Journal of Pharmaceutical Sciences xxx (2018) 1-12

Contents lists available at ScienceDirect

Journal of Pharmaceutical Sciences

journal homepage: www.jpharmsci.org

General Commentary

Past, Present, and Future of Bioequivalence: Improving Assessment

and Extrapolation of Therapeutic Equivalence for Oral Drug
Products
Rodrigo Cristofoletti 1, 2, Malcolm Rowland 3, Lawrence J. Lesko 4, Henning Blume 5,
Amin Rostami-Hodjegan 3, 6, Jennifer B. Dressman 2, *
1
Brazilian Health Surveillance Agency (ANVISA), Division of Therapeutic Equivalence, Brasilia, Brazil
2
Institute of Pharmaceutical Technology, Goethe University, Frankfurt am Main, Germany
3
Centre for Applied Pharmacokinetic Research, University of Manchester, Manchester, UK
4
Center for Pharmacometrics and Systems Pharmacology, University of Florida, Orlando, Florida 32827
5
SocraTec C&S, Oberursel, Germany
6
Certara, 1 Concourse Way, Sheffield, UK

a r t i c l e i n f o a b s t r a c t

Article history: The growth in the utilization of systems thinking principles has created a paradigm shift in the regulatory
Received 2 March 2018 sciences and drug product development. Instead of relying extensively on end product testing and one-
Revised 3 May 2018 size-fits-all regulatory criteria, this new paradigm has focused on building quality into the product by
Accepted 12 June 2018
design and fostering the development of product-specific, clinically relevant specifications. In this
context, this commentary describes the evolution of bioequivalence regulations up to the current day and
discusses the potential of applying a Bayesian-like approach, considering all relevant prior knowledge, to
Keywords:
bioequivalence guide regulatory bioequivalence decisions in a patient-centric environment.
pharmacokinetics © 2018 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.
regulatory science
dissolution
mechanistic modeling
pharmacokinetic/pharmacodynamic
(PK/PD) correlation

Introduction agencies across the globe to develop systematic toxicity testing

Historically, the development of the regulatory sciences has
been stimulated by public health catastrophes, as illustrated by the
chain of events in the United States. Before 1938, one could intro-
duce a drug product in the United States market without any
toxicity testing or demonstration of effectiveness. The deaths of 107
children in September and October of 1937, who were poisoned by
an elixir of sulfonamide which contained ethylene glycol intended
to solubilize the drug, symbolized the catastrophic flaw in the
caveat emptor approach.1 Almost immediately, in 1938, amend-
ments to the Food and Drugs Act of 1906 requiring toxicity testing
and safety studies before the introduction of any medicine into the
marketplace were enacted.1,2
The thalidomide tragedy, which originated in Europe in the
early 1960s, marked a further turning point in regulatory supervi-
sion of the pharmaceutical market, as it prompted regulatory
* Correspondence to: Jennifer B. Dressman (Telephone: þ49 (0) 69 798 29680).
E-mail address: dressman@em.uni-frankfurt.de (J.B. Dressman).
protocols.2 In the United States, for example, the Federal Food,
Drug, and Cosmetic Act was amended in 1962 by the so-called
Kefauver-Harris Amendments, which required not only more
stringent toxicological testing but also evidence of efficacy for
the product's intended use to support a new drug marketing
application (NDA), that is, premarket clinical trials.3,4 Because of a
grandfather clause, no further studies were required for drugs that
were already on the market before the 1938 law; however, efficacy
data for drugs that entered the U.S. market after 1938 became
mandatory.3,4 Some manufacturers contested the legality of the
Kefauver-Harris Amendments, but the U.S. Supreme Court upheld
the retroactive efficacy requirement.5 Owing to the lack of
resources to evaluate the efficacy of the several thousand drug
products that had been approved between 1938 and 1962, the U.S.
Food and Drug Administration (FDA) recognized that a transitional
period would be necessary and elaborated an abbreviated pro-
cedure to handle generic drugs. According to this procedure, if the
regulatory agency already had evidence supporting the safety and
efficacy of a certain drug substance in the dossier of the respective

https://doi.org/10.1016/j.xphs.2018.06.013
0022-3549/© 2018 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.
2 R. Cristofoletti et al. / Journal of Pharmaceutical Sciences xxx (2018) 1-12
innovator drug product, generic manufacturers would not have to
submit clinical trials. Instead, generics were required to be proven
equivalent to the respective brand name products to maintain their
marketing authorizations.4-7 For the first time, the focus shifted
from the drug substance to the formulation. Nevertheless, at that
time, bioequivalence (BE) standards were in a state of flux and had
not yet been established.4,5,8
The post-1962 scenario is even more complex. New clinical trials
had to be conducted even for generics and, thus, the cost of
developing generics of post-1962 drugs increased dramatically.7
This period was marked by intense legal battles among the regu-
latory authority, the innovative pharmaceutical companies, and the
generic manufacturers. Worth mentioning is the Roche Products Inc.
v. Bolar Pharmaceutical Co. case, in which the innovative company
questioned whether the generic manufacturers could conduct
testing on patented products solely for obtaining approval of
competitor products, and the United States v. Articles of Drug, appeal
of the Lannett Company, Inc. case, in which the generic manufacturer
contended that because the drug substances had already been
proven to be safe and effective by the innovative companies, clinical
trials should be superfluous for the amended new drug application.
Federal courts reached contrary decisions on these matters,
increasing the uncertainty around what was required to bring a
generic product to market.5,7 Consequently, as many as 150 brand
name drug products lacked generic versions, despite being
off-patent, in the following years.7,9
Paralleling the judicial disputes, academic concerns about the
in vivo equivalence of generics arose after some clinical results were
published showing that different formulations containing digoxin
led to different pharmacokinetic (PK) profiles.10-12 For instance,
Lindenbaum et al.11 investigated the systemic exposure of digoxin
released from 4 different formulations in healthy volunteers. The
authors found pronounced differences in the bioavailability (BA) of
different commercially available digoxin tablets. These differences
in BA results, in combination with the well-known narrow thera-
peutic index of digoxin, caught the attention of the regulators at the
FDA. Subsequently, Wagner et al.,13 under contract with the FDA,4,8
confirmed Lindenbaum's findings of lack of equivalence between
the plasma levels of digoxin released from multisource tablets. In
fact, Wagner et al.13 reported nonequivalence in the systemic
exposure of digoxin even with tablets that met the pharmacopoeial
standards for both potency and disintegration time. Similar obser-
vations were also reported for other drug products, including those
containing tetracycline,14,15 chloramphenicol,16 phenylbuta-
zone,17,18 and oxytetracycline.19 An editorial written by Levy and
Gibaldi20 in 1974 reinforced the urgency of developing an adequate
means of assuring the BA of oral drug products. Recognizing the
existence of BA problems in marketed products, the FDA Office of
Technology Assessment convened a panel of 10 senior medical
consultants who concluded that the then current compendial
reference standards and regulatory practices did not assure uni-
form BA.21
Finally, in 1984, almost 20 years after the Kefauver-Harris
Amendments, the U.S. Congress passed the Drug Price Competi-
tion and Patent Term Restoration Act (commonly known as the
Hatch-Waxman Act), which was intended to make low-cost ge-
nerics more widely available while simultaneously maintaining
adequate incentives for innovation, by, for example, extending
brand name market exclusivity. The Act authorized the FDA to
approve generic drug products through BE studies, fostering the
development of multisource drug products and generic-based
public health policies worldwide.6,9
In Europe, concepts for generic drugs and BE evolved at the
national level, with each country forming its own rules. The first
efforts to standardize European regulations regarding drug
approval in general were made before the formation of the EU, with
the passage of EC Directive 65/65/EEC in 1965.22 After the forma-
tion of the EU in 1993, there were significant changes, eventually
leading to a centralized procedure for new drug applications at the
European Medicines Agency (EMA) but continuing to allow
approval of generic versions at a national level. Today, generic drug
approvals within the EU can be obtained in one of the 3 ways: a
central approval which applies to the whole European market,
stepwise approval at the national level (which can lead to approval
throughout the EU once 3 countries have approved the drug
product, provided a certain mix of countries has been involved), or
approval only at the national level, which can be instigated for
example when the sponsor decides that the product will only be
marketed in that given country.23
Generics have played a role in many European countries for
many years, as in the United States. The first generic drug company
in Germany, Ratiopharm, was founded in 1973, and in that year,
generic products, such as acetylsalicylic acid and paracetamol
tablets, were introduced into the market. Other companies soon
followed (e.g., Stada in 1975), and by 1997, generics already
accounted for around 40% of prescriptions.24 This percentage has
continued to climb to over 80% over the ensuing 20 years.
Again taking Germany as an example, the regulation of generics
in Europe provides a variety of ways in which generic drug products
can be introduced into the market. In Germany, there are still
so-called “Standard Rezeptur” products on the booksdand some of
these are on the market. Standard Rezeptur products must be
manufactured according to a predefined composition and, if this is
the case, no BE testing is required to achieve a marketing authori-
zation.25 Furthermore, when a drug product has established itself
over a number of years in the market and scientific material addi-
tional to the approval package is available, a generic manufacturer
is permitted to rely on this database. For example, glucocorticoids
and endocrine hormones were introduced into the market in the
1950s and 1960s, and since then, many scientific papers and clinical
studies have been published regarding their use. Against this
background, it was deemed to be no longer necessary to perform
clinical studies for the approval of generic versions. However, this
type of waiver is only granted when the drug, excipients, and
dosage form are all identical with products already on the market.25
Another interesting modality for generic drug approval in Germany
is the literature-based approval, known as the “bibliographische
Zulassung” procedure. In this case, regulatory relief in the form of a
waiver of preclinical and clinical data is entertained. Requirements
are that the drug must have already been on the market for 10 years
and that the sponsor must produce thorough literature evidence of
efficacy combined with an acceptable side-effect profile. Both
positive and negative literature must be cited and discussed, with
peer-reviewed publication being accorded more weight than in-
ternal documents or nonepeer-reviewed articles.26 It should be
noted, however, that the usual procedure for approval of generic
products in Germany (and elsewhere in Europe) is to demonstrate
BE using PK studies in healthy human volunteers.
Outside the United States and Europe, there are many different
approaches to approval of generic drugs, ranging from very strin-
gent requirements and a low % of prescriptions filled with generics
in Japan (around 18% in 2007) to the situation in Iran, where,
because of a government decree shortly after the revolution, only
generic versions were permitted to be marketed. Still today,
approximately 96% of drugs available on the market are manufac-
tured by local generic companies.27,28 Interestingly though, 86% of
Iranians hold the perception that bioequivalent generics are
not therapeutically equivalent to the respective reference drug
products,29 a statistic which has driven both the government and
generic manufacturers to sponsor prospective, blinded efficacy, and
 R. Cristofoletti et al. / Journal of Pharmaceutical Sciences xxx (2018) 1-12
safety studies for head-to-head comparisons between approved
generics and branded versions.30,31
Initiation and Evolution of the BE Concept
The first U.S. legal definitions of BA and BE were stated in the
Code of Federal Regulation 21CFR320.1. For systemically acting
drugs, BA was defined in the Act as “the rate and extent to which
the active ingredient or active moiety is absorbed from a drug
product and becomes available at the site of action” and BE as “the
absence of a significant difference in the rate and extent to which
the active ingredient or active moiety in pharmaceutical equiva-
lents or pharmaceutical alternatives becomes available at the site of
drug action when administered at the same molar dose under
similar conditions in an appropriately designed study.” This statu-
tory definition of BA is purposely different from the academic view,
which defined BA as the percentage of an administered dose that
enters the systemic circulation in an unchanged form, being a
function of the fraction of the dose that is effectively absorbed into
the enterocytes (Fabs) and of that, the fraction escaping first-pass
elimination on passage through the gut wall (Fg) and liver (Fh)
before reaching the systemic blood supply32:
BA ¼ Fabs  Fg  Fh
As defined, for example, in the Biowaiver Guidances,33,34 oral
absorption refers to the movement of drugs across the mucosal
membranes of the gastrointestinal (GI) tract, and Fabs represents
the fraction of the administered drug that reaches the systemic
circulation as either a parent or metabolite. Thus, BA would equal
Fabs only in the absence of presystemic metabolism.32 Nevertheless,
by using measures of both the rate and extent of absorption to
define BA, the regulatory authorities emphasize the importance of
assessing the drug product performance in vivo.
By focusing on the “availability at the site of drug action,” reg-
ulators pointed out their concern about ensuring therapeutic
equivalence. From a clinical perspective, the metrics used to assess
BA and BE should be surrogates of drug efficacy and safety. In other
words, the BE metrics should bridge efficacy and safety data be-
tween compared formulations by showing that variations in BA,
which fall within an acceptable range, do not alter the clinical
outcome.
Indeed, the careful choice of the words used in the statutory
definitions of BA and BE highlights the 2 fundamental regulatory
aspects of the BE concept: biopharmaceutical and clinical. Inter-
estingly, the legislators did not specify which metrics should be
used to assess the rate and extent of drug absorption or what is
meant by “significant differences,” delegating this responsibility to
the regulatory authorities. The statutory definitions, thus, afford a
higher level of flexibility than has generally been practiced by
regulatory authorities, not only in the United States but also in
other jurisdictions.
Retrospectively, with a view to the delineation of drug/formu-
lation parameters from systems parameters, it is apparent that the
regulatory definitions of BA and BE do not stipulate the conditions
of the system under which BA and BE are to be assessed. This
omission implies that once BA or BE is established in one group of
individuals or under one set of conditions, it will be valid regardless
of any changes to the systems parameters. Based on a literal
interpretation of the aforementioned rules, one would postulate
that BA and BE may be extrapolated from healthy young adults
(who are typically used in such studies) to, for example, pediatric
and elderly patients without any adjustments.
The scientific evolution of regulations concerned with BE is
summarized in the following text and in Table 1.
3
Table 1
Historical Turning Points and Scientific Evolution of Regulations Concerned With BE
Year Major Events
1906 Food and Drugs Act
1937 Deaths of 107 children who were poisoned by an elixir of
sulfonamide
1938 Federal Food, Drug, and Cosmetic Act (FD&C)
1960s Thalidomide tragedy in Europe and further countries
1962 Drug Efficacy Amendment (also known as Kefauver-Harris
Amendments)
1973 First generic company in Germany
1974 Editorial on digoxin written by Gerhard Levy and Milo Gibaldi
1977 75/75 decision rule
1980s 80/20 rule or power approach
1984 Drug Price Competition and Patent Term Restoration Act (also
known as Hatch-Waxman Act)
1987 Two one-sided tests
1990s Bio-International conferences on Bioavailability, Bioequivalence,
and Pharmacokinetic Studies, held biennially
1999 Regulatory initiatives on individual BE and population BE were
stopped
2000 BCS-based biowaiver introduced for generics
e Narrowing acceptance interval for NTIs
e Reference-scaled average BE
e Partial AUCs
e Flexible BE criteria for NDAsdfocus on the clinical aspect of BE
e Prioritizing the biopharmaceutical aspect of BE in case of ANDAs
2017 Applying model-informed drug development approaches for
generics
ANDA, abbreviated new drug application.
Average BE
The first acceptance limits for declaring equivalence were
established in the early 1970s.35 A group of medical experts in
consultation with the FDA reached a consensus that only differ-
ences of greater than 20% for the PK metrics mean peak (Cmax) and
extent of exposure (AUC0-t) would be clinically significant for drug
products,36 and this criterion has since then been applied across the
board. At the time, some forceful arguments against using the same
PK metrics and setting the same, essentially arbitrary, acceptance
limits for every drug were voiced.12,37 Nevertheless, these voices
did not prevaildlikely because there was limited knowledge of PK/
pharmacodynamic (PD) relationships for many drugs, making it
difficult for regulatory authorities to define drug-specific BE re-
quirements and because there was pressure on the regulatory au-
thorities to increase the number of approved generics to support
the new policy. A consensus around the new paradigm for the BE
field, that is, the “± 20%” rule, was achieved within a shorter time
frame than would be usual in other scientific fields.38 The concerns
expressed by, for example, Levy12 about setting the same arbitrary
acceptance limit for every drug, regardless of where its usual dosing
lies on the sigmoidal exposure-response (E-R) relationship, were
unable to convince either the FDA or the majority of the scientific
community at the time. Some years later, even though he had
demonstrated that the delay between the time course of plasma
concentration and drug effect should be taken into account to guide
the decision about which exposure metrics would be clinically
relevant, Sheiner37 also failed to succeed in reopening the debate
on this field.
In contrast to the steadfast adhesion to the equivalence criteria
(“± 20%” rule), the statistical methods used to establish BE after
administering different formulations have evolved substantially
over the last decades. In 1977, the 75/75 decision rule was proposed
as an addition to the requirement that there should be a difference
of no more than 20% in the mean AUC and Cmax between the test (T)
and reference (R) formulations. This rule was proposed with the
aim of accounting for individual variability in BE. According to the
4 R. Cristofoletti et al. / Journal of Pharmaceutical Sciences xxx (2018) 1-12
75/75 rule, 2 formulations were to be deemed bioequivalent pro-
vided that the individual subject ratios (i.e., T/R for AUC0-t and Cmax)
exceeded 0.75 in at least 75% of the subjects studied, in addition to
the average comparison. Recognizing that this decision rule would
not ensure similarity in the safety profiles between the compared
formulations, it was further amended with the addition of an upper
bound on the ratios as well, that is, 0.75-1.25.39 However, this rule
garnered widespread criticism in the literature because of its high
sensitivity for drugs showing large intrasubject or intersubject
variability, and it was later abandoned.40,41 In the early 1980s, a
statistical method for assessing BE between formulations based on
conventional hypothesis testing (i.e., null hypothesis of no differ-
ence between the 2 compared averages; H00 : mT  mR ¼ 0) was
proposed. If the null hypothesis was not rejected, it was still
necessary to demonstrate that the study was large enough to
provide at least 80% chance of correctly detecting a 20% difference
of the R average; the so-called “80/20” rule or “power” approach.42
However, it became apparent that simple conventional null hy-
pothesis testing the equality of 2 formulations was inadequate for
BE purposes.43 First, because the main goal of BE studies is not to
assess whether there is a statistically significant difference between
T and R formulations but whether the difference is clinically rele-
vant or not. Second, because using the renowned “lady tasting tea
experiment,” Ronald Fisher had already demonstrated that “the
null hypothesis is never proved or established, but is possibly dis-
proved, in the course of experimentation. Every experiment may be
said to exist only in order to give the facts a chance of disproving
the null hypothesis.”44 Consequently, if one aims to demonstrate
equivalence between T and R formulations, the “equivalence
hypothesis” should be the alternative rather than the null
hypothesis.42,44,45
In 1987, Schuirmann attempted to resolve the issues of hy-
pothesis testing by recommending that statistical analysis of Cmax
and AUC0-t be based on a 2 one-sided test procedures.42 Briefly, the
alternative hypothesis was redefined as a range of values with an
equivalent effect after assuming a clinically significant difference of
±20% and a ¼ 0.05. The first condition tests whether the mean peak
and extent of drug exposure after administering the T are signifi-
cantly lower than the mean peak and extent of drug exposure after
the R formulation, while the second condition tests the opposite
scenario, that is, whether the mean Cmax and AUC0-t after admin-
istering the R are significantly lower than the Cmax and AUC0-t after
the T formulation. This analysis has the benefit of yielding the same
outcome regardless of which product is designated as T and which
as R. Because we conventionally note BE intervals as T/R, the second
BE limit is 1.25 (i.e., reciprocal of 0.8), illustrating the rationale
behind the 80%-125% confidence interval criteria. This approach is
termed average BE.42
Development of BE Requirements During the 1990s
While the United States was the driving force for the devel-
opment of the BE concept in the 1980s and early 1990s, Europe
and Japan started contributing significantly from the mid-1990s.
In fact, many developments in the BE field described in the
following sections of this article (e.g., measures of rate and extent
of absorption vs. measures of exposure, individual and population
BE, optimizing study design to increase sensitivity to detect
formulation differences, and so forth) were initiated and debated
during the Bio-International conferences that were held bienni-
ally in Europe, Japan, or North America and co-sponsored by the
FIP (International Pharmaceutical Federation), the FDA, and
several other regulatory authorities and pharmaceutical organi-
zations.46-48
Population and Individual BE
Already in the late 1970s, some authors had been expressing
concerns about the suitability of average BE to address
genericebrand name drug product interchangeability, especially in
patients who had already been started on treatments using the R
formulation. In particular, they were concerned that the average BE
approach focuses only on the comparison of population averages of
a BE measure of interest and not on the variances of the measure for
T and R products.49
To address the issue of interchangeability, 2 new concepts were
proposed, population BE and individual BE. Extensive reviews of
individual and population BE, including the respective statistical
methods, have been provided elsewhere.50-53 The population and
individual BE approaches aim at addressing 2 different aspects of
the interchangeability concept: prescribability and switchability.
Drug prescribability, which is assessed by population BE studies,
refers to the clinical setting in which a physician prescribes a drug
product to a drug-naive patient. In this setting, the prescriber relies
on an understanding that the average performance of the generic
has been well characterized and a comparative PK study has
bridged the safety and efficacy information from innovator's clin-
ical trials, that is, generic or brand name formulation can be chosen
to start the treatment. Besides comparing population averages, the
population BE approach assesses the total variability of the measure
in the population. Although population BE is an improvement over
average BE, it is still not sufficient to ensure that a given individual
will respond similarly to the 2 formulations.54-56
Drug switchability refers to the setting in which a practitioner
transfers a patient from a brand name drug product to a generic
formulation. Individual BE testing, that is, establishing BE on an
individual subject basis is needed for switchability, especially in
those instances in which the within-subject variability of the
generic product is greater than the brand name product55 or in
which the patient is switched from one generic product to another.
The advantages and challenges of using the individual BE approach
were intensely debated throughout the 1990s.36 Finally, in 1999,
the FDA decided to stop the regulatory initiative related to indi-
vidual BE testing because of the practical, economic, and statistical
challenges in its measurement and interpretation.51,57,58 Moreover,
refuting the claim that approved generics might exhibit non-BE
results when retested in special patient subgroups, the FDA Indi-
vidual Bioequivalence Expert Panel concluded that “at this time,
individual bioequivalence still remains a theoretical solution to
solve a theoretical clinical problem. We have no evidence that we
have a clinical problem, either a safety or an efficacy issue, and we
have no evidence that if we have the problem that individual bio-
equivalence will solve the problem” and “switchability is not a
problem for approved generics.”58 Thus, the average BE approach
remained the key method for the evaluation of therapeutic equiv-
alence and interchangeability. The conclusions reached by the
expert panel are examined in detail in the section on group-by-
formulation interactions. Hereinafter, the term BE will refer to
average BE unless otherwise stated.
Measures of Exposure Versus Measures of Rate and Extent of
Absorption
Because assessing the active moiety kinetics at the site(s) of
action is generally not possible and direct measurements in the GI
tract to quantify the rate and extent of drug absorption are rather
laborious,59 exposure metrics derived from the plasma
concentration-time curve are generally used, under the assumption
that the substance in the general circulation is in exchange with the
substance at the effect site.60-62 In this context, peak (Cmax) and
 R. Cristofoletti et al. / Journal of Pharmaceutical Sciences xxx (2018) 1-12
extent of exposure (AUC) are by far the most common metrics used
to assess BE. Although AUC is also a reliable metric for evaluating
extent of absorption,63 finding a unanimously acceptable measure
or parameter for describing the absorption rate has been the clay
foot of BE. Cmax is not a pure measure of absorption rate as it (a) is
partially dependent on the extent of drug absorption; (b) is
generally insensitive to changes in ka, especially when ka > kel; (c)
contains little mechanistic information about the drug absorption
process, and (d) cannot discern differences in Tmax or lag-time
between formulations.61,63-70 For these reasons, some authors
have argued that the regulatory authorities should concentrate on
the clinical relevance of the plasma PK profile by focusing on the
systemic exposure concept. In this context, Cmax is deemed a useful
metric, not because it is an accurate measure of absorption rate but
because it is a useful measure of the peak of systemic exposure, a
metric which has clinical relevance in terms of safety and efficacy of
the drug product with respect to, for example, maximum thera-
peutic response, adverse events, and dose-dumping.61,71 This line of
thinking has subsequently prevailed, and modern BE guidelines
have incorporated exposure concepts, to better focus on the clinical
aspects of BE.33,34
Optimizing Study Design to Increase Sensitivity to Detect
Formulation Differences
The FDA and the EMA, working together with prominent
pharmacokineticists, took the lead toward setting the most sensi-
tive study conditions to detect formulation differences. For
example, it was recognized that single-dose rather than multiple-
dose studies are generally better able to detect formulation ef-
fects. For the same reason, it was contended that BE studies should
be conducted under fasting conditions and that the parent drug,
rather than metabolites, should be quantified whenever possible.
Furthermore, to reduce variability that does not arise from product-
related differences, it was decided that BE studies should be per-
formed in healthy volunteers unless the drug is associated with
safety concerns that render studies in healthy volunteers unethical.
According to this paradigm, if the T and R drug products were able
to show similar in vivo dissolution and absorption in healthy adults,
assessed indirectly by means of comparing systemic exposure
surrogates, it can be considered highly likely that both formulations
would behave similarly in all target patient groups and clinical
scenarios. The aforementioned requirements for the study design
have been integrated in modern BE guidelines, focusing on the
biopharmaceutical aspect of BE.33,34
BE in the New Millenium
A major innovation in BE that was introduced in the early 2000s
is the possibility of waiving in vivo BE studies for certain generic
drug/drug product combinations. This is known as the BCS-based
biowaiver procedure, where BCS is the Biopharmaceutics Classifi-
cation System originated from the study by Amidon et al.72 The
underlying assumption of the BCS-based BE procedure is that if 2
drug products dissolve similarly under in vitro conditions repre-
sentative of the GI tract, they should result in indistinguishable
plasma profiles and thus be bioequivalent with each other. This
approach has been published in guidances from the FDA, EMA, and
WHO and is currently undergoing discussions at the International
Committee on Harmonization.33,34,73 Importantly, in vitro studies
assess product performance more directly than do conventional
PK- or PD-based BE studies because in vitro studies focus on
comparative drug release from T and R formulations, while in vivo
BE testing can be influenced by confounding factors because of its
indirect approach. Examples of confounding factors include
5
intraindividual and interindividual variability in gastric emptying,
transporters, and presystemic metabolism, as well as in post-
absorptive processes such as distribution and clearance.74
With respect to PK-based determination of BE, both the FDA and
the EMA, along with several other industrial countries, have pub-
lished guidelines containing recommendations to carry out BE
studies for specific products. While deviations from the standard
one-size-fits-all approach within the European jurisdiction are
limited to formulations containing narrow therapeutic index (NTI)
drugs (by narrowing the acceptance interval) and highly variable
drugs (HVDs; by widening the acceptance interval), requirements
to additionally assess partial AUC (e.g., various modified-release
formulations), compare within-subject variability (e.g., warfarin,
rivaroxaban, and dabigatran) and even assess subject-by-
formulation interaction variance (e.g., methylphenidate) are not
uncommon within the US jurisdiction. Moreover, PD end points
have been introduced for specific drug products along with tailored
acceptance criteria, for example, evaluation of the forced expiratory
volume in one second in the case of metered dose inhalers. These
product-specific amendments to the standard BE approach aim at
increasing the sensitivity of the in vivo test to detect formulation
differences and ensure therapeutic equivalence. In the following
sections, some detailed examples of additions/amendments to the
usual BE requirements are provided.
Partial AUC Approaches When ABE Metrics Are Insufficient to
Compare Drug Products
The regulatory approach of using systemic exposure metrics to
indirectly assess the luminal behavior of products is not straight-
forward because Cmax is not a pure measure of absorption rate and
systemic variability (i.e., drug related rather than formulation
related) also plays a major role in systemic drug exposure.61,71,74,75
In fact, concomitant plasma and GI fluid sampling in healthy adults
has revealed a lack of correlation between intraluminal and sys-
temic Cmax and AUC for immediate-release formulations containing
spironolactone, fosamprenavir, and simvastatin.76-78 Furthermore,
modified-release formulations containing budesonide, mesal-
amine, zolpidem, nifedipine, and methylphenidate illustrate that
satisfying the current BE criteria in terms of Cmax and AUC0-t does
not necessarily ensure similarity in the drug release characteristics;
these are better addressed by using partial AUC metrics derived
from the initial or later parts of the respective plasma profiles.79-86
Similarly, consideration of early exposure may be useful when a
rapid input is important to achieve an optimal efficacy profile, for
example, an analgesic effect.61,67,87
Group-by-Formulation InteractionsdRisks Associated With
Extrapolating BE Results From Healthy Adults to Special
Populations
Group-by-formulation effects have been reported for several
drugs in the literature. While on the one hand recent research
sponsored by the FDA demonstrated BE between a previously
approved generic lamotrigine and the brand name product, Lam-
ictal®, in “generic-brittle” patients with epilepsy who were already
taking lamotrigine,88 several clinical examples of subgroup-by-
formulation interactions have also been reported, for example, for
products containing methylphenidate, verapamil, or prasugrel.
Meyer et al.89 found that a T product containing methylpheni-
date was more variable than the respective R formulation and their
results further indicated a borderline subject-by-formulation
interaction, in line with the clinical observation that some
patients manifested a more rapid onset of action and a shorter
duration of drug response after taking the T formulation. As a
6 R. Cristofoletti et al. / Journal of Pharmaceutical Sciences xxx (2018) 1-12
result, the FDA recommended that BE between formulations
containing methylphenidate be based on Cmax, partial AUCs
and AUC0-∞ under fasting and fed state conditions. Similarly,
to ensure the switchability between Concerta® and generic
products, a subject-by-formulation test for each PK metric is now
recommended. The specific instructions in the Concerta® guideline
are illustrative of the trend to deviations from the traditional
one-size-fits-all criteria in applying average BE,86 noting that an
evaluation of the robustness of this particular guidance is outside
the scope of this commentary.
Carter et al.90 reported an age-based subject-by-formulation
interaction for an already approved generic verapamil immediate-
release dosage form. For one of the T formulations, the authors
observed average AUC0-t and Cmax values that were 43% and 77%
higher, respectively, in elderly than in young subjects, while these
values remained similar between the R and another T formulation in
both subgroups. In yet another case, the systemic exposure after the
administration of 2 different products containing a calcium channel
blocking agent was superimposable in male subjects but divergent
with respect to peak and extent of exposure in female subjects,
suggesting a gender-based subject-by-formulation interaction.91
Mechanistically, subject- or group-by-formulation interactions
can be viewed as a manifestation of a unique interaction among
characteristics of the anatomy, physiology, or biochemistry of the GI
tract (e.g., related to GI pH, gastric emptying rate, residence time in
the GI tract, and presystemic drug metabolism), or even lifestyle of
a specific patient/subpopulation, and the product dissolution
in vivo. In fact, some intrinsic or extrinsic factors that affect gastric
emptying and gastric pH have been reported to cause formulation-
dependent absorption, an effect that may not be observed in
traditional BE studies enrolling healthy young adults.92-98 For
example, Seiler et al.99 compared the systemic exposure of 2
different formulations containing prasugrel, a poorly soluble free
base (F1), and its hydrochloride salt (F2) in healthy adults with and
without proton pump inhibitor (PPI) pretreatment. Under normal
gastric acid secretion, the Cmax and AUC0-t ratios (F2/F1) were 1.2
and 1.1, respectively, whereas in PPI-pretreated volunteers, the
ratios were 2.2 and 1.4, respectively. The clinical relevance of these
findings persuaded the FDA to sponsor a project on the evaluation
of formulation dependence of drug interaction with PPIs, awarded
to Biopharma Services USA, Inc. in 2016.100
These examples suggest that the more heterogeneous the
sampled population of volunteers enrolled in fully replicated BE
studies, the higher the chances of detecting subject- or group-by-
formulation interactions. Bringing matters to a further level of
complexity, one should consider the possibility that effects from a
combination of factors could lead to different BE results. For
instance, while 2 formulations may be BE in healthy adults and in
elderly patients of Caucasian origin, this may not be the case in
Japanese subjects, where older patients are known to have a higher
incidence of achlorhydria.98
The statement made by the panel of experts in 1999 that
switchability is not a problem for approved drug products (see
section on population and individual BE) is clearly challenged by
the aforementioned examples. However, as permutations of
various conditions where a “formulation-by-condition” effect
might manifest itself are almost endless, it would be highly
impractical to recruit the whole range of possible patient subsets to
assess product performance over the entire range of physiologically
relevant conditions. An attractive and efficient alternative would be
to assess these permutations in relevant in vitro environments,57
with subsequent incorporation of the data into in silico models
(i.e., in vitro to in vivo extrapolation using physiologically relevant
PK models to investigate drug absorption in different virtual pop-
ulations (e.g., children, elderly, hypochlorhydric patients, bariatric
surgery patients and so forth). Indeed, virtual BE has already been
used by the Office of Generic Drugs at the FDA to set clinically
relevant specifications on the maximum amount of prasugrel free
base (i.e., 20%) in hydrochloride salt products to ensure BE of ge-
nerics in subjects that may be taking PPIs.101 By taking the lead in
this area, regulatory bodies, such as the FDA, will no doubt generate
even more interest in the use of modeling and simulation for
assessing potential group-by-formulation interactions.
Prescribable but Not Switchable?dBX Drug Products and
Hybrid Applications
Theoretically, a new drug product that has failed to demonstrate
BE to the innovator drug product can still be approved in the United
States (i.e., when coded as BX, meaning that the data reviewed by
the FDA are insufficient to determine that there is therapeutic
equivalence to the R formulation) or in Europe (i.e., hybrid appli-
cation based on article 10(3) of Directive 2001/83/EC), provided
claims of noninferior safety and efficacy profiles can be adequately
supported. Examples include the applications of Absorica® (iso-
tretinoin, NDA 021951), Triglide® (fenofibrate, NDA 021350), and
Sumavel® DosePro® (sumatriptan succinate, NDA 022239). All 3
drug products failed to show BE to the respective R formulation
under at least one of the required conditions: for Absorica®, it was
the fasted state PK; for Triglide®, it was the fed state PK; and for
Sumavel® DosePro®, it was when the injection site was the deltoid
muscle. Marketing authorizations in the US were granted for these
products, albeit coded as BX, based on favorable results in clinical
trials, or, subsequent to an adjustment to the SmPC (“a suboptimal
dose may be delivered when Sumavel® DosePro® is administered to
the arm, and therefore, the arm is not recommended as a site of
administration”).102
In the United States, some BX-rated products can be converted
in AB-rated drug products (e.g., DiaBeta, glyburide 5 mg, NDA
017532) and then become new R drug products (i.e., ABX, where
X ¼ 1, 2, 3…), to avoid shielding the former BX drug products from
direct generic competition.103,104 By contrast, in the EU, the current
legislation forbids the utilization of “hybrid” drug products as the R
product in subsequent BE studies to bridge safety and efficacy to
subsequent generic formulations.105 Examples include supra-
bioavailable products which have been approved and whose
dosage recommendations have been supported by clinical studies
or, alternatively, “hybrid” drug products which show a reduced food
effect.106 In summary, some drug products have been shown to be
safe and efficacious even though they were not pharmacokineti-
cally equivalent to the respective innovator's product under all
circumstances.
Can PD Information and Intrasubject Variability Be Used to
Set Clinically Relevant Acceptance Intervals for PK End Points?
During a meeting in 2004 of the Advisory Committee for Phar-
maceutical Science at the FDA, Prof. Les Benet declared the then
current BE guideline to be Procrustean.107 In Greek mythology,
Procrustes offered hospitality to travelers, who were invited to rest
on his special bed. Nevertheless, if the travelers were too tall or too
short to perfectly fit the bed, Procrustes would cut off their feet or
stretch them out. This is indeed a very good picture for the one-
size-fits-all criteria of BE guidelines. Benet emphasized that “the
BE requirement for each drug should be established based on
known measures of intrasubject variability and the PKPD rela-
tionship.”107,108 With the latter, it was highlighted that, in contrast
to the prevailing assumption that if 2 drug products have the same
PK this will result in the same PD, products with different PK can
also result in the same PD (depending where the administered dose
 R. Cristofoletti et al. / Journal of Pharmaceutical Sciences xxx (2018) 1-12
is on the E-R curve), or alternatively, products with the same sys-
temic exposure can produce different PD behavior (e.g., because of
formulation-dependent target site distribution).
Indeed, using PKPD information to set clinically relevant PK
limits would work analogously to replacing the classical toxico-
logical approach to estimate the “no observed adverse effect level”
by the benchmark dose approach, which involves fitting a mathe-
matical model to all the dose-response data within a study, and
thus enabling more biological information to be incorporated in the
resulting estimates of guidance values (e.g., acceptable daily
intakes).109
Perhaps, the most emblematic initiative in the field of BE is
the recognition of the necessity to integrate predictive dissolution,
physiologically relevant PK and PKPD models for decision-making
about generic drug BE standards, chosen as a priority for the fiscal
year 2018 based on inputs from the FDA, industry, and other
stakeholders.110 Indeed, this integrative approach has already been
applied to demonstrate that 2 pharmacokinetically inequivalent
formulations containing ibuprofen can nevertheless be therapeu-
tically equivalent.97,111 On the flip side, given the steep E-R rela-
tionship for efficacy end points of dabigatran and rivaroxaban, the
FDA took a more pragmatic approach and denied widening the
acceptance interval, although both are HVDs.112-115
Narrowing the Acceptance Interval Because of Considerations of
Safety
As the overall intrasubject variability is usually small for NTIs,
generic formulations of them can be approved in BE studies even if
the mean differences between T and R formulations are close to the
acceptance limits, leading to concerns about the transitivity of BE
testing (i.e., potential to drift to nonequivalence among generic
productsdthe NTI paradox).116 An additional concern arises from
the overlap between the exposure needed for efficacy and that
leading to unacceptable toxicity, so that a minor change in exposure
could lead to the patient experiencing unacceptable adverse drug
reactions. To address these issues, regulators have amended the 80/
125 rule by setting product-specific criteria. Mostly, this has
resulted in a tightening of the BE acceptance interval: for NTIs to
90%-111% (or similar limits) in Europe (for AUC and also for Cmax if
the latter is of importance for safety, efficacy, or drug level moni-
toring); in Canada (only for AUC); and in Japan (for AUC and Cmax).
According to the deductivism of Popper,117 the BE paradigm should
have been challenged in a more fundamental way by the NTI case;
instead, a “quick fix” approach has been applied to maintain the BE
status quo.
In contrast to other regulatory authorities, the FDA uses a
reference-scaled average BE (RSABE) approach to scale down the BE
limits for Cmax and AUC to the within-subject variability of the R
product (but only if it is <20%); in other cases, the 80/125 rule
should be applied.118 Even though such initiatives are welcome,
there are still some remaining drawbacks: (a) the lack of worldwide
harmonization,119 (b) the lack of a precise and broadly applicable
definition of NTI; and (c) the lack of consistent utilization of
available drug-specific PD/toxicity information, including PD vari-
ability, to set clinically relevant PK limits.
Widening the Acceptance Interval Because of Consideration of
Variability
In the other direction, the original BE paradigm has also been
extended to accommodate HVDs, that is, those with an intrasubject
coefficient of variation >30% (a cutoff value which was set by the
participants during the Bio-International '92 conference held in Bad
Homburg, Germany).47 The amplitude of the BE goalposts was
7
scaled as a function of the R drug product variability, to keep the
sample size needed for in vivo studies within a range that would be
feasible for the sponsors of the generic application to conduct.118
Under the assumption that an approved HVD must have a wide
therapeutic index, as otherwise there would have been significant
safety issues and lack of efficacy during phase III trials, the FDA
focuses only on the within-subject variability of the R formulation
to apply the RSABE approach to Cmax and AUC.120
On the other hand, the EMA only allows the use of the RSABE
method to widen Cmax acceptance range for HVD if there are no
clinically relevant implications, that is, decisions are made on a
case-by-case basis.33,106 For example, even though the EMA Phar-
macokinetics Working Party has acknowledged there are “some
data suggesting the existence of a plateau response in the inhibition
of platelet aggregation” for clopidogrel, which is a HVD, they
concluded that “it is currently not entirely clear what would be the
influence of variable clopidogrel concentrations on PD” and thus
“the widening of 90% confidence intervals for Cmax is not recom-
mended.”106 Extensive reviews on the statistical methods behind
different RSABE approaches are provided elsewhere.118,121 It is also
noted that the specifics of the RSABE method applied vary between
regulatory agencies and are in need of harmonization.
What Approach Should Be Taken When the Drug Is Dosed High on
the E-R Relationship?
Some authors have reported that it is not uncommon to have
phase III trials carried out using doses near to the maximum
tolerated doses derived from phase I trials to maximize the chances
of demonstrating drug effectiveness.122,123 Consequently, it is not
surprising to have approved therapeutic doses at or near the
plateau phase of E-RE-R curves, where the law of “diminishing
returns” plays a major role, for example, if the exposure is already
greater than the concentration producing 80% of drug effect (EC80),
a 90-fold change in PK will result in only a 9% change in effect.124 In
this scenario, using the traditional PK-based BE limits (80%-125%) to
indirectly assess therapeutic equivalence isdat first glanceda
rather conservative approach. However, widening the PK-based BE
limits using prior knowledge on E-R relationship may come with a
caveat: focusing on E-R curve for just one biomarker or surrogate
may not ensure similar efficacy and safety in all indications. For
example, fluctuations in PK which do not change drug efficacy may
still trigger safety concerns. Two different formulations, immedi-
ate- and modified-release products containing tacrolimus, an NTI
drug, were equivalent in terms of effectiveness in Caucasians but
turned out to be inequivalent with regard to safety in African-
American population.125 Therefore, care should be exercised,
especially in case of drugs prone to unacceptable adverse effects.
On the other hand, for well-characterized, wide therapeutic
index drugs, the available PKPD information may enable identifi-
cation of the most sensitive biomarker or surrogate end point to PK
changes (e.g., steepest E-R curve). In this case, the E-R curve could
be used to calibrate the PK-based BE limits (i.e., widening or nar-
rowing the acceptance limits), allowing generalization of PK-based
decisions on therapeutic equivalence and thus covering all in-
dications. Such a scenario could be applied to some nonsteroidal
anti-inflammatory drugs whose EC50 for different clinical outcomes
is well established, for example, if the EC50 for pain relief > EC50 for
antipyresis, when a wider PK fluctuation is not clinically relevant in
terms of pain relief it will not be relevant in terms of antipyresis
either.
In this context, the language to modernize the BE approach, at
least for new drug applications (NDAs), is already in place within
the U.S. jurisdiction. For example, the FDA emphasizes that
“exposure-response data can also support a new formulation that is
8 R. Cristofoletti et al. / Journal of Pharmaceutical Sciences xxx (2018) 1-12
unintentionally pharmacokinetically different from the formula-
tion used in the clinical trials… Rather than reformulating the
product or repeating the bioequivalence study, a sponsor may be
able to support the view that use of a wider confidence interval or
accepting a real difference in bioavailability or exposure would not
lead to a therapeutic difference.”126 Recently, the FDA reinforced its
position in a specific draft guidance on BA and BE for NDAs,
encouraging sponsors to “demonstrate that the differences in rate
and extent of absorption do not significantly affect the safety and
efficacy based on available dose-response or concentration-
response data.”120
An example illustrating the application of this policy by the FDA
is aliskiren. Quoting from the Web site ipsis litteris: “In the bio-
equivalence study (CSAH100A2102), the 90% CI for Cmax of aliskiren
(0.76,0.93) was not contained within the predetermined BE limits
of 0.8 e 1.25. However, this is not clinically significant, because of a
shallow exposure-response relationship.” Also noteworthy is that,
despite the massive negative food effect observed for aliskiren (i.e.,
mean AUC and Cmax are decreased by 71% and 85%, respectively),
the drug can be taken regardless of food intake, clearly suggesting
that the therapeutic dose is on the plateau of the E-R curve.127,128 It
is interesting to contemplate what ramifications there would have
been, had the food effect been to increase Cmax and AUC to the same
extent. In such a case, one would also have to consider the safety
aspects. Even if efficacy were to remain the same (i.e., dosing on the
plateau of the exposure-efficacy response curve), this may not
necessarily also be the case for the exposureetoxicity response
relationship. In summary, given that BE is meant to cover both ef-
ficacy and safety, analysis based on E-R relationships should focus
on whichever PD or clinical end point, including adverse events, is
most sensitive to PK changes.
Exceptions to the “golden rule” of average BE can also be found
in the realm of generics. For example, the United Kingdom Medi-
cines and Healthcare products Regulatory Agency approved the
registration of 3 generic products containing losartan even though
the 90% CIs for Cmax of the parent compound fell outside the BE
acceptance limits. In this case, the regulators took a pragmatic
approach and based their decisions on available evidence which
indicated that losartan Cmax has only a minor influence on its
clinical efficacy.129-131 Whether this represents a one-time initiative
of the British regulators or heralds a new regulatory trend is still
unclear, but such flexibility might be seen in the light of modern,
patient-centric approaches such as the Quality by Design and the
Biopharmaceutics Risk Assessment Roadmap,132,133 both of which
advocate departure from a theoretical, oversimplified one-size-fits-
all BE regulatory criterion in favor of product-specific, clinically
relevant specifications.
What If the Link Between Systemic Exposure and Response Is
Indirect?dSlow or Saturable Target Site Distribution Kinetics
Translating PK changes/differences under nonsteady state con-
ditions (e.g., single-dose BE studies) into clinical response during
chronic therapy may not be straightforward if the measured
concentration in plasma is not directly linked to the effect site
concentration. Indeed, 26 years ago, Lewis Sheiner had already
raised the question of “what should be measured in BE studies?,”
pointing out that the delay between the time course of plasma
concentration and drug effect should be taken into account to guide
the decision.37
Moreover, for drugs acting at intracellular targets protected by
biological barriers, the permeability into the cell can be the rate-
limiting step to the effect. In this context, saturable transport to
the site of action has been reported to cause concentration-
dependent hysteresis for alphaxalone in rats and capacity-limited
tissue distribution of methicillin in rabbits.134-136 In other words,
in case of a saturable cellular uptake process, nonequivalent
systemic exposure may still be translated into similar target site
distribution and hence therapeutic effectiveness.
Stepping away from oral dosage forms for a moment,
formulation-dependent target site distribution has already been
reported in both preclinical and clinical situations for parenteral
products. For example, 2 pegylated liposomal doxorubicin formu-
lations showed similar plasma PK profiles (in terms of total drug
concentration) but distinctly different tissue distribution in mice.137
Similarly, the liver-to-plasma AUC ratio was about 2-fold higher in
mice after administering a paclitaxel formulation made without
polyoxyethylated castor oil (Abraxane®) than with poly-
oxyethylated castor oil (Taxol®).138 Translating these results to
humans, it has been shown that polyoxyethylated castor oil causes
a profound alteration of paclitaxel accumulation in erythrocytes by
reducing the free drug fraction available for cellular partitioning,
due to drug trapping in micelle-like structures. The authors
concluded that changing the primary drug carrier in the systemic
circulation from the erythrocytes to micellar structures has a
substantial impact on how paclitaxel PK in humans should be
interpreted.139 Consequently, quantification of total drug concen-
tration in plasma was deemed to be inadequate for assessing
therapeutic equivalence. Instead, in cases where paclitaxel is
administered parenterally as a complex drug delivery system, the
standard proof for therapeutic equivalence requires not only that T
and R formulations have compositions that are qualitatively (Q1)
and quantitatively (Q2) the same, but also a demonstration of
in vivo BE for both unbound and total paclitaxel, as well as
demonstrating in vitro similarity in terms of particle size distribu-
tion using a population BE approach.140 An even more complex
product-specific BE criteria has been set for pegylated liposomal
doxorubicin hydrochloride, involving demonstration of in vivo BE
for both free and liposome encapsulated doxorubicin, in vitro
population BE for liposome size distribution, and further, stringent
demonstration of equivalent liposome characteristics.141
 -vis
Best Practice Evaluation of BEdClinical Aspects vis-a
Biopharmaceutical Aspects of BE
Overall, the regulatory authorities tend to prioritize one of the
BE aspects in lieu of the other depending on the type of the sub-
mission. BE in the world of NDAs focuses on the clinical aspect of
the concept, and regulators tend to be flexible with regard to
acceptance criteria, whereas in the realm of generics, the bio-
pharmaceutical aspects are prioritized, and these tend to be fol-
lowed more rigorously.142 Nevertheless, a new regulatory paradigm
envisioned by the FDA, focusing on applying model-informed drug
development approaches to both brand and generic drug products,
may be useful to extract the best from both BE aspects in a product-
specific manner. In this context, the workshop entitled “Leveraging
Quantitative Methods and Modeling to Modernize Generic Drug
Development and Review,” held by the FDA on October 2-3, 2017, is
an important step forward. While the classical one-size-fits-all
approach may well prove to be most suitable for many generics,
for others, including some of the examples discussed herein, such
as complex active ingredients, complex formulations, complex
routes of delivery, or complex drug device combinations, using a
Bayesian-like approach (in a conceptual sense rather than insisting
on its mathematical formalism), considering all relevant prior
knowledge to guide the definition of the test system (in vitro, on a
stand-alone basis or in combination with in silico tools, in vivo or
even a combination of both systems), including optimized experi-
mental design, metrics, acceptance limits, and so forth, in a
 R. Cristofoletti et al. / Journal of Pharmaceutical Sciences xxx (2018) 1-12
product-specific manner, should lead to better, more clinically
relevant BE decisions.
Forecasting the Future of the BE Paradigm Through a Pair of
Kuhnian Glasses
It has been demonstrated that the seminal work by Thomas
Kuhn on scientific revolutions and paradigm shifts can be applied
to various branches of natural sciences: physics, chemistry, and
biology.38,143-145 In this section, we scrutinize the evolution of BE
criteria in the light of Thomas Kuhn's theory of scientific revolution.
Importantly, one should always keep in mind that demonstration of
equivalence, in either the realm of NDAs or amended new drug
applications, is not a scientific fact per se but rather a judgement,
which (a) is based on facts (e.g., geometric mean ratios for Cmax and
AUC as well as intrasubject variability); (b) also relies on assump-
tions (e.g., setting of the acceptance limits at 80/125 on the
assumption that this would encompass the range of clinical simi-
larity); and (c) is prone to bias (e.g., status quo bias, heuristic bias,
hindsight bias, and so forth). Thus, decision-makers should always
ask “on what basis do we make these judgments?”
Since the early 1980s, when a consensus about using the one-
size-fits-all approach to compare product performances and
decide about therapeutic equivalence was reached, deviations from
the standard paradigm (i.e., the so-called “anomalies” using Kuh-
nian vocabulary) have been accumulating, as demonstrated by the
increasing numbers of drug products for which there are product-
specific recommendations for BE testing. The heretofore presented
risks of incurring false-positive results, for example, when failing to
extrapolate BE results from healthy adults to special populations, or
when Cmax and AUC do not necessarily ensure similarity in the
in vivo drug release characteristics, as well as the risks of false-
negative results by rejecting pharmacokinetically inequivalent
(but still having therapeutically equivalence) formulations, suggest
that at least some anomalies critically impact the main assump-
tions behind the biopharmaceutical and clinical aspects of the one-
size-fits-all BE paradigm. In the context of the Kuhnian lens, the
paradigm of average BE appears to be embedded in a crisis, and it
will be necessary to propose alternative pathways for ensuring
therapeutic equivalence across the entire range of generic products.
Indeed, there is already a movement toward applying systems
thinking in drug discovery, moving away from a reductionist, target-
centric approach of drug development into a new holistic systems
approach, for example, systems pharmacology.146 Under this
context, integrative approaches such as the model-informed drug
development may be useful to explore new clinically relevant,
product-specific BE recommendations in lieu of using the standard
one-size-fits-all approach, in cases where the latter is not adequate.
An important part of this process will be close communication by
regulators internationally to ensure that drug (product)-specific BE
requirements are harmonized across the globe. Efforts such as the
European Federation for Pharmaceutical Sciences (EUFEPS) Global
Bioequivalence Harmonization Initiative are already making
important contributions in this area.147
Acknowledgments
This article reflects the scientific opinion of the authors and not
necessarily the policies of the Brazilian Health Regulatory Agency
(Anvisa).
References
1. Seife M. Evolution of the principal US food & drug laws. In: Sharma KN,
Sharma KK, Sen P, eds. Generic drugs, bioequivalence and pharmacokinetics:
9
Proceedings of the Indo-US Symposium. New Delhi: University College of
Medical Sciences; 1990:7-14.
2. Kelsey FO. Thalidomide update: regulatory aspects. Teratology. 1988;38:221-
226.
3. Skelly JP. Bioavailability policies and guidelines. In: Martin A, Doluisio JT, eds.
Industrial bioavailability and pharmacokinetics: guidelines, regulations and
controls. College of pharmacy, drug dynamics institute. Austin, TX: University of
Texas; 1977:2-41.
4. Skelly JP. A history of biopharmaceutics in the Food and Drug Administration
1968-1993. AAPS J. 2010;12(1):44-50.
5. Abood RR. Pharmacy Practice and Law. 7 ed. Burlington, MA: Jones & Bartlett
Learning; 2014.
6. Carpenter D, Tobbell DA. Bioequivalence: the regulatory career of a pharma-
ceutical concept. Bull Hist Med. 2011;85(1):93-131.
7. Kesselheim AS, Darrow JJ. Hatch-Waxman Turns 30: Do We Need a Re-
Designed Approach for the Modern Era? Yale J Health Pol Law Ethics.
2015;15(2). Available at: http://digitalcommons.law.yale.edu/yjhple/vol15/
iss2/2. Accessed January 10, 2017.
8. Skelly JP. Bioavailability and bioequivalence. J Clin Pharmacol. 1976;16:539-
545.
9. Mossinghoff GJ. Overview of the Hatch-Waxman Act and its impact on the
drug development process. Food Drug Law J. 1999;54(2):187-194.
10. Manninen V, Melin J, Ha €rtel G. Serum-digoxin concentrations during treat-
ment with different preparations. Lancet. 1971;298:934-935.
11. Lindenbaum J, Mellow MH, Blackstone MO, Butler Jr VP. Variation in
biologic availability of digoxin from four preparations. N Engl J Med. 1971;285:
1344-1347.
12. Levy G. Bioavailability, clinical effectiveness, and the public interest. Phar-
macology. 1972;8(1):33-43.
13. Wagner JG, Christensen M, Sakmar E, et al. Equivalence lack in digoxin plasma
levels. JAMA. 1973;224:199-204.
14. Barnett DB, Smith RN, Greenwood ND, Hetherington C. Bioavailability of
commercial tetracycline products. Br J Clin Pharmacol. 1974;1:319-323.
15. Barr WH, Gerbracht LM, Letcher K, Plaut M, Strahl N. Assessment of the
biologic availability of tetracycline products in man. Clin Pharmacol Ther.
1972;13:97-108.
16. Glazko AJ, Kinkel AW, Alegnani WC, Holmes EL. An evaluation of the
absorption characteristics of different chloramphenicol preparations in
normal human subjects. Clin Pharmacol Ther. 1968;9:472-483.
17. Chiou WL. Determination of physiologic availability of commercial phenyl-
butazone preparations. J Clin Pharmacol New Drugs. 1972;12:296-300.
18. Van Petten GR, Feng H, Withey RJ, Lettau HF. The physiologic availability
of solid dosage forms of phenylbutazone. I. In vivo physiologic availability
and pharmacologic considerations. J Clin Pharmacol New Drugs. 1971;11:
177-186.
19. Barber HE, Calvey TN, Muir K, Hart A. Biological availability and in vitro
dissolution of oxytetracycline dihydrate tablets. Br J Clin Pharmacol. 1974;1:
405-408.
20. Levy G, Gibaldi M. Editorial: bioavailability of drugs: focus on digoxin. Cir-
culation. 1974;49(3):391-394.
21. Office of Technology Assessment (OTA). Drug bioequivalence. Recommenda-
tions from the Drug Bioequivalence Study Panel to the Office of Technology
Assessment, Congress of the United States. J Pharmacokinet Biopharm. 1974;2:
433-466.
22. European Union Law. Council Directive 65/65/EEC on the Approximation of
Provisions Laid Down by Law, Regulation or Administrative Action Relating to
Proprietary Medicinal Products. EUR-Lex; 1965. Available at: http://eur-lex.
europa.eu/legal-content/EN/TXT/?uri¼CELEX:31965L0065. Accessed January
30, 2018.
23. The European Parliament and the Council of the European Union. Regulation (EC)
No 726/2004. EUR-Lex; 2004. Available at: http://eur-lex.europa.eu/LexUriServ/
LexUriServ.do?uri¼OJ:L:2004:136:0001:0033:en:PDF. Accessed February 7,
2018.
24. World Health Organization. Drugs and MoneyePrices, Affordability and Cost
Containment. Geneva, Switzerland: WHO; 2003. Available at: http://apps.
who.int/medicinedocs/en/d/Js4912e/3.6.html. Accessed February 7, 2018.
25. Eckstein N, Haas B, Ro €per L. Arzneimittelzulassung in besonderen Fa €llen.
Deutsche Apotheker Zeitung; 2013. Available at: https://www.deutsche-
apotheker-zeitung.de/daz-az/2013/daz-22-2013/arzneimittelzulassung-in-
besonderen-faellen. Accessed January 30, 2018.
26. Bundesinstitut für arzneimittel und medizinprodukte (BfArM). Bibliografische
Zulassung. BfArM; 2011. Available at: https://www.bfarm.de/DE/Arzneimittel/
Arzneimittelzulassung/Zulassungsarten/BibliografischeZulassung/_node.html.
Accessed January 31, 2017.
27. Kuribayashi R, Matsuhama M, Mikami K. Regulation of generic drugs in Japan:
the current situation and future prospects. AAPS J. 2015;17(5):1312-1316.
28. Zargarzadeh AH, Emami MH, Hosseini F. Drug-related hospital admissions in
a generic pharmaceutical system. Clin Exp Pharmacol Physiol. 2007;34(5-6):
494-498.
29. Yousefi N, Mehralian G, Peiravian F, NourMohammadi S. Consumers'
perception of generic substitution in Iran. Int J Clin Pharm. 2015;37(3):497-
503.
30. Khoonsari H, Oghazian MB, Kargar M, et al. Comparing the efficacy of 8 weeks
treatment of Cipram® and its generic citalopram in patients with mixed
anxiety-depressive disorder. Iran J Psychiatry Behav Sci. 2015;9(2):e230.
10 R. Cristofoletti et al. / Journal of Pharmaceutical Sciences xxx (2018) 1-12
31. Tabatabaei O, Heshmat R, Taheri F, Larijani B. Comparison of a generic and a
brand metformin products in type II diabetes: a double blind randomized
clinical trial study. Daru J. 2007;15(2):113-117.
32. Gibaldi M, Levy G. Pharmacokinetics in clinical practice. I. Concepts. JAMA.
1976;235(17):1864-1867.
33. European Medicines Agency. Guideline on the Investigation of Bioequivalence.
London, UK: EMA; 2010. Available at: http://www.ema.europa.eu/docs/en_GB/
document_library/Scientific_guideline/2010/01/WC500070039.pdf. Accessed
February 15, 2018.
34. U.S. Food and Drug Administration. Waiver of In Vivo Bioavailability and
Bioequivalence Studies for Immediate-Release Solid Oral Dosage Forms Based
on a Biopharmaceutics Classification System. Silver Spring, MD: FDA; 2015.
Available at: https://www.fda.gov/downloads/Drugs/Guidances/ucm070246.
pdf. Accessed February 15, 2018.
35. Brodie BB, Heller WM. Bioavailability of Drugs: Proceedings. Basel, Switzerland:
S. Karger; 1971.
36. Yu A, Sun D, Li BV, Yu LX. Bioequivalence history. In: Li BV, Yu LX, eds. FDA
Bioequivalence standards. New York, NY: Springer; 2014:1-27.
37. Sheiner LB. Bioequivalence revisited. Stat Med. 1992;11(13):1777-1788.
38. Kuhn TS. The Structure of Scientific Revolutions. 3rd ed. Chicago, IL: University
of Chicago Press; 1996.
39. Dobbins TW, Thiyagarajan B. A retrospective assessment of the 75/75 rule in
bioequivalence. Stat Med. 1992;11(10):1333-1342.
40. Haynes JD. Statistical simulation study of new proposed uniformity require-
ment for bioequivalency studies. J Pharm Sci. 1981;70(6):673-675.
41. Haynes JD. FDA 75/75 rule: a response. J Pharm Sci. 1983;72(1):99-100.
42. Schuirmann DJ. A comparison of the two one-sided tests procedure and the
power approach for assessing the equivalence of average bioavailability.
J Pharmacokinet Biopharm. 1987;15(6):657-680.
43. Kirkwood BL, Westlake WJ. Bioequivalence testingea need to rethink. Bio-
metrics. 1981;37(3):589-594.
44. Fisher RA. The Design of Experiments. 9th ed. New York, NY: Macmillan Pub-
lishing Co.; 1971.
45. Anderson S, Hauck WW. A new procedure for testing equivalence in
comparative bioavailability and other clinical trials. Comm Stat. 1983;12(23):
2663-2692.
46. McGilveray IJ, Midha KK, Skelly JP, et al. Consensus report from “Bio Inter-
national '89”: issues in the evaluation of bioavailability data. J Pharm Sci.
1990;79(10):945-946.
47. Blume HH, Midha KK. Bio-International 92, conference on bioavailability,
bioequivalence, and pharmacokinetic studies. J Pharm Sci. 1993;82(11):1186-
1189.
48. Blume HH, McGilveray IJ, Midha KK. BIO-international '94 Conference on
Bioavailability, Bioequivalence and Pharmacokinetic Studies and Pre-
Conference Satellite on ‘In Vivo/In Vitro Correlation.’ Munich, Germany,
June 14-17, 1994. Eur J Drug Metab Pharmacokinet. 1995;20(1):3-13.
49. Hwang S, Huber PB, Hesney M, Kwan KC. Bioequivalence and interchange-
ability. J Pharm Sci. 1978;67:IV.
50. U.S. Food and Drug Administration. Average, Population, and Individual
Approaches to Establishing Bioequivalence. Silver Spring, MD: FDA; 1999.
Available at: http://www.fda.gov/OHRMS/DOCKETS/98fr/3657gd1.pdf.
Accessed January 10, 2017.
51. Hauck WW, Hyslop T, Chen ML, Patnaik R, Williams RL. Subject-by-formula-
tion interaction in bioequivalence: conceptual and statistical issues. FDA
Population/Individual Bioequivalence Working Group. Food and Drug
Administration. Pharm Res. 2000;17(4):375-380.
52. Chen ML, Lesko LJ. Individual bioequivalence revisited. Clin Pharmacokinet.
2001;40(10):701-706.
53. Chow SC. Bioavailability and bioequivalence in drug development. Rev Com-
put Stat. 2014;6(4):304-312.
54. Anderson S, Hauck WW. Consideration of individual bioequivalence.
J Pharmacokinet Biopharm. 1990;18(3):259-273.
55. Hauck WW, Anderson S. Types of bioequivalence and related statistical con-
siderations. Int J Clin Pharmacol Ther Toxicol. 1992;30(5):181-187.
56. Hauck WW, Anderson S. Measuring switchability and prescribability: when is
average bioequivalence sufficient? J Pharmacokinet Biopharm. 1994;22(6):
551-564.
57. Levy G. The clay feet of bioequivalence testing. J Pharm Pharmacol.
1995;47(12A):975-977.
58. Benet LZ. Individual bioequivalence: have the opinions of the scientific
community changed? Silver Spring, MD: 1999. Available at: http://www.fda.
gov/ohrms/dockets/ac/01/slides/3804s2_09_benet.ppt. Accessed January 10,
2017.
59. Lennerna €s H. Human jejunal effective permeability and its correlation with
preclinical drug absorption models. J Pharm Pharmacol. 1997;49(7):627-638.
60. Chen ML. Rate of absorption in bioequivalence studies: replace with the
concept of exposure? In: Midha K, Nagai T, eds. Bioavailability, Bioequivalence
and Pharmacokinetic Studies. Tokyo, Japan: Bio-International; 1996:191-195.
61. Chen ML, Lesko L, Williams RL. Measures of exposure versus measures of rate
and extent of absorption. Clin Pharmacokinet. 2001;40(8):565-572.
62. European Medicines Agency. Note for guidance on investigational bioavailability
and bioequivalence. London, UK: EMA; 2000. Available at: http://www.ema.
europa.eu/docs/en_GB/document_library/Scientific_guideline/2009/09/WC5
00003519.pdf. Accessed January 30, 2018.
63. Bois FY, Tozer TN, Hauck WW, et al. Bioequivalence: performance of several
measures of rate of absorption. Pharm Res. 1994;11:966-974.
64. Aarons L. Assessment of rate of absorption in bioequivalence studies. J Pharm
Sci. 1987;76:853-855.
65. Endrenyi L, Fritsch S, Yan W. Cmax/AUC is a clearer measure than Cmax for
absorption rates in investigations of bioequivalence. Int J Clin Pharmacol Ther
Toxicol. 1991;29:394-399.
66. Schall R, Luus HE. Comparison of absorption rates in bioequivalence studies of
immediate release drug formulations. Int J Clin Pharmacol Ther Toxicol.
1992;30:153-159.
67. Chen ML. An alternative approach for assessment of rate of absorption in
bioequivalence studies. Pharm Res. 1992;9(11):1380-1385.
68. Lacey LF, Keene ON, Duquesnoy C, et al. Evaluation of different indirect
measures of rate of drug absorption in comparative pharmacokinetic studies.
J Pharm Sci. 1994;83:212-215.
69. Reppas C, Lacey LF, Keene ON, et al. Evaluation of different metrics as indirect
measures of rate of drug absorption from extended release dosage forms at
steady-state. Pharm Res. 1995;1995(12):103-107.
70. Endrenyi L, Al-Shaikh P. Sensitive and specific determination of the equiva-
lence of absorption rates. Pharm Res. 1995;12:1856-1864.
71. Rostami-Hodjegan A, Jackson PR, Tucker GT. Sensitivity of indirect metrics for
assessing ‘rate’ in bioequivalence studiesemoving the ‘goalposts’ or changing
the ‘game.’ J Pharm Sci. 1994;83:1554-1557.
72. Amidon GL, Lennern€ as H, Shah VP, Crison JR. A theoretical basis for a bio-
pharmaceutic drug classification: the correlation of in vitro drug product
dissolution and in vivo bioavailability. Pharm Res. 1995;12(3):413-420.
73. World Health Organization (WHO). Multisource (generic) pharmaceutical
products: guidelines on registration requirements to establish interchange-
ability. Geneva, Switzerland: WHO; 2017. Available at: http://apps.who.int/
medicinedocs/documents/s23245en/s23245en.pdf. Accessed January 30,
2018.
74. Polli JE. In vitro studies are sometimes better than conventional human
pharmacokinetic in vivo studies in assessing bioequivalence of immediate-
release solid oral dosage forms. AAPS J. 2008;10(2):289-299.
75. Davit BM, Conner DP, Fabian-Fritsch B, et al. Highly variable drugs: observa-
tions from bioequivalence data submitted to the FDA for new generic drug
applications. AAPS J. 2008;10(1):148-156.
76. Bønløkke L, Hovgaard L, Kristensen HG, Knutson L, Lennern€ as H. Direct esti-
mation of the in vivo dissolution of spironolactone, in two particle size ranges,
using the single-pass perfusion technique (Loc-I-Gut) in humans. Eur J Pharm
Sci. 2001;12(3):239-250.
77. Brouwers J, Augustijns P. Resolving intraluminal drug and formulation
behavior: gastrointestinal concentration profiling in humans. Eur J Pharm Sci.
2014;30(61):2-10.
78. Geboers S, Stappaerts J, Tack J, Annaert P, Augustijns P. In vitro and in vivo
investigation of the gastrointestinal behavior of simvastatin. Int J Pharm.
2016;510(1):296-303.
79. Endrenyi L, Tothfalusi L. Do regulatory bioequivalence requirements
adequately reflect the therapeutic equivalence of modified-release drug
products? J Pharm Pharm Sci. 2010;13(1):107-113.
80. Lionberger RA, Raw AS, Kim SH, Zhang X, Yu LX. Use of partial AUC to
demonstrate bioequivalence of Zolpidem Tartrate Extended Release formu-
lations. Pharm Res. 2012;29(4):1110-1120.
81. Nicholls A, Harris-Collazo R, Huang M, Hardiman Y, Jones R, Moro L.
Bioavailability profile of Uceris MMX extended-release tablets compared with
Entocort EC capsules in healthy volunteers. J Int Med Res. 2013;41(2):386-394.
82. Yu A, Baker JR, Fioritto AF, et al. Measurement of in vivo gastrointestinal
release and dissolution of three locally acting mesalamine formulations in
regions of the human gastrointestinal tract. Mol Pharm. 2017;14(2):345-358.
83. U.S. Food and Drug Administration. Draft Guidance on Budesonide. Silver
Spring, MD: FDA; 2017. Available at: http://www.fda.gov/downloads/drugs/
guidancecomplianceregulatoryinformation/guidances/ucm426317.pdf. Accessed
January 10, 2017.
84. U.S. Food and Drug Administration. Draft Guidance on Nifedipine. Silver
Spring, MD: FDA; 2017. Available at: http://www.fda.gov/downloads/Drugs/
GuidanceComplianceRegulatoryInformation/Guidances/ucm118368.pdf.
Accessed January 10, 2017.
85. U.S. Food and Drug Administration. Draft Guidance on Zolpidem. Silver
Spring, MD: FDA; 2017. Available at: http://www.fda.gov/downloads/drugs/
guidancecomplianceregulatoryinformation/guidances/ucm175029.pdf.
Accessed January 10, 2017.
86. U.S. Food and Drug Administration. Draft Guidance on Methylphenidate.
Silver Spring, MD: FDA; 2017. Available at: http://www.fda.gov/downloads/
Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM320007.
pdf. Accessed January 10, 2017.
87. Cristofoletti R, Dressman JB. Bridging the gap between in vitro dissolution and
the time course of ibuprofen-mediating pain relief. J Pharm Sci. 2016;105(12):
3658-3667.
88. Ting TY, Jiang W, Lionberger R, et al. Generic lamotrigine versus brand-name
Lamictal bioequivalence in patients with epilepsy: a field test of the FDA
bioequivalence standard. Epilepsia. 2015;56(9):1415-1424.
89. Meyer MC, Straughn AB, Jarvi EJ, et al. Bioequivalence of methylphenidate
immediate-release tablets using a replicated study design to characterize
intrasubject variability. Pharm Res. 2000;17(4):381-384.
 R. Cristofoletti et al. / Journal of Pharmaceutical Sciences xxx (2018) 1-12
90. Carter BL, Noyes MA, Demmler RW. Differences in serum concentration of
and responses to generic verapamil in the elderly. Pharmacotherapy. 1993;13:
359-368.
91. Lesko LJ. Mechanistic aspects: the subject-by-formulation interaction. Amer-
ican Association of Pharmaceutical Scientists (AAPS) Workshop, ‘Scientific and
regulatory issues in product quality: narrow therapeutic index drugs and
individual bioequivalence’; 1998 Mar 16-18; Arlington, VA.
92. Aghazadeh-Habashi A, Jamali F. Pharmacokinetics of meloxicam administered
as regular and fast dissolving formulations to the rat: influence of gastroin-
testinal dysfunction on the relative bioavailability of two formulations. Eur J
Pharm Biopharm. 2008;70(3):889-894.
93. Almukainzi M, Jamali F, Aghazadeh-Habashi A, Lo € benberg R. Disease specific
modeling: simulation of the pharmacokinetics of meloxicam and ibuprofen in
disease state vs. healthy conditions. Eur J Pharm Biopharm. 2016;100:77-84.
94. Mitra A, Kesisoglou F, Beauchamp M, Zhu W, Chiti F, Wu Y. Using absorption
simulation and gastric pH modulated dog model for formulation development
to overcome achlorhydria effect. Mol Pharm. 2011;8(6):2216-2223.
95. Dreyfuss D, Shader R, Harmatz JS, Greenblatt DJ. Kinetics and dynamics of
single doses of oxazepam in the elderly: implications of absorption rate. J Clin
Psychiatry. 1986;47:511-514.
96. Badawy SI, Gray DB, Zhao F, Sun D, Schuster AE, Hussain MA. Formulation of
solid dosage forms to overcome gastric pH interaction of the factor Xa in-
hibitor, BMS-561389. Pharm Res. 2006;23(5):989-996.
97. Cristofoletti R, Patel N, Dressman JB. Assessment of bioequivalence of weak
base formulations under various dosing conditions using PBPK simulations in
virtual populations. Case Examples: ketoconazole and posaconazole. J Pharm
Sci. 2017;106(2):560-569.
98. Doki K, Darwich AS, Patel N, Rostami-Hodjegan A. Virtual bioequivalence for
achlorhydric subjects: the use of PBPK modelling to assess the formulation-
dependent effect of achlorhydria. Eur J Pharm Sci. 2017;109:111-120.
99. Seiler D, Doser K, Salem I. Relative bioavailability of prasugrel free base in
comparison to prasugrel hydrochloride in the presence and in the absence of
a proton pump inhibitor. Arzneimittelforschung. 2011;61(4):247-251.
100. U.S. Food and Drug Administration. FY 2016 Awarded GDUFA Regulatory
Research Contracts and Grants. Silver Spring, MD: FDA; 2016. Available at:
https://www.fda.gov/downloads/ForIndustry/UserFees/GenericDrugUserFees/
UCM526210.pdf. Accessed January 31, 2018.
101. Fan J, Zhang X, Zhao L. Utility of physiologically based pharmacokinetic ab-
sorption modeling to predict the impact of salt-to-base conversion on Pra-
sugrel HCl product bioequivalence in the presence of proton pump inhibitors.
AAPS J. 2017;19(5):1479-1486.
102. U.S. Food and Drug Administration. Drugs@FDA: FDA Approved Drug Prod-
ucts. Silver Spring, MD: FDA; 2017. Available at: http://www.accessdata.fda.
gov/scripts/cder/daf/index.cfm. Accessed January 31, 2017.
103. Lachman Consultant Services, Inc. Citizen Petition to the US-FDA Docket No.
2004P-0504/CP1 requesting that the Commissioner of Food and Drugs desig-
nate DiaBeta Tablets, 5 mg, as a second RLD for the purposes of submitting an
ANDA for a duplicate version of that RLD product. Silver Spring, MD: FDA; 2004.
Available at: http://www.fda.gov/ohrms/dockets/dailys/04/nov04/111204/
04p-0504-cp00001-vol1.pdf. Accessed January 31, 2017.
104. U.S. Food and Drug Administration. Response to the Citizen Petition to the US-
FDA Docket No. 2004P-0504/CP1. Silver Spring, MD: FDA; 2005. Available at:
http://www.fda.gov/ohrms/dockets/dockets/04p0504/04p-0504-pav0001-
vol1.pdf. Accessed January 31, 2017.
105. Vogel AM. Hybrid or mixed marketing authorization application in the
European Union: not a trivial decision in new development programs for
established drugs. Drug Information Journal. 2012;46(4):479-484.
106. European Medicines Agency. Questions & Answers: positions on specific
questions addressed to the Pharmacokinetics Working Party (PKWP). London,
UK: EMA; 2015. Available at: http://www.ema.europa.eu/docs/en_GB/
document_library/Scientific_guideline/2009/09/WC500002963.pdf. Accessed
February 1, 2017.
107. Benet L. Why highly variable drugs are safer, FDA Advisory Committee for
Pharmaceutical Sciences and Clinical Pharmacology meeting transcript. Silver
Spring, MD: FDA; 2004. Available at: http://www.fda.gov/ohrms/dockets/ac/
04/transcripts/4034T2.htm. Accessed February 1, 2017.
108. Benet LZ, Goyan JE. Bioequivalence and narrow therapeutic index drugs.
Pharmacotherapy. 1995;15(4):433-440.
109. Filipsson AF, Sand S, Nilsson J, Victorin K. The benchmark dose
methodereview of available models, and recommendations for application in
health risk assessment. Crit Rev Toxicol. 2003;33(5):505-542.
110. U.S. Food and Drug Administration. GDUFA Regulatory Science Priorities for Fiscal
Year 2018. Silver Spring, MD: FDA; 2018. Available at:https://www.fda.gov/
downloads/Drugs/ResourcesForYou/Consumers/BuyingUsingMedicineSafely/
GenericDrugs/UCM582777.pdf. Accessed January 31, 2018.
111. Cristofoletti R, Dressman JB. Use of physiologically based pharmacokinetic
models coupled with pharmacodynamic models to assess the clinical rele-
vance of current bioequivalence criteria for generic drug products containing
Ibuprofen. J Pharm Sci. 2014;103(10):3263-3275.
112. Stangier J, Rathgen K, St€ ahle H, Gansser D, Roth W. The pharmacokinetics,
pharmacodynamics and tolerability of dabigatran etexilate, a new oral direct
thrombin inhibitor, in healthy male subjects. Br J Clin Pharmacol. 2007;64(3):
292-303.
113. Mueck W, Becka M, Kubitza D, Voith B, Zuehlsdorf M. Population model of the
pharmacokinetics and pharmacodynamics of rivaroxabanean oral, direct
11
factor Xa inhibitorein healthy subjects. Int J Clin Pharmacol Ther. 2007;45(6):
335-344.
114. U.S. Food and Drug Administration. Draft Guidance on Dabigatran etexilate
mesylate. Silver Spring, MD: FDA; 2017. Available at: http://www.fda.gov/
downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/
UCM308030.pdf. Accessed February 13, 2017.
115. U.S. United States Food and Drug Administration. Draft Guidance on Rivar-
oxaban. Silver Spring, MD: FDA; 2017. Available at: http://www.fda.gov/ucm/
groups/fdagov-public/@fdagov-drugs-gen/documents/document/ucm461150.
pdf. Accessed February 13, 2017.
116. Lionberger R, Jiang W, Huang SM, Geba G. Confidence in generic drug sub-
stitution. Clin Pharmacol Ther. 2013;94(4):438-440.
117. Popper KR. The Logic of Scientific Discovery. 15th ed. London: Routledge; 2002.
118. Davit BM, Chen ML, Conner DP, et al. Implementation of a reference-scaled
average bioequivalence approach for highly variable generic drug products
by the U.S. Food and Drug Administration. AAPS J. 2012;14(4):915-924.
119. Davit B, Braddy AC, Conner DP, Yu LX. International guidelines for bioequiv-
alence of systemically available orally administered generic drug products: a
survey of similarities and differences. AAPS J. 2013;15(4):974-990.
120. U.S. Food and Drug Administration. Guidance for Industry Bioavailability and
Bioequivalence Studies Submitted in NDAs or INDsdGeneral Considerations.
Silver Spring, MD: FDA; 2014. Available at: http://www.fda.gov/downloads/
drugs/guidancecomplianceregulatoryinformation/guidances/ucm389370.pdf.
Accessed August 11, 2016.
121. Tothfalusi L, Endrenyi L. An exact procedure for the evaluation of reference-
scaled average bioequivalence. AAPS J. 2016;18(2):476-489.
122. Cross J, Lee H, Westelinck A, Nelson J, Grudzinskas C, Peck C. Postmarketing
drug dosage changes of 499 FDA-approved new molecular entities, 1980-
1999. Pharmacoepidemiol Drug Saf. 2002;11(6):439-446.
123. Heerdink ER, Urquhart J, Leufkens HG. Changes in prescribed drug doses after
market introduction. Pharmacoepidemiol Drug Saf. 2002;11(6):447-453.
124. Holford NH, Sheiner LB. Understanding the dose-effect relationship: clinical
application of pharmacokinetic-pharmacodynamic models. Clin Pharmacoki-
net. 1981;6(6):429-453.
125. Kuypers DRJ. Tacrolimus formulations and African American Kidney Transplant
Recipients: when do details matter? Am J Kidney Dis. 2018;71(3):302-305.
126. U.S. Food and Drug Administration. Guidance for Industry. Exposure-
Response RelationshipsdStudy Design, Data Analysis, and Regulatory Appli-
cations. Silver Spring, MD: FDA; 2003. Available at: http://www.fda.gov/
downloads/drugs/guidancecomplianceregulatoryinformation/guidances/
ucm072109.pdf. Accessed February 13, 2017.
127. U.S. Food and Drug Administration. Clinical Pharmacology and Bio-
pharmaceutic Review. Application number:200045Orig1s000. Silver Spring,
MD: FDA; 2010. Available at: http://www.accessdata.fda.gov/drugsatfda_
docs/nda/2010/200045Orig1s000OtherR.pdf. Accessed February 13, 2018.
128. U.S. Food and Drug Administration. Tekturna label. Silver Spring, MD: FDA;
2017. Available at: http://www.accessdata.fda.gov/drugsatfda_docs/label/2
012/021985s023lbl.pdf. Accessed February 13, 2018.
129. Medicines and Healthcare Products Regulatory Agency. Public assessment
report of the Medicines Evaluation Board in UK. Losartan potassium 25, 50
and 100 mg tablets, PL 16002/0077e79, PL 16002/0081e83. London, UK:
MHRA; 2008. Available at: http://www.mhra.gov.uk/home/groups/l-unit1/
documents/websiteresources/con014176.pdf. Accessed January 12, 2017.
130. Medicines and Healthcare Products Regulatory Agency. Public assessment
report of the Medicines Evaluation Board in UK. Losartan potassium/hydro-
chlorothiazide 50/12.5 mg tablets, PL 24701/0001/PL 24701/0002. London,
UK: MHRA; 2008. Available at: http://www.mhra.gov.uk/home/groups/l-
unit1/documents/websiteresources/con014706.pdf. Accessed January 12,
2017.
131. Medicines and Healthcare Products Regulatory Agency (MHRA). Public
assessment report of the Medicines Evaluation Board in UK. Losartan
potassium/hydrochlorothiazide 100/12.5 mg tablets, PL04416/1197e1199,
Sandoz Limited. London, UK: MHRA; 2011. Available at: http://www.mhra.
gov.uk/home/groups/par/documents/websiteresources/con132112.pdf. Accessed
January 12, 2017.
132. Selen A, Dickinson PA, Müllertz A, et al. The biopharmaceutics risk assessment
roadmap for optimizing clinical drug product performance. J Pharm Sci.
2014;103(11):3377-3397.
133. Yu LX, Amidon G, Khan MA, et al. Understanding pharmaceutical quality by
design. AAPS J. 2014;16(4):771-783.
134. Visser SA, Smulders CJ, Reijers BP, Van der Graaf PH, Peletier LA, Danhof M.
Mechanism-based pharmacokinetic-pharmacodynamic modeling of
concentration-dependent hysteresis and biphasic electroencephalogram ef-
fects of alphaxalone in rats. J Pharmacol Exp Ther. 2002;302(3):1158-1167.
135. Gengo FM, Schentag JJ, Jusko WJ. Pharmacokinetics of capacity-limited tissue
distribution of methicillin in rabbits. J Pharm Sci. 1984;73(7):867-873.
136. Danhof M. Kinetics of drug action in disease states: towards physiology-based
pharmacodynamic (PBPD) models. J Pharmacokinet Pharmacodyn. 2015;42(5):
447-462.
137. Cui J, Li C, Guo W, et al. Direct comparison of two pegylated liposomal
doxorubicin formulations: is AUC predictive for toxicity and efficacy? J Control
Release. 2007;118(2):204-215.
138. Smith NF, Acharya MR, Desai N, Figg WD, Sparreboom A. Identification of
OATP1B3 as a high-affinity hepatocellular transporter of paclitaxel. Cancer
Biol Ther. 2005;4(8):815-818.
12 R. Cristofoletti et al. / Journal of Pharmaceutical Sciences xxx (2018) 1-12
139. Sparreboom A, van Zuylen L, Brouwer E, et al. Cremophor EL-mediated
alteration of paclitaxel distribution in human blood: clinical pharmacoki-
netic implications. Cancer Res. 1999;59(7):1454-1457.
140. U.S. Food and Drug Administration. Draft Guidance on Paclitaxel. Silver
Spring, MD: FDA; 2017. Available at: http://www.fda.gov/downloads/
drugs/guidancecomplianceregulatoryinformation/guidances/ucm320015.
pdf. Accessed January 10, 2017.
141. U.S. Food and Drug Administration. Draft Guidance on Doxirrubicin hydro-
chlorhide. Silver Spring, MD: FDA; 2017. Available at: http://www.fda.gov/
downloads/Drugs/.../Guidances/UCM199635.pdf. Accessed January 10, 2017.
142. Endrenyi L, Blume HH, Tothfalusi L. The two main goals of bioequivalence
studies. AAPS J. 2017;19(4):885-890.
143. Hoyningen-Huene P. Thomas Kuhn and the chemical revolution. Found Chem.
2008;10(2):101-115.
144. Portin P. The development of genetics in the light of Thomas Kuhn's theory of
scientific revolutions. Recent Adv DNA Gene Seq. 2015;9(1):14-25.
145. Strohman RC. The coming Kuhnian revolution in biology. Nat Biotechnol.
1997;15(3):194-200.
146. van der Greef J, McBurney RN. Innovation: rescuing drug doscovery: in vivo
systems pathology and systems pharmacology. Nat Rev Drug Discov. 2005;4:
961-967.
147. Chen ML, Blume H, Beuerle G, et al. The global bioequivalence harmonization
initiative: summary report for EUFEPS international conference. Eur J Pharm
Sci. 2018;111:153-157.