chemicals and
obtain more information on the predictabnity of the eutanwus i r d -
tancy potential sf such chemicds.
Mdnly for these reasons, in vitra systems are being developed
to study cutaneous toxicity, Such systems include mammdian cells
and tissues m&ntained in d t r o by using standard tissue mlture
techniques, as well as cell-free systems prepared from l i ~ n gtissue
for assay of subcellular changes caused by xenobiotics, Valuable
information is currently being generated with these systems. One
of the more promising contfibutions of in vitro systems anticipates
the use of human biolo@c& materid iin designing mode-ls for study-
ing toxicity.
Correlations between in d t r o cytotoxicitg and local irdtancy in
viva were described by Schmidt, et al, (51, who used established
cell lines of mouse liver and connective tissue, for a system assay-
ing anesthetics, Another early developed cell culture system was
descAbed by Guess et al, (8) in which cylotoxicity was determined
after exposing agar-overlay monolayer cultures to chenxicd irstants
of varying potency. The test was reported to correlate well with
implantation, htradermal , dermal, and intraocular tests fn rabbits .
Some investigators have used cells derived. from the skin in de-
velaping cutaneous toxicity in vitro assays. It is felt that the re-
sponse of cells f~omthis tissue would more c3ssf31y refleet skin irri-
tation. Early exgerjirnents by N u and co-workers (I) demonstrated
a correlathn between in ~ t r cgtoloxicity
o of topical medicants to
keriltinwyte cultures and local irritancy in vjivo. Kao and asmci-
ates (8) employed an organ culture system composed of full-thick-
ness mouse skin to quantitate cutaneous toxicity. The 48-hr organ
culture was treated topically with a xenobhtic and deletedous mor-
pholo@cal changes were correlated with the incorporation of radio-
labeled DNA, and protein precursors, as well as with the leakage of
intracellular enzymes into the culture medium,
A s a result of these, and other earlier studies, some desired
cfnaracteristics of mammdian cell systems have come t a the forefront.
For example, it has been established that, in some instances, cell
types that become permanent lines with indefinite mortality through
genetic dterations , lose much of the metabolic aetidty expressed by
their progenitors. A s a result, many cell culture systems currently
under study use pdmary cells, which generally have retained many
of the od@na_l genetic markers of the parent cell. Another example
of the desirability of specialized in vitro cellular a c t i ~ l yis the ob-
servation that cells cultivated and maintained in a microenv?ironment
simsar to that in d v o tend to display simnar morpholo@ctsl and met-
abolic activitie~. Tissue cell types growing as monolayers sub-
Epidermal Culture to Study Cutaneous Toxicity 2 73

merged in growth medium generally do not display the homeostatic


interaction seen in cell cuftures that have orgmizationd structure
similar to that observed in vivo. As a result of these and other
observations, some investigators are developing in vitro models us-
ing primary cells isolated from living tissue. Attempts are also made
to reconstruct the in situ microenvironment of the original tissue as
closely as possible. It is hoped that this approach will contribute
to obtaining more valid information in assaying the toxicity of xeno-
biotics .
The skin is divided into two main parts: the dermis or support-
ive connective tissue containing blood vessels, nerves, etc. ; and
the epidermis, which is the outermost structure of the skin and
which is most responsible for protecting the body from external
chemical and physical agents. We feel that an ideal in vitro system
for studying the mechanism of action of chemicds, and for screen-
ing chemicals to determine their potential toxicity to humans, would
be one that mimics the human epidermis both in morpholo@cal and
in biochemical stmcture. Such a system, therefore, should meet
the following criteria:

1. MorphoXogically , it must exhibit all of the stages of epidermal


differentiation as seen in intact skin. This should result in
keratinization of the outer cell layers as a result of ordered,
.
step-by -step differentiation or arthokeratosis Also, most of
the unique morpholo@cal markers of the epidermis should be
produced.
2. Biochemical markers of epidermal differentiation should be ex-
hibited as it takes place in situ. Among those changes are

a, Progressive cell surface alterations in the keratinocytes as


they differentiate
b, The production of specific keratin peptides found in termi-
n a y diffemntlated keratinoqtes
c, The synthesis of a specific epidermal histidine-rich protein
found only in that tissue

The cell type that is most respons.ible for the organizd epider-
mis is the keratinocyte. Other cells such as melanocytes, Langer-
han cells, and macrophages are also present, but they are not in-
volved in forming the anatomical barrier that is characteristic of the
epidermis. The basal cells, which are the only keratinocytes capa-
ble of pmliferation, form a single layer above the dermis. This
layer is connected to the underlying dermis by the basement mem-
brane. Basd cefls differentiate to spinous cetts which form multi-
layers immediately above the basal layer. Keratinocytes in the next
stage sf differentiation are called granular cells because they contain
keratohydin vanules , an important structure in the formation of
keratin. The cells at the ternrind stage of differentiation are the
cornifled cells that form the outermost desquamated l ~ y e r sof the
epidermis. ft is beEeved that tonofsarnents and keratohydin gran-
ules combine in same way to farm the keratin that characterizes the
terminally differentiated keratinocyte. Another cfraractedstic unique
to the epidermis is the formation of desmasomes-structures that
form very. tenacious bonds between the cells in various layers, New
idesrnosomes are bonds between the bas& keratinocyte and the bass-
ment membrane,
Therefore, any chemical that contacts the epidermis is presented
with all of the cellular components of the epidermis, We have devel-
oped a difff-lrentiated keratinocyte culture system that lFsrms a strat-
ified squamous epithelium with morphols@cal. and biochemicd markers
like those observed in an epidermis &med in situ, Keratinocytes
are obtdned as p ~ m a r yisolates (to retain as much of the od@n&
metabolic actidty as possible), seeded on an apgropriste substratum,
allowed to attach and proliferate, then, raised to the air-medium
interface to stimulate further growth and differentiation, This meth -
od creates a microenvironment simgar to that in which a nsrmd epi-
dermis develops, A s a msult , the cells differentiate and farm a
homeostatic relationship with each ather that results .in a tissue that
would be expected So react to a xexliobiotic applied to itits surface in
a manner simBar to the epidermis. We have &so developed methods
of applying test substances to the surface sf this keratinocyte cul-
ture system and have obtdned. data concerning the effects that eer-
tain classes of chemicals have on its metabolism of macmmdecules .
Epidermal cultures of this type have been established using both
rat a d human basal keratinocytes isolated from epidermal fragments,
The system currently used for studying the effects of xenobiotics
applied to the surface of the eulture has as its substratum a corn-
mereiafly a v ~ l a b l enylon membrane, rather than biolo&cal materids
such as collagen.
The early, h i t i d molecular events that t&e glace after topical
application of bis( 6 -ch'foroethyl)sulfide (B CES) to the surface of the
ezllture has been identiffed using this culture system,

11, DEVELOPING CULTIVAT1ON PROCEDURES;


A. 1soIat"io-n of Rat ar Human Keratinocytas for Cultivation
Two types of cells are the major stmeturd components of the nor-
m& mammalian full -thickness skin : fibroblasts and keratinoe-ytes .
Fibroblasts, found in the dermis and responsible for the formation
Epidermal Culture to Study Cutaneous Toxicity 2 75

of collagen, can be cultured in d t r o and subcultured in sa bas& me-


dium with semm as the only supplement (9). Keratinacy tes, which
form the epidermd stmcture, do not proliferate as well in ~ t r in o
primary, wltivation or a f e r passage when semm is the only supgle-
ment to bas& medium. Qnly the basal cells harvested from the liv-
ing epidermis engage in pmliferative acti-vity (10,11), The more
differentiated cells appear to have active DNA that engaged in re-
pair but not semiconservative synthesis (12).
Sever& systems have been developed to promote increased s e d d
.
cultivation of keratinocytes Improvements resulted when eells were
grown on collagen ( 13,14) , on feeder layers of irradiated fibr?sblasts
( 15), or in conditioned medhnt ( 16). Later reports indicate that
these systems may not be necessary (16-18) and that seriaf, cufti-
vation of keratinocytes can be achieved by supplernentfng the medi-
um with growth promoters such as epidermal growth factor ( 19,20)
and hormones (17,21). We repor.ted a descdption of the marpho-
loecaL3 and ctxltural charactefistics of rat keratinocytes after supple-
menting the growth medium with various concentrations and combina-
tions of hydrocorYtisone and insulin (22).
In our investigations into successfully cultivating keratinacytes,
full-thickness skins from both newborn rats and human biopsy
specimens were obtained from which the desired cells were isolated,
using essentiauy the same pmcedures, The skins were aseptiealligr
removed, placed in a petri dish and loose connective tissue and
blood vessels scraped from the underside af the dermis, This scrap-
ing also caused the skin to sbetch slightly, The resulting tissue
was chnled to 4OG and covered with cmde trypsin (1:250) pre-
chaled to the same temperature. This combination of stretching and
tqpsilnlzing the skin at 4OC caused the dermis and epidermis to
separate at the basement membrane and the loosening of desmasomd
attachments between basalt and lower ~lpinouscells, These cells,
mostly in the form of agpeg&es, were removed from the epidermd
fragment by bmshing them into growth medium contdning .fetal bo-
vine semm (FBS), The remaining epidiernd fragments containing
differenltiated ceUs m d stratum eorneum were discarded. Bmshing
the cellular components into qowth medium also suspended single
bas& cells, fibroblasts, and cellular d&ris, These were removed
by layering the resulting suspension on 10%Ficoll foXlowed by cen-
trifugation lFor 5 min at 13 x g (22). The pellet cont&ning aggre-
gates of basal cells was resuspended in fresh msc2.ium, and this
washing pmcedure repeated seven times to remove most of tkrs single
cells and debAs. Cell aggregate ~ a b n i t ywas determined using a
dye, exelusion test with 0.5% e q t h r o s b B (231, The ~ a b i f i t yof
these suspensions was routinely greater than "38%as indicated by
this method,
Figure 14.1 Attachment and growth of pdmary cultures of rat bas&
keratinocytea 24 h r after reeding on a plastic substratum, Aggre-
gates of basal keratinocytes are forming ~ o w t hcenters (arrows).
(Phase-contrast micrograph, x 540 1

The clafified suspension of basal keratinocytes was pelfeted at


53 xg for 10 min to obtiitlin a packed volume and a 0.2% (vollvd)
suspension prepared in complete growth medjtum, These suspensions
contsined 5 x 105 cellslml as determined by counting dissociated
.
cells electroniedly When these suspensions were seeded onto ap -
propriate substrata and incubated at 35°6: in an atmosphere of 5%
CQ2 and 95% air with a humicli$y of 95%, the aggregates of basal
cells and the few rerndning spinous cells attached within 12 ta 18
h r (22). Each aggregate of basal cells formed ai gmwtfl center from
which cells migrated and began proliferation as shown. in P i p r e 1,
Although the spinous cells attached initidly-, they did not prolifer-
ate and were detached and removed at the first change sf growth
medium ( 2 4 h r aRer seeding). Continued incubation for I days
pmduced a confluent monolayer of closely connected epithelial cells
(Fig. 2).
Epidermal Culturn to Study Cutaneous Toxicity 277

Figure 14.2 Proliferation of primary. cuXtures of rat basal keratino-


cytes 7 days after seeding orr a plastic substratum, A canfIuent
monolayer has developed. (Phase-contrast micravaph , x 140. )

B. Selection of an Appropriate Substratum for Attachment and


Growth of Keratinocytes
1, Plastic Substrata
When sufficient amounts of cells were seeded initially on eommercli81fy
prepared plastic ctrlture vessels, monolayera formed in 5 to 7 days
(example: 15 x 105 cells in a 60-mm plastic petri dish). The ker-
atinocyte cultures just descdbed (shown in Figs. 1 and 2) were
seeded on plastic vessels. Some stratification and differentiation
eautd also be observed 1.5 to 2 weeks after culture initiation.

2. Collagen Substrata
Various kinds of collagen gels were used as substrata and the re-
sulting attachment and proliferation of keratinoeytes after appropd-
ate seeding observed, The geIs were usually formed in the bottom
of plastic petri dishes and the cells seeded on the surface of the
gels in suffident medium to completely cover the cells.
When rat keratinocytes were seeded onto Vitrogen 100 bovine
dermal collagen, which is composed of 95% type I and 5% type 11 col-
lagens, firm attachment to this substratum resulted (24). However,
3 days after seeding, growth resulted in clusters of cells rather
than in a uniform proliferation over the substratum surface as seen
with plastic substrata. At this time, holes were also produced in
the collagen gel, and the surface began to dissolve. If the culture
was incubated longer, the entire! collagen Iayer disappeared, and
the cells were seen growing directly on the plastic surface of the
contdner.
When rat tail collagen was used as a substratum for the growth
of rat keratinocytes, attachment was observed 6 h r after plating,
compared with 18 to 24 h r on commerciaI plastic vessels (24). By
24 h r , 75% of the surface of this substratum was covered with kera-
tinocytes that had spread and proliferated. A confluent monolayer
was formed in 4 days. However, gels formed from this collagen are
very soft and difficult to handle.
Gels formed from mixtures of Vitrogen 100 and rat tail collagen
were &so investigated as appropriate substrata (24). A mixture
contdning equal amounts of the two types (1: 1, vol/vol) supported
attachment and growth of rat keratinocytes as well as rat tail col-
lagen alone, and also resulted in a firm substratum that would allow
relocation of the entire culture to the dr-medium interface (a pro-
cedure that will be described in a later section). This collagen
mixture also supported stratification and differentiation of submerged
keratinocytes to a greater extent than plastic substrata.

3. Synthetic Membranes as Substrata


Xn a concentrated study, 11 commerciaily available and experimental
synthetic membrane filters were selected and evaluated as appropri-
ate substrata for rat keratinacytes (25). In Table 1, the source of
each membrane and the material from which it was constmcted are
listed. Membranes TCM200, TCM440, HA-TF, and RA-TF were se-
lected for study because they are specifically treated to remove de-
tergent and are more suitable for tissue culture use. The Aero-
shidd D (ACRO) and silicone-polycarbonate mgolymer (MEN) were
included because they are transparent and wouId allow cell growth
to be continuously monitored by phase-contrast microscopy. A11
membranes used in this particular study were 13 mm in diameter and
were placed In the wells of 24-well plastic culture vessels before
seeding a suspension of keratinocytes on their surfaces. Wells
without membranes were also seeded and served as a basis for com-
paring the resulting attachment and w w t h of cells on each mem-
brane (plastic controls).
Table 14. I Synthetic Membranes Selected as Substrata in Studies of Attachment and Pmliferalion of Rat
Keratinocytes In Vitro

Pore size
Membrane (vm) Matehl Source

P200 Nylon (Pumpor) Gelman Sciences,


Ann Arbor, Mich.
P450 Nylon (Pumpor) Gdman Sciences,
Ann Arbor, Mieh.
TCM200 Cellulose triacetate Gelman Sciences,
Ann Arbor, Mich.
TCM450 Cellulose triacetate Gelman Sciences,
Ann Arbor, Mieh.
HA-TF Cellulose nit rate Malipore, Bedford,
Mass.
Cellulose acetate
RA-TF CeUulose nitrate Malipore, Bedford,
Mass.
Cellulose acetate
RT Polysulfone (200W) Gelman Sciences,
Ann Arbor, Mich.
ACRO Polyethylene Gelman Sciences,
Ann Arbor, Mfch.
MEM Silicon polyearbonate General Electric,
..
Scheneetady , N Y
Initid attachment and proliferation of rat keratinocytes on se-
lected membranes and plastic culture vessels have been compared
and reported f 25). The extent of attachment and growth of cells an
opmue membranes were determined after fixing the cells and sub-
stratum in f 0%phosphate-buffered li)rm&in or Carnoyls solution (90%
butanol, 10% acetic acid), The f k e d culture was Stained with he-
matoxylin , dehydrated, cleared with xylene , and mounted on micro-
scopic slides with Permount and coverslips, Quantificatkn of initial
attachment; and subsequent pmliferrzlion was accomplished by count-
ing stdned nadei per unit area of the substratum. This was per-
formed electronicdly with the Bioioquant Image Andysis system on the
Apple ZIE: micrmomputer and accesso&es. Cross sections for micro-
seopfe examination were obtdned by e r s t %ing selected cultures in
10% buffered firmdin, embedding in paraffin, sectioning with a ro-
t a microtome,
~ and mounting sections on Klicroscopie slides and,
ffnally, stajlning with hematoxflin and eosin (N&E).
EpiderrnaX cultures grown on synthetic membranes were also ex-
amined at the ultrastmnctur& level, Cultures were fixed in Rarnov-
sky's fixative, washed, and prestained with 2% osmium tetroxide ,
sf ~ n e denblock with uranyl acetate, dehydrated, and embedded in
resin. The embedded specimens were polymerized, sectioned, post -
stdned with uranyl acetate and lead citrate, and observed with an
AEX Gorj;nth 275 transmission dectron microscope (TEM) .
F i p r e 3 Blustrates the initid attachment (see Fig, 3A) and en-
suhg pmliferatian (see Fig. 3B) of rat keratlnocytes an synthetic
membranes, compared with growth on a plastic substratum. The
data in this f i p r e indicate that cultivathn on some of the selected
membranes was supmior to that on plastic mlture vessels used as
controls. Observation of cultures aAer a f -dily incubation showed
that aggregates of keratinaeytes had attached m d spread out,
forming growth centers, The large variation in cell number per
unit area seen jin Fimre 3A reflects the different size and unequd
distribution of w w t h centers in early cultures, After IQ days of
incubation, monolayers had formed with varying degrees of compact-
ness, depending on the substratum (see Fig. 3B). Membranes spe-
cially prepared for tissue culture procedures (i.e. , TGM200, TCM-
450, HA-TF , and RA-TI") supported attachment and pmliferation
equal to a r better than the plastic substratum. The increased at-
tachment and proli.EFtration of keratinocytes on Pumpor nylon mem-
branes (PZQO and P 45 0 ) over controls was st atistically significant and
pmliferation on F200 was supedor to the others examined, There-
fore, this nentbrene was selected for further study in producing
stratified, differentiated keratinocy-lte cultures in d t r o (25).
EpidemaZ Culture Lo Study Cutaneous Taxicity

4000
a:
LJ
b-
Ld
3000
A
,A
.'.".
2
Lcl
ar 2080
4
3
0
U)

I000
.A
A
LLJ
C)

0
Control P4SQ TCM.I.50 RA-TF ACRO
P200 TCM2QO HA-TF H-T MEM
SUBSTRATUM

Control P450 ICM450 RA-IF ACRO


P200 ICEut200 HA-"T F--T MEM
(B) SUBSTRATUM
Figure 14 . 3 Attachment and praliferation of rat kcj~atinoeytesan
steaized synthetic membranes. (A) Attachment and growth lFor 1
day; ( B ) pmliferatiron for 10 d w ~ i r . Data are preaenhed as mean +
SD, n = 4-5,
4. A t tackment Fae tors Inzres tiga ted in the Development of Epz'demat
Cultures
Two attachment factors were studied for their abnity to enhance the
attachment and pmliferation of rat keratinocytes seeded on synthetic
membranes (25). Human fi't'bronectin (WFPJ) is a @ycuprotein isolated
from human s e a m and was found to be essential for promoting the
attachment of epithegal cells in viva ( 2 6 ) . Laminin (LMN) is a pro-
tein that is part of the structure of the basement membrane of the
epidermis (27) and is commerdally avdable, as is HFN,
The membranes greeoated with HFN did not suppost; the attach-
ment and proliferation of keratinocytes better than untreated mem-
branes (P 3 0.05) as shown in Pimres 4A and 4B. On membranes
P200, TGM260, and TCM450, there appeared to have been less pro-
lihsation on HFN-coated than on uncoated membrmes as observed
aAer 10 days of incubation (see Fig. 4B). The coating of mem-
branes with LMN enhanced the initid attachment, compared with un-
treated control membranes after 1 day of incubation f see Fig. 4A) .
Enhanced attachment and spreading of seeded keratinoeytes across
the substratum was particularly e ~ d e n twhen LMN pretreated
P u r u p r nylon membranes were compared with untreated membranes.
However, only slight, although significant, differences in the totaj
number of cells on untreated or LMM pretreated nylon membranes
were observed after incubation for 10 days (P < 0.01; see Fig. 4B).
Therehre, precoatirtg nylon membranes with ZMN appeared to en-
hance initial attachment but did not support proliferation mere than
did the untreated membrane. The LMN-coated HA-TF and RA-TF
membranes &so supported keratinocyte sowtfi substantially better
than uncoated membranes (P < 0,001). However, detachment of rat
keratinocytes as sheets from the membranes and wlture vessds
precoated with LMN was observed in most cultures after incubation
for approximately 1 4 days, Because of these observations, neither
HFN nor LMPJ were used In further studies af the cultivation of"rat
keratinocytes on synthetic membranes (25). Studies of the effect
of attachment factors on the cultivation of human. keratinocytes are
incomplete.

it 1, STRATfFICATfQN AND DIFFERENTIATION


OF KERATIMOCYTES CULTURED ON
NYLON MEMBRANES
A. Two-Stage Cultbation Pracdures
Isolated keratinocytes were seeded on untreated nylon membranes
and incubated as submerged cultures untn monolayers resulted, or
up to 7 days, The cultures were then raised to the dr-medium in-
Epidermal CUElure to Study Cutaneous Toxicity

6000
a L J Untreated
it-I
-i ELXI LMN
3 rni-iFN
,"'.A
4000
f
LJ
Bf
4
I3,
0
fyl 3,000
\
V1
J
.A
tJ
C_)

Q
Control P450 TCM450 RA--IF" ACRO
P200 TCCM280 HA-TF H-T MEM
SUBSTRATUM

8000
cCJx Untreated
r--
u
mLMN
SO00 rnHFN
3
5%
LLJ
sr: 4000
Q
2
(CT
V1
3;
A
2000
W
0
Q
Control P450 TCM45Q R A - I F ACRO
P2QQ PCM200 HA---TI"' t-4-T MEM

Figure 14.4 Attmhmnt and praliferrttbn of rat keratinocytes on


synthetic membranes that were untreated (UNTR), precoltted with
LMN, or precoatsd w i t h HFN. (A) Growth .for 1 day; (B) grawth
far 10 days, Data are presented as mean .t SD, n = 4-5,
Tap Petrr Ptate (Tf

Membranes {MI

Petri Otsh (PD)

Assembled Side View M

Figure 14.5 Illustration of the rnethad used in lifting keratinoeyte


monolayers to the air- medium interface, Papopor nylon membranes
contdning the monolayer are placed on the surface of glass fiber
fsters saturated with growth medium.

terface by using the pmcedure nlustrlzted in F i v r e 5, The mern-


braxles contdning the monolayer s f keratfnwytes were transferred
to the surface of a glass fiber iFjfter positioned at the bottom of a
petri Clish and soaked in growth medium, This permits the culture
to be fed from the underside wh2e the cells are in contact with the
atmosphere above (in the same manner as an organ culture).
This two-stage cultivation pmceilure (25) resulted in differenti-
ated mult.ilayers o f keratinocytes on the synthetic nylon membrane,
VaAous stages of differentiation were also observed that paralfeled
those observed in the epidermis formed in vivo. Multaayedng and
differentiation increased with the age of the XiRed culture, up to
approximately 12 to 14 days, after whieh time the number of layers
remaining and the extent of differentiation did not change. Con-
Epidermal Culture to Study Cutaneous Toxicity 285

Figure 14.6 A, TEM of an epidermid culture pown an a nylon mem-


brane submerged for 7 d w s and, lifted to the dr-medium interface
far 14 days. Severd layers of nucleated cells (n) formed zzbove the
membrane fm) and numerous layers of enucleated cells farmed a
straturn corneum at the culture surface f s ) f x 2200).

siderable stratiacation and differentiation: took glace between 4 and 8


days of incubation at the air-medium interface and after 8 days,
four to five nucleated layers and I5 to 20 enucleated l~tyerswere
produced,

B , UItrastructurat Examination of Resulting Cultures


Fimre 6 is a transmission electran micropaph (TEM) af a cross
section sf a culture incubated submerged for 7 days and then lifted
to the air-medium interface for am addition& 14 days. Five to six
nucleated and 15 to 20 enucfeated layers of differentiating kerrlilino-
cytes were pmdtxcsd ,
Various markers a f epidermal differentiation were clearly demon-
strated in TEMs at higher map@ications, A culture liRed for 14
days showed abundant desmosomes, taxlofilaments, and mitochondria
fn the bas& layer f Fig. 7 ) , and electron-dense panules, morpho-
lo@cdly simnar to in situ keratohydin panules, and cells resembling
those in the in sltu stratum comeurn were eddent in the upper fay-
ers of lifted cultures at this age (Fig. 8). Many of these epiderrnd
markers do not develop in submerged wltures,

IV, EXPERfMENTATIQN WITH SULFUR MUS-


TARD tN CUTANEOUS TOXICITY STUD-
IES USING THE EPIDERMAL CULTURE
A. Togicaf Application of a Xenobiotic ta t h e Surface of
the Lifted Cultures
This section describes the expedmentation to measure the toxic re-
sponse of epidernd cultures after topical appficatian of the sulfur
mustard, bis( f3 -chlorosthyl) sulfide (B GES) , Beta2s of the methodol-
oery have been reported ( 28). Aliqtuots of BGES dissolved in meth-
ylene ckrlofide (MG) (10 mg/ml) were f u ~ h e dnuted
r to selected
concentrations in one of the folfowhg solvents : 2,4-pentanedfone
(acetyl acetone), ethanol (ETQH) , hexane, or &methyl sulfoxide
(DMSQ). A, t s t d of 0.04 ml solvent or of solvent contdning BCES
was applied to the s u r f ~ c eof an epidermal culture grown on. a 13-
m m disk of P200 and lifted for 14 days, This amount covered most
of the culture surface without flowing over the side onto the sup-
poding glass-ffber fater, The amount of matedal applied was ex-
pressed as nanornoles per square centimeter of the culture surface
(nmolf erne). Solvent or solvent-BGES solutions were =moved from
the culture surface by gently washing in Earlets batanced s d t solu-
tion (EIBSS), The washed cultures were placed on the surface of
supporting glass-fiber fiXters saturated with fresh growth medium
and incubation continued for preselected periods, The cultures
were next transferred to the appropriiate medium conta_infmg radio-
labeled precursors when experiments were conducted to observe ef-
fects on macromolecular metabolism subsequent to vadous treatment
(28) *
Data were obtdned from preliminary expedments to select an
appropriate solvent for topied application of xenobiotics to the sur-
face of epidernzaf cultures. AcetylI acetone, hexane, and MC were
a31 highly toxic as indicated by sipibeartt reductions in the incor-
poration of PHI thymidine into DNA, when compared with untreated
controls. In further studies, it was observed tharndiltkns of
1ETBX-X and DMSQ in distilled water were much less toxic and were
&so appropsiate sogvents for BCES, Table 2 shows the results of
both [ 3 ~ thymidine
1 and [ 1 4 ~ leucine
1 incorporation into DNA and
Epidermal Culture to Study Cutaneous Toxicity

Figure 14.7 A %"EM of an epidermal! culture wown on a nylon xnern-


$ran@ submerpd for 7 days and lifted to the dr-medium interface
for 14 clays. This i s a higher magnification of the Iawer layers
next to the membrane (m) showing LanoEiXaments (t), and desmo-
aomes ( d ) ( x 12,0001,
Figure 14 . 8 A TEM o f an epidermal culture grown an a nylon mem-
brane submerged far 7 dztys ~ n lifted
d to the air-medium interface
far 14 d a y s . This is a higher napifitcation of the upper layers of
the culture showing keril-tohyaEn ~anu1c;rs(k) and cornified cells
(el f x 12,000).
Epidermal Culture to Study Cutaneous Toxicity 289

Table 14.2 Incorporation of HI Thymidine and [ 14Cl~eucineby


Epidemal Cultures After Topical Exposure to DMSQ and ETOH

Thymidine (GPM) Leucine (CPM)

Exposure time (hr) Exposure time (hr)

Treatmenta 2.5 24 2.5 24

Untreated eontrolb 1190 t 279 798 2 125 2.69 t 0.6 1.87 t 0.19
70%DMSO 658 L 138 564 t 110 1.27 t 0.03 1.42 t 0.29
40%ETOW 836 1 8 2 542 2 162 2.05 L 0.78 1.12 + 0.24
Heat killed 83 t 21 0.20 L 0.06

aCultures treated by topical application of 0,04 m l solvent,


bculture controls treated by topiwl application of 0.04 mf saline.
The values are the mean 1 SD (n = 3).

protein, respectively, after topical appucation of these diluted sol-


vents to culture surfaces (28). These data demonstrated that ex-
posure of the cultures to 40%ETOH or 70% DNSO affected metabolic
activity to a similar extent. There was no signExcant difference in
the incorporation of either thymidine or leucine by cultures exposed
to the two solvents for 2.5 hr or 24 hr (P > 0.01 in all compaAeons),
The table also compares the effects of the two solvents with those
04 untreated and heat-killed controls. It was decided to use 70%
DMSO, which was not toxic at a higher percentae than ETON, as
the solvent in subsequent studies with BCES.

B. The Effect of Sulfur Mustard on Macromolecular Metabolism


in Exposed Epidermal Cultures
In all experiments to determine the effect of BGES on maemmolecular
metabolism of exposed epidermal cuItures, two controls were em-
ployed: untreated cultures handled identically with treated cultures
but not exposed to the xenobiotic or to the solvent used to dissolve
the agent; and solvent controls that were similarly handled including
exposum to the solvent without BGES. Three cultures were used
for each experimental observation. Counts per minute per millipam
of protein or DNA were determined for each observation and results
expressed as a percenta* of untreated controls (28).
1, The Optimal Period for Exposing Epidermal Cultures to Sulfur
Mustard
Cultures were exposed for 0,5, 1, and 2 h r to 50 to 500 nmol/crn%
of BCES dissolved in 70%DMSO, The effect of tkis agent on D N A ,
R N A , m d protein metabollism, as detected, by the use of the radio-
labeled precursors thymidine, [%] uridine, and leucine,
respectrively , is summa~zedin F i p r e s 9A (thymidine) , 9B (uddine) ,
and 86= (leucine). These data incTiicate that the major effect of
BCES t&es place within 30 rnin after topied application. At that
time, the incorporation of all three precursors was decreased after
exposure to d l concentrations of BCES when compared with solvent
controls (P < 0.001). The most drastic decrease in maeromolecular
rnetaboliam was in the incorporation of [ 3 ~ thymiAjne
] (see Fig. 9A).
Decreases in the incorporation of the other two macwmolecular gre-
cursors after exposure was also eddent, Both uridine and leueine
incorporation was inhibited by exposure lo 200 and 500 nmollcn2 of
BCES (P < 0,001) (see Figs, 9B and 9C), However, at lower con-
centrations, much less inhibition resulted, when compared with in-
hibition of DNA, metabolism at those coneentra-tions,
These data demonstrate that a 30-nin exposure to BCES was
sufficient time to initiate early lesions in cellular function of the
epidermd culture as determined by inhibition of macromdecular syn-
thesis .
2. Dose-Dependent Inhibition of Macromolecular Metabolism After
Exposu~eof Epidermal Cultures to Sulfer Mustard
Experiments were performed to identify the minirnd concentration of
BCES that would cause measurr?ible alteration in macromolecular me-
tabolism. Concentrations of B GES were selected r a n e n g from 0.01
to 500 nmol/ema, and the data resulting Trom several experiments
were combined, The cultures were exposed for 30 rnin to the se-
lected concentrations of BCES dissolved in 70% DMSO, and the agent
washed away with EBBS, Treated cultures were transferred to the
labding medium immediately (0 hr) or incubated, in fresh growth
medium for 4, 8, or 24 h r a a e r exposure (postexposure pedod) be-
fore transfer to labeling medium, The purpose of tkis approach
was to determine the minimum concentration of BCES that would re-
sult in a significant inhbition of metabofism in the treated culture
and to observe the extent of this inhibition, or possible recovery,
over a periad. Figures POA and 30I3 show the results of 6383 thy-
midine incarporation. Fimres llh and. l1EI show results of incor-
poration of [ 3~ ] usidine and [ 1 4 3 Ieueine,
~ respectively.
The data in Figure 10A show the incorporation of radiolabeled
thymidine by the cultures after exposure to BCES , 0.0, 0.01, 0,1,
and O.5 nmoll/cm"fi There: was an initial stimulatbn of thymidine in-
CPNIJpG PROTEIN (96 CONTROL) GPM/pG PROTEIN (WONTROL)
TIME. POST EXPOSURE (hr)

Figure 14.10 Incorporation of [ 3 ~ thymidhe


1 by epidermal cultures
after a 30-min exposure to BCES in 70%DMSO. The cultures were
pulse-labeled after removal of BCES followed by selected periods of
postexposure incubation, BCES concentrations ( urn01lcm2) were
(A) 0.0 (solvent control), open circles; 0.01, open square; 0.1,
closed circles ; 0.5, closed squares: 03) 0.0, open circles ; 1.0,
open squares; 10, open triangles; 50, closed circles; 200, closed
squares; 500, closed triangles. Results are expressed as percentaw
of untreated control J- SD f n = 9).
Epidermal Culture to Study Cutaneous Toxicity

TIME, POST EXPOSURE (hr!

Figure 14.11 Incorporation of (A} [%I] uridine and (B) 11465 leu-
cine by epidermd cultures after a 30-min exposure to BCES in 70%
DNISQ, The cultures were pulse-labeled after removal of BCES fol-
lowed by selected periods of postexposure incubation. BCES con-
centrations (umo~ /crn2) were: 0.0 (solvent control) , open circles ;
0.01, open squares; 1.0, open triangIes; 10, closed circles; 50,
closed squares ; 200, closed triangles. Results are expressed as
percentage of untreated control t SD (n = 9).
corporation by both treated cultures and solvent controls 4-hr PO&-
exposure, pwbably caused by stimulation by the solvent. By 8-ktr
postexposure, all concentrations of B CE S included in this b w r e
caused inhibition of incorporation (P < 0.001). A concentration of
0.01 nmollcm2 caused inhibition for at least: 8-hr postexposure, but
mltures exposed to this dose showed a recovery of metabolic activ-
ity by 24 h r , when compared with the solvent control, However,
concentrations of 0.1 and 0.5 nmol lcmg of BCES resulted in signif -
icisnt hhibition of incorporation of radiolabeled thymidine as ob-
served at 4- and 24-hr postexposure. Fiwre IOB hclcldes csncen-
trations of 1 to 500 nmollcm2 of BCES and enxmmar;izes their effects
on thymidine incorgora~onby epidermal cultures, These concen-
trations caused sipificant inhibition at 4 h r and at 24 h r f P .:
O.001 for both observations), compareid with solvent contmls,
F i p r e 1 l A shows the results of [ 3 ~ uridine
1 incorporation by
cultures after exposure to BCES. Xncorporation of this precursor
after exposure to 0.01 nmatlemz of BCES appezrred to be stimulated,
when compared with solvent controls ( P < 0,OQI). This was evi-
dent 4 h r after exposure and remdned for the full observation peri-
.
od ( 24 h ~ ) A concentration of l. Q nmol / m a caused mild, but sf g-
nificmt , inhibition of hcorporation 24-hr postexposure, after an
initid stimulation of incovoration 4-hr postexposure. The higher
concent rations ( 10- 200 nrns~lcrnz) caused subst antis1 inhibition of
incorporation of this precursor, Similar results were o b t ~ n e din
the heorporation of radiolabeled leucine 'by lifted cultures after ex-
posure to BCES (see Fig. 11B). Exposure to 0.01 nmol /cm2 of
BCES stimulated the incorporation of radiolabeled feucine, compared
with solvent controls (P < O.001). Sipificant inhibition of leucine
incorporation occurred only after exposure to $0 and 200 nmol /ern2
of BCES.
The data in these experiments indicated that only at cornpara-
tively high concentrations of BCES was there sipincant interfer-
ence with RNA and protein metabolism. Inhibition in the uptake of
radiolabeled urYidine and leucine at these concentrations was proba-
bly a result of generalized cytotoxicity caused by massive DNA d-
kyXation and irreversible molecular destruction, Thymidine metabo-
lism, on the other hand, was affected at much lower concentrations.
Theee data suppor2: the condusion that DNA is the most important
target of BCES and are in agreement with observations of a number
of investigators. Crathorn and Roberts f 29) found that exposure
to a concentration causing DNA, damage and interference with repli-
eative abaity in BeLa cells did not bterfere with RNA or protein
synthesis. Lawley and Bmokes (30) observed that the cytotoxic
action. of sulfur mustard was not associated with the inhibition of
growth of BseherSchia colt as measured by RNA. or protein synthesis,
but with inhibition, of DNA, synthesis,
Epidermal Culture la Study Cutaneous Toxicity 295

This study demonstrated that the epidermal culture formed with


rat keratinoqtes behaved sim3arly to other in vitro systems in its
initial reaction to topically applied BCES. More directed studies
may elucidate what happens to the dkylated DNA mofeeule upon ex-
posure to the agent and how this initid molecular event is related
to subsequent cellular destmction leading to vesication. The mor-
phalo@cd markers (25) and bioehennicaj markers (24) of this culture
system p r o ~ d ethe opportunity to simulate what happens in the in-
tact epidermis, It may as a result be possible to identify the initid
molecular event leading to vesication and, also, beeause of the or-
dered differentiation of the culture, there can be a determination sf
which cells of the epidermis are affected initially,

V e SUMMARY AND CONCLUSIONS

The studies descAbed in this chapter demonstrated that keratino-


cytes athch, pmliferate , 8nd differentiate on sdeeted synthetic
membranes that are untreated except for stesization. Growth on
Puropor nylon membranes was supedor to that on other membranes
but there were others that supported powth q u a 1 to, or better
than, cornmereidly obt&necP plastic culture vessels. Membranes pre-
coated with RFN gave inconsistent results, hdicating that using
this attachment f'aetor offered little addition& value in supporting
rat keratinocyte attachment and proliferation in vitro, Precoating
membranes with. LMN hitially ssugported attachment and ~ o w t haf
rat keratbocytes better than untreated membranes, but extended
-incubation of eeas growing on that substratum proved unsuccessful.
Attachment and proliferation of keratinocytes on the groper sub-
stratum was essential in developing an epidermal culture for toxi-
eolo@c& studies, However, differentiation of cultured keratinocytets
was equaBy if not more important. The homeostatic relationship be-
tween keratinocytes in vadous stages af differentiatian may have
sf piBcant effects an the abEty af an agent to cause tissue damage.
Thus, establishing a microenvironment that fosters differentiation as
well as controlled gmlifc3ration is mandatory.
Another characteristic a successful in ~ t system
m must have if
it is to be used for studying the toxic effects of xenobiotics on the
epidermis ia simple, repmducibfe eonstmction, This is necessary if
data obtained from different Xabaratories are to be used collectively
h evauating the effects of toxic agents. The successful cultivation
m d differentiation of rat keratinocytes on nylon membranes helps to
fulfB1 these cdteda, Puropor nylon is a synthetic mated81 that can
be manufactured with minim& variation in its chenrticd nature, where-
as the preparation of collagen, for example, can vary simifieantly
from lot to lot and, thus, is difficult to standardize, Comparatively
296 Vaughan

large amounts of the nylon membrane can be sterilized by various


methods that will not change its chemical structure, and no further
treatment is necessary for optimal attachment and growth of primary
keratinocytes.
The studies described in this chapter also demonstrated that
the epidermal culture formed from isolated rat kerathocytes be-
haved similarly to other in vitro systems in its initial reaction to
topically applied BCES. DNA was the main target of this agent in
this system, the morphoIoEfical and biochemical structures of which
mimic the in situ epidermis. More directed studies may elucidate
what happens to the alkylated DNA molecule upon exposure to this
a p n t and how this initial moIecular event is related to subsequent
cellular destruction leading to vesication. This approach can also
be used to verify earlier observations in whole skin that the first
cella destroyed after BCES exposure are basal and lower spinous
cells, and that this Ieads to blister formation.
Studies of the untoward actions of drugs on the skin after top-
ical appucation are necessary before these chemicals can be released
for human use. Investigation of percutaneous absorption are fre-
quently performed in vitro using membranes of animsl or human ca-
daver sk& that bear little similar"ity to the living human epidermis
and suffer enormous variabnity. A s a result, they may show little
resemblance to the clinical performance of a given applied drug
(31). Therefore, what is needed is an in vitro model that possesses
barrier characteristics similar to those of the intact living epidermis
and that offers highly reproducible characteristics. The epidermal
culture developed in this laboratory resembles the in situ human
epidermis both morphologicdly and biochemically. With further re-
finement this system may become a valuable toof in investigations of
percutaneous characteristics pertaining to human health. Studies
have been initiated to evaluate the effectiveness of this in vitro
system in measuring transepidermal water loss (TEWL) .
REFERENCES

1, 3. H. Draize, 6 . Woodard, and R. 0,Calvery, Pharmacol.


E x p . Ther., 82:317, 1944.
2. C , S, Weil and R . A , Scala, Toxicol. A p p l . Pharmacol., 18:
276, 1971.
.
3. G . A. Nixon, C A . Tyson, and W. C. Wertz, Toxicot. A p p l .
Phamacol, , 31:481, 1975,
4. K. J, Falahee, C. S. Rose, R . E. Seifried, and D. Sawhney,
in Product Safety Testing (A. M, Coldberg, ed.) . Mary
Ann Liebert, New York, p. 131, 1983.
Epidermal CuZCure ta Study Cutaneous Toxicity 297

3. I;, Schmidt, P. C. Mcfntire, I3, L. Martin, M, A , Raw-


. .
thorne , and R K Richards, Toxicol, AppE. Pharmacot. , f :
454, 1959.
W. I;. Guest, S. A , Rosenbluth, B , Schmidt, and 3 , Autjian,
J. P h a m , Sc9. , 54: 1545, 1965.
F. Hu, C, S. L i ~ n g g s o d ,and J , F, Hadebrand, J , Invest,
DematoZ, , 26: 23, 1955.
F, Raa, 3. Hall, and F, M e Holland, Toxicol. A p p l . Pkamacal,,
68: 206, 1983,
W, R. Earl, Fed, Proc, , 17: 967, 1958,
IM. Karasek and T, Moore, J, Invest, Dematol, , 56: 205, 1971.
F. L , Vaughan, and 1 - A , Bemskin, J, Xnvest. Dermatol.,
56: 454, 1911,
F, L , Vaughan, R. S. illlitpa, and I , A , Bemstein, J , Invest,
Dermalal, , 66: 355, 1976.
M , Karasek and M. F , Charlton , J , Invest, Dernzatol, , 56:
205, 1971.
A , E. Freeman, H I J. Ingef, B. J , Werrman, and R , N.
Kleinfeld , In Vitro, 12; 352, 1976.
3. G , RheinwaXd and W. Green, Cell, 6: 331, 1975,
a , M, Peel and R e G , Ham, In Vitro, f8:516, 1980,
E . Elisinger, 6 , S. Lee, J , M , Hefton, Z, Darzynkiewiez, 3 , W ,
Chiao, and E. DeWarven, Proc, Natl. Acad, Sei, U S A , 76:
5340, 1979.
33, M, Peehl and R , C. H a m , In Vitro, 26~526,1980.
S. Gahen, and C , R. Savage, Recent Prog, Worn, R e s , , 31:
551, 1974,
J. G . RhehwaXd and H . Green, Nature, 2635: 421, 1971.
1, Hayashi, J , Lasher, and G . Sato, In Vitro, f 4 : 2 3 , 1978.
F . L , Vaughan, 1;. L. Kass, and 3 . A. U z m m , In Vitro, f 7:
941, 1981.
3. EI. PkrilEps and 2 , E , Terwbersy, E x p . Cell Res,, 13:34P,
1957.
1;. I. Bemstarn, F". L . Vrrugharr, and X I A. Bernstein, Xn
Vlitro, 22: 695, 1986,
F. L , Vaughan, R, W , Gray, and I. A. Bernstein, In Vitro,
22: 141, 1986,
R . J, KXcttbe, Nature, 250: 248, 1974,
R . Timpl, H . Rhode, P, G . Robey, S , I , Rennard, 2, M.
Faidart, and G . R . Martin, J , Bz'ol. Cfiem, , 254: 9933, 1979,
F. El Vaughan, S. Zrtman, R e Seavarelti, and I , A. Bemstein,
S, Toxicol, Envfron. Wealth, 23: 507, 1988,
A , R, Cr&lhorn and J. J. Roberts, Nature, 211r150, 1966,
P. D. Lawley and B, Brookes, Nature, 2063480, 1951,
€2. L . Flynn, E , M. Topp, and 6. L . Amidon, in Percutaneous
Absorption CR. C. Bronaugh, and B, X . Mdbach, e d s . ) ,
Mareel Dekker, New Vark , p. I?, 1985,
NONTRADITIONAL TOPICAL
DRUG DEUVERY FORMULATIONS
The Microsponae: A Novel Topical
Programma'bleDelivery System

SERGIO NACHT and MARTIN K A T Z Advanced Polymer- Systems, Inc.,


Redwood C i t y , California

1. THE NEED FOR BETTER SKIN DELEVERY


SYSTEMS

Over the past 40 years, the ab%ity to control the delivery rate of
active agents to a predetermined site in the human body has been
one of the biggest challenges -nzel by continued innovative solu -
tions-that has faced the rnedicd profession and d m g i n d u s t q .
During this time, some areas of pharmweutical research have
been focused on the controlled delivery of systemic dmgs. SeveraZ
predictable and reliable systems were developed for sylstemic drugs
under the heading of transdermal delivery using the skin as a gor-
tail of" entm ( I ) , Transdermal patches, developed in the 1970s, im-
proved the delivery of drugs such a s nitroglyceAn and scopolamine,
resulting in better control of therapeutic doses, simfler dosage reg-
imens, and fewer side effects than the more traditional oral or par-
enteral administration of the same d m g s , In. general, these deliv-
ery systems have improved, the efficacy and safety of many d m g s
that now may be better administered through the skin.
Although transdermal ddivery systems can be efficient in sup-
plying drugs for systemic effects, they are not practical for eon-
trolling the delivery af materials whose final target is the skin it-
self.
Controlled release of drugs onto the epidermis, with assurance
that the d m g remains primafly. localized and does not enter the
sytstem in significant amounts, is an area of research that has only
recently been addressed with success. No efficient vehicles have
been developed for the controlled and ZocaBzed defivew sf dmgs
300 Nacht and Katz

into the stratum corneum and underlying skin layers. Yet, there
are many instances when epidermal localization of a drug is desira-
ble, but absarpt;ion beyond the epidermis is not.
Corticosteroids are a typical example, Although effective for
skin disorders, their topical application can result in sipificant
systemic &sorption; this may lead to unwanted side effects such as
adrenal suppression or interference with immune functions (2).
Similarly, with sunscreens it is necessary to maximize the amount
of time that the active ingredient is present on the skin surface or
within the outer layers of the epidermis whjile minimizing its trans-
epidermal penetration into the body,
Another problem with the application of topical drugs is that
many vehicles, sueh as ointments, often prove aesthetically unap-
geafing, Greasiness, stickiness, or even discolorations in clothing
can m&e dany wear unpleasant. Thfs frequently results in the pa-
tient" slack of compliance with tredment.
Many of these conventbnd vehicles require high concentrations
of active agents for effective therapy because of their law efficidency
ae delivery systems. As a consequence, irriitation sr allerdc re-
sponses can be elicited in a sipifiicant percentage of users. Qther
disadvantages af existing topical d m g formulations can be uncon-
trolled evaporation of the active ingredient, unpleasant odor, and
the potenthl hcampatibitity of one a r more drugs with each. other or
with the vekiide ,
Thus, the need exists for systems to maximize the amount of
time that an active ingredient, such as a sunscreen, is present,
either on the skin surface or within the epidermis, while minimizing
its transdermd penetration into the body.
Such a new system would pssibly increase the efficacy of topi-
cdly active agents while enhancing product safety,
The ~ierosponge@ polymeric microsphere system utliwely fulfjlls
these requirements, These tiny spongelike spheric& particles (Fig.
1) can entrap active ingredients and then release them onto the
skin over time and in response lo a trigger. They are bioXo@cally
inert, norrirritathg , nonmutagenic , nondllergenic , nontoxic, and
nonbiodegradable. They can extend pmduet stabnfty because of
their unique configuration and improve its aesthetic properties,

II, MICROSPONGE TECHNOLOGY

Microsponges are patented ( 3 ) polymeric deXiverg systems consisting


of pornus mterospheres that can entrap a wide range of active in-
gredients, sueh as emollients, fragrances , essential oils, sunscreens,
an"r-infective, antifungal, and anti-inflammatory agents. These
A ProwmmabEe Topical Delivery System 301

entrapped active agents can then be incorporated into many pmduct


forms, such a s creams, lotions, powders, and soaps. After the
praduct is applied, the entrapped matesals are then delivered to
the skin in a controlled tine-release pattern or a preprograrnned
manner thrmgh the use of sever& different "triggers": mbbing or
p r e s ~ h gthe microsponge after it has been applied to the skin; el@-
vating skin surface temperature ; introducing solvents for the en -
trapped material, such as water, alcohol, or even perspiration ; and
eontrolgng the rate of evaporation.
Like a t m e sponge, each microsphere consists of a myfiad of
interconnecting voids within a noncollapsibfe structure with a large
porous surface. The size of the Microsponge can be vaded, usuaXly
from 5 to 300 um in diameter, depending upon the degree of smmth-
ness or after-feel required for the end formula, Although the Mi-
csosponge size may vary, a typical 25-urn sphere can have up to
250,000 pores and an intern& pare structure equivdent to 10 1"1: in
length, providing a totd pore volume of about 1. dig. This re-
sults in a large reservoir within each Mkrosponge, which can be
loaded with up to its awn weight in active agent.

Itl. MICROSPONGE SYNTHESIS

Microsponge particles are conveniently formed by suspension poIym-


erizatian In a liquid-liquid system (3) *
Xn general, a solution, is made cornpHsing the monomers and the
l'unctional or active ingredient which is immiscible with wafer, This
phase is then suspended: with a@tation in an aqueous phase, usu&Iy
containing additives, such as s~rfact8llt 8 and dispersants, to pro-
mote suspension,
Once the suspension is established with discrete droplets of the
desired size, polymerization is effected by activating the monomers
either by catalysis, increased temperature, or irradiation.
The result is a sedes of polymer "ladders" wrapping around one
another into solid microspheres (Fig, 2). As the polymerization
process continues, a sphedeaJ, stmcture is produced containing
thousands of microspheres bunched together like p a p e s , forming in-
terconnecting reservoirs in which the gorogen is entrapped. These
reservoirs open onto the surface of the sphere as pores through
which the active ingredient can be released when triggered.
Once polymerization is complete, the solid, padicles that result
from the pmcess are recovered from the suspension. The particles
are then washed and p m e s s e d untita they are substantially ready
for use.
Particle formation and heorporation of the functiond substance
is thus performed as a single step. This may be termed a one-step
procedure.
302 Nacht and Katz

Figure 'I5.1 Scandng electron micrographs of Microsponges, Mag-


nification: ( A ) x 240; (B) x 1200; CG) x 6000; (ff) x 24,000; CE)
x Freeze-.f"racture section x 6000; (F) x Freeze-fracture section x
24,000.

Function& substances that can be entrapped in this manner


must meet the following criteria:

I. They are either fully miscible with the monomer mixture or ca-
pable of being made fully miscible by the addition of a minor
amount of a non -water- miscible solvent.
2. They are immiscible with water or, at most, only slightly solu-
ble in it.
3. They are inert ta the monomers and stable when in contact
with any polymerization catdyat used and when subjected to
conditions needed to induce polymerization (such as temperature
.
and radiation)

Because of these requirements , often the functional substance


itself nag not serve as the porogen or pore-forming agent, In
(6)

Figure f 5.1 (Continued)


(El
Figure 15.1 (Continued)
A Programmable TopfcaE Delivery System 305

(Fl
Figure 15.1 (Cantinued)

these cases, an inert liquid, fully nnisdble with the monomers but
immiscible w&h water, is used during palymerisation to form the
pore network. The liquid that has served as a porogen and aecu-
pies the pares of "the formed particles can then be removed leadng
preformed dry micmspheres ,
Subsequently, placement of the functional substance inside the
reservoirs may be achieved by impregnation of the preformed dry
porous polymer particles aceording to techniques, such as contact
ab~osption,assisted, when necessaq , by sdvents to enhance the
absorption rate. The find product is thus prepared in two sequen-
tial steps. First, polymerization is performed using the substitute
p ~ r a g e n . This substila* iis then removed and, replaced by the
function& substmnce.
MatedetIs suitable as substitute porngens are liquid substaxlees
meeting the foregoing cr;Eteda and h a ~ n gthe further chmracte.I.istic
o f being readlily extracted from the particlesF pore network once
polymerization 1s complete, This covers a wide range of substances,
in particular inert, nonpolar o r e n i e salvents,
NacZlt and Katz

2 )***
MONOMER A MONOMER B

Step One: Selecting the Monomers Step Two: Making Monomer Chains

Step Three: Crosslinking Step Four: Wrapping-up the Polymer Ladders


the Chains (Polymers) into Micrdspheres

Step Five: Gathering Bunches of Polymeric Step Six: Binding the Bunches
Microspheres to Form a Polymeric
Microsponge", the
Basis of Command
Release" Tee hnology

Figure 15.2 The m&hg of a polymeric Microsponge system.


A Programmable TopScaI DeEivery System

IV. PHYSICAL CHARACTERIZATION OF


MICROSPONGES

The Microsponge is physicaXly characterized through the use of qual-


ity assuranee techniques to measure particle size and pore stmcture
parameters, as well as diameter, volume, and release characteristics.
The most valuable of these techniques is mercury intmsion
pomsimetry [4). During testing, a random sampling of polymer
beadlets is placed in a vacuum chamber and submerged under a pool
of mercury contained within a volume-calibrated cell, As pressure
is padually increased on the cell, mercury is forced into progres-
sively smaller pores of the microsphems. Thus, the apparent vol-
ume of mercury within the calibrated cell is reduced as it pene-
trates into the Miemsponge pores. Values comparing the applied
pressure to the apparent mercury volume are reduced by computer
into data on pore distribution by volume and diameter (Fig. 3).
Changes in surface area and pore diameter can be easily de-
tected with the use of mercury intrusion pomsimetry, providing a
reproducible method to monitor these parameters consistently and to
c o r ~ l a t ethem with pracessing variables or release rates of the ac-
tive inpedient from the Nlicrosponge .

.010 ,016 .025 ,040 ,063 .lo0 .250


Pore diameter (microns)
Figure 15.3 Pore volume distribution versus pore diameter as de-
termined by mercury intrusion porosimetry,
Nacht and Katz

Table 15.1 Typical Microsponge Entrapment


System Physic& Profile

Characteristic Vdue

Sample: CH-274
Total intrusion volume 1.22 cclg

Total surface area 74.5 m2jg


Average pore diameter 0.023 u m
Bulk particle density 0.708 glee

Table 1 illustrates a typical set of values for a sample of po-


rous microspheres .
V. PROGRAMMABLE PARAMETERS

Several primary characteristics, or parameters, of the Microsponge


can be altered during the production phase to obtain spheres tai-
lored to specific product appwations and vehicle compatibility.
These parameters include particle size, pore diameter and volume,
resiliency, and monomer composition. Each can be predesigned at
the time of manufacture by altering the components and polymeriza-
tion conditions required to produce a Microsponge.
Pmselection of the programmable parmeters of the Microsponge
results in a truly tlcustom-made" paparticle designed to meet the spe-
cific needs and requirements of the product formulator.

A. Particle Size
Free- flowing powders with fine aesthetic attributes can be created
by tightly cantrolHng the dkmeter of the microspheres at the time
of polymerization. Typically, diameters between 5 and 300 vm are
feasible, Small particle-sized systems provide the feel of a super-
fine talcum powder, even when they contain substances such as
mineral oil at a 50% payload. Larger-sized particles can be pre-
pared according to the ultimate needs for the products. Particle
size &so has some influence on the release rate of the active ingre-
dient from the Microsponge. The smaller the particle size, the
slower the rate of release (Fig. 4).
A Pragvmmable Topical Delivery System

20
Time (min)
Figure 15. Y Effect of particle size on release rate, Release rates
of declczne from two Microsponge systems of different particle size:
1.4181mh (25 um. squares) and I, 8l%lmin (300 vm, circles) .
B, Pore Diameter and Volume
Pore volume (void volume) determines the amount of active ingredi-
ent that can be entrapped withh the microsphere, Conversely ,
pore dhmeter can have a siwificant effect on the rate of release of
the ingredient. Fimre 5 demonstrates the effect of pore diameter
on the volat2ization rate of menthol. Both pare volume and diameter
are dtal in conlmlling the intensity and duration of effectiveness of
the active ingredient. Pore diameter also affects the m3pation of
the active ingredient from the Microsponge into the vehicle in which
the mated& is dispersed, A s a result, the diiameter of the pores
can have direct impact on the stabirity of the final formulation.

C . Resiliency
By aftedng the d e p e e of monomer cross-linking at the time of po-
lymerization, resniency (dscoelastic properties) of the microsphere
can be modified to produce a beadlet that is softer or firmer accord-
ing to the needs of the find formulation, Cross-linking in excess
of 10%is efficient in mast ~ystems. This allows; ~uffieients t r e n o b
fsr the Mic~ospongeto retain its shape after some or all of the ac-
310 Nacht and Katz

Pore Diameter (Microns)


Figure 15.5 Effect of pore diameter on rate of release. Rates of
release of menthol from Microsponge systems of same polymeric com-
position and particle size with varying pore diameters.

tive ingredient has been removed from the pore network. Increased
cross-linking tends to slow down the rate of release (Fig. 6).

D. Monomer Compositian
Selection of the monomer is dictated bath by the characteristics of
the active ingredient ultimately to be entrapped and by the vehicle
into which it will be dispersed (Fig. 7). Polymers with varying
electrical charges or devees of hydropfiiXicity or lipophilicity may
be prepared to provide Bexibnity in the release of such materids as
lipids, humectants, moisturizers, sunscreens, vitamins, insect repel-
Eants , fragrances, and a variety of pharmacoXo~callyactive ingredi-
ents. Onee entrapped, these ingredients can be formulated into
virtually any product form: powders, gels, ointments, lotions,
creams, liquids.

VI. RELEASE MECHANlSMS


A. Sustained or Time Release
Xn the development of a sustained-release Microsponge, many varia-
bles must be considered.
0 20 40 60 80 100 120 140 160
Time (min)

Figure 15.6 Release as a function of cross-linking. Release rates


of decane from two polymeric systems with different cross-linking:
Z.lZ%/min(4% cross-linking, squares) and O.Ql%/min (30%cross-
.
Iinking , circles)

10 20 30
Time fmin)
Figure 15.7 Release of decane from two Micmsponge systems of
different polymeric composition (MMAIEGDMA : Methyl-methacrylatel
Ethyleneglycoldimeth8cry1ate; St /DVB : Styrene/Divinylbenzene).
31 2 Nachl and Katz

Physic& xlfcfienzicd properties of the entrapped active ingredient, fn -


clu&ng volatuity , Tsiscosity , and solubilZity of the entrapped ma-
ted&
Physicdlchernicd properties of the pdyrnefie mierasponge, such as
pore aameter and vdume and resfiiency of the polymerjlc sphere
Vehicle and produet form in which the Microponges d l be used
(ofls, lotions, creams, powders and other physic& forms)

B, Release on Cornnand
El y proper manipuf ation of the aforementioned programmable parame-
ters, Microsponges can be d e s i p e d to rdease @ven amounts of ac-
tive ingredients over time in response to one or more externaf trig-
gers,

1, Pressure Release
Like m ordinary sponge, Microsponge systems release fluid when
pressed or squeezed, '' "ereby replenishing the level of entrapped
f f

taetive ingredient onto the skin, This action results In renewed


product efficacy (Pig. 8). The amount of ingreaent released rnay
d s o depend upon the pressure applied to the sponge and the re@-
iency of the microsphere,

2, Temperature Change
Some entrapped materials, such as sunscreens and emsll%ents,can
be too .viscous at rmm temperature to now sgontmeously from the
Microsponge onto the skin. When warmed by the skin temperature,
the sun or other heat source, their dscosity rnay decrease, result -
ing -in an increased Row rate f Fig. 9).

3, Solubility
By taking into account the solubility of the entrapped ingredient,
the Iblicmsponge system can be? programmed to respond to water,
perspiration, or other solvents. For example, dry Microsponges
loaded with a water- soluble ingredient, such as &ntiperspirants or
antiseptics, will release that ingredient in the prcl;senee of water
(Fig. 1a), Release ean also be activated by diffusion, taking into
consideration the partition coefficient of the ingredient between the
Mierasponge and the outside system.

Vf l, SAFETY SUBSTANTIATEON

Mice.ospange systems are made of Isiolo@eafly inert polymers. More


than 30 safety studies, ineluang skin irdtation (in rabbits and hu-
A Programmable Topical Delivery System

to!.ion
%vitk
Microsponges @

time (hr)
Figure 15.8 Comparison of human s k h emouieney induced by min-
eral oil in a Mierasponge system vs. microcapsules, Measured with
the gas bearing ellectradynamametes ( R e f , 5) ,

m a s ) , eye irritation (in rabbits), oral toxicity (in rats), mutagen-


icity {in bacteda), and dlergenicity (in p f n e a pigs) demonstrate
that the polymers are nanirdtating , rtonmutagenic , nondlergenic ,
nontoxic, m d nonbiodegradable-the body cannot convert tkem into
other substances, Furthermore, preliminary data indicate that upon
injection inta the skin, same Microsponge systems are not recapized
as f o r e i p bodies and cause n s reactions by the body in response
to their presence.
Because af their size, these mierospheres are too large to pass
through the stratum comeum. They rem&n on. the s k h surface,
gradually releasing their eontents on command or over time. This
release pattern prevents excessive aceumulatian o f active agents In
the epidermis and, as a result, mag enhance the safety of topical.
drugs ,
Nacht and Katz

0.5 1.0 1.5


Time (hr)

Figure 15.9 Effect of tempereture on the rate of release of octyl


dimethyl PABA.

2 4 6
Time (hr)
Figure 15. f 0 Solvent-activated release: release of an antiperspi-
rant in the pmsenee of water from two different Micmsponge sys-
tems.
A Programmable TopieaZ Delivery System

FIgure 15.1 1 Reduction in skin irdtancg . Results from a 21-day


cumulative irdtancy test in rabbits with formulations contdning two
different levels of benzoyl pemxide, either freely dispersed or en-
trapped in a Mierasponge system,

A reduction in the irdtancy potential of the antiacne drug ben-


zoyl peroxide is a demonstration of the increased safety achievable
by the Microsponge entrapment system, fn a 21-dfly cumulative ir-
ritaneg study, lotions eonttllning 2.5% and 5% concentrations o f free
or entrapped benzoyl peroxide were prepared and applied to rabbits.
Free benzoy'l peroxide at both concentrations produced much ktigher
irdtancy thm the entrapped dmg (Fig, 111, Effictiveness of the
benzoyl peroxide was not impaired as demonstrated by an in viva
human antimicrobid test (Fig. 12). Both PmpionibacteM'um aenes
&nd aembic bacteria showed reducthns in their skin number well
within the range required in the over-the-counter (01'C) Food and
Drug Administration (FDA) manograph,
Allergexlicity can also be reduced when the sensitizing ingredi-
ent is entrapped jin a Microsponge system, Cinnamic aldehyde, a
component of chnamon ail commonity used in pharmaceuticaf and css-
NachE and Kalz

Baseline 2 Week
Figure 15.12 In viva antimicrobial efficacy of benzayl peroAde la-
tion. (Lotion eontdnfng 5 percent BPQ entrapped Jin. a Miemsponge
system tested as descdbedi by Williamson and Kfigman (from Ref, 8).

metie applicieatians, is a known, albeit m%d, sensitizer, Free and en-


trapped cinnamic aldehyde was tested. at 345, 4, $%, and 6% levels in
the same base, At each cancentration, the number af itnimats sen-
s;itized was significantly lower with the entrapped system (Fig. 13).
Another safety concern i s the potential bacterial contamination
of the matedals entrapped in the Microsponge, Because the size of
the gore diameter is smdler than bacteria, ran@ng from about 0.007
urn to 0,2 pm, bacteria cannot penefrate into the tunnef structure
of the Micmrspange , Nffcrckbfolam tests conducted, over a 12-month
period show that- Microsponclfe systems are nonsuppartive of micrabial
growth m d require ns preservative system,
Therefore, formula stability may &SO be extended by entrapping
unstable a r bacteria- sensitin ingredients in the Microsponge, en-
trapment will keep the ingredient protected until release f r o m the
Miemsponge is effected.
A Progrmmabile Topical Deltvery System 31 7

Figure 15.13 Reduction in. skin sensitizatiEon, Results from a Draize


skin sensitization test in guinea pigs with preparations containing
three different levels of cinnamic afdehy.de, either freely dispersed
or entrapped in a Miemsponge system,

VIII, INCREASED EFFICACY A N D AESTHETIC


APPEAL OF DRUG FORMULATIONS
Skin care formulations can have increased efficacy and higher aes-
thetic appelll to the consumer w%h the use of active ingredients en-
trapped in Microsponge systems.
In an experiment, the gas-bear-iing efectrodynamometsr (QBE)
technique was used to measure skin sxlfiening (5). Microsponges
cont&ning 50% mjnertsl ail were dispersed at a 5% concentration in a
typic& oil-in -water skin lotion. A simaar prepmation containing a
urea-farmafdehyds micrwapsule system with mineral ail was used
as a contraX. Bath products were appBed ta specjimens of excised
human skin and readings of skin softening were made with the GEE.
A t the end of 3 Inr, both sites s f application were gently mbbed.
The Microsponge system produced a greater initial increase and a
repeated enhancement of skin soRen3ng. This softening effect was
31 8 Naeht a n d Katz

significant even 6 h r after application, the time at which the micro-


capsule system had r e t u n e d to baseline (see Fig. 8).
A dose-response relationship was o b t ~ n e din another test con-
ducted in ~ V O . A Microsponge system was prepared with a pmpX"i-
e t a v emollient entrapped at 50% concentration and incorporated into
a standard oil-h-water emulsion at 0,5%, I%,and 3%1eve3s, These
creams were applied to the back of the hand of human subjects and
p e ~ o d i cmeasurements of skin softening were made with the GBE*
An increase in skin suppleness proportion& to the concentration of
emol;tient was obtdned (Fig, 14).
Cortieosteroids produce a IscaEzed blanching effect on the skin
caused by vasoconstriction (6). To determine the release of cor-
tieosteroids entrapped in Microsponges, an in viva vasoconstdction
a s s q was conducted. A system containing 0.05% entrapped fluo-
cinolone aeetonide was applied to intact skin on both forearms of all
test subjects, Blanching effects were measurrsd at 8, 24, and 32
h r after application, One-half hour before reading, one arm was
gently rubbed to effect a renewed release of the entrapped ingredi-
ent. Observations on the nonmbbed arm indicate that sustained
release was effective even at 32 h r (Fig. 15). However, at each
time-point , the blanching observed in the site rubbed 30 nin earlier
was greater than in the nonrubbed site.
Although most topical drugs must penetrate into the stratum
comeurn to be effective, sunscreens have greater efficiency when
they remain on the skin surface to absorb ultradolet light from the
sun, After penetration, sunscreens can absorb radiation belaw the
skin surface, but damage is afready oceurfing that reduces sun-
screen efficacy. Sunscreens can also be irdtants and even skin
sensitizers. X f these ingredients can be kept on the skin surface
for a longer len@h of time, and in controlled quantities, an in-
creased efficacy and a reduced sensitldty to these mateAals would
result. This could overcome the need for repeated applications and
pmv;ide enhanced aesthetic apped for the consumer,
To determine whether or not sunscreens entrapped in a Nicro-
sponge system would still be av&lttble for sun protection, octyl di-
methyl PABA (Padimate Ef) was entrapped at 50% payload; the Micro-
sponges were incorporated into an 03-in-water emulsion at a 2.8%
level to prorride a 1.48 sunscreen concentration. The sun protection
factor (SPF) of this formu2ation was compared with that of a similar
formulation contdning the same amount of sunscreen that was freely
dispersed in the base, To assess the possible additive effects of
free and entrapped sunscreen, a product cont&ning 1.4% free and
1.4% entrapped Padirnate O was also prepared, The lotion base &one
and one containing 8%free homosallate were used as negative and
positive controls, The SPP testing was performed on the backs of
A Programmable Topical Relivery System

% Emollient

Figure 1 5 . 1 4 Maximum skin softening as a function of totd emolXient


concentration. Measured an human skin with gas-bearing electro-
dynamometer.

a Not Rubbed
Gentle Rub

Hours After Application


Figure 15. IS Demand and sustained release of a cortieostemid from
a Microsponge system as measured by the vasoconstrictor effect.
NacZll and Hatz

Figure 15.16 Sun protection factor of formulations containing octyl-


dimethyl PABA, (ODP) either freely dispersed in the vehide or en-
trapped in a Mjcrosponge system,

human vdzrnteers using a xenon -arc solar simulator as recommended


in the corresponding FDA. OTC tentative find monograph for sun-
screens ( 7).
Results obtained from this test (Fig, 16) indicate (1) entrapped
sunscreen is indeed avaaable for protection and (2) the addition of
free and entrapped d m g results in additive SPF,
Further tests have canfirmed that higher SPF values Cup to 15
or more) can be obtdned by prepadng suitable formulatians con-
tdning mixtures of UVB and UVA absorbers,
An in d t r o study was conducted on excised human skin to de-
termixle the amount af sunscreen remaining on the skin surface when
.formulations containing [ 1 4 octytdimethyl
~ PABA, either free or en-
trapped in a Microsponge system, were applied to it. The test was
A Programmcrble Topical Delivery System

,
,Lotion
micraentrappc?d
with
active
I \ m---a Lotion with "free" active

1/4 112 1 2 3 4 5 6 7
%me After Application (Hours)

Figure 15.17 Amount of ~ ~ ~ ~ l o c t y l d i r nPABA


e t h ~recovered
l from
the superfieid skin layers of disks of excised human skin after ap-
plication of lotions containing sunscreen either freely dispersed
(squares) or entrapped in a Microsponge system (circles). At each
time point, the surface of individual skin disks was quickly rinsed
with a stream of 95% ethanol and the stratum eorneum was then
stripped three times and counted.

conducted as described elsewhere (9). Disks of excised human


skin were utilized and one disk was used for each data point. The
amaunt of sunscreen remaining on the skin surface was determined
by quickly rinsing the surface of each disk with a stream of ethanol
to remove nonabsorbed material. For mass balance, the rinses were
collected and counted for radioactivity. The stratum corneum was
then stripped three times with cellophane tape and the strips were
counted fox radioactivity, The results presented in Fig, 17 show
that the amount of sunscmen present on the epidermal layers of
these skin disks treated with the "freew sunscreen product gradually
decreased as a consequence of percutanwus absorption, Converse-
ly, those disks treated with the sunscreen in the lMicrosponge sys-
tem consistently had a higher surface concentration of active ingre-
dient a-id, when the skin surfaces were rubbed just before assay,
that amount went back almost to baseline.
322 Machf and Katz

Table 15.2 Consumer Panel Evduation of Two Final Products

% Preference

Microscopic sponge Encapsulation


Attdbutes entrapment system system

Less oily 72 28
Less greasy during
application
Disappears faster
Less greasy afterfeel 78 22
Less residue on skin 67 33

Aesthetic apped plays an important role in the use of topical


drugs. Users tend to reject those products they judge to be too
oily, too greasy, slow in absorption, or leave a noticeable residue
on the skin.
A double-blind crossover panel test with 200 subjects was per-
formed with the same two products used in the expelrlment described
in Figure 8. Subjects tested the two formulations on a blind basis
for aesthetic appeal (Table 2 ) . The product with the Microsponge
system rated significantly higher in each of the &%tributesevaIuated
by the panel.

IX. POSSIBLE APPLICATIONS FOR MICRO-


SPONGE SYSTEMS

Desimed expressly for applications in which the skin itself is the


tapget organ, Microsponge topical delivery systems offer the formu-
lator a range of alternatives limited only by one's own ima@nation
to devdop drug or cosmetic products with enhanced safety or ef-
ficacy.

A. Sunscreens
High levels of sunscreens can be utilized for longer-lasting product
efficacy to increase protection against sunburn and sun-rel&ted In-
judes (aging and skin cancer). Even at elevated sunscreen con-
centrations, oiliness, irdtancy , and sensitization could be reduced.
A, Programmable Topical Delz'vey System

Efficacy of antiacne ingredients, such as benzoyl peroxide, can be


maintdned , whereas skin i r ~ t a t i o nand sensitization are decreased.

6. Skin Depigmentation
Skin bleaches, such as hydroquinone, can be stabilized and protect-
ed from oxidtnlion as eddenced by the lack of diacalaration of the
product. This results in impmved efficacy and aesthetic appeal.

Long-lasting anti-infiantmatory a c t i ~ t gcan be attained for dmgs,


such as hydrocortisone, ta prodde extended benefits in. the reduc-
tion of skin allerm responses and dermatoses,

E. Antipruritics
Extended actidty is achieved for anti-itching compounds and local
anesthetics used to treat prmr;itus caused by dry skin, eczema,
hemorrhoids, or poison. ivy.

F, Antidandruff
The unpleasant adars af zinc pysjithione and selenium sulfide, as
well as other antidandnB ingredients, are reduced, Irfitation can
afss be lowered, whereas safety and eflictlcy. are extended,

All-day treatment and symptomatic relief of h n g d problems, such


as athletes fwt, can be prodded by the sustained release of the
active ingredients ,

Prolonged actidty with less irdtarxq are okztdned with these active
agents, accompanied By a reduction in odor and peasiness of the
product.
In each of the foregoing cases, the final pmduct can be de-
s i p e d to give distinct and perceivable benefits to the user that
were not p r e ~ o u s l yfeasible because of formulation restr;ictions.
Such statements as '"ease?less," "all-day relief ," and ""new pkesing
frapancef' can now be achievable Consumer beneAts for these types
of products,
Nacht and Katz

X. ADVANTAGES FOR INDUSTRY

For the pharmaceutical indu stry , Microsponge topical d e 1 i t . e ~sys-


tems p m d d e a wide range af formulating advantages.
Entrapment offers a way to control undesirable ingredient char-
actedstics that have restricted the use of some ingredients i~ the
past. Liquids with such: undesirable properties as peasiness or
stickiness, can be transformed into nonaay powders, PormuE&tions
can be developed using otherwise incompatible hgsedients through
separate entrapment of one or more of those ingredients. New
product forms can be developed as a result of entrapment.
Shelf life and product stabnfty can be prolonged without the
use of chemical preservatives. And, because the pores of the Mi-
crosponge are smaller t hwn bacteda , the entrapped ingredients are
autarnaticdly protected from bactelt7ierX contarninathn.
Finfly, safety can be increased when irritating and sensitizjng
ingredients are entrapped, Programmed release can both control
the amount of delivery to the targeted site and minimize entry into
the lower skin layers, where undesirable side effect are mast likely
to occur,
Microsponge topical delivery systems elearly answer the needs
of industry to maimize the amount of time an actjve ingredient is
present on the skin" ssurhce or within the epidermis, wfiife minimiz-
ing its penetration through the dermis and, therefore, into the
body.
In surnnxaq, these systems can pm.vide increased efficacy for
topiedly active agents, enhanced safety, extended product stabirity
and imprmved aesthetic properties in an efficient and novel manner.

REFERENCES

I, A. f". Kydonieus, and B, Berner, Transde~malDelivery of


Dmlgs. CRC Pmss, Boca Raton, 1987,
2, J . P. Hater, and A. C . De Grsot, Unwanted Effects of Cosmel-
lies and Drugs Used In De~matotoay, EXseder, New York,
1985.
3. I%. Won, U , S . Patent 4,690,825, Sept, 1, 1987,
4. Poresizer Model No, 9310, Micromer%ticsInstrument COT, , Nor-
cross, Georgia,
5. M, S, Chdstensen, 6 , W, Hargens 111, SL Nacht, and E. W,
Gans, J , Invest, Lfematol,, 69:282, 1977,
6 . M. Katz, and B. 3, Poulsen, 2 , Gosmel, Chem., 23:565, 1972.
1, Sunscreen Drug Products for Over-the-Coun ter Human Drugs,
Fed, Reg, , 436166) :38259, (Aug, 25, f 9781, Dept, of Health,
Education and Welfare, Food and Bru g Administration,
A Programnab E
e Topical Delivery S y s tern 325

8. P , VVBBamison, and A , Kligmaxl, J , Invest, Dermatol, , 45: 498,


1965.
9, T1. Yeung, S , Macht, D, Bucks, and H, 1 . Mailbach, J, Am,
Acad, DsrmatoE, , 9: 920, 1983,
Liposome-Based Vehicles
for Topical De

PAUL S . USTER Liposome Technology, Xne, , Mento Park, Gattf-ornia

Liposomes are micmscopic vesicles consisting of anphipathie lipids


.
arranged in one or mare concentric bi-layers These thermacfynam-
ieally stable, lamellar stmctures farm spontaneously when lipid i s
brought into contact with an aqueous phase. Unlike micelles , emul-
sions, and microemulsions , lipssomes have an entrapped, discontbu-
ous aquesus phase separated by 4-nm thick, bilayered Xamel?alaefrom
the continuous aqueous phase, Liposomes were developed sd@nAly
to modd bioiolo@caf membrane functions such as transport phenomena
(1). However, it was soon recopized that the phenomenon of
spontaneous compartmentation might be used to improve the tk-xera-
peutic index of drugs by targeting active ingredient to the appro-
priate site of action and rnoctulating drug release from the vehicle.
Because lfposames are a new vehicle to the pharmaeeutied in-
d u s t w , the scope of this review will be to introduce the concept of
liposomes, discuss some aspects af BarrnuIaHng a topied product with
liposomes, and review the literature an Xiposome-'based topic& deliv-
ery systems.

Liposorne preparations can be distin wished according to size and


number of lamella@,lipid camposition, the encapsulation volume, and
the percentage entrapment (Table 1).
328 Uster

Table 16. t Liposome Nomenclature

Vesicle Captured
diameter Number of volume Percent
Acronym (vm> lamellae (ml fg) entrappeda

MLV > 0.1 -> 5 -> 0.6 6 .-


c <70
SUV 0.03 .c 0.06 1 0.6-1.9 6 < < 19
LCrV > 0.06 1 -> 1.Q 19 -
< < 70
OLV > 0.06 l c -< 5 -> 0.6 -
6 < <70

aCalculated for 10%lipid (wlv),

A. tlposome Specific Parameters


The encapsulation volume or captured volume is defined as the vol-
ume of entrapped discontinuous aqueous phase per unit mass of Ii-
pid phase, and is typically expressed in units of microliters per
micmmle lipid (pllpmol) or mill8iters per gram lipid (mlfg), It is
often desirable to maximize drug entrapment with minimum lipid,
and this can be achieved by formulating for the greatest possible
captured volume. For a given quantity of lipid, the captured vol-
ume is increased by increasing the diameter of the Iiposomes and re-
ducing the number of lamellae, The theoretical relationship between
liposo&e diameter and captured volume of ideal, uniformly sized
unnamellar liposomes is kllustrated in Figure 1, These values repre-
sent the upper limit of captured volume. For lipasomes with more
than one lamella per vesicle, the captured volume will be reduced as
a function of number of hilayers and the interbiIayer repeat dls-
tance. In practice, the captured volume is determined by using a
water-soluble marker (such as radiolabeled sucrose, inulin , or the
drug of interest) of known concentration in the hydrating buffer.
This buffer is added to a known amount of lipid. After the lipid is
completely hydrated, the liposome-entrapped marker is separated
from unentrapped material by centrifugation, column chromatography,
didysis, or some other means. Final marker concentration and lipid
concentration in the cleaned up preparation aze determined by suit-
able methods. The captured volume is calculated from

M/(C x E) = captured volume 111

where M is the total mass quantity of marker in the cleaned up


liposome preparation, C is the original concentration of marker in
Eiposome-Based Vehicles for Topical Delivery

-
E 8.3
0.02 0.10 1.00
Vesicle Diameter (lurn)

F igure 16.1 Thearetic& relationship of captured volume to diameter


of unaamellar vesicles.

the 'hydrating buffer, and L is the total mass quantity of Epid in


the cleaned-up prepmation,
Another Ifposomal propexrEy, percent entrapment or percent en-
eapsulalian, is an impodant: parameter far chmacte&zing the effi-
ciency of the hydrating p m e s s . It is dtefhed as the fraction of
the total aqueous compartment sequestered witilhira liposomal mem-
branes and is cafeulated from

Captured volume x lipid concentration x 100%= % entrapment


f 21
These I s et practical upper limit (apprax~nrately70%, see Table 1)
to the percent entrapment of a highly water-soluble marker in pure-
ly aqueous phase, This is due to marnetAc Emits of paeking sphe-
roids into a @ven volume during the hydrating process.
This gearnetfie constrdnt does not apply if the active ingredient
adsorbs to the bifayer or partitions into the lipid phase, Under
these conditions, it is more accurate to descdbe the apparent effi-
deney of d m g association with Iiposomes as the percent z'ncarpora-
tion. of active ingredient,
In addition, estimates of parameters such as the? average num-
ber of larnelXae per vesicle and interIamelIar spadng (2), and the
number of vesicles and tot& surface area of st liposorne suspensbn
( 3 ) can be d e ~ v e dProm the captured volume, average vesicle size,
and lQid concentration.
Liposome size is usually determined by Caulter counting for
vesicles larger than I prn diameter, or quasi-elastic laser light scat-
tering if smaller than 1 i;lm diameter. Relative particle size of small,
vnilamellar vesicles and the degree of contamination ~ t heteroge-
h
neous large particles can also be measured turbidirnetrieally (4).
Electron microscopy remains the only accurate, albeit tjme-consuming
method of determining absolute Piposome size and mean number of
larnellae per lipid vesicle. However, a great deal can be learned
with a good phase-contrast microscope equipped with polahzing fil-
ters, It is an indispensable tool of liposomolou. Both ~ s u a and l
microscopic inspection provide the investigator with a good idea of
the approximate size and "laarnllarity'bf the preparation,

Size and number of lamellae have p i d e d the standardization of lip-


osome nomenclature. Each Xiposome category, p n e r d methods I"or
producing them, and visud properties will be discussed later, and
the reader is referred to appropriate references for details of each
method. Also, recent re-viteuvs are avaaable that discuss methods of
liposome preparation and ckraractefization in greater detaa than will
be discussed here (5-8; and also Val. 1-3 of Liposome TechnoEo,~
.
( G Gregoriadis , ed ,) ,
MultBamellar vesicles, known as MkVs, are distinmished by be-
ing larger than 0.1 .ttm in diameter and generdly k a ~ n gmore than
five lamellae enclosing the aqueous core, They are prepared simply
by dwing a thin f 2 m of lipid on a surface and hydrating with. buf-
fer ( I ) , Thin fnms are used preferably because the large surface
area of exposed lipid faci-litates liposome hrnzation and efficient en-
trapment of aqueous phase. Alternatively, NLV tlisgersians can be
formed by homogenizing or sonicating dried lipid vanules suspended
h aqueous solution. Care must be taken to ensure that such a
preparation is not contaminated with unhyclrated lipid, To the naked
eye, an MLV dispersion appears rnil;ky and opaque at 0.1% lipid
.
f w lv) or p e a t e r The MLVs appear as cross- sectioned, anionlike
stmctures of variable size and shape under high magnification in
the phase-contrast Xight microscope, These preparations can dis-
play a Maftese cross birefAngent pattern when observed under
crossed po1as;izing faters, if there are sufficient larnellae per ves-
icle and the interb2ayer djstance is relatively small and evenly
spaced.
The limiting size of a fiposome is about 0.03 urn diameter be-
cause of the radius of curvature imposed by ligid pornetry (Fig. 2).
QP necessity, these smaflest of vesicles have only one lipid b2ayer
and, thus, are known as small un2amellar vesicles (SUVs) , The
Liposome-Based Vehicles fir Topical Delivery

Figure 16.2 Size relationship between a SUV of limiting radius and


a LUV of 0 . 1 - p m diameter. Bilayer thickness is 0,004 urn.

S U V s are produced by sonicating an MLV preparation to terminal


clarity ( 9 , 2 0 ) . Beeausc? they are sa small and do not scatter trisible
light, an SUV preparation at about 0.1% to 1%lipid (vvlv) will ap-
pear perf"ect1.y transparent to the eye, It i s not possible to observe
them by light microscopy.
During the 1970s considerable effort went into developing a new
class of liposomes that were much more efficient (per unit mass of
lipid) at entrapping aqueous solutes. This is because the lipid
phase displaces 60% of the total volume of a 0,03-pm diameter SUV,
but only 2 . 4 % of a 0,1 - ~ r ndiameter liposomca, The gad was to make
the largest possible diameter liposome with the fewest number of
larnelae, preferably a large, uniliamellar vesliele (LUlr). For LUVs
OeI vim or larger in diameter, doubling the vesicle diameter approxi-
mately doubles the captured volume (see Fig. 2 ) . Mote that for
vesicles smaller than 0.1 prn in diameter, this relationship is not
linear bmause of the volume displaced by the lipid phase, Prom
the literature, S U V s are generally eonsidered to be no more than
0.06 urn in diameter, and 0.1. urn is oBen accepted as an a r b i t r a ~
lower size limit far LUVs, I propose that the line of demarcation
should be at the limiting diameter of a vesicle of two or more lamel-
lae (&out 0.06 wm) because it is necessary to distinpish between
LUVs and oligolarnellar vesicles ( OLVs) .
Numerous methods have been developed that are capable of mak-
ing LUV preparations on the research bench. scale, and can be
grouped into three major approaches, Puston-extrusion techniques
often start with SUVs that are then fused into larger structures by
addition of divdent cations ( 11) , freezing ( 12,131 , dehycfration ( 7) ,
or rapid pH shift ( 14). These larger, usudly 01igolarneBar vesicles
are usually sized through polyearbonate Enters to reduce vesicle
size heterogeneity , Another gensrrrl s t r a t e w can be charactedzed
as organic phase addition--removal. Dissolved lipid in organic sol-
vent is introduced to an aquwus buffer. The organic solvent is
removed durjing or after liposome formation by- evaporation ( 15,16)
or didysis (17). StBI a third approach, d e t e r g e n t - d i a l y s h , is to
make mhed micelles of lipid and detergent, followed by detergent
removal by column chromatography f 18) or dialysis (19-21). A
0.1% to 1%lipid (tvlv) LUV preparation i s clear and opalescent to
the eye. True LUVs, having only one lipid bnayer with which to
scatter light, are essentidly impassfile to observe by phase-con -
trast microscopy.
Xt is noteworthy that unless errpedmental parameters are care-
fully charactedzed and optimized, the casual application of these
LUV methods will result in a preparation of predominantly oligola-
mellar, large vesicles known as OLVs. Other acmnyms have been
developed that derive from the specific method used (REVS,
LUVETS, SPLVs, etc.) but these methods d l result in preparations
that are predominantly LUV to OLV in character. Under phase-
contrast light microscopy, QLVs larger than 0 . 4 yrn in diameter can
be resolved as phase p a y '"hooststV that are not birefringent under
crossed polarizing fBters,

111. FORMULATING TOPICAL, PRODUCTS


WlTH LIPOSOMES

The eornmerci&ization of liposorne technolow began in the early


1 9 8 0 ~and
~ it is a rapidly evoldng field. There are now four pub-
licly held companies in the 'United States exclu8ively devoted to
product development using liposome-based vehicles. Many large
p2-xarmaceuticaf hauses and cosmetic firms are also engaged in active
research and development,
Cosmetic skin creams containing Xiposomes have been attanable
for severd years ; however, the first therapeuticafIy acceptable
produet remains to be introduced. The mjiiestones for a successful
liposome-based pharmaceuticd product are no less than those of
Liposome-Based Vehicles for Topical Delivery 333

conventional dosage forms. Efficacy, safety, scaleup, stability,


raw materials supply and purity, process control, and vdidated
analytical methods must aU be achieved. Selected parameters that
are germane to formulating with liposomes will be noted in the fol-
lowing section.

A. Vehicle Selection for the Active ingredient


An active ingredient can be situated in the aqueous compartments,
hydrophobic phase, or oriented at the water-bilayer interface. The
selection of the appropriate liposome vehicle (NILV, SUV, or other)
should be based on a careful analysis of the physical p r o p e ~ i e sof
the drug and its tendency to associate with the lipid components.
Additional faetors , such as buffer composition, pH, ionic strength,
and lipid composition, a l l wBl influence the distribution of active
ingredient. Cost of goods is also germane when selecting a ISposorne
vehicle. A very costly, water-soluble ingredient may require maxi-
mum captured volume of the formulation. A process that results in
an LUV to OLV preparation should be employed. However, NLVs
are more cost-effective at incorporating lipophilic active inpedients
at considerable savings, because a simpler process can be used and
a clean up step to remove unincorporated drug may be omitted.

B. Process Selection
Liposomes have their own unique problems of scaleup and process
control. These are concerned with adequate supplies of phospholi-
pid raw materials, lipid hydration to achieve the desired capture
volume and particle size, use of pharmaceutically acceptable organic
solvents, removal of solvents (organic phase addition -removal proc-
esses; 22) or detergent (detergent dialysis; 20), and clean up of
unentrapped drug.
Selection of a process should be based on the solubility proper-
ties of the drug in aqueous solution and appropriate solvents, the
liposome / buffer partition coefficient, and the required percantap
entrapment or incorporation.
Topical liposome formulations have the advantage of not requir-
ing the stringent particle size specifications of parenteral products.
If liposome size or heteropneity does not affect in viva performance,
the specifications can be kept quite broad and the process simplified
considerably.

C, Characterization
The characterization of bulk properties of liposome formulations, such
as color, pH, rheolosry , are very similar ta those commonIy employed
for other types of dispersions and suspensions.
When formulating emu'Isions and microemulsions, considerable ef-
fort is placed on making appropriate blends of oil and surfactant-
cosurfactants to achieve a stable dispersion. With phospholipid-
based liposomes that spontaneously form a stable structure, more
emphasis is placed on characterizing liposome-specific properties.
Determination of vesicle size, captured volume, percent entrap-
ment and percent incorporation have already been discussed. One
other prope&y unique to liposomes is the active ingredient release
rate or efflux rate, which has important Implications for the per-
formance and stability of a formulation, It is most simply measured
by monitoring the change in percent entrapment or percent incor-
poration of a formuIation over time under a given set of conditions
(temperature, other excipients , ionic strength, pRsence of biologi-
cal fluids). Average liposome size of a formulation is an important
factor in determining the release rate of small nonelectrolytes (23).
However, mathematical model fitting from experimental determinations
of glucose release indicates that the degree of liposome size hetero-
geneity (polydispersity) does not affect the release rate s i p i f i c m t -
XY ( 2 4 ) .

D. Antimicrobial Preservatives
Topical products are usually preserved by the addition of one or
more antimicrobial preservatives. Preservatives are intrinsicalIy no
different from an active ingredient, which is to say they can be en-
trapped in Xiposomes or incorporated into the bilayer or interfacial
region. The end result is that microbial deterrence at a given con-
centration will be different from that of a simiIar aqueous solution
without liposomes ,
For instance, the commonly used parabens are inactivated by
liposomes as a consequence of partitioning into the bilayer (25).
The extent of inactivation will be a function of partition coefficient
and the lipid concentration. Consequently, the total quantity of
lipophilic preservative must be increased to achieve a satisfactory
germicidal effect. If the "free" concentration (Cfree) required for
the germicidal effect of preservatfve P , the formulation% liposome-
water partition coefficient (Kp) and the total lipid concentration are
known, the totaI concentration of preservative (Ctotal) required in
the formulation can be cdculated from a modification of Bean's
equation (26.) :

where 4 a ratio of lipid to aqueous phase (w /w) , and


K = ( g Plg lipidlltg P i g aqueous phase)
P
Giposome.-Based Vehicles far Topical, Delivery 335

Mote that this equtation nay &so be used to predict loading of


1ipophBic active ingredients in a liposome vehicle,
For water - soluble compounds, inactivation can occur if suf dcient
preservative of interest is entrapped and has such a low bnrz~rer
permeab3it.y that, after the ini"cial mjicrobial chdlenge , it Is u n a v d -
able for subsequent challenges. If such is determined to be true,
addition of the preservative should be delayed unt-il liposome forma-
tion is completed,

E. Stability
Formulation stability e m be s u b d i ~ d e dinto chemical, phy sicat, and
f ipossome- speddc characteristics.

1. Chemical
For ehemica3. stability of phospholipid-based liposomes, lipid permi-
dation and fatty acyl-ester hydrolysis are the primav concerns.
Phospholipid peroxidation is accelerated by oxygen, transition metal
ions, elevated temperature, and light. ft is usually controlled by
the addition of chelating agents to bind iron salts (271, antioxidants
such as a-tocopherol ( 2 8 ) , by using phospholipid that has saturated
or mono-unsaturated fatty acids, a s combinations, Cholesterol ,
anather primary component of lipasome formulations, is mininndly af-
fected by oxidation in liposomes because the phoaphatidylchslines
are very effective inhibitors ( 29) .
Fatty acid ester hydralysis is a time-, temperature-, and pH-
dependent process, Hydrolysis results in the formation of lysopkos-
phatidylcholine and free fatty add, Depending on the active ingre-
dient, changes in liposorne size, efflux rate, parcutaneous a b m q -
Lion, and in d v s performance may be affected. Fmkjer et ale (30)
have published a v e v excellent, systematic study of the chemiczil,
phgsicd , and release rate stabifity of plnosphatibylehollne-choles-
terol Eposomes. The pseudo-first-order rate constants for phos-
phatidylclholine hydrolysis at 10°G showed a pH minimum at 6.5.
Only 6% fiydroly sis of distearylpkosphatidg1~holinc!to ly saphaspha-
tidylcholine was observed aAer 300 days at pH 6.5 and 2 s 0 6 ,
Thus, ester hydrolysis can be minimized by formufating in the vi-
cinity of pH 6,5.

2, Physical
Changes in the bulk physical properties of a liposorne preparation
usually reflect changes in vesicle agpegation or hereasing fiposome
size.
Cross aggregation of liposomes can be observed naerascoplcafly
as "creaming?' or floccula~onof the ori@na"lly homogeneous disper-
sion. Occasionally, controlled flocculation of liposomes can be put
to good purpose in eontrolling drug suspensions f 31), but, sFor the
most past, aggregation is a nuisance from the rriewpoint of cosmetic
acceptabiility., and it may have undesirable effects on formulation
properlies.
Vesicle aggregation can be controlled by imparting an electro-
.
static charge to the liposomes Typicdly , negatively charged lipids,
such 8s pho~phatidyfglycemfand phosphatidflsefine, are added in
small quantities (about 2%- 10%by weight of total, phospholipid).
The addition of amphipaths, such as steary-larnine or eetgltrimethyl-
ammonium bromide, impart a positive surface charge to liposomes ,
but cationic lipid toxicity usually precludes use of positively
charged lipids in. pharmaeeutica?l products, A chelating agent, such
a s ethylenediaminetetraacetic acid (EBTA) ( 0.01%-O, 1% w / v) should
d s o be included to prevent divalent metal cation induced a g o e g a -
tion.
Growth of mean Xiposome size can &SO be inhibited by imparting
an efectrostatic charge, the inclusion of trace amounts of chelating
agent and by avoiding formulations based on SUVs, Because the
high radius of curvature in StlVs cause them to be thermodgnamical-
ly sts&ned, such. preparations show net vesicle p o w t h , which is
particularly rapid when the formulation is incubated a t , o r thermally
cycled through, the clcystdline-liquid c q s t d f i n e phase transition
of the lipid (32).

3, Liposome-Speeiffc Parameters
Changes in the chemical or physic& stability of a liposome h m u l a -
tion wB1 oftentimes be refieeted by changes in the liposome-specific
parameters. Appreciable changes in mean particle size may &feet
the release rate of active fngredient ( 2 3 ) . The captured volume
and the percentage entrapment will be changed if vesicle dkrneter
increases or decreases.
Thermodynamic equilib~umwill drive the flux of entrapped ma-
te&& into the continuous aqueous phase of a liposome vehicle.
Therefore, the desired percent entrapment shelf life of a liposorns
formulation may be difficult to achieve if unentrapped materid must
be removed and the release rate in the storage buffer is not mark-
edly different from that of the in viva release rate, By example,
consider a Eiposorne 6i3rmulatlion af a water-soluble d m g that is de-
s i p e d to release half its contents within 2 days of application to an
ophthdmic surface. A cleaned up formulation must have a storage
buffer release rate at least three orders of mamitude slower than
the in viva rate to mhieve 2 years of percent entrapment stabaitg.
Liposorne-Based Vehicles f"or Topical Deltver~y

IV, RESEARCH STUDIES

Therapeutic appacations of ligosomes have been enrrlisioned since the


early 1 9 7 0 ~hut
~ research on the topicat route of administration was
not pubuslled until the 1980s. The current body of literature con-
cerned specifically with topical. applications of liposornes can be
~ o u p e dinto ophthalmic and dermal studies.

A. Ophthafmie Studies
In 1981, Smolfn et aJ, (33) used free and liposorne-atrapped idoxu-
ridine, an antiurird agent, to study the treatment of herpes sirnpXex:
ksratitis in rabbits, An aqueous solutlian of 0.1% (wIv) idoxu~dine,
0. I % idoxuddine in phosjphatidylcholine-phosphatfdylglyeerol --eho-
lesterol LUV-OLV Xiposornes (35%entrapped dmg) , or the liposome
placebo, were applied three times d a y to rabbits that had been in-
fected p r e ~ ~ ~by s l corned
y abrasion and application sf a titer of
d r u s . Eiposome -entrapped idoxtl~dinereduced the size of viral-
induced corneal defects sipificantly, compared with free drug or
liposome placebo.
Lee and colleagues (3$,35) investigated the delivery of free and
liposorne-entrappa inulin and epinephrine XICI , bath, water-soluble
model compounds of substantially different permeabaity, In d t r o
release expeAments revealed that epinephrine HCI crossed liposome
baayers readily; 50% af osgi-nally- entrapped epinephrine HCX was
released from MLVs within 1 h r , whereas more than 90%of the en-
trapped inulin remained entrapped after 4 hr, In vivo studies In a
rabbit model Indicated that liposomes did not slow down ep-inephdne
HCl removal from tear f'fufd, Fufihermore, 45 min aAer instalation
of the liposome vehicle, epinephdne concentrations in eye tissues
(conjunctiva, cornea, aqueous humor, and iris-cniaw body) were
significantly- lower than aqueous solutions of epinephrine HC1.
However, when using Xiposome-entrapped fnulin in viva, marker
levels were increased many-fold over aqueous soXut.jlans In the con-
junctiva, cornea, and iris-cniary body. Inugn could not be de-
tected art; dl in the aqueous humor when entrapped in liposomes,
Thus, depending on the drug, Iiposome entrapment changed marker
.
dist ribution s i ~ i f i c a n t l y Liposonres were able ta increase localized
cancentrations of a water-soluble marker with low membrane perme-
abnity, perhaps by adhering to ocular tissues.
These results were vefibed by Ahrned and Patton (36) who &so
found that transcorneal u p t k e of MLV liposome-entrapped inuIin
liposomes is suppressed and noncorneal uptake was enhanced when
compared with an aqueous inuBn sdution, Aliquols of free or Xip-
osornal-entrapped tritiated hulin were instnled in rabbit eyes and
assayed for radiolabel 20 min later. Increased inulin levels were
found in the cornea and irls-c3iaw body when Eposorne-entrapped
inulin was instailled, Whereas free inulin was able to penetrate into
the aqueous humor to some smd1 d e p e e , liposome-entrapped inulin
was not detectable in the aquews humor. The authors suggested
that liposonle dosage forms can be used to promote d r u g absorption
selectively to noncomed oeular tissues,
In subsequent studies, Lee and co-workers found that Iiposome-
entrapped inulin u p t a e into intraocular tissues was indeed depend -
ent on adhesion of Bposomes to corneal epithelium ( 3 7 ) . The reten-
tion of MLV (phospatidylckroline-cholesterol) liposomes in tears, and
their interaction with corned and conjunctival, sur.flaees was studied
using tritietted-inulin (5%, w l v ) as a marker for entrtapped contents.
The rate constant for dearing these neutral MLVs was only 30%
slower than that compared with free inulin, and this rate constant
was insensitive to volume of instzed dose* Inulin-loaded liposornes
competed with empty gposomes for binding sites in the corneal and
eonjunetival surfaces, Pretreatment with empty MLVs reduced in-
ulin-loaded liposome binding by greater than 75% for both surfaces.
Neve&heless, liposomes did not kind 8~d;X-yto these surfaces; dm-
ing removed them effectively, The authors suggest that dssorption
into tear fluid is the major pathway of clearance,
Another factor inETuencing the reduced concentrations of inulin,
in various eye tissues was the slow liiposomzrl release rate of inulin
(about I% per hour in this study]. Inulin is a large-moleeular-
weight (about 5000 MW), water-soluble compound and is not expected
to have a rapid e f f l u x rate.
.
Lee et al f 37) concluded that topicdly applied, neutral MLV
liposomes are removed from the corned surface by tear drainage.
The extent of entrapped inulin absorption seemed to be controlled
by the number of adsorbed liposomes 8s opposed to the number of
instilled liposornes, Adsorption was not sufficiently strong, and ef-
flux from liposomes was too slow, La sustdn inulin concentration in
intraacular tissues. Thus, improved ocular cLte1ive;rzy. of water-soluble
entrapped materials requires enhanced liposorne-binding affinity and
a rapid release rate,
K r o h and ca-workers attacked the problem of i m p r o ~ n goeular
retention of liposomes by pretreating isolated rabbit corneas with
wheat germ agglutinin (WGA) and instnling p'bosphatidylcholine-
m h e d b r d n ganglioside MLV or SUV liposomes (38). Were, lectin-
mediated adhesion of liposomes enhanced binding to rabbit corneal
epithelium 2.5-fold. Transcorned ilux of radiolabeled carbachol en-
trapped in such ganglios-ide-containing liposomes was studied in an
in e t r o model using excised corneas with simulated "teefar Bow."
Unentrapped earbachol was not removed from the Iipossme formufa-
tion. Thirty minutes after instnlatlian there were no significant dif-
ferences in carbachol actidt y in cornea or transcorned receiver
.
compartment fir liposome-entrapped or free carbachol However,
90 min after instillation the liposome vehicle was able to m&nl;&n
significantly higher earbachol concentrations in the cornea and in
the transcorned receiver compartment. The authors propose that
the liposome vehiclet@aabjlity to sustain drug levels at 90 m h in the
receiver compartment was due -lo the continuing presence of drug at
the corneal surface.
Schaeffer and Krohn ( 3 9 ) studied in d t r o lipasorne-corned
binding mediated by electrostatic adsorption. Stearylarnine or di-
cetyl phosphate was used to modify the net charge of phosphaedyl-
ehoEne-cholesterol Iiposornes. Rank ordedng of liposome binding
to excirsed rabbit corneas showed significantly greater affinity
fabout twofold) of positively charged liposomes compared with neg-
atively charged Iiposornes, In turn, negatively charged lipasomes
bound twofold greater than neutrtzf, liposomes.
Recently, Gua and co-workers (40) have been able to demon-
strate that the in vivo retention of radioiodinated liposomes in rab-
bit eyes is increased many-fold by using MLVs containing positivr?ly
charged cholesteryl esters. They observed that liposome binding in
the eye was saturable, and the half-life of Xiposorne retention cauld
be manipulated by varying the charged cholesteryl es"tr/phosphdi-
pid ratio, the length of the spacer arm separating amino function
from the lipid head group, and lipid phase transition (fiuidi_Ly).
Schaeffer and Krohn (39) investigated the in d t r o transearned
flux of the water-soluble drug penienlin G in MLV or LUV lfposome
vehicles of quditatively different electrostatic charge, A11 liposome
vehicles tested showed significantly improved, drug delivery over an
aqueous penicalln G solution. Penicalin G flux: wross; rabbit cor-
neas J h r after inst3lation exhibited the same charge-dependent
rank ordering as lipid binding. Free d m g added to preformed,
empty liposomes did not show increased. d m g flux, suggesting that
d m g encap;rsulatian and liposome binding was a prerequisite for i m -
proved delivery,
The transcorneal Elux of the lipaphnie dmg hdozrole incorporat-
ed in neutral phosatidylcholine liposomes or in a palysarbale 86 die-
pession was evaluated in 6 v o wMh rats (39). A t f. h r after instiXla-
t.ion, a 1.Q rng indoxole per miXX9iter liposome vehicle preparation
prodded. the same drrrg concentrations in rat aqueous hurnor (about
20 vglml) as a 10.0 mg indoxole per mnlniter polysorbale 80 vehicle,
Furthermore, Schaeffer and Krohn discovered that the Iiposome ve-
hicle caused none of the histoia@cal changes induced by polysorbate
80 ocular toxicity (unpublished results). Thus;, in addition to im-
pmving the solubility of the anti-inflammatory agent indoxale, ve-
hicle safety was improved.
Shiota et &I , reported a sever-&-fold improvement of lfpo-
(41)
phnic dtamin A deliveq in viva. Increased specific a c t i ~ t y in
. var-
ious rabbit eye tissues persisted for up to 8 h r after instalation of
dtamin A in positively charged liposomes, when compared with 14-
tamin A in a simple vegetaible oil vehicle.
The consistent conclusion from ophthallmic studies is that lipo-
somes provlide improved local concentrations of active ingredients
proTajlded the compound is lipophaic, or it is water-soluble but not
very membrane-permeabXe, The indoxole study &so suggests that a
liposome vehicle may improve safety by replacing other more toxic
excipients used for drug solub;Ilizatian, The foregoing studks ali~so
indicate that liposornss improve transcornetrl Rux of lipophnic mole-
cules and for, at least, some water-soluble moleales such as peni-
cillin G.

8. Dermal Studies
The first results of topicdly applied liposomes were reported by
Mezei and Gulasekharam (42) in which tdamclindone acetonide en-
trapped within, ~pdmitoylpE.losphatidylch01i~~.e-cholestero1 MLVs re-
sulted in sipificantly Increased d m g levels in rabbit dermis and
epidermis, reduced blmd and reduced u ~ n concentrations,
e Xn a
series of reports (43-45) Mezei and co-workers used MLVs to dem-
onstrate the in viva transdermal delivergr of cortieosteroid drugs,
such as triamcinolone acetonide, tAamcino3lone acetanide 2 I-pdmitate ,
and progesterone, in rabbits and guinea pigs. Controls were con-
ventionail dosage forms such as ointments, although it is not clear
that control and liposome preparations had similar thermodynamic ae-
t i d t y to drive steroid partitioning out of the vehicle. Prepariltions
were compared tissue by tissue, For tdantcindone acetonide and
its more lipophilie pdmitate ester deAvative, liposorne preparations
were reported to consistently @ve simificantly higher epidermd and
dermal d m g concentrations, no significant differences in systemic
tissues (except for blood, which was significantly lower), and re-
duced u ~ n w yexcretion. Mezei a r m e s that this is indicative of
selective tsiarncinolone acetonide delirrev to skin tissues.
This generalization was tilsa reported for econazole base and
econazofe nitrlrLe (44). For this active ingredient, all skin tissues
showed increased concentrations of radiolabeled. drrxg, The pathrn
of in viva disposition of seven different ecanazole-liposorne prepa-
rations is simBar to triamcinolone acetonide; but; different lipid corn-
positions delivered different quantities, Multiple dosing wfth control
and econazole-loaded Iiposome preparations increased econazole ae-
cumulatfon silf"ni;Flcantly when lipssome vehicles were applied,
However, progesterone-loacled Xiposorne preparations skawed in-
creased concentrations in all skin tissues, and with the exception of
Liposome-Based Vehicles for Topical Delivery 341

blood, increased d m g concentrations in biopsy specimens &om all


systemic organs (44).
Mezei concludes that liposomes can prodde a vehicle that is most
useful for selective delivery of d m g to local sites, with reduced
systemic actidty. This would be a highly desirrcible feature for cer-
t&n d m gs (corticosteroicls by Mezeils example) in which systemic
effects must be avoided, At least for tdamcinolone acetonide and
ecanzlzole, it seems possible to achieve s i ~ i 6 c a n t l yhigher skin lev-
els of drug without increased systemic levels,
Liposome preparations of progesterone and hydrocortisone have
been studied in some detan, Flynni and c o l l e a ~ e s(46,47) made in
d t r o ftux measurements to determine the mechanism of percutaneous
absowtion of progesterone, hydrocortisone, and glucose acmss
h&rles% mouse skin. A mathematied treatment was derived, which
was later validated by Ertel and Carstensen (48). Incorporation in-
to a liposome vehicle did not change the skin permeahaity coefficient
of progesterone and hydrocortisone significantly, but the skin per-
meabi3ity coefficient of glucose was reduced sever& orders of nag-
nitude when entrapped in liposomes, Active ingredient flux: across
lipid b2ayers was the rate-limiting step of percutaneous absorption
for highly water-soluble molecules like glucose. Membrane-permeant
lipoph%ie molecules such as progesterone and hydrocortisone were
hypothesized to move across by a ""direct transferrvmechanism from
the bnayer hydrophobic phase to skin hydrophobic phase.
Knepp et al. (49) investigated the transfer mechanism of pro-
gesterone in more d e t d by comparing the transdermal Rux of free
progesterone zznd steroid in liposorne vehicles. The lipssome vehi-
cles exhibited zem-order kinetics of transdermd f2ux for over 24
.
h r Furthermore, there were no appreciable differences in trans-
dermd flax of progesterone from liposornes in suspension a r pro-
gesterone-containing liposomes immobilized in a $&, Because of
these similar kinetics and the physical inpossibifity of liposornes in
a gel coming into intimate contact with the skin, the authors con-
cluded that interfacid release of progestemne into the surrounding
aqueous medium was actually the rate-limiting step of the transfer
mechanism .
A sirngar mechanism may he operating for hydrocortisone as well
(46,47). tasch and Wohlrab (50,511 studied the concentration pro-
file of hydrocortisone in human cadaver skin after treatment with
hydrocortisone ointment or hydrocortisone-loaded liposomes , At 30
or 300 rnin afller application of hydrocortisone in ointment or lipo-
some vehicle, the skin was tape-stripped to remove layers of the
stratum comeurn, or it was clermatomr~tdto section the epidermis and
dermis. A s a whole, there were no differences in the quantity of
steroid found in tho stratum corneum with either vehicle. However,
at 30 min, the liposorne vehicle prolrjided about an eightfold and 14-
fold improved hydrocortisone concentration in the epidermis and
dermis, respectively, A t 300 min , the differences between liposome
and ointment were about four-fold for the epidermis and nhe-fold
for dermis. Lasch and Wohlrab concluded that Iiposomes prodele in-
creased eoncentratbns of corticosteroid in skin tissues.
It is woHh noting one study that is in disagreement with the
general results on steroid-containing liposomes , Vermarken et al .
(52) investigated the biolo@cal effect on hamster flank organs of
topically applied 5a-dihydrotestosterone. An acetone or liposome ve-
hicle (uncharacterized) was compared at equal androgen coneentra-
tions for its abnity to increase the flank organ size. The acetone
vehicle, it was discovered, elicited a p e a t e r effect than the dfiy-
drotestosterone-loaded liposomes. The authors suggest severd rrea-
sons for the disparity between their results and that of Mezeits
work with triamcinolone acetonide, Different steroids, schemes of
application, and animd species were used ; Vermorken and colleagues
also measured a bioXo@c& response, not tissue concentration aE
dmg.
Also, it is not known whether or not the liposome vehicle was
saturated with d m g . The actual dmglfipid ratio may not have fa-
vored significant partitioning of dihydrotestosterone from the vehi-
cle into skin layers, The importance of considering the effect of
this ratio on quantity af active ingredient absorbed was irlustrated
with butylparaben by Komatsu et al. ( 2 5 , 5 3 1 . The quantity of bu-
tylparaben absorbed per unit time across excised guinea pig skin
was inversely proportiond to the Iiposorne lipid /parsiben ratlio,

C, Miscclltaneous Studies Using Phosphalipid-Based Vehictes


Mishihata et A , (54) studied the in ~ t r ando in vivo percutaneous
absorption of the nonsteroidal anti-inflammatory~compound sodium
diclofenae with aqueous d m g solutions end "'phospholipid gelse''
Both in d t r o and In viva transport across rat skin was increased
by the addition of hydrogenated soy phospholipid, The in vitro
penetration rate of dicXofenac increased with increasing soy pkosphw
lipid content up to 0.5% ( w l w ) lipid, The phospholipid gel systems
also promoted significant accumulation of diclofenac in skin tissue;
the cumulative amount was proportional to the penetration rate. The
authors proposed that the improvement was due to the surface-active
nature of the phospholipids, I[ntravenous administration of diclo-
fenac resulted in no sipificant skin tissue concentrations , and top-
icdly applied diclofenac was not as bioavailable systenieally as in-
travenous injection, The authors concluded that aqueous phospho-
Xipid gels of dielofenac are more suititable for lmalized than for sys-
temic treatments.
Vehicles 343

freeze-frac-
ma~n.eutcresonance
mixtures as in a tran-
of pure phOSI>h~l­
of pure
Nishihata's

ab-

perC1Jt~:mE~outs ahsott'nt1ton have also been


studied in solvent vehicles in which are not pres-
ent. Kato et al. (56) studied the effects of 1% lecithin in
on the in vitro percutaneous of bunazo-
or isosorbide dinitrate across excised hairless
skin. The addition of lecithin increased the amount of
in vitro at 24 hr, for all , with the
for bunazosin Lecithin, at concentrations used in this
\.<.uo.J.LJ;I'."" bunazosin in
of bunazosin (about 23
with lecithin was COIDPl~re~d
to rabbits. In vivo pene-
tration was undetectable in but the addi-
tion of lecithin was able to effect levels
the 24 hr of observation. Because the formulations used were all
saturated solutions of ( sim:ilar and be-
cause lecithin did not bunazosin saturation, the authors
propose that the mechanism of lecithin action the
permealt>il1tty barrier of the skin. This is since the
of the lecithin was not characterized it is im1DOf3Silt:>le
that of its
These
to

0. Conclusions
The results of many studies may be summarized the
servations that

1. to some surfaces can be mediated


electrostatic interactions.
2. Topical application of dmg-loaded liposornes promotes delivery
of the active inpedient to local tissues and, for at least some
cases, may reduce systemic d m g levels.
3, Transdermal and transocular delivev of lipophnic molecules is
definitely enhanced by liposome incorporation,
4. TransdermaX or transocular topical, delivev of hydrophigc mate-
r i d s is considerably more problematic, Most small, membrane-
impernreant markers show no improvement, yet the occasionaf
success with some malecules like penicalin G suggests that more
studies to discover structure-actfdt y correlations are warranted,

Xt is perhaps not surprising that transdermd delivem of water-


soluble membrane-impermeant compounds is further impeded by add-
.
ing yet another permeabsity barder , namely, the liposome Cer-
tainly, there is no eddence for g,enetratfc.n of intact liposornes with
their payload through the skin and, indeed, eddenee to the eon-
tram. Specific rnonitodng of radiolabeled phospholipid markers
shows no eddence for phospholipid penetration through the skin un-
der in d t s o ( 4 1 ; Uster, unpublished data) or in viva conditians
(25). The penetration of intact liposomes through the stratum eor-
neum can be laened to the pmbabillity of passing a basketball
through a chdn-link fence without ruptufing it.

V. FUTURE DIRECTIONS
A. Basic Research
Tapiedly applied, dmg-loaded fiposomes can substantially imprave
d m g loading, drug deliveq, and sustained release, thereby offer-
ing cfearcut advantages over traditional dosage forms. These ad-
vantages are particularly eddemt for the more lipaphi3ic aetive in-
gredients, in which elevated locd concentrations are conaistently
demonstrated and that for, at least, some dmgs reduces the sys-
temic concentrations, Because there is no evcidence for liposome
penetration through intact topical surfaces, such as the stratum
corneum, it is perfectly lo@cal that water-soluble, membrane-imper-
meant active inpedients face an additional barder to increased bio-
avaSabi2ity when entrapped within liposomes , For topical surfaces,
the stmeturd integrity of which Is not compromised, a second gen-
eration of lipssome vehicles vvnl have to be developed,
However, water-soluble drugs entrapped in convention& lipo-
somes may play an extremely useful mle in treating conditions ;In
which the i n t e e t y of the stratum comeurn or other topied surface
has been damaged. The study by Smctlin and co-workers on vciral
infection. of abraded rabbit corneas is espedslly encoura@ng (33).
Liposome-Based Vehicles for Topical Delivery 345

The entire area of drug-liposome treatment of "broken" skin or


other compmmised topical surfaces has yet to be explored in depth.
Almost all of the research investigations that have been cited on
topical applications have facused on phosphatidylcholine or phospha-
tidylchoane-cholesterol mixtures. Usually, measurements of impor-
tant lipomme variables such as mean size, captured volume, and
percent encapsulation are not reported, For the most part, lipid
raw materids and the liposome formulations themselves have not been
carefully characterized for actual lipid composition or for depadation
products, such as free fatty acid and Iysophosphatides. Careful
biochemical and physicochemical characterization is warranted be-
cause differences in these pmpePties may lead to substantial dis-
crepancies in results from preparations that nominally appear to be
the same. Indeed, the precise role of lipid composition in facilitat-
ing drug delivery is another research area that requires gystematie
charting.
Liposome interactions with the immediate skin microenvironment
also have not been ~xploredin any detail. Fundamental questions,
such as the structural integrity of liposomes in intSmate contact with
skin components, remain unanswered.
The potential discrepancy between bioassay and marker monitor-
ing, as highlighted by Vermorken et al. (521, raises a general issue
about liposome formulations that is especially pertinent to water-sol-
uble , membrane-impermeant active ingredients. Namely, do tissue
concentrations of drug, as measured by marker or chemical assay,
represent biologically avdable drug, or is the drug permanently
masked within its Eposome carrier? Future research should carefully
evduate, either by use of by an appropriate bioassay or by demon-
strating the physical separation of active invedient from llposome
carrier at the site of action, whether or not the increased tissue
concentrations of the active ingredient actually is bioavailabfe.

8. Liposome Formulations in Product Development


From the fiteraturr?, it seems clear that Epophilic drugs are the most
likely candidates for conventional liposome vehicles in potential ef-
ficacy. Lipophilic drug candidates also have advantages during for-
mulation and process development, most notably by improving drug
loading and obviating the need for a cleanup step.
A s with any new technology, there have been unique problems
that must be overcome to introduce a pharmaceuticay awegtabIe
product to market. Many of these challenges have already been met
successfully because there are now more than a half-dozen parenter-
al liposorne formulations in clinical trials worldwide. The rapid
progress in lipasorne technology that has been made during the past
346 Uster

few years suggests that topical Eposome products are practicd as


well as useful. It appears to be simply a matter of time before a
topical formulation &I also be in clinical trials.

ACKNOWLEDGMENTS

I would like to thank Drs. R, Abra, L. Guo, and IM. Woodle for a
critical reading of this manuscript and for their many useful sug-
gestions.

REFERENCES

A . D. Bangham, M. M. Standish, and J. C. Watkins, J. Not.


Biol. , 13: 238, 1965.
C. Pidgeon, A. H. Hunt, and K. Dittrich, Pharm. Res,, 3:23,
1986.
C. Pidgeon, and C. A. Hunt, J. Pharm. Scf., 70:173, 1981.
D, A. Barrow, and B. Ft, Lentz, Biochim. Biophys. Acta, 597:
92, 1980.
F. Szoka, JP., and D. Papahadjopoulos, Annu. Rev. Biophys.
Bioeng., 9: 467, 1980.
F. Szoka, and D. Papahadjopoulos, in Liposomes: from Physi-
cal Structure to Therapeutic Applications (6, Knight, ed. ) ,
Elsevier /North-Holland Biomedical Press, Amsterdam, p 51,.
1981.
D, W. Deamer, and P. S, Uster, in Liposomes (M, J. Qstro,
ed.). Marcel Dekker, New York, p, 39, 1983.
P. R. Cullis, M. J. Hope, M. El. Bally, T. D. Madden, L. D.
Mayer, and A. S. Janoff, in Liposomes: from Biophysics to
Therapeutics (M. 3. Ostro, e d , ) . Marcel Dekker, New York,
p. 314, 1987.
L. Saunders, J. Perrin, and D. B, Gammaek, J. Pharm,
Pharmacol, , 14: 567, 1962.
IM. B. Abramson, R . Katzmann, and H. B. Gregor, J . BioI.
Chem. , 239: 70, 1964.
D. Papahadjopoulos, W. J , Vail, K. Jacobsen, and G, Poste,
Biochim, Biophys. Acta, 4: 483, lS75,
U. Pick, Arch. Biochern, Biophys., 212: 186, 1981,
M. J. Nope, M. B. Bally, G , Webb, and P. R, Cullis, Bio-
chim. Biophys. Acta, 812: 55, 1985.
W. Li, and T . H , Haines, Biochemistry, 2537477, 1986.
D. W. Deamer, and A. D. Bangham, Biochim. Biophys. Acta,
443: 629, 1976.
Liposome-Based Vehicles for Topical Delivery 34 7

.
F , Szoka , J r , and P) , Papahadjopoulos , Proe, Matl, Aead,
Sei, , U S A , 75: 4184, 19.78.
9, M. N, Kremer, IM, W. J, v. d , Esker, 6 , Pathmamanohoran,
and P. H. Wiersema, Biaehemistry, 17:3932, 1971,
W . G. Enoeh, and P, Stdttmatter, Proe, Na21. Acad, Set, ,
USA, 76: 145, 1919.
3 . Pkriffppot , S . Mutaftschiev, and J. P. Liautard, Biochim.
B t o p k y ~ ,Acea, 734:137, 1983.
R, 111, Sehwendener , Cancer Drug B e l i v e v , 3: 123, 1986,
Q , IZurnBuehl, and H. C. Weder, Blochim, B h p h y s . Aeta, 640:
252, 1981.
R. 9. XClimchak, and R . P , Leuk, Biopharm. Manuf"act, , 1: 1 8 ,
1988.
R , H Guy, S , Wadgraft, M e J. Taylor, and I , W, KelXaway,
J, Pharm, Pharmaeot, , 35: 12, 1983.
Z. T. Chowhan, T. Yotsuyanae, and W. L, Higuchi, Biaefiim,
Biophys. Aeta, 266: 320, 1972.
M. Kornatsu, K . Hig8ki, W e Okamoto, M, Miyagawa, M. Hashida,
and H, S e z e i , Chem, Pharm. Bull, 34: 3415, 1986,
N, S . Bean, and S M, Hernan-tickahl, J , Pharm. Pharmacol, ,
I

f6:58t, 1964.
3, EM. C. Gutteddge, FEBS Lett., 17%:245, 1984,
6 . A , Hunt, and S , Tssang, fnt. 6 , Pkarm., 8: 101, 1981.
N, I>. Weiner, W. C . Bmning, and A . Felmeister, J, Pharm,
Sci, , 62: 1202, 1973,
S. Frokjes, E . L. Hjosth, and 0. Worts, in Optimization of
Drug DeZilaeq, Alfred Benzon Symposium 17 (HeBundgailrd,
A. B. Hanson, and W. Kofod, eds.). Munkagaard, Coperr-
h a p n , p. 384, 1982,
8.-L. Law, W. -YeLo, and G,-W, Teh, J . Pharm, S C ~,, 76: 545,
1987.
.
F , 2. Martin, and R , C MaeDonaXd, cSr"oche?mistv, 15: 321,
1976.
G , Smolin, M. Qkurnoto, S. F d e r , and E), Condon, Am, d .
Qphlhalma!, 91: 220, 19.81.
.
It. E. Stratford, J r , , B, C Yang , M , A. Redell, and V, H,
L e Lee, Int, J, Pharm, , 13: 263, 1983.
E . S t r a t f o ~ d ,J r . , B. 43. Yang, M. A , Redell, and V . W ,
L , Lee, Curr, Eye Res, , 2~317, 1983,
I , Ahmed, and T. F. Patton, I n t , J. PPlarm., 343163, 1986,
V, H. L. Lee, K, A . Takernoto, and D. S. limoto, Curr, Eye
Res. , 3: 585, 1984,
H. E . Schaeffer, J, M, Breitfeller, and D L L . Krohn, Invest*
OphlhaEmsE, Vise Set, , 23: 530, 1982.
EX. E . Sehaeffer, and D. E , KroEm, Invest, OphthaZmol, V i s ,
Sci, , 22: 220, 1982,
L. 8. S , Guo, 6 . T, Redemann, and R . Radhakritshnan, In-
varst, Qphthafmol, V i s , Set. , 28: 72a, 1987.
R . Shiiola, A. Sarris, R. Radkit&rlshnan, and L. S. S, Guo,
Invest, OphthalmoE, Vis. Sei, , 28: 159a, 1981,
M, Mezei, and 17. Gulasekharlam , L i f e Scr". , 26: 1473, 1980,
M , Mezei, and V. Gulasekharam , J. Pharm, PharmacoZ, , 34: $73,
1982,
M, Mezei, in Topics in Pharmaceutical Sciences (D, D *
.
Breimer and P Speiiser , eds .I, Elseder Science Putbliahers ,
Amsterdam, p. 345, 1985.
El. B. Harsanyi, 3, C . HBcf-rie, and M, Mezei, J. D e n t , Res,,
65: 1133, 1986,
M. G . Ganesan, N, D, Wetner, C, L. PXynn, and N. F", H,
No, lnt, J. Pharm., 20: 139, 1984,
N . F. H. N o , M. G. Canesarr, M. D. Weiner, and G. L. Flynn,
J. ConlmlZed Release, 2~61,1885.
K . 13, Ertel, and 3. T, Carstensen, Int. I , P h e w , , 3 4 ~ 1 7 9 ,
1986.
TI, M, Knepp, fit. S , Hbz, F. C . Szoka, Jr., and XE. R . Guy,
J , Controlled Release, 5: 211, 1981,
3, Lasch, and W , Wohlrab, Biomed, Btoclzim, Acla, 45: 1295,
1988,
W , Woltfrcab, and J, Lasch, Demalo2ogica, 174: 18, 1987,
A . J , M, flermorken, M. W e A , C. Nukkelhoven, A . M . G ,
Vermeesch-Markslag, G . M. A. A. Gms, P . Wir-tz, and 2,
Zlegenmeyer, J. P h a m , PharmacoE, , 35: 334, 1984.
H, Kornatsu, 15. Olramoto, K , Miyagawa, M, Hashida, and W.
Sezaki, Chem. Pharm, Bull. , 34: 3423, 1986,
.
T, Niahihata, R Kotera, V. N&ans, and M. Pamazakf , Ghem,
Plzamm, Bull, , 35: 380q 12987,
S , W . Nuf, T , P , Stewart, P. E . Ueaglle, and A , E). Albert,
A r c h , Bioehem, Biophys. , 20rt 227, 1981,
A , KaPo, Y, Ishibashi, and Y , Miyaki, J. Pharm, Pharmacol,,
39:399, 1986.
SurtaGtant Association Coiloids
as Topical Drug Delivery Vehicles

DAVID W. OSBQRNE The Upjohn Company, Kalamazoo, Michigan


ANTHONY J. I . W A R D * and KILIAN J. OBNEiLL University College
Dublin, Dublin, Ireland

The addition of a surface active agent to a drug delivery vehicle


can result in improved drug stability and cIinical potency, increased
drug absorption, and decreased drug toxicity (1). If the drug ve-
hicle containing a surfactant is further optimized to form a micro-
emulsion or lyotropic Iiquid crystsffine system, the additional advan-
tages of potentially increased solubilization of a poorly soluble drug
and thermodynamic stability are rerdized (2). For micmc?mulsions,
these advantaefes stem fram an apparent particle size in the range
of 100 to 1000 A compared with the 5000 to 10,000 A minimum parti-
cle size, typical of macroemulsions. The use of micellar systems,
micmemulsions, and lyotropic Xiquid crystals as topical drug delivery
vehicles will be the focus of this chapter. Collectively, these mo-
lecularly structured vehicles are called surfactant association colloids.
A description of these vehicles and current progress concerning
their formulation will be followed by a discussion of in vitro perm-
taneous absorption experiments that have been used to evaluate sur-
factant association colloid vehicles.
Before proceeding with this chapter, it is important to define
the term microemulsion. Unfortunately, this term has different
meanings depending upon whether the investigator has a colloidal
chemist~ybackground o r a pharmaceutical background. In this
chapter we wifl consider a micmemulsion to be a system of water,
oil, and amphiphile(s) that forms a single phase, transparent or
translucent (opalescent), fluid, isotropic dispersion that is presum-

*Current affiliation: Clarkson University, Potsdam, New York


350 Osborne, Ward, and OWelFI

ably thermodynamically stable ( 3 ) . Lyotropic liquid crystals are


likewise f'armed &om a system s f water and mphiphile(s) that may
hclude an oil, These solvent-dependent liquid crystals may be ei-
ther ~ s c o u sanisotropic or stiff isotropic dispersions. Frequently,
a water, oil, and amphiphae system w n l cont&n concentration ranges
that can be defined as mieellarlmx"e~oemuZsionand &so contain can-
centration ranges that form liquid crystals (i,e., higher surfactant
or lower oil concentrations) .

Surfactant molecules may be descdbed in terms of their hydrophgie


and lipophilic natures, ft is the bdance between these two confiict-
ing properties that determines the range of phenomena from surface
effects to colloid& b e h a ~ o rf"ound in these systems, Qverdl, the
basic thermodynamic requirements of minimum free e n e r w and maxi-
mum entropy in the system apply, Briefly, the unfavorable entropy
consideration assczeiated with the aggregation of amphiphirle molecules
is balanced by the lower system free e n e r m obtained by mmimizing
the Izydrophsie-water contacts.
Advances in understanding the phase behador of surfactants
and lipids have been made (4-7) by using thermadyrrilmic approaches
incorporating the geometric requirements o f the molecular packing.
Qbservations of agpegated systems that use novel optical (8) and
c3Xectron microscopic (9) techniques both support the nature of the
aggregates expected and confirm their highly dynamic nature.
In Chapter 10 of this text (Sect. 11-A. B ) the t e r n a q represen-
tation of phase b e h a ~ o rwas introduced, and the usefulness of this
representation described. In light sf the "thermodynamic approaches
just @ven as references, -it is useful to systematically consider the
mo'lecular events that occur to cause phase changes. For the sur-
faetant systems considered in this text, the solvent is water and
will be positioned on the lower left corner of the t d a n m l a r plot,
whereas the surface-active component ( surfaet ant) will be positioned
at the lower right corner, The remaining component, usually an
oil, will be positioned at the top of the triangle. First consider the
b i n a v system in which a surfactant is added to water.
Although sodium dodecyl sulfate I'SDS) is not acceptable for use
in topical pmduct s ( 102 , the well-characterized; definite series of
phase changes that occur for this surfactant, as the surfactant!
water ratio increases, serves as an instructive example. Note that
increasing the surfactantlwater ratio means the earnposition changes
from the lower-left corner af the ternary diagram to the lower-right
corner o f the ternary diagram, Initially, the addition of SDS to
Surfae tan t Association Colloids as Vehicles 352

water at 2s06 results in a clear, colorless isotropic solution (11).


Macroscopicdly this mktmre does not change appearance until a 36%
SDS solution i s obtained, Molecularly, however, dramatic changes
are occurring. The first change occurs at the e d t i c d micellization
concentration (CMG), Before the GMG, the SBS molecules in the
bulk water phase can be considered as indiddually dispersed and as
acting as simple electrolytes, Above the CMG, many of the solu-
tionis physiezll properties change. These correspond to the forma-
tion of norm& micelles (Fig, l) in which approfimatelg 60 amphi-
ph2es form mughly sphedcd rni~elleswith a hydrated radius of 25
GI Between the CMC and 25% SDS the micelfes are approdmately
constant in size and mdntain a relatively eonstant number of bound
counterions (80%). Above 25% SDS , however, a transition from
spherical micelles to rodlike (prolate ellipsoid) shapes rapidly oc-
curs. This effect, sometimes known as the third C M C , is tempera-
ture-sensitive ( 1 2 ) , occurdng at 15%SDS when the temperature is
raised to 7 0 ° 6 ,
The addition of electrolyte also causes a transithn from spheres
to rods to occur at lower SlDS concentrations (13). At 2S0C, a
6,9 x 1 0 " "M~ SDS ~ in 0.15 M HaCl solution contdns rods with a
semiminor axis equal to the radjlus of sphe&cd micelles (25 A ) and
a semimajor =is slightly larger, with a length of 33 A . 'Phis is
dramaticdly contrasted with a 6.9 x 10-10 R4 SBS solution in 0.6 M
NaCl in which the semimajor axis increases to a len@h af 208 A ,
whereas the semiminor axis remains unchanged. When NaGl is pres-
ent in the SDS-water system, temperature has the apposite effect*
Bere, decreasing temperature causes micelles to elongate with the
greatest distortion occurring at the highest MaCl cancentrations (136,
The next phase encountered with further addition of SDS fol-
lows directly from the formation of the rodlike micelles, Before the
addition of 36%SDS, the rod-shaped miceltes are disordered and,
therefore, isotropic, After the addition of 36%SDS, the disordered
rodlike micelles come into equnibrium w&h the simnar, ordered hex-
agonal liquid cryst& phase (see Fig, 21, From 39% to 50%SDS, alZ1
of the SDS amphiphifes become ordered to produce the one-phase
hexapnaf liquid c v s t a l re@on, At soap concentrations &ove 50%
SDS, there is no longer enough water to act as a solvent, and the
cqstalUne soap is in. e q u 2 i b ~ u mwith the hexagonal liquid c v s t a l .
Thus , the baseline of' ternargr -phase dianams for crystaaine ionic
surfactants frequently show this provession of micelles, to hexrag-
anal liquid crystals, to erystdline surfactant, For nonionic sur-
factants of the polyoxyethyleneglycol type the baseline phase pro-
gression is from normal mieelle, to lamellar liquid ewstal, to inverse
micelle or surfactant- Ach solution ( 3) .
Upon addition of a third component to an anionic surfactant
syst ern, a different characteristic phase pra~fressianresults, The
Osborne, Ward, and QWeiZX

Figure 17.1 Idealized sta"tic structures far surfactant association


colloids: (a) normal micellar, (b) hexagonaf Equid cryst&, (c) la-
mellar liquid crystd p (d) inverse mieel-flar, Ce) reversed hexagonal
Iiqufd clfystd, and (f) a cubic liquid crystaf in which the crass
seetian has been split to show that either a norrnaf or inversed orpi-
entatian can be dsudfzed. Other structures have been proposed
for cubic liquid crystalline phases. Additions to a w a t e r - ~ c hsingle
taBed anionic surfactant, system that will result in movlirrg from one
phase t s another are noted by the arrows,
Surfoe f ant Association Galloids as Vehicles

Figure 17.2 Phase diagram according to Ekwafl (12) Tor the system
water-octanol-sodium octmoate at 25OC. The single-phase re@ons
are denoted by L1 for the normal micellar, E for the hexagonal Xiq-
uid c q s t d , D for the lamellar Equid crystal, and L2 for the inverse
mieelfar, This phase behavior is typical. for a single-tajil: anionic
surfactant in combination with a medium chain alcohol. (From Ref.
11, used with permission of Aeadernie Press),

thoroughly characterized three-component system water-octanoX-


sodium caprylate shows the typic& phase behador in Fimre 2 (11).
At high water eontent, normal miceXles form, As oetanol is added,
the lamellar Bqufd crystalgne phase is encountered and, uftimately,
at highest oetanol eontent, the inverse micells re@on occurs (see
Pig, I). This trend, which is characledstic of ionic surfactant
sy sterns with medium -chain length. alcohols a8 eosurfactant s , can be
reasonably explained by packing considerations ( 6) ,
The paeking ratio (PR) far these systems i s defined as

where v and lc are appraximately the volume and the length of the
hydr0cam;bon portion of the amphiphae, and acl, is the optimum area
354 O s b o m e , Ward, and OVJeilt

per headgroup. The method used to cafculate the packing ratio


uses Tanford's relations to calculate v and I, as shown in the fol-
lowing (7):

where n is slightly less than the number of carbon atoms per am-
phiphiIe chdn. Determination of a. is accomplished by x-ray dif-
fraction measurements on the lamellar liquid crystal.
Defined this way, the values for packing ratios related to the
different stmctures are:

Structure Ratio

Spherical micelles PR < 113


Cylindrical micelles 1 / 3 < PR < 112
Bilayers (or vesicles) 1/2 < PR < 1
Inverted structures PR > 1

The use of these principles can explain the transition of normal


miceEes , to bitayers , to inverted structures upon cosurfactant (oc-
tanol) addition. Ninham argues that a cosurfactant acts principally
to increase the volume per surfactant molecule, without significantly
affecting % or (6). This is because the uncharged cosurfactant
is able to pack between the ionic head groups without causing re-
pulsion. Thus, as octanol is added to the normal micellar region,
the packing ratio increases until it falls In the range of the lamellar
liquid crystal, thereby implying that packing considerations find
this assaciation structure more favorable. Analopusly, further ad-
dition of octanol increased the packing ratio toward and above uni-
t y , indicating that inverted structures are more favored,
To further investigate this idea, Friberg and Flaim (14) evalu-
ated the packing ratio for a number of lamellar systems at their
highest cosurfactant concentrations. For all systems studied, the
packing ratio approached unity at that point at which the system
entered into equilibrium with the inverted phase. Despite the sim-
ple geometfic arguments used in the packing ratio concept, the
method provides a straightforward, conceptural method of relating
association structure formation to the physic& eharactegstics of the
component compounds. From these arguments, the phase progres-
Surfactant Assoeiatisn Calloidls as Vehicles 355

sions of normal micelle, to cyX;indricl;ll micelle, to hexagonal phase,


to crystdline s o w and normal micelle, to lamellar (bilayer) liquid
crystals, to inverse miedles, can be related to a serYies of molecular
events that are based on the molecular structures of the surface-
active matell.i&s involved. Therefore, it 1s v e v reasonable that
water-afcohoX-ionic surfaetant systems exhibit sirn2ar, but not iden-
tical, phase behaear, whereas nonionic surfactant systems will show
different behador, but sirnaar phase progression. These trends
become very useful for predicting how surfactants will interact with
solvents, even when p r e ~ o u scombinations of these components
have not been mked, An empirically dedved schematic description
sf the transition from normal to reversed structures (15) is shown
in Figure 1.

l (I. FORMULATION Q F TOPICAL SURFAC-


TANT ASSOCfATlQN C O t l O l D S

Although brief, the foregoing description, of surfactant assoeiatkn


colldds does p r o ~ d ethe bzrekgmund necessary for a more cletaserrl
redew of the literature that discusses tapical micmemrxlsions and
liquid crystals, This section will describe formulations that are
suitable for topicd use, but that have not been evalluated by per-
cutaneous transport methods to measure release of a particular dmg
or model compound. Some of these formulations mag have been evd-
uated for parental administration and, thus, could also be ccmsid-
ered for use topically. Formulations that have been evaluated using
in ~ t r permxtaneous
o trmspo& techniques will be described in
Section I V .

A s stated in the introduction of this chapter, microemulsion. vehicles


have the potential advantages of both improved stabsity and solu-
bsization characteristics compared with macroemulsion topieals .
These advantages stemmed from the smaller particle size character-
lstie of microennulsions, It is noteworthy that particles of 1000 A
and smaller are less than one-fourth the wavelength of dsible light,
resulting in transparent or translucent [opalescent) systems, Al-
though transparency may provide a conceptual apped to the can-
sumer, it prorrides other advantages to the furmulator. A clear
system guarantees mixing has been adequate to solubnize afX of the
d r u g or other solid materials in the formulation. Milky white creams
typical of macroemulsion-based topicals obscure the presence of un-
dissolved solids, This is particularly important when the d m g is a
356 Osborne, Ward, and QfNeitE

solid, because bioavdlabilit y of a crystalline drug su spension may


v a v dramatically from that of a solubilized drug. For m2ky white
macraernulsion creams or lotions i_P is difficult to determine if the
Anal p r d u c t requires addition& mixing time to pmperly disperse or
soIublfize the drug* However, for a micmemulsion formulation, mix-
ing unts all of the solid is dissolved or solubaized, as eddenced
by a clear pmduct, proddes a readay discernible endpoint for
produet manufacture.
The ability of micrwmulsion systems to solubnize poorly soluble
materials is fundament& to the extensive study of these systems
over the years. Because drugs frequently cannot be delivered to a
therapeutic d o s a p because of insufscient solubility in solvents that
can be ingested, injected , or applied topically ; solubiljzation af the
d m g within a mier~emulsionmay be particularly impor-.tant. The al-
ternative prodrug approach of improrring solubaity by molecularly
chan@ng the drug to a more so1ubXe andogue requires toxicols@cal
studies to be completed on each a n d o w e , which may result in un-
acceptable time delays or increased expense, Thus, soldng prab -
lerns of low salubaity by using surfactants can complement prodmg
appmaehes by satisfying solublBty or solubilization demands in the
short -term, while the ideal prodrug is being developed, character-
ized, and screened for safety,
Thermodynamic stability is important because this charaeteriIstic
results in (I) a formulation the properties of which are not depend-
ent upon process (2. e , , shear rate, forced cooling, and such) , and
( 2 ) a formulation that will not phase-separate, prodded temperature
and pressure eanditioxls =main reasonably constant. It is useful to
further consider what it means fir a system to be thermodyrtarnic&ly
s.t;rble. For a microemulsion, thermodynamically s t a l e means the
system is coneeptudly andogous to an equjilibrated salt solution be-
low saturation, As long as the system remains closed (i,e, , no
evaporation or chemicral. depadation) and the temperature and pres-
sure remains the same, the properties of the system will remain un-
changed indefinitely. However, if the temperature sf the solution
is decreased sufficiently to cause salt crystals to precipitate, then
the single-phase solution changes to the two-phase system of solid
cx~rstaIsin equilibAum with saturated salt solution , Returning this
system to the o ~ @ n atemperature
l will reverse the phase change,
and at equ3ibriu;m will result in a solution with properties and corn-
position identical with the ori@nal salt solution, Analogously, a
microemulsion once mixed (equilibrated) will not separate into its
component phases provided that the temperature and pressure re-
main eonstant. Xf changes in storage conditions cause the micro-
emulsion to split, it will re-form once it returns lo the os@nal con-
ditions, Importantly the ori@nal formation or reformation of the
Surfie t ant Association Colloids as Vehicles 35 7

single-phase microemulsion is independent of process, being soXdy


dependent upon adequate mking to b d n g the system to equsibrium.
Comparison of thermodynamically stable microemulsions with kinetical-
ly stabilized macroemulsions emphasizes the g m e s s i n g advantages
gained by formulating microemulsions, Macroemulsions are generdly
very process-dependent , as such require the use of homogenizers
or scrapped surface coolers to produce stable pmduets. If condi-
tions change, resulting in separation, simple mixing will not be suf-
ficient for reformation of a macroemulsion system, Likewise, even
the slightest change in processing may produce a difference in the
praperties of the macmemtllsion system, Based upon these manu-
facturhg considerations , tlrermodynanicaffy stable microemulsion far -
mulations require less capital investment because only the most basic
mhing equipment is required.
In addition, microemufsions that have an inverse mieel,Iar struc-
ture may be less comedogenic than either creams or solutians. This
spewlation is based on studies of the lipids charaetewlistic of hair
follicles (16). Such fouicular ljpids are rich in fatty acids, When
.viewed in the light of recent information on the effect that partidly
neutralized fatty acids have on baayer-structured epidermal lipids
f 173, it seema possible that bsayers may also form fmm follicular
lipids. The increase in viscosity that accompanies bnayer formation
could be the be@nning of the comedo '?plug," lf Xfbillayer formation
were to occur, water-based emulsions would tend to swell the bi-
layers, whereas an bverse micellar system may tend to destabilize
the bilayers,
The current disadvantages of microemulsion systems for topical
dmg delivev can be related ta the limited number of topfcrzl: micro-
emulsion systems studied. Because most pubgshed m5ememulsion
studies are concerned with detergency (18) or tertiary oil recovem
( 19) , the ol%,phases tend to be either short - to medium-chain alco-
.
hols (butanof to dadecanol) or hydrocarbons (a, g , decme , hexa-
dacane , benzene), Only a f e w surfactant asssciaticzn colloid systems
have been charactedzed that are pharmaceutically applieilble Thus, .
rather than current disadvantages, it is more accurate to list these
as currtlnt challenges in formulating topical microemulsions, namely,
(1) determining systems that are nontoxic, ~ o n i r d t a t i n g ,noneome-
dogenic, and nansensitizing , and ( 2) formulatin p cosmeticdly ele-
gant micrsernuls$o:on topieals ,
Although only a few of the abrernentfoned pharmaceuticalXg ap-
plicable studies are limited to topical systems, more general texts
provide useful, discussions concerning the implications of d w g de-
livery. systems that contain surfactants, In particular, the text by
Attwood and Florence (20) and a very readable redew by Ehar-
grave et al. (21) concerning use of microemulsions as oral, fnjeeta-
358 Osborne, Ward, and O'Neill

Table 17.1 Seifety Considerations of Short - to Moderate-Chain


Length Alcohols

Qral LDZ0 in
Eye Skin
Alcohol rat (mgikg) irritation irritation

2-Butanol (least irri- 6480 Moderate Threshold cone.


tating of four isomers) 7.8%
Hexanol 720 Severe Mad
Octanol 1790a Moderate Threshold cone.
12%
Decanol 472 Severe Severe

aOral LDs0 for mouse, oral LDS0 for humans 0.5 to 5 glkg.

ble, and topicd dosage forms. Studies that have been completed to
identify a microemulsion formulation that will be pharmaceutica1:y ae-
ceptable, that i s , nontoxic, nonirritating, and so on, will be the
focus of this section.
The formation of micmemulsions has traditionally been dependent
upon the addition of a medium-chain length alcohol to function as a
cosurfactant. Because alcohols of this chain length range tend to
be skin o r eye irritants (Table 13, their use In topical. formulations
is very limited. An early study that attempted to find a more suit-
able topical microemulsion compared the solubilization of h y d m o r t i -
sone in a traditional microemulsion with the solubilization in a micro-
emulsion contetining pharmaceutically acceptable surfactants (22)-
The traditiona1 mieroemulsion consisted of sodium stearate-sodium
myristate surfactants and moderate-chain l e n e h alcohols (2.e.. n-
butanol t h r w g h n-heptanol) whereas the pharmaceutical microemul-
sion was a mixture of Brij 35 and Arlacel 186 combined with iso-
propanol. For both the traditional and pharmaceutical microeml-
sions, solubility of hydrocartisone was found to be independent of
the waterlofi ratio studied (0.2-0.5) and independent of the oil
chain length (n-alkanes Cg, (214, Cf5, C16). The pharmaceutical
formulation speciEed in this study consisted of 10 ml n-alkane (oil) ,
4 ml isopmganol, 4 g Arlacel 186, 2 g Brij 35, and 1 to 5 ml water,
A more recent study by this research group (23) indicated that
the combination of the anionic surfactant docusate sodium (dioctyl
sodium sulfosuccinate , which is the USP grade of the substance so-
dium 1.4-bis(2-ethylhexy1)-sulfosuccinate and is marketed as Aero-
sol-OT, most commonly abbreviated as AOT in the chemicall litera-
Surfactant Association Colloids as Vehicles

ARLACEL 2 0

w AUI
WATER

Figure 17.3 Ternary phase behavior for the water-Arlacel 20-AQT


system at 25OC. The single-phase =@on8 are denoted by L for a
clear, colorless, isotropic liquid, D for the lamellar liquid crystalline
phase, 12 for the cubic liquid crystalline phase, and F for the in-
verse hexagonal liquid c r y s t a i n e phase.

ture) and the nonionic surfactant sorbitan monolaurate results in a


mieraemulsion capable of obtaining a molar water/surfactmt ratio of
approximately 50. Both of these surfactants are suitable for topical
use. and the phase behavior for AOT has been extensively studied
(24.25).
To further evaluate this system, a comprehensive study of the
phase behavior resulting from this surfactant mix when combined
with water and oil was completed [(25,27). Figure 3 pmvides the
base diagram for the water-Arlacel 20-AOT system. The most
striking feature of this diagram is that less than 5%(wlw) of A r -
lacel 20 can be solubilized in the water-AOT lamellar structure at
equilibrium. For the water/AOT weight ratios from 4: 1 to 3:7, ad-
dition of 4% to 8% Artacel 20 resulted in separation of an isotropic
phase that could be detected microscopicalIy after approximately 1
week. The addition of larger amounts of Arlacel 20 had a more im-
360 Qsborne, W a r d , and QrNeilE

mediate destabazing effect. This base diagram proedes insight


concerning the surfactant interactions that allow the formation of an
deohol- free microemulsion, The addition of Grlacel 220 (sorbitan
laurate) destab2ized the structure of the AQT-water lamellar phase,
This disordering eEect decreases the lamellar re@on, whereas eorn-
plementa&ly increasing the size of the isotropic single-phase region.
Thus, for the water-Arlaeel 20-r90"r base diapam, it appears that
Arlacel 20 functions as a cosurfactant sintaar in mechanism to that
of medium-chain aleahols in traditional anionic soap systems, Addi-
tion of the cosurfaetant destabsizes the bilayer-stmctured lamellar
Xiquid crystalline phase, resulting in an increase fn the composi-
tional range over which the microemulsion is pwsent.
In contrast with the anionic soap systems (113, medium-chain
alcohols do not function as cosurfactants when added to the aqueous
AQT system. A s seen in Figures 4 through 6 (28), 'the size of the
lamellar I-iquid crystal phase is not reduced to the degree seen upon
addition of Arlacel 20. Although the micmemulsion re@on is large
when either hexanol or octanol is added as the third component,
the most dominant features of these diaqams are the large cubic
and inverse hexagonal liquid crystdline redons. The stabaitg of
these stiff phases shape the boundaAes of the microemulsion redon,
The water- butanol- AOT diagram shown fn Figure 4 is eharactedzed
by a continuaus solubility re@an between the bu-lanol and water
corners. This phase behavior is typjlcal far butanol systems, and
the resulting nuid bieontinuaus phase is speculated to be less l&el;y
to have the desired ability to solubgize both lipophilic and hydro-
1phiJie drugs,
The mmiman water mlub2ization obtainable for three Arlileel
20/AOT ratios in hexadecane is shown in F i p r e 7, The 73:27 Axl-
lacel 20 /AOT surfactant ratio that gave maximum water solubiBzation
for the base diagram (see Pig. 3) did not eve mafmum water solu-
bilization after addition o f more t h m 20%Ihexadecane. SimflaAy,
the 85: 15 Arltacel 201AOT surfaetant ratio had limited abnity to sslu-
bflize water, However, increasing the amount of AOT to a 60: 40
Arlacel 320/GOT surfactant ratio provided, a large isotropic redon
capable of salubilfzing more than 60%water, More importantly, this
system was able to soXubr"lXfzesignifkantly more water at higher hex-
adecane concentrations than the systems with larger Arlacel 2Of sur -
faetant ratios. Thus, addition of hexadecane to the AOT-Arlacef
20 system pmvides for a systematic charaetedzation of the alcohol-
free micrmrnuls~on, A s seen in F i ~ r e7, the surfaetant ratio has
a simificant Ilnfiuence on the water solubilization capacity of the
sy~tem. Xf high water solubnization over a range of mixed surfae-
tantloB ratios is required, then the Arlacl~;?l 201AOT ratio should be
less than 73: 27 (i,e. , the surfactant ratio that results in the maxi-
mum water solubiXity in the base diagmm ; see Fig. 3).
Surficlan l Association Colloids as Vehicles

WATER

Figure 17.4 P h a ~ ediavam for the system water-butanol-AOT at


24 ;1: P C . L is continuous solution ~ e @ o n 2), is the lamellar liquid
e q s t a l , I2 is the cubic liquid crystal, and F is the reversed hex-
agonal. liquid crystal.

The low solubility of the w a y serniaolfd AOT" fn wetter and the


subsequent formation of liquid ~ q s t a l l i n ephases at higher AQT
compositions ha%been well charaete~zedin the literature (24,25),
Initial phase behador studies of the AOT-sorbitan laurate micro-
emulsions indicated a severe practicd itfmitatfoxt of the system. Long
mking times were required for the w s y semfsoXid surfactant AOT
to mix with the viscous liquid Arlacsl 20, In attempts to avoid tho
slow mixing of the surfactants, the c~mmerbialXy.avagable liquid sur-
factant AOT-"I was used instead of AQT. AOT-75 is a mlutian
containing 75% AOT, 20"$ater, and 5% ethancil. It i s a liquid be-
cause the addition of 5 wt%ethanol to the 75:263 AQTlwater ratio
asborne, Ward, and fl%~eiEi

WATER AQT

Figure 17.5 Phase diagram for the system water-hexanol-AOI" at


24 -i. 2@6. L2 is the inverse mieclllar (micmemulsion) reeon. Other
phases are labeled the same as in Pig. 4.

prevents the formation of the cubic tigrnid eqstall.line phase that


exists from 75% to 80% AOT in water. Thus, A0T-75 remains a liq-
uid (viscosity approrsirnately 200 cp) , provided ethanol. is not lost
through evaporation. Unlike the medium-chain alcohols, ethanol
shows very low irritation and taxicity, Thus, in these small
amounts, ethanol i s completely acceptable as a pharmaceutical ingre-
dient. By determining the rheol.o@cd p m p e ~ i e sof the miscible
AOT- 75-sorbitan Iaurate mkture , and characterizing the phase be-
hador of this surfactant mixture when combined with hexadecane
and water, optimal compositions for a pharmaceutical low-alcohol mi-
croernulsian were determined,
As seen in Fiwre 8, addition of Arfa~eX20 ( ~ s c a s i t yapproxi-
mately 5100 ep) does not dramatically increase the viscosity of AOT-
Surfactant Association Colloids as Vehicles

Figure 1'7.6 Phase behador for the water-octanol-AOT system at


24 2 g°C. Phases are labeled as in Fig. 4. Cornpositlions circled
are tho813 described in Table 4.

75 until less than 40% AOT-75 is present (i.e, , greater than 60%
Arlacel 20). Thus, the mklng problems characteristic o f the w a x y
200% AQT are not encountered with A0T-75, Fudfiermare, the rhe-
slam studies indic&e that the most readily combined binary surfac-
tant mixtures will contain p e a t e r than 411% AOT-75. Because micro-
emulsions are thermsdynamica.lIy stable and their formation is revers-
ible with temperature, the mixing of 100% AQT could be facilitated by
heating.. However, studies indicating hydrolysis of AQT upon long-
term storage ( 29) diseaurages this approach,
The extent of the mieraemulsion re@on between 100%AOT- 75
and 50: 411 AQT-751Artacel 22 is shown in Elwre 9. This one-phase
re@on grows in area smoothly. A11 surfactant ratios except the
Qsborne, W a r d , and QtNeill

WATER SURFACTANT MIX


Figure, 17.7 Pseudo--.temav-phase b e h a ~ o rflop the water-hexadee-
me-Arlaeel 20-AOT system, in which the ArIaeeX 2 Q l A Q T weight
ratio is 60:4Q ( 1; 1 2 . 5 : 2 7 . 5 (---); and 85: 15 (---I, The
single-phase clear, isotropic microemulsion side of the diagram is la-
beled by L ,

100UAa-75 are completely miscible with hexaclecane , and tho,


amount: sf water ~alublein the surfactant-krexadecilne solutions
p w s as the amount of Arlacel 20 increases. Figures 10 and 11
skaw the transition reeon near the mmirnurn water solubility micro-
emulsi~n. The mmimum water solubility appears to occur between
55: 45 and 60: 40 AOT-75lArlacel 220 ratios and is rtpproximately the
same for hexadeeane /surfactant mixture ratios between 20: 80 and
50:50, A t 54% AOT-75, a two-phase redon appears inside the one-
phase micmemulsion re@on. The samples in this two-phase re@sn,
and nearby in the one-phase re@on, were temperature-sensitive and
were examined in a temperature-eantroll8d air chamber. As the
fraction af AOT-75 decreases further, the miememulsion re@on also
decreases in size (Fig. 32).
Surfactant Association Colloids as Vehicles

0 Spindle Q 1

A Spindle $2

0 Spindle a3

Figure 17.8 Viscosity of AOT-75-Arlacel 220 mixtures at the con-


stant apparent shear rate of 64.6 sec-l.

Comparison of Figures 9 through 12 with Figure 7 shows that


replacing the waxy semisolid AQT with the liquid 80%-'75 does not
alter the trends in the phase behavior. The small amount of ethanol
present in the microemulsion appears to have minimal effect on the
microemulsion. Although the narrow extensions found in Figure 9
can ineorporate approximately 55%water, the broad-roundkg bound-
ary characteristic of 57%to 70%AOT-75 is much more desirable be-
cause maximum water content ean be maint&ed over a wider range
of added hydraearbon. FOP AOT-75 concentrations above the 70%:
30%AOT-75/ArlaceI 20 mixture, the amount of water solubilized de-
creases, once again indicating that It is the blend of these two sur-
factants that produce the microemulsion region capable of the great-
est water incorporation.
The unique phase behavior seen for both Figures 10 and 11 de-
serves further comment. It is not surprising that a mixed surfac-
tant system would be characterized by an ideal surfactant ratio for
incorporating water into the miememulsion. A s shown in Figure 11,
the two-phase region found within the single-phase microemulsion
suggests that critical points may be present in this region of the
Qsborne, W a r d , and OrNeiTE

HEXADEGANE

Figure 17.8 Pseudoternaq phase b e h a ~ o rfor the water-hexadee-


ane-AOT-75-Arlacel 20 system at 5 2°C. The ratios of AOT-75
(75%sodium- l,4-bis~2-ethylhexy1~sulfosucci~1ate, 20% water, 5% etha-
nol) to Arlacel 20 (sorbitan laurate) are as follows: 100%
AQT- 1 5 ; 90: 10 AOT- "IEilArXaeel 20; 80: 20 AOT-751
Arlaeel 220; - - - 70: 30 AQT AQT-75lArlacel 20; - 60:40 AOrS-
651ArIaceI 20,

quaternary plot ., f t is also the 54: 46 ratio of AOT-?5/ArlaceX 20


that provides the maximum incorporation of willer. However, the
extreme temperature sensitivity of the phase b e h a ~ o rat this sur-
factant ratio m&es the 50% to 60% A0T-75 in Arlacel 220 systems of
limited practicd use. Because single-phase systems are desired frtr
stabre gharmaceuticd formulations, slightly larger amounts of AOT-
15 CCiQ%-70%]should be mixed with Arlaeelt 28 to formulate a phar-
maceutical micrmmulsion capable of incorporating Xarge amounts of
water, B y thoroughly characterrizing the phase behaGor aE this
multiple-componenI system, further efforts toward the formulation
of 13; low-alcohol pharmaceutical mieroemulsian is possible, Xt skxoulld
be noted that such thorough charactedzatian of a bur-component:
Surfactant Association Colloids as Vehicles

Figure 17.10 Pseudotemav phase diagram for the water-hexadec-


ane-AQT-75--Arlaeel 26 system at 2 5 O Lt O,Z°C. Dashed line is for
57: 43 AQT-'ISIAraeel 20 ratio, whereas the solid line is for the 54:
46 AOT-151ArlaettX 20 ratio, Note the two-phase re@on within. the
single -phase redan far the 54: $6 A Q T - 751Arlaeel 20 system,

system, esgeclaXly one of practical concern, is relatively rare in the


open literature ( 2,30,31). A l s o , note that A 0 T does not require
the presence of a cosurfaetant to form a micmemufsion. The micro-
emulsion redon resulting from water-hydmcarbon-AOT systems
has been extearsive2-y studied by scattelel"ng techniques (32,331 .
These systems do not tend to incorporate as much water as when
Arlaeel 20 is used as a cosurfactmt, Furthermore, the highest
water incorporation for these t hree-component micraenulsions tends
to be relrrtivefy narrow extensjioions toward the water comer. Thus,
the three-component microemulsions , &though prabblibly well suited
for some applications , are not expected to be a s widely useful a s
the ArXaeel 20- t40T micrwmulsions,
Another surfactant assmiation eolloidaZ system that has received
active investigation as a pharmaceutical vehicle consisjts of lecithin -
Osborne, Ward, and O'Neill

HEXADECANE

Figure 17.1 1 Pseudoternary phase diagram for the water-hexadec-


ane-AOT-75-Arlacel 20 system at 24 2 2% for the 50:50 AOT-751
Arlacel 20 mixture.

bile salt mixtures. These systems have been studied for use as in-
jectable~ (34,35) and togicals (36). Additional studies on the phase
behavior of these systems have been completed in efforts to better
understand dissolution-solub1lization of cholesterol @llstones (37,
38). Evaluation of this system for topical drug delivery included
comparison of totally hydrogenated lecithin with natural lecithin that
had been "purifiedt' but not hydrogenated, and replacement of wa-
ter in part by other polar solvents (39).
The results of this study are shown in Figure 13. The striking
feature is the difference bet ween the totally hydmgenated lecithin
and the unsaturated purified lecithin. The distinct decrease in the
ability of the sodium cholate micelles to incorporate the totally hy-
drogenated lecithin can be attributed to the stiffness of the leci-
thin's saturated chains. This stiffness would be expected to raise
the transition temperature of the bilayer, thus pmventing the for-
Surfaetcmt Association Colloids as Vehicles

WEXABECANE

WATER AQT-75IARLAGEL 20
Figure 17.12 Pseudoternary phase diagram for the water-hexadec-
ane-AOT-75-ArlaceX 2Q system at 24 2 2%. The AOT-"151ArlaceI
20 ratios are as follows: 40: 60, AOT-751Arlacel 20;
30: 70 AOT-751Arlacel 20; 0: 80 AOT-751ArXacel 20;
90 AQT-751Arlact31 20; and ------ - 100% Arlacel 20,

matlion, of the La phase at 30°C. This is confirmed by the phase


behador of the system, because the excess phase for the totally
hydrogenated lecithin is crystalline rather than Equid crystaIline,
The micellar phase re@on for the purified lecithin was sirnsar
to that found in the literature ( 3 1 ) . The transition from sphedeal
to obfate, ellipmidal structure (38) was noted in the pudfied Xeci-
thin system for lecithin /sodium cholate ratios greater than unity.
This was inferred by obseming the extremely slow solubilization of
the Xedthin in this transition re@on.
The use of a I: 1 mixture of tvaterlpropylene glycol as the sol-
vent &lawed slightly greater solublifizettfon of the Xecithln. This la
sead2y attributed to the bilayer destabnizing nature of prop5l;fene
glycol (40) * This s m e effect accounts for the rapid solubilization
of lecithin into the mdium ckxolate miceIles far the propylene glycol-
Qsborne, Ward, and O'NeiZE

Figure 17.13 Pa&id terntaw diagram for the system water-sodlixm


cholate-totally hydrogenerated lecithin. (-- - - -. -.-), and for
three systems using purified lecithin--13ocfiumcholate and either water
(------- ) , waterlprapylene glycol I: 1 ( b),or waterlbutanediof
1:1 ( ) The rnkellar re@on was determined at 30°C.

water system, compared with having water as the solvent. For the
water system, a lamdlar Xquid cqstalline surface film immediately
formed around the added ledthin, severely h-indedng solubilfzation,
The bilayer destabsidrxg action of the pmpylene glycol prevented.
this film formation resulting in rapid disfsolutisn of the lecithin in
the sodium eholate- water mieellar system. Replacement of half the
water with butanedial had an effect simnar to the use of propylene
glycolI

Be Topical Liquid Crystals


tyotropic liquid crystdline vehicles have genera3ly the same advan-
tages that were Usted for microemulsfons, These mesaphases are
thermodynamied stable and have unique, often superhr , solubBi-
zat-ion pmperties. Unl&e micrcrenrulsions, some of the liquid crys-
tafBne systems may be sllllghtly cloudy, rather than clear. In addi-
tion to the advantages listed for micmernulsions, %quid crystals may
Surfactan t Association Colloids as Vehicles 3 71

release drugs over a sustained period either because of slower dif-


fusion from the liquid crystilt3 or because of an increased solub2ized
residence time for the drug after topical application*
Most work published, to date, describing delivery from liquid
c r y s t d s focuses on cubic -structured phases. This liquid c q s t a l
has been deaeribed (41) as being either erodible or nonerodible de-
pending upon its phase b e h a ~ o rin exceas water. For the golyoxy-
ethylene surfactant-water system (42) two liquid elpystdline systems
were d e s c ~ b e d , The cubic phase eontahing low anaunts of surfac-
tant was formed from closely packed micelles and, thus, readny
"emdedf9nto a micellar system when added to water, The cubic
phase that formed at high surfactant content passes through a vis-
cous hexagonal phase upon addition of water, then farms the closely
packed micellar cubic liquid c ~ s t a upon
l further addition of water..
Thus, erosbn will be much slower for the cubic phase containing a
@eater itmount of surfactant. Although this concept of erosion is
more pertinent to parental or oral liquid crystalline delivery, the
use of the complete phase diapam to understand the delivery of a
drug is equdly important for topieds .
The cubic liquid crystalfine systems studied above (41) included
60%rnonwlein- 4lt"tiater, 48%monoolein- 12% lecithin-40% water, and
60% rnonooleh-48 to 10%glycero1-.3fi% to 30%water, (All percent-
ages were based upon weight.) Methylene blue was used as the
ma~kerto establish sustained release from each of these vehicles,
Other pharmaceuticd applications include the use of a eubic liquid
crystlrils to act as a parenteral depot for peptides that protects
against biodegradation (43) and to depress the rate of drmg hydrol-
ysis even in the presence of a gastrointestinal-Xke buB phase (44).
Lamellar liquid crystals have also been evaluated for possible
use as topical drug dellivev systems. Work by Wahlgren et al.
(45) showed that lecithin-water systems could dissolve up to 5 wt%
hydrocortisone, and that the diffusion rate of the d w g within this
liquid crystalline vehicle was four orders of magnitude higher than
the diffusion rate of hydrocortisone across the skin. Additional
lamellar liquid crystalline systems were characterized by El-NoktlZy
et al. (46,47) in which the water was replaced by either alkanediols
a r oligomers of polyethylene glycofs. Although some of these sol-
vents (e * g. , ethylene glycol) are not suitable for topied use, many
of the compounds are, and these systems should be considered for
evaluations as topied drug deliyew formulations. A n additional
nonaqueous lamellar liquid clrystalline system that was evaluated for
topical use combined triethanolarnine vv3h oleic acid and other polar
and nonpolar solvents ($8). This study eharactedzed the interac-
tion of various components of human sebum with the liquid crystal-
line vehicle.
3 72 Qsbarne, W a r d , and Q%eitl

IV, DRUG BEtfVERY FROM TOPICAL SUR-


FACTANT ASSOCIATION COLLOIDS

Although relatively few microernuf sion and liquid crystalline systems


that are suitable for topical use have been descrjbed in the litera-
ture, even fewer systems have been evaluated by percutaneous
transport techniques. Xn this section, the systems that have been
studied will be described,

One of the first, and certainly most interesting studies of topical


drug delivery from a microemulsion, compared with a gel and cream
of ~ImBarcompo~ition,was reported by Zliegenmeyer and Fuhrer in
1980 (49). Xn this study, the three farmulations evaluated contained
the same amount of dodeeane and water. The amount of decanol was
varied, and polyoxyethylene 7-lauryf ether was the surfactant used.
In ~ t r experiments
o using skin membranes and 1%tetracycline XXGI
formulations showed that much less active agent penetrates through
the skin from the cream than from the gel and that both of these
formu1ati;ons are definitely exceeded in effect by the microemulsion.
In viva studies using fluorescent tetracycline confirmed that the mi-
croemulsion was supeAor to the other systems in abgity to promote
penetration.
A more eonprehensivc? study of tapicd drug delivery h r n a mi-
eraemulsion was recently undert&en by oar research groups (50-
52). This study focused on the microernulsion and liquid e q s t d l i n e
regions that result from the water-aetanol-tlOT system, As seen
in Fimre 6 , the phase behavior of this system is truly unique, es-
pecially considering the large extension sf the cubic liquid crystd-
line (i.e, , ringing gel) phase. The close proximity of the liquid
c ~ s t d l i n eregions to the microemulsion reeon &lows evaluation of
the effect of vehide structure upon percutaneous transport of a
drug. Far this unique system, this evaluation can be completed for
vehicles that v a q in component campos;ilion by only a few weight
percent. Also noteworthy is the abifity of the 58 :42 AOT loctanol
weight ratlis to incorporate @eater than 70%water into the micro-
emulsion re@on, Thus, microernufsions with the same AOT loctanal
ratio, but with widely ranging ( 15- 70 w t u amounts of water were
evaluated for drug delivery..
The initid study (50) limited itself to examining the flux of
tdtiurn-labeled water from four microemulsion vehicles across human
cadaver skin. Each of the vehieles has the set ra"to of 58: 42 Aalr /
octanol and rmged in water content as shown in Table 2. In an
attempt to separate vehicle effects upon water transport from the
Surfactant Association Colloids as Vehicles 373

Table 17.2 Water Self-Diffusion and Normalized In Vitro Transder-


mal Flux Values for the Microemulsion Formed by Addition of Water
to a 58: 42 Weight Ratio of AOTIOctanol

% Water X Water D/Dw Normlized flux (ave)

15 0.68 0.035 0.89, 0.78 (0.84)

skin barrier effects associated with the stratum corneum, the diffu-
sion of water within the vehicle was measured using a pulsed field
gradient fourier transform NMR technique (53). The unitless value
D/Dw is the ratio of the self-diffusion coefficient for water in the
microemulsion vehicle divided by the self-diffusian coefficient for
water in water. Therefore, a DlDw value of 0.100 means than, on
the average, water in the 45%water micmemulsian environment dif-
fuses at one-tenth the rate that a water molewle would diffuse in a
totally aqueous environment. The initially low values for DlDw for
microemulsions of low water content, and subsequent increase with
addition of water, can be attributed to binding of the water mole-
cules to the surfactmt headgroup. Thus, for the 15%water micro-
emulsion, most of the water in the micrmmulsion is bound to the
surfactant headgroup and is not available for transport across the
skin. The result is that transport of water across the skin from
this particular microemulsion is less than the transport of water
from neat water, that is, normalized flux is less than unity. Pre-
treatment studies using octanol, AOT , and AOT loetano1 (58: 42 ra-
tio; Table 3) further indicated that the enhancement in water pene-
tration from the high water content microemulsions was a result of
a synergistic effect between AOT and octanol and was not due to
the vehicle's microemulsion strmcture.
Additional studies that evaluated the in vitro transdermal trans-
port of labeled glucose from the microemulsions across cadaver skin
(511 showed that the glucose essentially crossed with the water.
Thus, microemulsion systems that showed high water transport, also
demonstrated high glucos@transport when an infinitely small amount
of labeled glucose was spiked into the microemulsion.
3 74 Osborne, Ward, and OtNeill

Table 17.3 NormaIized In Vitro Transdermal Flux Values for Neat


Water Following Pretreatment

Substance Duration (hr) Normalized flux

Octanol
AOT (from EtOH)
AOT foetan01 58: 42

8, Liquid Crystals
The close proximity of the liquid crystalline regions to the micro-
emulsion region allows comparison of drug transport from the struc-
turally different, but compositionally similar, surfactant association
colloids. A summation of this comparimn is given in Table 4. Two
different study designs were used to evaluate in vitro transdermd
flux of water across human skin after application of a micmmulsion
or Iiquid crystal. Study design A utilized surgical skin obtained
from radical mastectamy eases, with the skin being used within 24
to 48 h r after procurement* This skin was full thickness, stored on
wet ice, and used after removal of the subcutaneous fat (50). Re-
sulting fIux values were normalized against flux values for tritiated
water using skin from the same donor in an attempt to eliminate in-
dividual variation. For study design A, the vehicles were prepared
by dilution of tritium-labeled water with distiUed water, and then
addition of the appropriate amount of AOT dissolved in oetanol. A
single-label liquid scintmation-counting study was used to assay the
receiver phase of the Bow through transdermsl cell. In contrast,
study design B used dermatomed cadaver skin that was shipped on
dry ice, and that remained frozen until the time of use. Five rep-
licates of each formulation were tested using skin from a single in-
a ~ d u s l , Although normalization of flux values was not required to
minimize variation, the necessary transport data was determined to
allow direct comparison with the results from study design A . Sam-
ples were prepared with tritiattld water as described in design A
with the exception that a 4-pCi spike of glucose D - E ~ ~ C ( U(5.7
)I
mCi/mmol) was dissolved in the water before addition of the AQT-
octanol solution. For study design B, a dual-label liquid scintilla-
tion-counting study was used to assay the receiver phase of the
flow through transdermal cell.
Considering the differences in experimental design bet ween
these two studies, the normalizd flux of water across the skin after
Surfactant Association Cotloids as Vehicles 375

Table 17.4 Water Transport from Structurally Different Surfactant


Association Colloids

Normalized flux values

Colloid type Design A Design B

15%water microemulsion 0. 8 0.4


35% water micmemulsion 2.5 2.0
67% water microemulsion 4.5 5.0
Lamellar liquid crystalline 2.4 1.1
Inverse hexagonal liquid crystdline 0.7 0.3
Cubic liquid crystalline 2.5 1.3

delivery from microemulsions or Iiquid crystals is remarkably consis-


tent. Note that delivery from the microemulsions follows the same
trends for both desims and can be described in terms of water mo-
bility, as discussed earlier. The high water content liquid crystd-
line phases give approximately equivdent water transport that is
approximately half the vaIue characteristic of the highest water con-
tent micmemulsions. This decrease in transport from the liquid
crystalline phases is attributed to the involvement of octanol and,
although probably to a smaller extent, AOT in the formation of the
Iiquid crystalline structure. The cubic liquid crystalline phase 2-
lustrates this point. Without a doubt, the stiff amphiphllic films
that are necessary to produce the cubic liquid crystal results in the
dramatic decrease in self diffusion of actanol and BOT. The diffu-
sion of octanol within the vehicle likely becomes slower than the dif-
fusion of octanol aeross the skin. Thus, a component of the vehicle
that was required to produce enhancement is no longer limited in
its effect by slow diffusion across the stratum corneum-dermis bar-
rier, rather the enhancer is limited by slow diffusion out of the cu-
bic Iiquid c~ystallinevehicle. Note that the self-diffusion cseffieient
of water within the cubic phase is equivalent to the self-diffusbn
coefficient of water within the high water content microemulsions,
whereas the transport of water across the skin is approximately half
for the cubic phase compared with the microemulsion.
The low water transport value for the lameUar liquid crystal is
not readily explained in terms of decreased diffusivity of octanol,
The rrmphiphilic film for the water-oetanol-AOT bilayer region is
p r o b a l y more fluid than for the cubic liquid crystal. For the la-
376 Osborne, Ward, and Q t N e i l l

mellar phase the diminished water transport may be attdbuted to


the AOTlactanol ratio being less than the synergistic optimum ratio.
Thus, the 58 :42 AOT loetano1 ratio proTJides greater enhancement
than the 83: 17 AQT lactanol ratio necessary to form the lamellar liq-
uid crystal.
Explmation of the very low water transport from the inverse
hexagon& liguid cmstsi fallows directly from the preceding discus-
sion of the 15%water microemulsion. Par these systems, the water
is bound to the headpoups of the surfactant and, thus, is less
available far transport. The glucose also will presumably reside in
this polar re@on of the vehicle, and will simnarly be limited in its
transport.
Sokoloski et al, (54) have also examined the skin transport of
drugs of different polarities ddivered from a water-gropylene gly-
col- AOT lamellar liguid cry-stal, For this liquid cvstdXine vehicle,
hahless mouse skin permeabuities of minoxidil and p-nitrophenol
were determined and compared against delivew of the same t w o mod-
el drugs from a PC-water mllxture. The use of the liquid eqstrzlt-
line vehicle increased the transport of minoxidil, but decreased the
transport of g-nitrophenol. These results emphasize the need to
understand, in detaif , how each component of the formulation inter-
acts with the skin, and how the drug intertncts with each component
of the formulation.

V. COhtCLLlDtNG REMARKS

Drug delivery from surfactant assmiation, colloids appears to be


unique when compared with delivery from traditional formulations,
The mobaity of the drug within the vehicle, and the mobBity of ve-
hicle components that enhance percutanwas transport must be con-
sidered. Highly stmetured vehicles, such as inversed hexagon&
and cubic liguid c r y s t a s , may exhibit sustahed-release properties,
either by binding the water t i . @ , , internal phase) or by stiffening
.
the amphiphitie film (i e ,, the penetration enhancer components)
.
within the formulatkn Alternatively, micraemulsions , if properly
formulatc;d, can prodde optimum d m g delivery, For microemulsion
systems the phase b e h a ~ o rshould be adequately characterized to
assure that sufficient solvent is a v a a b l e to solvate the amphiphges.
In summary, the studies that have evaluated topical drag de-
livery by mieroemulsians or liquid crystdline formulations indkate
that surfactant associtilt-ion colloids are often superior vehicles for
topical d m g delivery. MicroemuXsions can be formulated so that
theis molecularly dmamic nature is maimized , thus allowing maari-
mized delivery of both drug m d penetration enhancer. Liquid
Surfactan t Association Colloids as Vehicles 377

crystals can be formulated with either ffuid or bound-water cores,


and with surfactant films of varying stiffness, Because of their
structure, liquid e q s t d s provide a rmge of release propertks for
both the ts~olventn portion of the structure (i.e,, the drug) and the
penetration enhancer. As tradition& enhancers are themselves ab-
sorbed into the sy stemic cirezllati~n, substained release from the ve -
hicle could grodde percutaneous penetration enhancement for ex-
tended periods. These d m g deliver37 results when combined with
increased drug solubnization and thermodynamic stability strongly
suggest that association cdloid s w i l l become irzcreasingly important
as specidized topical drug delivery formulations,

REFERENCES

A , T , Florence, Drugs Pharm, Sci, , f 2; 15, 1981.


S. E. Frfberg and R, L. Yenable, in Encyclopedia of Emulsllon
Technology, Vol, 1. ( P . Becher, ed.) Marcel Dekker, New
York, pp. 287-356, 1985.
K , Shinoda and S , Priberg , Emulsions and Sotubi!izatton,
Chap, 1. John WBey lk Sons, New Vork, 1986,
. .
J, N fsraelaclhfii, D. J, Mitehell, and B W . Ninham, J.
Chem, Soc, Fawday Trrms, 11, 72: 1525, 1996.
J , N. Israelaehdi, Proc. Xnt. Sch. Phys. tr'Enrico Ferniff,
( P h y s , Amphiptzii'as) , 90: 24, 1985,
B , W Ninham, and D, J , Mitchell, d , Cltem, Soc. Faraday
I

Trans. PI, 17:601, 1981,


C. Tanford, f. Phys, Chem,, 76:3020, 1972.
D. I). Mlaler, J.. R. Bellare, a. P", Evans, Y, Talman, and
B. Mf. Ninharn, J . Phys. Cktem., 91:674, 1987,
d. 3%. Bellare, T . Kanedo, and LZ. F, Evans, Langmuir, 4:
1066, 1988,
A. W . Fulmer, and 6 . J . Kramer, J , i n v e s t , Dermatol, 863598,
1986.
P , Ekwall, in Advances in Liquid Crystals, 1701. 1. ( E . H.
Brawn, e d . ) . Academic Press, New Vork, 1975.
.
P , Reiss-Efusson , and V Luzzati, J. Bhys. Chem. , 68: 3504,
1964.
PJ, A. Mezer, (I;. B, Benedek, and M. C I Carey, J , Phys,
Ghem., 80: 1075, 1916.
E . Toultou, and A . H e Goldberg, European Patent Application
84109963.1, June 22, 1984.
S . E l Friberg, and T, f), Flairn, in Inorganic Reactions in Qr-
ganl'zed Media, ( S . L. HoIt , ed.). 1468 Symposium Series
177, Washington, B ,C ,, 1982,
Qsborne, W a r d , and OfNez"ll

P , EkwtaXX, L. Mandell, and K . Fonteft, Mol. C r y s t . Liquid


Cryst., 8: 157, 1969,
K . M. Nordstrorn, J. N . Labows, N;. J. McGinfey, and J. J.
LeycZen, J , Invest, Dermatol, , 86: 706, 1986,
S. E. Friberg, and D. W. Bsborne, 3 , Dispersion Sci. Tech.,
6:485, 1987.
F. Comeffes, C . Solans, N . Azemae, S . L e d , and J. L, Parma,
I, Dispersion Sei. Technol, , 1 :369, 198s.
D , 0 . Shah, ed., Surface Phenomena in Enhanced Oil Reeov-
ery I Plenum Press, New York , 1981.
B, Attwmd, and A , T , Florence, Surfactant Systems: Their
Chemistry, Pharmacy and Biology. Chapman and Hall, New
York, 1983,
H. N, Bhargava, A , Narurkar, and L . M , Lieb, Pharm. Tech-
nol., 1 1 ( 3 ) : 4 6 , 1987,
A . Jayakrishnan, K . Kalaiaiizrasi, and 23. 0 . Shah, J , Soe,
Gasmet. Chem, 3 4 : 3 3 5 , 1983.
fC A,. Johnson, and D. 0 . Shah, b e Colloid Interface Sei, ,
I

101:269, 1885,
d. Rodgers, and P. A , Winsor, J. Colloid Interface Sei, , 30;
247, 1969..
. .
E 1 , Frances, and T S , Hart, J . Colloid Interface Sci, , 94:
2, 1983.
B. W . Osborne, C A . Middletan, and R . L. Rogers, J . B ~ s -
I

persion Sci. Teeh. 9: 415, 1988.


D. W . Bsborne, R . 3 , Chipman, and C, V. Peschek, Xnl. J.
Pharm. Tech, Prod. Manuf, (in press).
D. W, Osborne, A . J. I . Ward, and K , J . OfNeill, unpublished
resulh. Doeusate mdiurn USP was obtained from Amedean Cy-
marnide ; butanol, hexanol , and octanol were the highest purity
available f ~ o mAldrieh, These matedals were used as received,
Samples were mixed and equil.ibrated flor at teast 2 weeks be-
fore phase b e h a ~ o rcharacterization using polarized light mi-
croscopy *
P. Delord, and F. C . Larche, 3 . Colloid Interfirtee Sci., 98:
277, 1984,
S. M. Ng, and S, G . Frank, J. Dispersion Sci. Tech., 3:271,
1982.
R . 6. Baker, A. T , Florence, T. F , Tardos, and R . M. Wood,
J . Colloid Xnterface Sci, , 100: 311, 1984,
B . W . Robinson, 6. Toprakeioglu, J. G . Dore, and I?. Cheiux,
I, Chem* See, , Faraday Trans. 1, 80: 13, 1984.
NI, Kol-tzarchyk, S. H, Chen, J, S, Nuang, and M. W . K i m ,
Bhys, Rev. A , 29: 2054, 1984.
Xlaffmann-LaXtoche i f G , European Patent 133- 258-A, February-
20, 1985,
Su~lfa~tcln
t Association Colloids as Vehicles 379

American Lecithin, U. S , Patent 4,252,193, February 24, 1981,


I. Lyon, and N. Lyon, U .S, Patent 4,115,3113, September 19,
1978,
G I Lindblom, P. 0 . Er&son, and 6 , A r d d ~ o n ,Hepatolorn 4
(suppl. 5) : 1295, 1984.
K , Muller , Wepatology , I(suppf. 5) ;1345, 1984,
B e liV , Osborne, Unpublished results. Lecithin was obtained
from American Lecithin Company, Atlanta, Ga. , sodium cholate
was obtained fmm Sigma Clremied Co. , St, Louis, M a .
A . J. I. Ward, C . Marie, L. A . S y l ~ a and
, M. A . Phillfpi, J .
Dispersion Sei, Tecihnol. , 9: 149, 1988.
S. EngstrGm, K . Larsson, and B , Lindman, International Pat-
ent Application WCl 84102076, June 7, 11984.
D. J , Mitchell, G , J, T . Tiddy, E , Waring, 4". Bostoek, and
M. P. McDonald, J , Chen, Soe, Famday Tvsans, I, 78: 975,
1983.
.
B Eriksson, 5 . Learder, and M. Ohlin, 15th International
Symposium on Controlled Release of Bioioaetive Materials, paper
219, B asel, Switzerland, Auwst 1988,
G . Rehmberg, A . Andreassan, and J , E , Lafroth, 25th. Inter-
national Symposium on Controlled Release of Efioactive Materi-
als, paper 218, Basel, Switzerland, A u p s t 1988.
S, Wahlgren, A . L. Lindstrom, and S, E.. Friberg, b. Pharm.
Sei. 73: 2484, 1984.
M. A . El-PJokaly, L. D , Ford, S. E. Friberg, and D. W.
Larsen, 9, Colloid Interface Sci, , 84: 228, 1981.
M e El-Nokdy, S. E. Friberg, and D. W. Larsen, in Liquid
Crystals a n d Ordered Fluids, Vol. 4 , ( A . C . Griffin, and J.
F * Johnson, eds,). Plenum, New York, pp. ""11-450, 1984.
S = E. Friberg, 6 . S, Wohn, and F , E . Eockwasd, 6 . Pharm,
Sei,, 74:771, 11985.
S. Ziegenmeyer, and C. Fuhrer, AcEa Piharm. Techno!, , 26:
273, 1980. (Translated from German)
L). W. Dsborne, A . d . I , Ward, and K . S . Ofl\lelll, Drug. Dev.
I n d . Pharm. , 14: 1203, 11988.
f). W e Osbome, A , J , I , Ward, and K . J, O%eilX (in prepara-
tion),
B. VIT. Qsborne, A . 2 . 1 . Ward, and K . J, O%eill (in prepara-
tion).
B. Lindman, and P . Stabs, in Microemulsion Structure and
Dynamics. CRG Press, Baea Raton, pp, 119- 152, 1987.
T. B. Sokoloski, D. W. Osborne, R . G . StehXe, IC. J.
Stefanski, N . F. IX. Ho, and P), Gleasan Paper PP 826, AAPS
National Meeting, October 30, 1988, Orlando, Fla,
Gel Dosage Forms: Theory,
Formulation, and Processing

LORRAINE E, PENA The Upjohn Campany, Kalamazao, Michigan

1. THEORY

Gels are trmsparent to opaque semisolids containhg a high ratio of


solvent to gelling agent. When dispersed in an appropriate solvent,
gelling agents merge or entangle to form a three-dimensional colloidal
network structure. This network limits fluiff flow by entrapment
and immobBization of the solvent moleculesl The network structure
is also responsible for a gel" resistance to Piefarmation and, there-
fore, its viscoelastic properties.
A variety of structures are associated with gel networks. Fig-
ure 1 illustrates some of the most common ones. Random coils are
the least ordered and occur most frequently with synthetic polymers
sueh as resins and cellulose derivatives. The h e l k i s a more or-
dered structure farmed from the intertwining of two polymer cha-ins,
Xmtkan gum and starch are typical, examples. Stacks, or the egg-
box model, as it is sometimes clscflect, results from cross-linking of
polymer chains by divdent cations. Cdcium alginate is a classic
example. The house of cards structure is characteristic of gel-forming
colloidal particles sueh as bentonite and Veegum. In the case of
Veemrn, thestructure results from the agpment of the positively
charge edges with the negatively charged fht surfaces; af the clay
pwtides (1). Because most gels that are used in the pharmaceutiestl
industry are associated with the rmdom coil network, further dis-
cussion will be centered around that structrxre,
Random coil gelation meehanisns are rooted in the polymer-
polymer m d polymer-- solvent; interwtians , With a @ven polymer,
the gel network forms through successive increases in c~ncentration.
RANDOM COIL

STACKS HOUSE OF CARDS


Figure 18.1 Gel structures,

This results in a reduction of the interparticle distances, which


subsequently leads to chain entanglement and the development of
cross-links, As the number of cross-links increases, the chains
lock, solvent mohalty is reduced, and a gel farms. Continued pol-
ymer addition strengthens the gel network m d results in increased
resiliency and viscoelasticfty .
Although the gel network is basically formed through polymeric
interactions, the nature of the polymer- solvent affinity actually de-
termines the integrity of the gel, Clctssicd gel theory distinguishes
between three categories of solvents: (1) free solvent that is v e v
mobge; ( 2 ) solvent bound as a solvation layer, usualXy through hy-
drogen banding; md (3) solvent entrapped within the network
structure. Tha ratios of the three solvent types in, a given. gel are
dependent an the polymer concentration and the solvent affinity for
the polymer. Solvent affinity governs extension of the random coilil,
The greater the solvent affinity, the more the coil expands and en-
tangles with adjacent coils to form crass-links, In a good solvent,
the poly rner chains are interpenetrated by solvent molecules, and
the solvation layer is enhmced, This facilitates random eail expan-
sion and network formation. In a poor solvent, the polymer chains
cantract to minimize solvent contact, thereby reducing the ef&ctive
number of cross-links and weakening the gel network structure.

I I. FORMULATION

Gelation theory can be readgy applied when ftormulating gel products.


However, before discussing this topic, presentation af some desira-
Gel Dosage Forms 383

ble gel characteristics is in order. For optimum consumer appeal*


the gel should have good optical clarity and sparkle. A high vis-
cosity and a high-yield v&ue are essential, but how high is p ~ m a '
rily a matter of intended product application. To preserve product
integrity, the gel should maintain its sscasity at all temperatures
that may be encountered during shipment and storage. During for-
mulation, there is frequently a trade-off between these optimum
eharaeteristics and the chemicd requirements of active ingredients.
Carbomer is a commonly used gelling agent that produces gels
having a number of these desirable characteristics. The grade,
carbomer 934 I) N F , is most commonly used in the pharmaceutical in-
dustry and has been selected as the exemplanry. polymer for formula-
tion discussion. Chemically, carbomer 934 P is a crass-linked acryl-
ic acid polymer having a molcewlar weight of approximately 3 x 106
(2). The gelation mechanism depends on neutragzation of the car-
b x y l i c acid moiety to form ac s ~ l u b l esalt. The polymer is hydro-
phfGc and produces sparkling clear gels when neutralized. Although
carbomer tolerates large amounts of alcohol, it does so with de-
creased dscosity and clmity , Gel viscosity is strongly dependent
on pH and the presence of electrolfles. A, maximum of" approxi-
mately 3%electrolyte can be tolerated before precipitation occurs as
a rubbery mass, Carborner gels possess good thermal stability in
that gel ~ s c o s i t yand yield vdue are essentially unaffected by tem-
perature, As a topical pmducl, carbomer gels possess optimum
.
rheolo@cal propervties The inherent pseudopfastic flow permits irn -
mediate recovery of escosity when shear is terminated and the high-
yield vafue and quick break make it ideal for dispensing.
The ascosity building effects of carbomer are readay apparent
when examining the gelatian mechmism in association with the col-
loid& network structure. As Figure 2 illustrates, before neutraliza-
tion, carbomer in water exists in its un-ionized form and yields a
thin opalescent dispersion of approximately pH 3 (3). At; this p H ,
the polymer is very flexible and behaves like a random coil, Addi-
tion of sodium hydroxide or a neutralizing amine to the dispersion
shif"ts the ionic egu8ibrium in favor of the soluble salt farm, This
results in ionic repulsion of the carboxylate groups and the polymer
becomes stiff and ri@d, thereby increasing the ~ s c o s i t yof the wa-
ter. Qverneutralizatlon and excess salts reduce the dscosity of
carbomer gels or cause precipitation by the counterion effect.
Figure 3 shows the dramatic effect of pH on. the viscosity de-
velopment af car"bomer gels, As pH increases and the carboxylic
acid moieties of the polymer are neutralized, viscosity and clarity
increase. Acceptable gel clarity and viscosity occur at approximate -
1y pH 4.5 to 5.0, but optimum dseosity and clarity are at pH 7,
Overneutralization results in a decrease in viseasity that cannot be
reversed by addition of acid lo tower the pH because an electrolyte
is formed.
UNNEUTRALIZED
CARBOMER

0.
U

NEUTRALIZED
CARBOMER n
D
0 9

Figure 18.2 Carbomer g d network structures,

Figure 18.3 pH-viscosity pmfile of 0.5% (wlw) carbomer 934 P NF


in water.
Get Dosage Fams 385

Table t 8.1 Alcohol Effects on Carbomer Gels

Carbomer Alcohol Viscosity


% (w/w) 8 (wiw) PH (ema

ashear rate 7.61 see-I.


b ~ s o p m ~ m o70% .
l (vlv)
CAIcohol USP, 95% (v/v) ethanol,

Although carbomer can be used to gel formulations with a large


proportion of alcohol, the dehydration effects of the alcohol on the
polymer are still substmtial. As Table 1 indicates, at pH 5 . 5 , the
use of 50%isapropanol. in a formulation requires 0.5% more carbomer
to produce a viscosity equivalent to that of an aqueous gel. At pH
8.2, doubling the concentration of alcohol USP requires 0.35% more
carbomer to produce an equivalent gel viscosity. The viscosity
responses of the gels to alcohol may be interpmted by its action as
a nonsolvent. Because the salvent affinity is reduced, the polymer
contracts, with a consequential increase in the interparticle distance
and subsequent decrease in the number of entanglements and cross-
finks. To decrease the interparticle distances and restore the in-
tegrity of the gel network structure, a greater concentration of
polymer must be used. The reduction in solvent affinity and change
in the polymer conformation results in increased haziness of the gel
as the alcohol content increases.
The rheograms of carbomer 934 P gels in hydroalcohoIic and
aqueous formulations are presented in Figure 4 using 0.5% (w/w)
.
and 1.0% (w f w) carbomer , respctctively The characteristic concav-
ity of the rheogram toward the shear rate axis indicates that the
gels exhibit pseudoplastic flow. This pseudoplasticity results from
a colloidal network structure that aligns itself in the direction of
sheer, thereby decreasing the viscosity as the shear rate increases.
The rheograms also show that the gels exhibit substantial yield val-
ues at 666 and 866 dynes/cm2 for the hydroalcoholic and aqueous
formulations, respectively. The yield value is an indication of the
extent of formation of a three-dimensional ealloidal network strue-
ture. The variation in the yield values of the two gels is a reflec-
tion of the higher casbomer concentration in the aqueous formulation,
SHEAR STRESS (dynes/cm2f

Figure 18.4 Rheogrzlms of carbamer 934 P MF gels, ( A ) hydroal-


~olrlolicgel; ( B ) aqueous gel.

O 10 20 30 40 50 60
TEMPERATURE (%)

Figure 18.5 Viscosity-temperature plots of cslrbamer 934 P NF gels:


circle, ky drodcoholic gel ; t riam gle , aqzxeou s gel ,
Gel Dosage Forms 38 7

The viscosity-temperature relationships of the aqueous and hy-


droalcohoXic gejs are shown in Figure 5. The hydroaXcoholic gel
shows a plateau at 3T°C, whereas the aqueous gel continues to show
a general decline in viscosity. A s the graphs indicate, the magni-
tude of the viscosity decline is not substantial over the temperature
range covering refrigerated to elevated storage conditions.

II1. PROCESSING

To manufacture clear, uniform, air-free gels, certain key processing


characteristics must be provided. The nature of carbomer requires
initial high-shear mixing to form a uniform smooth dispersion, fol-
lowed by low-shear planetary mixing during the neutralization-gel-
ling process. Air entrainment during the neutralization process
can be minimized by subsurface addition of liquids in conjunction
with Iow-shear mixing. In addition, mixing under vacuum, if avail-
able, will withdraw entrapped air from the dispersion during manu-
facture and prevent further air entrainment by incident& surface
breaks. Minimization of air entrainment is necessary from the aes-
thetic standpoint and, most importantly, from the aspect of control-
ling fill weights during packaging operations.
A large variety of planetary mixers with and without vacuum
capability are available for gel manufacture. The Nauta mixer ( 4 )
has a conical mix tank and consists of a lumpbreaker, located at the
base of the tank, for high-shear mixing and a rotating auger on a
pivot arm for low-shear planetary mixing. The lumpbreaker is flush
with the side of the tank and consists of four flat blades spinning
at either 1800 rpm or 3600 rpm, The auger can be operated to spi-
ral either upward or downward, depending on whether one desires
to deaerate or incorporate materials from the surface into the batch.
Both the auger and lumpbreaker are controlled separately. A i r en-
trapment is minimized with the Nauta by mixing under vacuum and
by subsurface addition of materials through a port near the base of
the mixer.
The Agi-mixer (5) has a hemispherical mix tank. High-shear
mixing is provided by the homomixer located at the base of the
tank. Counterrotating paddles equipped with scraper blades pro-
vide the low-shear planetary mixing action. The homomixer can be
operated at only two speeds, whereas the paddles are infinitely var-
iable over the range 0 to 60 rpm. A s with the Nauta mixer, aera-
tion can be controlled by mixing under vacuum and by subsurface
addition of materials through a basal port.
ACKNOWLEDGMENTS

The assistance of 6 . M. Horton , f3, L, Lee, and 4. F. Stearns in


obtaining some of the earbomer gel data presented here is greatly
apprechtect.

REFERENCES

1. B. 6 . Carson, Cosmet. Toiletries, 92(1):81, 1977.


2. Carbopol Water Salubte Resins, Bulletin Number GG-67, B . F a
Goadrich Company, Cleveland, Ohh .
3. C . A . Diltrnar, flmg Gosmelt, I n d , , 811:44?, 1951.
4. Day M h h g Company, 4932 Beech St. , Cirrehnati, Ohio 45212-
2397.
5. Greerco Corporation , Executive Drive, P " 0 , Box 187, Hudson,
New Hsmpshhe 03651.
19
Using Silicones in Topical Products

MICHAEL S. STARCH Dow Corning Corporation, Mt'dland, Michigan

I. INTRODUCTION
A. What Are Silicones?
The term silicone refers to a class of ~yntheticpolymers that are
based on dternating silicon-oxygen, or siloxane (-Si-0-1 units.
The silicones that are most commonly used in topical pmducts are
polydiorganosiloxanes in which two organic groups are bonded to
each silicon atom. For commercial polydiorganosiloxanes , the organic
groups are neady always methyl groups, and such materials are re-
ferred to as polydimsthylsiloxanes (PDMS) . Other types of silicones
that are used in topical products can be thought of as derivatives
of PDMS in which some of the metfiyl groups have been replaced
with other organic groups,
Polymethylsiloxanes are produced in one of two forms: linear or
cyclic, Both are widely used in topical formulations and together
account for most of the volume of silicones umd in these applications,
Linear PDMS is a ooIorless, odorless oil that is available in a wide
range of viscosities, The viscosity of linear PDMS is directly related
to molecular weight, and can range from less that 1 centistoke to
over 1 million centistokes (cs), Cyclic PDMS is an odorless, color-
less, low-visoosity , volattle oil. Commercially produced cyclic PDMS
is available in a fairly narmw range of molecular weights, corre-
spanding to the cyclic species with four, five, or six dimethylsilox-
ane units in the ring. A s expected, volatility of cyclic PDMS is in-
versely related to molecular weight. Unlike linear PDMS , cyclic
PDMS is a fairly good solvent, and this is one reason for its popu-
larity among formulators.
The chemical and physical properties of silicones that contain
organic substituents other than methyl vary according to the nature
of the organic substituent. For example, substitution of some meth-
yl groups on PDMS with polyethylene oxide chains dramatically in-
creases the viscosity of the material because of the polarity of the
polyethylene oxide. The chemical properties of the organic substit-
uents are generally not affected by the siloxane backbone. Conse-
quently, polydiorganosiloxanes exhibit the chemical reactivity that
would be associated with the organic substituents, The two types
of silicones with nonmethyl substituents that are used in topical
products are discussed in Sections 1I.C and I1 .D.

B. Benefits of Silicones in Topical Products


Silicones exhibit a unique combination of pmperties that, to a large
extent, arise from the inorganic siloxane backbone. The silicon-
oxygen bonds that form this backbone are quite strong, yet they
are very flexible. This results in polymers that are particularly
stable and exhibit unusual surface activity because of the ability of
the siloxane chain to present its pendant organic groups at inter-
faces (1). In PDMS, the inorganic siloxane backbone is covered by
nonpolar methyl groups, leading to extremely low intermolecular
forces. A s a result, PDMS is a liquid over a wide range of molecu-
lar weights. In fact, PDMS will exhibit viscous flow even at molec-
ular weights above 700,000. Low intermolecular forces are also re-
sponsible for the low surface tension of PDMS, which is in the
range of 20 to 21 dynlcm. The low surface tension and relatively
high molecular weight of PDMS give rise to a number of properties
that are beneficid in topical products: excellent spreading, detacki-
fitation, defoaming, and water repellency. A PDMS that has been
substituted with nonmethyl organic groups will provide these bene-
fits to varying degrees, depending on the amount of substitution
and the nature of the nonmethyl organic groups.

11. TYPES OF SILICONES USED I N TOPICAL


PRODUCTS A N D TI-IEIR PROPERTIES
A. L i n e r Polydimethyfsiloxanes
Ingredients that are used in topical products are often referred to
by nomenclature established by the Cosmetics Toiletries, and Fra-
grance Association (CTFA). These CFTA names ( 2 ) have been es-
tablished primamy for the purpose of ingredient labeling. The
CTFA has chosen the term dimethicone for linear PDMS that con-
form to the general structure given in Figure 1. This structure is
representative of the linear PDMS available from all silicone manu-
facturers.
Use of Silicones in Topical Products

Figure 19.1 Dimethicone.

Dimethicone was the first silicone used in a commercial topical


product, and it continues to be widely used. Until the late 1970s,
however, dimethicone was generdly used only in small amounts (less
than 0.5%) for its defoaming properties, At this level, dimethieone
is very effective in preventing the objectionable foam generated
when cream or lotion is spread onto the skin. This foaming or lath-
ering is a problem, particularly for creams and lotions containing a
soap (e.g., triethanolamine stearate), A small amount of dimethi-
cone in such formulations effectively disrupts the foam because the
silicone rapidly spreads over the foam, causing it to break f 3 ) .
Beginning in the mid-l870s, formulators began to take advantage
of the other benefits of dimethicone that become apparent at higher
use levels. The general approach was to substitute dimethicone for
a portion of organic emollients traditionally used in topical products
such as mineral oil, petrolatum, and fatty acid esters, Although
this increased the cost of the formulation, it could be justified on
the basis of improved aesthetics, Dimethicone provides a less-greasy
feel on the skin compared with organics of comparable viscosity, an
effect that can be attributed to the higher molecular weight of the
silicone and to its surface properties. The use of dimethieone also
helps to reduce the stickiness associated with some ingredients be-
cause it tends to form a film over the other ingredients when the
formulation is applied to the skin. Another consequence of the high
molecuhr weight of dimethicone relative to organic ingredients is
improved water repellency and wash-off resistance, This is partic-
ularly important for topical products, such a s sunscreens, for which
it is desirable to retain the active ingredient on the skin during
water immersion.
The increasing use levels of dimethicone in topical p ~ o d u c t shas
been prompted, in part, by the regulation of over-the-counter (QTC)
drug products by the Food and Drug Administration (FDA). In
1978, the FDA issued the first draft of proposed regulations (4)
based on recommendations from their advisory panel for OTC topical
analgesic, mtirheumatic , otic , burn, and sunburn prevention and
treatment drug products. These regulations categorize dimethicone
392 Starch

as a safe w d effective inpedient for such products. Specifically,


the panel concluded that dimethicone was a safe and effective ingre-
dient for a topic& drug product used to provide temporary relief of
mhor skin ir&tations when. used at a concentration from 1% to 30%.
AAer reviewhg comments from the public, the FDA issued a notice
of proposed rule-m&ing in 1983 in which the status of dimethicone
as a safe an effective ingredient was retained (5).
Ths dimethicones most commonly used in, topic& formulations are
in the 6scosity range of 100 to l0OO cs. This represents a com-
promise between ease of formulation and the benefits that are typl-
c&ly sought by the inclusion of silicone in the formulation. High-
viscosity dimethicones have increased skin protection benefits and
are more substantive than low-viseasit y dimethicones, but they are
more difficult to incorporate into topical formulations. Rimethicones
with triscosities below 100 cs are sometimes used when optimum
spreading and lubricity are desired, but cyelic PDMS has largely
replaced dimethicone in these instances because of its ease of for-
muXation.
To the formulator who is attempting to incorporate unfamniar in-
gredients into a topical formulation, ane useful rule is that additives
that are soluble in one or more ingredients of a known formulation
can generally be included in smdl amounts without making any other
formulation changes, The rule applies to both homogeneous (solu-
tion) and heterogeneous (emulsion) products , Vaughan has pub-
lished a solubiZity parameter for dimethicone of 5.92 ( 6 ) . A general
rule (also @ven in the same publication) is that materids with solu-
bility pmameters within 2 units of the materid of interest will be
soluble. This suggests that dimethicone should be soluble in sever&
nonpolar oils cornmanly used in topical formulations : mineral oil and
petrolatum. In practice, however, this is not true for amethicone,
because solubility parameters do not work well for predicting the
solubility of high polymers owing to faetore arising from the entropy
of mixing. Xsapropyl myristate is one of the few commonly used
organic ingredients that is miscible with dimethicone, The solubility
of dimethicone in other orgmic ingredients i s generally poor. The
only exceptions are very low molecular weight dimetfiicones with ~ 3 s -
eosities af less than 5 cs.
Despite the inaolubifliLy of most dimethicones in typical topied
formulation ingredients , they are not especidy c2.rfficult to include
in tapicd formul~tions. Par emulsion products, a number of factors
work in the formulators favor, First, the emulsifiers commonly used
to stabilize topical formulations can usually accommodate a s m d
amount of dimethicone, even though it has different solubillity prop-
erties than the other oils used in the formulation. Second, most
topical formulations are sufficiently viscous to st ab;iXize the added
Use af Silicones in Topical Produels 393

dimethicone, which might separate from a thinner product. Finally,


the low intermolecular forces in dimethicone cause it to be broken
down, to very small droplets under the mking conditions norntdly en-
cauntered in the manufacture of a topical emulsion product, Stoke1@
Taw predicts that reducing the size sf emulsion droplets leads to a
more stable emulsion, and this works to the, advantage of the firmu-
lator when using sgieones. For the few topical pmducts that are
solutions, it is often possible to use isopropyl myristate or another
cosolvent to get dimethicone into the product.

The GTFA (a has established the name eyelomethfcane for v c l i c


polydimethylsiloxanes that are used in topical products, The chemi-
cal structure for the cycliomethicone tetramer i s shown in Figure 2.
The cyclomethicone tetramer is the most widely used because it has
the highest evaporation. rate of the homologous s e ~ e s but
, the q -
clornethicorxe pentamer has somewhat better ~1ub;ifityproperties. A
cyclomethicone Itrimer does exist, but it, is a solid at
tuxle m d is not used in topical products. The higher hornolomes,
such as the cycXomethicone hexamer, are normdTy present in corn-
mercw eyelornethicone, but in smdl amounts; hence, their lower
volat2ity generdly is not a concern, Typic& cornmereid cyelomethi-
cones are mixtures of the cyclomethicone tetramer and pentamer
which are made by distillation of crude polydimethyls2oxanr;s.
Gyclometkricones pm-FI-idebenefits that would be expected from a
low-moXwular-weight PDMS: law toxicity, Xow odor, exceXlent spread-
ing, and lubricity. These properties, together with their volatility
and mild solvent properties, have made cyclomethicone the highest-
volume silicone product used in the cosmetics and toiletries industry.
They were first used in the late 1970s as a vehicle for antiperepi-
rant salts in underarm topical products. Since that ime, cyclomethi-
cones have been used in virtually every category of topical; product.
One reason for the popularity of cyclomethicones as a formulation
vehicle is their low heat of vapolrization. The heat of vaporization
for the cyclomethicone tetramer is only 32 cal J g, which contrasts
sharply with that of water (520 calfg) and ethanol (210 callg).
This means that topical formulations based on cyclomethicone have a
pleasant "dry" feel when they are applied to the skin.
Unlike the dimethicones, eyclomethicones are soluble in a variety
of invedients commonly used in topical products. Table 1 lists the
solubility of eycXomethicone in a number of these ingredients. For
ingredients that are solids at ruom temperature, the solubility given
in Table I is for a mixture that has been heated to approximately
80°C. Solubility of cyclomethicone is generally limited to predomi-
nantly nonpolar organic mralerias, although it is soluble in some po-
lar organic solvents such as ethanol and isopropanol. Because cy-
clomethicone is more soluble in other ingredients than dimethicone,
it is easier to formulate into an existing topical product. This is
pariicularly true for formulations that are solutions. When cyclo-
methicone functions as the primary formulation vehicle and is the
major component in a topical formulation, it is often necessary to
employ some special ingredients to accommodate the active ingredi-
ents.

C, Phenyltrimethylsiloxane
PhenyE trimethicone is the name that the CTFA (2) has assigned to
silicones that conform to the structure shown in Figure 3. Commer-
cially available phenyl trimethicone is a &ture of species corre-
sponding to the structures indicated by Figure 3 for which the value
of n ranges from one to three. The increased number of organic
substituents on the silicon In phenyl trimethicone greatly increases
its solubility in organic ingredients relative to dimethyl silicones.
Phenyl trimethicone is the only silicone that is miscible in aU pro-
portions with 95% ethanol. It is also miscible with all commonly used
nonpolar organic ingredients in topical formulations.
Because of its solubility in other ingredients, phenyl trimethi-
cone is easy to incorporate into existing topical formulations. Typi-
caUy it provides the same benefits a s other low-viscosity silicones.
Phenyl trimethicone has the additional benefit of a somewhat higher
refractive index than dimethyl silicones ( 1.46 vs. 1.39). Hence, it
is used in topical formulations for which gloss is desired, such as
in a hair dressing.
Use of Silicones in Topical Products

Table t 9.1 Solubility of Topical Formulation


Ingredients in Cyclomethicone

Ingredient ~o~ub~ity

Mineral; oil Soluble


Petrolatum Soluble (80°C)
Paraffin wax Soluble (80°C)
Stearyl alcohol Soluble (80°C)
Stearic acid Soluble (80°C)
Impropy1 myristexte SolubIe
Isopropyl palmitate Soluble
Glycerol monostearate Soluble (80°C)
PEG 8 stearate Insoluble (80°C)
PEG 40 stearate Insoluble (80°C)
PEG 400 Insoluble
Glycerin Insoluble

CH3
Figure 19.3, Phenyl trimethicone.
Starch

D, Pol ydirnettcrylsiIaxapte-PoIyaIkyfeneQ x ids Copaiymers


When some of the methyl p a u p s on PDMS are substituted for side
chains of palydkylene oxide, a v e q surface-active class of saicones
is produced. For commerci& sBicones of this type, the polydkglene
oxide are polymers of either ethylene oxide (EO), prapylene oxide
(PO), or copolymers of ethylene m d propylene oxide (EOIBO). The
CTFA ( 2 ) has established the term dimethkane capolyoE for these
rnatefids. A generd structure for the most .commonly encountered
dimethicone copolyols is shown in F i p r e 4. The large difference in
polarity between the PDMS and polydkylene oxide segments is re-
sponsible for the surface activity of the dirnethicone copalyols, Un-
like conventiond nonionic surfme-active agents, which generdly
consist of a single hydrophobic (hydrocarbon) segment connected to
the polycilkylenie oxide segment, dimethicone copolyola often, have
multiple polyalkylene oxide segments attached dong the ssoxane
chain. This e v e s rise to a large variety sf possible dirnethicone
copolyol structures, with concomitant variatians in physical pmper-
ties and surface activ;ity, Dimethicone copolyols range from liquids
to w a y solids, depending p&mar2y on molecular weight and degree
of substitution with potydkylene oxide,
Dimethicone copolyols that contain large amounts of polyethylene
oxide are water-mluble, m d these are most often the ones used in
topical formulations. Water-soluble dimethicone eapolyals are easy
to add to aqueaus topical farmulathns, and they do not exhibit de-
foaming b e h a ~ o r . They can also be used in certain situations a s
.
emulsifiers for other saieones The disadvantage of using water-
soluble dimethicsne copolyofs is "tat they generally prodde few of
the benefits associated with dimethyl sniconies.

F Egu re 19.12 Dimethicone copolyal .


Use of Silicones in Topical Products

IIl, FORMUCATiNG SILICONES INTO CON-


VENTIONAL TOPICAL FORMULATIONS
A. Mow to Use the Example Formulations
In the following sections, examples are provided that illustrate how
silicones discussed in the preceding sections can be used in topical
formulations. These examples are relatively simple formulations that
have proved to be stable and easy to prepare using mixing equip-
ment commonly available in a formulation laboratory. They demon-
strate how silicones can be used with organic ingredients and are
intended to be used as a starting point for formulators who wish to
develop a silicone-containing topical formulation. Alternatively, the
example formulations can suggest which silicone may be appropriate
for a known formdation that is similar to one of the examples,
The CTPA nomenclature has been used for all the ingredients in
the example formulations. The CTFA Ingredient Dictionam (2) pro-
vides a chemical description and supplierltradename reference for
all the materials listed in the example formulations. A tradenme
and supplier has been indicated for materials that are avdable in
several grades or molecular weights. Tradenames are also indicated
when the ingredient is a blend of several components. Here, sub-
stitution for the indicated material by an similar material from an-
other vendor may be satisfactory; however, this may affect formula-
tion viacosity and stability. If no tradename is indicated, then any
suitable grade of material (e ,g, , USP , NF) conforming to the CTFA
description can be used .
In many cases, the examples are similar to commerciaf. formula-
tions that contain silicones; however, they are not finished formula-
tions that would be suitable for commercial use, because several im-
portant hgredient types have been omitted in the interest of sim-
plicity. The most significant omission is preservatives, which are
added to virtually all commercial topical formulations to prevent the
growth of micraorganisms in the product. The selection of a proper
preservative or preservatives depends on a number of factors, in-
cluding the intended shelf life, storage conditions, and the types of
organisms that might contaminate the product. Perfumes and colors
have also besn omitted because proper selection requires knowledge
of the intended use for the formulation. Such considerations are
beyond the scope of this chapter.

6. Compatibility of Silicones with Other Fornulation Components


Silicones that are used in topical formulations are chemically nonre-
active under conditions encountered during the preparation and use
of these formulations. Consequently9 there are no ingredients that
are incompatible with silicones because of formation of undesired re-
action products, or degradation of an ingredient by" sflieone. The
only troublesome interaction of sBiconera with other ingredients is
their well-known defoaming properties, which can be v e q undesira-
ble in a product that is intended to foam! XE foaming is an impor-
tant formulation attribute, then dimethicone eopolyol should be used
because this class of sBicones exhibits defoaming b e h a ~ o ronly at
elevated temperatures.

6. The Use of Silicones in Topical Creams and Lotions


The term Eation is generdly used to describe topicd formulations
that m e pourable emulsions, Creams are similar to lotions, but
higher in vitscositily and not pourable. Often a lotion can be modi-
fied to produce a cream simply by increasing the propartion of 02s
used in the formulation, In. describing the preparation of topical
creams and lotions, it is convenient to p o u p the ingredients into
two categories: water-insoluble ingredients, which are referred to
collectively as the oil phase, and water- soluble ingredients , which
are collrzclively referred to as the water phase. Topic& creams and
lotions contain emulsifiers that allow stable mktures of the 031 phase
and water phase to be prepared. Although emulsifiers, by their
nature, are partially soluble in both phases, they are usually added
as part of the oil phase,
Table 2 lists the ingredients for a lotion formulation based on
the widely used stearate.--cet yf alcohol emulsifier combination. The
primary emulsifier for this formulation is triethanolamine stearate,
which is formed from triethanolamine m d stearic acid when the for-
mulation components are mhed together. Cetyl alcohol is included
as a seeondaw emulsifier to improve the stability of the lotion and
provide a smooth texture. The other ingredients: dimethicone, min-
ersll oil, and petrolatum could be considered the '?&activeingredient s,If
as these are the mast commonly used emollients in commercial skin
care lotions, The same OTC panel that concluded that dimethicone
is safe and effective for use in skin protectants (see Sect , ZI , A )
also approved minerd oil and petrolatum for this purpose,
The basie lotion formulation is made by weighing the an-phase
ingredients and water-phase ingredients in separate containers.
Each phase should be heated to approximately 70°C, and mixed unllil
uniform. The hot ojit phase is then added to the hot water phase
whae stirring with a pmpeller-type mher. The mixer speed and
addition rate should be such that the oil phase is rapidly dispersed.
in the water phase during the addition, However, the mixer speed
should not be so high that a vortex is created that wB1 draw air
bubbles into the mkture. A i r bubbles will interlfere with, the emul-
slification of the oil phase. Mixing must be continued a s the lotion
Use of Silicones in Topicat Products

Table 19.2 Basic Lotion Formulation

Ingredient Wt% Tradename /supplier

Oil phase
dimethicone ( 500 cs) 3.0
mineral oil 1.5 Rlearol I Witco Chemical.
petrolatum 1.0 Sono-Jell No. 9 I' Witco ChemicQ
stearic acid (50%min.) 3.0
cetyl alcohol 1.0
Water phase
water 89.3
triethanolamine (99%) 1.2

is allowed to cool until it reaches 40°C. Heat-sensitive ingredients


(e,g., fragrance) can be mixed in at this point befoi the formula-
tion is packaged.
The viscosity of the formulation in Table 2 can be adjusted, to
a certain extent, by chan@ng the waterloil-phase component ratios.
To increase the formulation viscosity, the amount of water is re-
duced, This provides a way to make a cream formulation variant
that is based on the same ingredients. Another topical cream for-
mulation containbe; silicones is given in Table 3. The Arlacel 186 is
the primary emulsifier for this formulation. Addition& stabilization
and thickening are provided by the ozokerite and beeswax. The
formulation is a suitable starting point for a "cleansing creamn that
would be used to remove oily soils (e. g. , cosmetics) from the skin.
A novel feature of this cream formulation is that it is a water-in-oil
emulsion and, therefore, should be appropriate for delivering oil-
soluble therapeutic agents. The preparation is essentially the same
as for the previous lotion formulation, except that the water phase
is added to the oil phase while mixing.
The formulation listed in Table 4 is for a topical. antiperspirant
that is designed to deliver the active ingredient, aluminum chlorohy-
drate, to the skin, where it will suppress the activity of the sweat-
producing glands. The formulation is a low-viscosity emulsion suit -
able for packaging in rolling ball (roll-on) applicator. The inclusion
of cyclomethicone eliminates the stickiness that is characteristic of
aqueous solutions of aluminum chlomhydrate and other antiperspi-
400 Starch

Table 19.3 Water-in-Oil Cream

f ngredient Wt% Tradename I supplier

Oil phase
cyclomethicone 4.0 345 Fluid / Dow Corning
dimethicone ( 1000 cs) 1.0
synthetic beeswax 2.0 Syncrowax BB4 / Croda
ozokerite wax 2.0
glycepyl oleate (and)
propylene glycol 2.0 Arlacel 186 / XCI Amerieas
mineral oil 15.0
Water phaae
water
glycerin

Table 19.4 Aqueous Antiperspirant "Roll-on"

Ingredient Wt% Tradename / supplier

Mapesium aluminum silicate 1.0 Veegum I R. T. Vander-


bilt Go.
Water 33.2
Aluminum chforohydrate 50.0
(powder)
~1y;epyl stearate (and)
PEG 100 stearate 8.0 Arlacel 165 / ICI Americas
Glycerin 0.8
Cycbmethicone 7.0
Use of Siticanes in Topical Products 401

rant safts. Arlacel 165, a nonionic emulsifier, is used to stabaize


the cyclomethicone, an4 an inorganic thickener (Veepm) is included
to improve emulsion stabnifily, This type of formulation would be
suitable for delivering many other water-soXuhle active ingredients.
The antiperspirant formulation is prepared by first dispersing
the Veepm in hat (75OC) water using a Cowles mking blade or some
other type of high-shear stlirrer, The Veewm and, water should be
mked at 7fi°C for about 1 h r to ensure that the Veegurn becomes
f"u1ly hydrated. Hext, the alurninum ehlorohydrate is dissolved in
the hot mixture, followed by the glycerin and Arlacel 165, The
mhture is allowed to cool, with continuous m i x h g , until the temper-
ature drops to 30°C. Finitfly, the cyclornethicone is emulsified by
slowly pouring it into the warm mhture with Yigorous stirring.

D, The Use of Silicones in Anhydrous Topical Formulations

Anhydrous topical. formulations are, in many respects, less complex


than the emulsion forrnulati~nsdiscussed in the predous section,
They are often homogeneous blends that are prepared by simply
blending the ingredients. Organic oils serve as the vehicle to de-
liver therapeutic or other active ingredients to the skin, The key
to successful formulation of this type of formulation is to find the
an, or combination of oils that wIl1 solubBize the desired active in-
gredient. Tables 5 and 6 Xist two sample formulations: a sunscreen
oil m d an oil base. In the sunscreen, the octyldimethyl-PABA is
an absorber of ultradolet light whieh grorrides proteeli~nfrom the
sun. It is soluble In both the cyclomethicone m d the isopropyl
myristate, so the amount of protection can be easily adjusted by
changing the concentration in the formulation* The formulathn in
Table 6 is a base farmulatian that was developed as the starting
poht for a scented bath oil, Pheny-1 trimethicone and the lactate
ester blend (Ceraghyl 41) are used to soiubilize the fragrance in
the cyclomethimne. This base mkture wBl accommodate up to 15%
fragrance or other relatively nonpolar inpedients.
The last example of a homogenmus oil-based formulation is @ven
in T a l e 7 , Xn this formulation, the active ingredient, benzocaine,
is solubiXizect in the eyelomethicone whicle d t h two fatty Etlcohol poly-
propylene glywl (PBC) ethers, This formulation is m alternative
to the more traditional approach in which the benzocdne is delivered
by a hydmaleoholic vehicle, The advantage of m alcobl-free vehicle
is that the stirr@ng and drying effects are eliminated.
The formulation given in Table 8 is a special ease in the cate-
p~y of anhydrous topical formulations. Here the active ingredient
is an antiperspirant salt that is partially suspended in the c;yeb-
402 Starch

Table 19.5 Clear Sunscreen Oil

Ingredient Wt% Tradename I supplier

Cyclomethicone 16.0 344 Fluid I Dow Corning

Mineral oil 68.0 Carnation I Witco Chemical


Octyldimethyl-PABA 3.0

Table 19.6 Clear Oil Base

Xnpedient Wt$ Tradename I supplier

Cyclomethicone 17.6 344 Fluid I Row Corning


Phenyl trimethicone 6.0
Mineral oil 58.8 Carnation I Witco Chemical
C 12- 15 alcohols lactate 17.6 Ceraphyl 41 / Van Dyk &
Company

Table t 9.7 Clear Benzocaine Lotion

Ingredient Wt% Tradename / supplier

C yclomet hicone 36.4 344 Fluid I Dow Corning


PPG-15 stearyl ether 43.9 Arlarnol E 1 ICI Americas
PPG-10 cetyl ether 13.6 Procetyl 10 1 Croda
Ben zocaine 6.1
Use of Silicones in Topical PP-oducts 403

Table 19.8 Anhydrous Antiperspirant wRoll-onw

Ingredient Wt% Tradename I supplier

Dimethicone (TO cs) 5.0


Cyclomethicone (and)
Quaternium 18 hectorite
Bentone Gel VS-5 /
(and) ethanol 3.0 NL Chemicals
Cyclomethicone 70.0
Ethanol (SDA 40) 2.0
Aluminum chlorohydrate (powder) 20.0

methicone base. The salt will settle to the bottom, but is easily re-
dispersed by shaking. This type of formulation has improved aes-
thetics relative to antiperspirant shown in Table 4 because the elim-
ination of water gives the formulation a pleasant dry feel when it is
applied to the skin. Other active ingredients that are availabIe in
the form of finely divided powders could be substituted for the alu-
minun cfiiorohydrate to produce a variety of different formulations.

E. The Use of Silicones in Clear Aqueaus Formulations


Unlike the water-based formulations that were discussed in Section
III.B, the formulations in this section are intended to deliver active
ingredients that are completely soluble in water. Such formulations
are less common than emulsion forms, because often the active in-
gredients are not water-soluble. The formulation listed in Table 9
can be thought of as a vehicle to deliver glycerin to the skin; how-
ever, other ingredients that are glycerin-soluble can be included.
The formulation is a gel that is thickened with carbomer 934, a form
of polyacrylic acid, It is prepared by first dispersing the carbomer
powder in water. The temperature is not sul important factor for
this step, but good mixing is essential because the carbomer tends
to form lumps when it hydrates. Once the carbomer is dissolved,
it will be somewhat hazy, but the solution viscosity w 2 be low.
Thickening will occur when the carbomer is neutralized with the
triethanolamine. The triethanolamine should be dissolved in a small
portion of the water to facilitate mixing it with the carbomer solu-
tion. The other ingredients are then stirred into the gel. Di-
methicone copolyol is inctuded in the formulation to reduce the
sticky feel associated with the glycerin,
Table 19.9 Clear Aqueous Gel

Ingredient Wt% Tradename I supplier

Water 59.8
Carbomer 934 0.5 Carbopol 934 / B .I?. Goadrich
Triethanolamine (99%) 1.2
Glycerin 34.2
Propylene glycol 2.0
Dimethimne copoXyol 2.3 193 Surfactant I Ilow Corning

The active ingredients for the formulation listed in Table 10 are


surfactants designed for cleaning, The formulation is similar to a
shampoo but is suitable for cIeanlng skin as well because the sur-
factant combination used has very low skin irritation. Dimethicone
eopolyol is the only type of silicone that is suitable for use in this
type of formulation because it wiU not interfere with foaming. The
viwosity of the formulation is controlled by the amount of sodium
chloride used. If a higher viscosity is desired, then the sodium
chloride level should be increased. The formulation is prepared by
first dissolving the sodium chloride in the water, and then mixing
in the other ingredients in the order listed, The two surfactants,
sodium laureth sulfate and cocarnidopropyl betaine, are usually sold
as water solutions. For this formulation, the specific concentrations
of surfactants in water are given. Surfactants from other manu-
facturers may differ in concentration, so adhstments should be
made to earnpensate.

F. The Use of Silicones in Solid Delivery Forms (Sticks)


In certain categories of topical formulations, solid, or stick formula-
tions are well suited for delivering the active ingredient. This is
especfally true for antiperspkants and deodorants. The formulations
given in Tables I 1 and 12 are based on a deodormt and an anti-
perspirant, respectively. The clear stick is suitable for delivering
active ingredients that are soluble in one or more of the liquid in-
gredients. Sodium stearate is the gellant for the clear stick. A
typical deodorant stick is simply a solution of biocide (e.g., tri-
closan) in a mixture of water and ethanol gelled with sodium stea-
rate, To accommodate the cyclomethicone while maintaining clarity,
a combination of propylene glycol, isostea~ylalcohol, and PPG 10
Use of Silicones in Topical Products 405

Table 19.10 Foaming Cleanser

Ingredient Wt% Tradename I supplier

Sodium chloride 2.0


Water 31.0
Coeamide DEB 2.0
Dimethicone copolyol 4.5 193 Surfactant I Dow Corning
Cocmidopmpyl betaine (31%) 14.0 Lonzaine C I Lonza Inc.
Sodium Iaureth sulfate (28%) 46.5

cetyl ether is used. Because the formulation must be heated to


dissolve the sodium stearate, it should be prepared in some type of
covered container equipped with a condenser to prevent significant
losses of volatile components. The formulation is prepared by mix-
ing together all the ingredients except the sodium stearate and
heating them to 65OC. The sodium stearate is then dissolved by
stirring it in the hot mixture. The formulation should be poured
into molds while it is hot because it gels as it cools.
The other stick farmulation, listed in Table 12, is for an anti-
perspirant. This formulation is similar to that given in Table 8,
except that the antiperspirant salt is suspended in a mixture of mol-
ten inqedients that form a stick when cooled. Because this formu-
lation must be heated, the same precautions recommended for the
prevlous formulation to minimize loss of volatile material should be
made. To prepare this formulation, the stearic acid and cetyl aleo-

Table 19.1 1 Clear Stick Base

Ingredient Wt% Tradename 1 supplier

Propylene glycol
Ethanol (SDA 40)
PPC 10 cetyl ether 10.0 Procetyl 16 I Cmda
lsostearyl alcohol 19.8 Adol 66 1 Sherex
Cyclomethicone 39.8 345 Fluid / Dow Corning
Sodium stearate 8.0
hol m e heated to about 80°C, and the aluminum chlorohydrate is
stirred In. After the aluminum chlorohydrate is dispersed, the cy-
clomethicone is added, and the mixture is stirred untn it Q uniform.
The formulation should be allowed to cool, with mixing, until just
above the mlidification temperature (50-60°C) and then poured into
molds.

IV. NONTRADITIONAL TOPICAL FORMULA-


TIONS BASED QN SILICONES

The topical formulations containing silicones discussed in the previ-


ous sections are for the most part, variations of commercbl formu-
lations that havd changed little over the last 10 to 20 years. This
is especidy true for the emulsion formulations discussed in Section
I1l.C. Recently, the development of new silicone surfactants (di-
methicone eopolyf2ls) has provided the formulator with emulsifiers
that allow novel silicone-based formulations to be made. One such
emulsifhr is a dimethicone copolyol that can be used to make a vwi-
ety of water-in-92 emulsions based on cyclomethicone (7). These
formulations are quite different from water-in-oil emulsions, such a s
the one given in Table 3 , that must be thickened to provide a stable
formulation. Formulations of the type listed in Tables 13 and 14,
on the other hand, are stable over a wide viscosity range, giving
the formulator more flexibility in producing a formulation with the
desired viscosity. These formulations can accommodate ingredients
that are soXuble in either water o r cyclomethicone.. Because the
external phase of the emulsion is cyclomethicone, all have the pleas-
ant dry fed associated with the silicone,
The formulation given in Table 13 is an example of a skin lotion
that delivers mineral oil and glycerin a s the active ingredients,
Other emollients such a s petrolatum can be added to the oil phase,
or substituted for the mineral oil. A small amount of electrolyte is

Table 19.12 Antiperspirant Stick

Ingredient Wtai Tradename / supplier

Stearic acid 15.0


Cetyl alcohol 15.0
Aluminum chlorohydrate (powder) 20.0
Cyclomethicone 50.0 345 Fluid I Dow Corning
Use of Silicones in Tspical Products 407

Table 19. t 3 Water-in-Oil Skin Lotion

Ingredient Wt% Tradename / supplier

Oil phase
cyclomethicone (and)
32256 Formulation Aid /
dimethicone ccrpolyol 7.2 Dow Corning
cyclomethicone 9.6
mineral oil 7.4
pareth- 15-3 0.4 Tergitol 15-S-3 /
Union Carbide
Water phase
glycerin 20.2
water 53.2
sodium chloride 2.0

Table 19.14 Water-in-Oil Antiperspirant Lotion

Ingredient Wt% Tradename I supplier

Oil phase
eyelomethicone (and)
32256 Formulation Aid /
dimethicone copolyol 6.0 Dow Corning
cyclomethicone 27.0
polymrbate-20 1.0 Tween 20 / XCI Amedcas
Water phase
aluminum chlorohydrate
(powder) 20.0
water 46.0
needed for emulsion stabBity, and sodium chloflde has been used jn
this example, The nonionic surfactant pareth- 15-3 is included to
improve emulsion stability. The viscosity of the formulation can be
controlled by adjusting the ofi-phaselwater-phase ratio, Preparation
of the formulation is sirnirar to other topical emulsion formulathns,
except that no heating is needed. The oll phase and water phase
are mixed in sepamte containers and the emulsion is made by slowly
addbg the water phase to the oil phase with rapid mking. A high-
shear mher, such 8s an Eppenbach or Ssversan, is desirable for
makhg this type of emulsion, but a stirrer equipped with a Codes
blade can be used. For large quantities of emulsion, the best pm-
cedure is to prepare the batch with a conventional stirrer and then
pass it through a colloid ma1 to ensure that the entire batch i s ade-
quately mhed,
Table 14 lists a starting f"srmulation for an mtiperspirant emul-
sion.. in contrast with the formulation given in Table 8, the anti-
perspirant salt will not settle out because it is dissolved in the wa-
ter phase. No additional electrolyte is required in the formulation
because of the high level of antiperspirant salt used. The formula-
tian is prepared using the same method given for the predous ex-
ample.

REFERENCES

1. M, J, Owen, Chemtech, 11:288, 1981.


2. C T F A Cosmetic Ingredient Dictionary, 3rd ed. The Gasmetic,
..
Toaetlry., and Fragrance Assocktion, Washington, D C , 1982.
3. Encyclopedia o f Polymer Science and Engineering, Val. 2, 2nd
ed. John Wifey & Sons, New Uork, p. 66, 1985.
4. Federal Register, 43: 3468, 1978.
5, FederaZ Register, 48: 6820, 1983.
6, 42, D, Vauglrrm, J. Sse, Cosmet. C h e m , , 36:329, 1985.
7. A, J. DiSapio, M. S. Starch, Cosmet, TotZetries, 963(8):55,
1981,

APPENDIX
Suppliers
Witco G hemicd Corporation
Sonneborn Didsion
520 Madison Avenue
Mew York, Mew Uork 10022
(212) 605-3908
Use of Silicones in Tapfcat Products

Daw Corning Carparation


22Q West Salzburg Road
Midlmd , Michigan 48640
(517) 496- 4000
Croda Incorporated
183 Madison Avenue
;New Uork, New Vark 10016
(212) 683-3089
ICI Americas
Wgntineon, Delaware 29891
(800) 441-1757
.
R , T Vanderbilt Company Xncoleporated
Main rt WUliam Streets
Bellevjllle, New Jersey 871013
(201) 159-3225
ML Xndustries Incoworated
NL Chemcals Divzision
VVycoff MBls Road
Wightstown, Hew Jersey 08528
(609) 443- 2000
B e F. Gaodrieh Chemical Company
6100 Oak Tree Boulevard
Cleveland, Q h h 44 131
(216) 524- 0200
Union Carbide Corporation
Old Ridgebury Road
D a n b u v , Connecticut 068 17
( 203) 194- 2000

Sherex Chemical Company


P.0. Box 646
Dublin, Ohio 43017
(614) 164-6500
Zlonza Incorporated
22-10 Route 208
Faklawn, New Jersey 07410
(201) 791- 7500
Absorption, 215 APG , see Antigen-presenting
Acid ligase, 17 eeUs
Adsorbed liposames , 338 ApoePine glands, 70
Aesthetic appeal, 311 Application thickness, 231
Agglomeration, 206 Aqueous formulations, 403
Agi-mher , 387 A r e a - ~ X a rgroup, 52
A i r - medium interface, 274 Arlacel 20, 360
Alkyl polyoxyetktylene glycol Arrhenius plots , 203
ether, 47 At tachrnent factors, 282
Allergenieity, 315 Autoraaopaphy , 82, 282
Amphiphile, 42 Aaone, 43, 48
Andlrogens, 78
Anhydrous topical formula-
tlions, $01 23 cells, 91
Anti-inflammatory, 323 Bacterial eontamination, 3 16
Anti-inflammatory response, Barrier function, see Strfl-Cum
241 carneum barrier
Antiacne, 323 Basal cells, 273
Antibacte~al, 77 BCES, see bis(6-ehlaroelbyl)
Antidandruff, 323 sulfide
Antifungal, 77, 323 Benzodiazepines , 113
Antigen-nonspecific inflarnma- Biconthuous phase, 360
tory reactions, 91 Bilayer fluidity, 4 1
Antigen - presenting cells Bilayer water retaining capacity,
(APG), 91 55
Antiprurities, 323 Bilayer-structured lipids, 70
AQT, see Dioctyl sodium Bhary component vehicle, 259,
suf fosuecinate 260
Binary diagrams, 146, 160 Container and clssure system
napththalene-benzene , 149 selection, 203
white petmlaturn-mhydmus Container-closure system, 207
lanolin, 153, 154, 155 Containers, see Laboratory ro-
Binarcy systems, 146, 156 botie techniques
Bioavaiitability, 232 Cantinuous solubility re@on, 36Q
1E3 iochemical markers, 2 95 Cantour plot, 113, 120, 124
Birbeck grmules, 94 Cornification , 15
bis(B-ehloraethyl) smlfide Cornified cell, 16, 274
(ECES), 286, 290 Cosurfaetant , 358
Bragg relation, see BraggTs Creams, 398
law Critical micellization eoncent ra-
Zfraggrs law, 31, 52 tion, 351
Broad lameffae, 15 Crj,licaX point, 163, 148, 365
Broad membrane bihyers, 17 Critical, pmduct characteristics,
143
Capping, see Laboratory robotic Critieai, quality parameters, 204
techniques Crass-linking , 381
Caprylic leapric t &glyceride , Crystd growth, 206
10 CrystaUine prtcking , 40
Captured volume, 328 GrystaXline struct ure , 3 1
Casttd level, 72 Cubic liquid crystal, 311, 372
Cell aggregate ~ b i l i t ,y 275 Cultivation procedures, 294
Cement somes , 16 Cutaneous irritation, 7
Ceranxides, 1 7 , 78 Cutaneous steralogenesis, 19
CH , see Contaet hypersen- Cutaneous toxicity, 272, 286
sitivity Cyclomet hieane, 393
Chemical actidty , 2 Gytoklnes, 92, 93
Chemical potential, 2
Ghemical properties, Z99 d- spacings, see Xnterlsyer spac-
Clnennicd stabilities, 1 ing
Cholesteml, 35 Refoaming, 390
Cholesterol sulfate, 17 Devadsltion pathways, 206
Clinical supplies, 204, 205 Deleterious excipient s , 202
Glonidine , 109 Delivery rate, 9
Closed patch irritation, 11 Dendritic epidermal T cells, $9,
Cloud point temperature, 169 98
Coexistence curve, 149 Densities for liquids, 180
Collagen substrate, 279 Dermal mast cells, 96
Component ranges, 145 Dermis, 70, 88, 273
Camposition rmge8 174 Desmosomes , 274, 286
Computational methods, 124 Desonide, 10
Consumer appetll, 383 Ilesquamation , 19, 26
Contact mgles, 83 Detackification , 390
Contact hyperserrsitidty (CH) , Deuterium EJMR , 49
94, 88 Dialysis membrme , 234
Contact sensitizers, 95 Dgferential, extraction, 136
Differential salubBization, 136 Epidermal localization, 300
Differentiated keratinocyte c ~ 1 - Epidermal morphology, 27 3
ture, 254 Epidermis, 88, 273
Difdlerentiating keratinocytes , Epidermotmpic lymphc=tcytes, 89
285 Epidermotmpic blactd-borne cells,
Diflusion, 128, 236, 237 97
Diffusion ceU, 38, 215 Equ-ilibriurn solubilization capae-
Diffusion coeffidents , 228, 236 ity, 132
Diffusion mechanisms , 129 Esctterkhh "acoli, 294
Dimet hieone, 390 Essential fatty acid deficiency,
Dimethicone copolyol, 396 21, 24, 40
Dimethyl formamide , 43 Established product, 209
Dimethyl sdfoxide , 43 stage, 198, 208
Dbftrofluorobenzene (DNFB), Estradiol, 106
95 Eutectic, 151
Dioctyf sodium sulf~suceinate, Eutectic mkture, 150
358 Eutectie point, 150
Diseased s k h , 8 Exagenous lipaphsc materials,
Disperse systems, 145 77
Distdbutian-type process, 249 Experiment, 188
DNFB , see DinitmEluombenzc3ne Experiment optimization, 188
Draize skin test, 2 7 1 Experimental design, 190
Drug analogs, 109, 113 Experimental flux values, 112
Drug deliverzy, 221, 229 Expiration dating, 206
Drug diffusion, 230 Expiw date, 197
Drug loadhg, 344
Drug-skin interacthas, 2 2 1 FaXse-positive , 7
Drug- vehicle interazctions, 22 1 Feedback mechanism, 77
Dynamic structure, 49 Fetal. boene serum, 275
Fibroblasts, 90, 274
Ecerine sweat glands, 70 Fick% first law, 246
Eczema, 24 Fiek" second law, 224, 230, 236
Edema, 90 Fickian dif fusion , 111
Effective t hermadynamic aclidty , Faaggrin, 16
239 Film thickness, 5
Effector eefls, 96 Finite dose, 3, 217
E f a u rate, 334 &iffusion cell, 231
Electron-poor , 3 1 Fluidizing lipids, see Bilayer
Electmn-rich , 30 fluidity
Emxllsion, 146, 327 Flux, 112
Encapsulation volume, 328 Foam, 146
Epidermal bimhemical markers, FoXLicular reservoir, 72
273 Formaation, 206
Epidermal differentiatian, 15, comparison, 202
19, 273 of development stage, 198, 201
Epidermal growth factor, 275 Fragment constants, 251
Epidermal lipogenesis, 22 Free fatty acids, 17
Free sterols, 1 7 Hydrolytic enzymes, 16
Freeze-fracture, 343 Hydraphitic base, 10
H ydrophiiicity , 310
Gel, 146, 381 Hydrophobic compounds, 218
formulation, 382 H ydrot ropes, 4 1
viscosity, 383 Hypodermis, 88
Gelling agent, 381
Generalized cytotoxicity , 294 Ichthyoses, 26
Glucose, 373 Immediate-type hypersensitivity
Glycerol, 11 responses, 97
Glycolipids , 1 7 Immune responses, 90
Glycosidases , 17 lmmune system, 87
GlycosphingoEpids , 17 Immunwompetent cells, 90, 94
Granular cells, 274 Immunolo~caUymediated re-
Group contribution methods, sponses, 90
113 Immunosurve~ance, 91
Xmpaimd stratum corneum, 222
Hair follicle, 70, 71 In vitro cytotoxicity, 272
Nand additionst, see Laboratory In vitro evaluation, see In vitro
robotic techniques percutaneous absorption
Haptens, 95 In vitro models, 273
Helix, 381 In vitro pemutaneous absorption.
Hemidesmosomes, 294 2 13
Heterogeneous structural model, cell design, 215
111 dose, 215
Hexagonal liquid crystal, 351, equilibration, 217
371 finite versus infinite dose, 217
High endothelid venules, 93 material babnce, 218
High-shear mixing, 387 membranes, 214
Hadebrand-Scatchard equa- open versus occluded, 217
tions, 248 receptor Ruid, 217
Hlildebrand solubility coeffi- temperature, 215
cient s , see Solubility time, 217
paramter In vitro permeation, see In vitro
Histidine-rich protein, 16 percutaneous absorption
Histological organization of the In vitm release, 232
skin, 88 from vehicles, 222, 232
Homeostatic mechanisms, 271 In vivo studies, 240
Homogeneous film, 13 In vivo vasoconstriction, 239,
Hormones, 275 318
House of cards, 381 Infinite dose, 217
Human abdomen stratum corneum , Inflammation, 90
20 Inflammatory cells, 90, 97
Human fibronectin , 282 Inhibition of metabolism, 290
Human skin, 214 Insoluble peripheral envelope,
Hydrocarbons, 17 16
Hydrogen-bonding interactions, Instilled liposomes , 338
47 Xnterfacial kinetics, 139
Interfacial liquid crystal forma- Lipid phase transition, 339
tion, 128, 136, 140 Lipid-protein compartmentalim-
Interlayer spacing, 32, 52 tion, 14
Interleukin- 1, 93 Lipids, 17, 2 1
Investigational new drug (IND) , Lipophilic base, 10
197 Lipophilic drugs, 7
InvoXucrin , 16 Lipophilic molecubs , 6
Ionic surfactants, 134 Lipophilic pathways, 110
Irritancy potential, 315 Lipophilicity , 310
Isopropyl alcohol, 43 Liposames
Isotropic liquid, 4 1 characterization, 47, 327, 333
preservatives, 334
Kerathization , 15 process selection , 333
Keratinocytes, 98, 273, 275 stability, 335
differentiation, 282 vehicle selection, 333
stratification, 282 Liquid crystal, see Lyotmpic
Keratinosomes , 16 liquid crystal
Keratohyalin deposition, 16 Liquid crystal diffusion, 37
Keratohydin granules, 274 Liquids dispensing, 179, also
Keratolinin , 16 see Laboratory robotic tech-
niques
Labeling, 182 Local in%mmatory responses,
Laboratory robotics, 143, 174 87
Laboratory rabotie techniques, Local irritancy , 272
145 Lacalized delivefy , 109
capping, 182 Long-lasting reservoir, 11
containers, 182 Lot-to-lot variability, 144
hand additions, 182 Lotion, 398
liquids dispensing, 179 tow-angle x-ray diffraction, 30
semisolids dispensing, 181 Low-shear mixing, 387
Lag time, 138 Lower critical points, 148
Lamellar b d i e s , 16 Lubricity, 394
Lamellar liquid crystal, 44, 47, Lymphocytes, 97
145, 351 Ly mphokines , 92
Lamellar structure, 30 Lyotropic liquid crystals, 31, 40,
Laminin, 282 47, 136, 236, 349
Langerhan cells (LC), 94, 273 Lyotropic mesophases , see Lyo-
Large unilamellar vesicle (LUV), tropic liquid crystals
331
Laurncapram, 48, 55 Rilacrmmdsion -based topicals ,
Layered structure, 29 335
Level of the variable, 189 Macmmolecular metabolism, 286
Linoleic acid, 76 Macrophages, 273
Lipid bilayers, 30 Mass transfer, 127
Lipid biosynthetic padients, 19 coefficient, 228
Lipid catabolic enzymes, 15 Mast cells, 90, 97
Lipid organization, 29 Material balance, 218
Matrix, 19a Non- hydrogen - bonding solvent,
contml, 229 125
Matrix-receiver interface, 226 Nonelectmlytes , 125
Maximum water content, 365 Nonionic surfactants, 134
Melanocytes, 89, 273 Nonpolar material, 175
Membrane-coating granules, 16 Nonpolar lipids, 23, 24
Mercury intrusion porosimetry , Nonpolar pathway, 6
307 Nonsaponifiable lipids, 19
Merkei cells, 89 Normal micelles, 351
Micellar solutions, 129, 327, Nuclear magnetic resonance
350 (NMR), 30, 48
Micellar volume, 134 Nylon membranes, 282
Micrwmulsion, 337, 349
Microsponge, 300 Octanol/water partition coeffi-
composition, 310 cient, 111
particle size, 308 Ocular retention, 338
polyneeie microsphere, 300 Odland bodies, 16
pore diameter, 309 Oil phase, 398
release on command, 312 Ointment efficacy, 232
resiliency, 309 Oleic acid, 43
synthesis, 301 Oleic acid palmitic ester, 78
systems, 322 Qleyl alcohol, 48
volume, 309 O1igo;olamellar vesicles ( OLVs) ,
Mitochondria, 286 332
Mitomycin C (MMG) analogues, Opthalmic studies, 331
124 Optimization techniques. 188
Model skin surface lipid, 78 Optimized formulation, 189
Modulating drug release, 327 Optimum area per head group,
Molar attraction constants, 354
248 Qrder parameter, 36, 49
Monocytes , 97 Order profile, 49
Morpho1oEr;ical markers, 295
Multilamehr vesicles, 330 Packing constraints, 50
Multllaminate pathway, 111 Palmitic acid, 37
Multiple-phase regions, 146 P a r a f f ~ i chydrocarbons , 76
Mycoplasma arthritidis, 100 Partition coefficients, 113, 124,
228, 246, 250
n -dodecylpentaoxyet hylene Partitioning p m e s s e s , 247
glycol ether, 48 Passive diffusion, 132- 133, 221
N-methyl-2-pyrmlidone, 43 Passive drop-on-fiber, 128
Nauta mixer, 387 Pathophysiologic implications, 24
New product stage, 198, 208 Penetration enhancement, 5, 221
New drug application (NDA) , Percent encapsulation, 329
197 Percent entrapment, 329
Nitroglycerh, 109, 299 Percent incorporation, 329
NMR, see Nuclear magnetic Percutanmus absorption, 127,
resonance 213, 219, 241, 249, 321
Index

Percutaneous delivery, 189 Palydiorganosiloxanes , 389


Percutaneous drug t r a n s p o ~ Polyitispersity , 334
(see Table 7, 1) Polyethylene glycol, 11
benzene, 216 Polymer-solvent affinity, 382
5- BuarouracB , 255 Poly morphonucleiir cells, 97
9-bis-acybxymethy.2-6-mer- Polypeptides, 125
captopurine derivatives, 261 Pslysorbate 60, 10
nitoranfine, 2 16 Pore diameter, 309
paraquat, 2 15 Pore volume, 309
PCB, 216 Porogen, 301
siitlicy-llic acid, 9, 253 Porous microspheres, 300
6-mereaptopurine , 254 Porous polymer, 3Q5
sprinonohctone , 2 19 Powders dispensing, 178
theophyllhe, 253 Predicted flux values, 112
t sicielocarbon , 2 16 Preformulation stage, 198
water, 39, 215, 375 Preliminary formulations, 20 1
Pereutaneaus flux models, 110 Pressure release, 3 12
Pereutaneous permeation, 120, Pretreatment, 250
213 Prislane, 76
Percutaneous transport, 13 Praeess variables, 144
methods, 355 Prodrug, 109, 113, 250, 262
Permeability coefficients , 250 Product pmfiling , 197
Perturbed skin, 21 Product r e ~ s i a n stage,
. 209
Phagocytic cells, 90 Product stability, 198
Phase, 184 ProfBe studies, 199
beXtav;ior, 145, 186 Profiling, 197
boundaries, 146 Prolonged delivery, see Sus-
diagrams, 111 tained delivery
evolution, 140 Proposed eontainer- closure sys -
Phenyl trimethicone, 394 tems, 206
Phesplzatidylet hanolamine, 32 Pmposed pmduct , 206
Phospholipase A , 17 stage, 198
Phosphofipids , 17 Propylene glycol, TO, 43, 48
Phosphorous nuclear magnetic Prostaglandins, 93
resonance, 34 3 Pseudo-steady-staw , 228
Phosphorylcholhe , I 6 Pseudoplastic Elaw , 38 3
Physical propedias , 199 Pseudoternary diagrams, 171,
Phy skal stability, 1 165, 185
Pilosebaceous unit, 7 2 Pseudoternary systems , 155
Plastk substratum, 280 Pulsed delivery, 8
Plastic wrap hypothesis, 13 Pyramidal diavams , 17 1
Polar tipid, 23
Polar mzrterial, 175
Polar msfecules, 6 Quadrupolar coupling constant,
Polar pathway, 6 49
Palydimethy1sil.~xanes( PEtMS ) , Quadrupolar splitting, 36
38 9 Quaternary diagmms , 171
[Quaternary diagrams] Semisolids dispensing, see Labor -
water-butztnediol-sodium atory robotic technology
cktohte -lecithin, 370 Sepia2 cultivation of keratina-
water- hexadecane- sorbitan cyles, 275
monolaurate- diwt yl sodium Shelf life, 197
sulfosuccinate , 364 Significant quality attAbutes,
water -oleyl dcohol-propylene 208
glycol-C 12E0 5, 62 Ssicone, 389
water-pmpylene glycol-sodium Silicone rubber membrmes , 236
cholate-lecithixl, 370 Silicone surfactants, 406
Quaternary systems, 155 SingXe- component vehicles, 253
Single phase, 184
Random coils, 381 Single-phase compositions , 158
Receptor fluid, 217 Single-phase regions, 146
Reconstituted stratum corneurn , Sink conditions, 4, 226
37 Skin, 237, 271
Refatling process, 1 2 associated immune responses,
Regular solution theory, 241 87
Relative clearance, 232 associated lymphoid tissue
Release rate, 334 (SALT), 91
Replet ion, 2 1 care formulations, 3 17
Reservoir fluid, 214 delivery systems, 299
Reservoir function, I 6 depigmentation, 323
ResiBency , 309 distribution, 216
Resistant membrane, 227 grading, 8
Responses, 189 histological organization, 88
Revised product stage, 198 hydration, 2 2 1
Ringing gel, 372 metabolism , 2 14
Rodfike mkelles, 351 permeability, 237
Rotating disk, 128 permeation, 222
Rubefacient s , 323 structure, 88
surface lipids, 69, 77
Salicylic acid, 124 surface lipids composition, 73
SALT, see Skin-associated Small unilamellar vesicles (StfVs),
lymphoid, tissue 330
Saturated vehicles, 3 Sodium xylene sulfonate , 4 3
Saturation-type process, 249 Solid delivery farms, 4Q4
Scopolamine, 109, 299 Solids dispensing, 118
Sebaceous glands, 70 Solubilities, 259
Sebum, 72 Solubility parameter, 83, 240,
Sebum model, 8 1 245, 257, 264, 392
Sebum refatting, 78 Solubilization, 127, 130
Sebum-rich zones, 69 Solute-solvent , 245
Seburn-seleetive vehicles, 69 Solution ointment, 224
Second applkation studies, 258 Solution theory, 299
Self - diff usion coclfficient , 37 3 Solvent affinity, 382
Semipermeable membrane, 233 Solvent -solvent, 245
Sorbitan laurate , 360 Ternary diagrams, 163
Sorbitan stearate, 10 propylene glycol-white petro-
Sphingolipids , 19 latum-glycerol monostearate,
Sphingomyelin , 76 166
Sphingomyelinase, 17, 76 pmpylene glycol-white petm-
Spinous cells, 273 latum-sodium stearoyl lacty-
Spontanwus compartmentation, late, 166
327 propylene glycol- white petro-
StabGity , 146 latum-sorbitan monolaurate ,
limiting factor, 206 166
studies, 198 water-azone-C 1$05, 56
testing, 197 water-butanol-dioctyl sodium
testing protocol, 197 suXfbsuccinate (AOT) , 36 X
Stacks, 381 water- hexanol-dioctyl sodium
Steady-state flux, 5 sulfosuccinate (AOT) , 362
Steroid sulfatase, 17 water-lipid mixture-triethanol-
Sterol esters, I7 amine, 168
Storage conditions, 206 water-octanol-dioctyl sodium
Stratum corneum, 14, 17, 30, sulfosuccinate (A0T), 363
70, 89, 222, 231, 239, 252, water-octanol-sodium octanoate ,
318 353
barrier, 109 water-oleyl alcohol-C 12EC85, 56
lipids, 29 water-pmpylene glycol-C l@Qg,
structure, 13 56
S t r u c t u ~ ,30 water-sodium cholate-lecithin ,
Substituted melamines, 112 165, 370
Substratum, 274 water-sorbitan monolaurate-
Sulfur mustard, 286 dioctyl sodium sulfosuccinate ,
Sun protection factor (SPF) , (AOT), 359
318 Ternary plot, 155
Sunscreens, 318, 322 Ternary representation, 159
Surfactant, 127 Tetracycline HCI, 372
association colloids, 349 TEWL , see Transepidermal water
hydration, 55 loss
Suspension ointment, 223 Theoretical partition coefficients,
Sustained delivery, 7 251
Sustained release, 310, 344 Thermodynamic activity, 217,
Synthetic membranes, 278 238
substrata, 278 Thermodynamic stability, 356
Systemic drug levels, 109 Thin film, 3
Systemic immune m s p n s e s , Thrw-component system, 155,
87 164
Three-parallel-pathway model,
T cells, 91, 93 111
Tanfordis relations, 354 Three-phase region, 161
Targetkg sebum, 82 Tie line, 147
Terminal hairs, 70 Time release, 310
Tine scale, 127, 140 Variables optimization, 188
Tirnolol prodrugs, 113 Vasoconstrictor assay, 11
Tonofaament s , 274, 286 Vehicle
Topical &sorption, 222 d e s i e , 1, 65
Topical aquicf, crystals, 310 evaporation, 3
"f oxicalogy supplies, 20 1 performance, 7
Tracer perfusion studies, 15 skin interactions, 221
Transdermal defivery , 299 Veiled cells , 96
Trwsdermal flux, 110, 112, Vellus hairs, 70
113, 114, 118, 120, 122, Vesicles, $1
34 1 Viable epidermis, 22, 10
Transderml patches, 299 Viseoelastic pmperties , 30 9,
Trasepidermaf water loss, 23, 39 1
296 Visking membrane, 235
Transient chemical modification, V;i-tamin.f) precursor, 77
250 Void volume, 309
Trial optimization, I88
T r h n m l a r prism, 173
Triglycerides , 17 Water phase, 398
TrioXein, 78 Water repeBney, 390
Trypsinized skin, 38 Water solubility, 119
T wo-compartment model, 1, 243 Weak1y amp hiplnilic molecules,
Two-parallel-pathway model, 60
6 , 110
X-ray diffraction, 48, 343
Unllamellar liposornes , 328 flenobiotic, 212, 286
Upper critical points, 148
Upper consdt boundary, 148 Yield value, 383

You might also like