PHARMCEUTICA

ACTA
HELVETIAE
ELSEVIER

Pharmaceutics

Acta Helvetiae

70 (1995) 125-131

Dry-powder inhalers: evaluation of testing methodology and effect of

inhaler design
J. Broadhead

a,*, S.K. Edmond

Rouan

b, C.T. Rhodes

a Department of Pharmaceutics, University of Rhode Island, Kingston, RI 02881, USA

b Smith&Tine Beecham Pharmaceuticals, 709 Swedeland Road, King of Prussia, PA 19406, USA
Received

11 October

1993; accepted

21 October

1994

Abstract
We have previously developed a spray dried formulation
of a model protein (P-galactosidase)
of a size suitable for evaluation
in dry powder
inhaler devices. In this study, we wished to evaluate the roles of various methods available for the laboratory
testing of dry powders for inhalation
(cascade impactor, twin impinger, aerodynamic
time of flight and image analysis). Secondly we wished to compare different inhaler devices using
formulations
with and without a carrier. Both the cascade impactor and twin impinger were appropriate
methods for the testing of dry powder
inhalers,
and gave comparable
estimates
of respirable
fraction.
Image analysis and aerodynamic
time of flight were suitable methods for
determining
the particle size of the dry powders, although the former was considerably
more time consuming than the latter. The four inhalers
evaluated differed greatly in terms of in vitro deposition properties.
The presence of a carrier significantly improved respirable fraction with the
poorer inhalers, but was less critical to the performance
of the more efficient devices.
Keywords: Spray drying;

Impactor;

Impinger:

Particle

size; Inhalation;

1. Introduction
Bronchodilators
and other drugs for asthma have
traditionally
been administered
by the inhalation
route,
usually in the form of suspension,
metered
dose inhalers (MDIs).
More recently
dry powder
inhalers
(DPIs) have been developed.
These enable the pulmonary delivery of higher dose, locally acting, drugs
such as sodium cromoglycate.
They also offer an alternative delivery system to patients who are unable to
synchronize
the discharge
and inhalation
of MDIs
(Ganderton
and Kassem, 1991).
Whether a drug is to be delivered by a metered dose
or dry powder device, control of particle size is critical.
Particles
of around
5 pm generally
deposit
in the
bronchiolar
region of the respiratory
tract, whereas

* Corresponding
author.
Present
address:
Creative
BioMolecules,
35, South Street,
Hopkinton,
MA 01748, USA. Fax: +l (508)
4350454.
Elsevier Science B.V. All rights
SSDIOO31-6865(95)00005-4

reserved

Protein

those in the l-3 pm size range can be expected to

reach the alveoli (Gupta and Hickey, 1991). Delivery of
drugs to the systemic system via the pulmonary
route, a
potential
future use of inhalation
therapy, will be dependent
on achieving alveolar deposition
so that absorption can occur.
Evaluation
of inhaler devices in pharmaceutical
development
has traditionally
relied heavily on the use of
cascade
impactors.
Conventional
cascade
impactors,
originally intended
for air sampling, consist of a series
of metal stages with holes of decreasing
size as the
stack is descended.
A vacuum pump draws the sample
in the air stream through the stack. The smaller particles are retained in the air stream longer and therefore
reach the lower stages; conversely the larger particles
will deposit on the upper stages of the stack. For
pharmaceutical
purposes
a chamber
is generally
attached to the top of the impactor to provide a closed
system into which the inhaler is discharged.
The inhaled dose is thus divided into size fractions; typically

126

J.Broadhead et al. /Pharmaceutics

six or eight. This enables the respirable

fraction and
mass median
aerodynamic
diameter
(MMAD)
to be
calculated.
However,
use of this apparatus
is time
consuming
and tedious for carrying out routine tests.
Furthermore,
cascade impactors are typically intended
for the analysis of dilute aerosols,
and may not be
appropriate
for the analysis of concentrated
powder
systems (Bell et al., 1971). In order for the cascade
impactor
to be adapted
for testing DPIs, the plates
must be coated with an adhesive medium to prevent
particles bouncing off them and becoming re-entrained
in the air stream (Allen, 1990). Thus the extraction of
the drug from the plates is cumbersome,
and there is
also potential for the coating medium to interfere with
the assay.
A simpler testing device is the twin impinger. This is
a glass apparatus
which divides the dose into two
fractions;
respirable
and non-respirable.
The drug is
deposited
directly into the assay solvent which need
only be diluted to volume prior to assay. This device
was developed
to provide
a more straightforward
method for routine
development
and quality control
studies, and also to enable testing of both MDIs and
DPIs using the same apparatus
(Hallworth
and Westmoreland,
1987). To meet the latter criterion, the twin
impinger
operates
at an air flow rate of 60 I/min,
which is necessary for most DPIs to function correctly.
In contrast,
most cascade impactors
are designed
to
operate
at much lower flow rates. Both the cascade
impactor
and twin impinger
are now official USP
methods for the determination
of the fine particle size
fraction of metered dose inhalers (Pharmacopeial
Forum, 1992).
There has been much discussion
in the literature
concerning
the methods available for testing metered
dose inhalers, but evaluation
of dry powder inhalers
has received far less attention.
Thus the objective of
the present study was to compare the twin impinger
and cascade impactor
as methods for evaluating
the
performance
of dry powder inhalers. We also wished to
assess the roles of particle size analysis methods (image
analysis and aerodynamic
time of flight) in the development and testing of these devices. We have previously
developed
a spray dried formulation
of a model protein (p-galactosidase)
of a suitable size for evaluation
in DPIs (Broadhead
et al., 1994). This formulation
was
used in the current investigation.
A further goal of these studies was to evaluate the
delivery of the spray dried protein from different dry
powder devices. DPI formulations
are typically blended
with a coarser carrier material,
usually lactose. This
functions to improve the flow of the formulation
and to
facilitate dispersion
of the micronized
drug during in-

Acta Hebetiae 70 (1995) 125-131

halation. Thus in this study we evaluated

the in vitro
deposition
of the spray dried P-galactosidase
alone
and when blended
with an equal weight of carrier
material.

2. Experimental

procedures

2.1. Materials
Lyophilized
P-galactosidase
derived
from
Aspergillus oryzae was obtained
from Enzyme Development (New York). Reagents and standard for the BCA
and micro BCA protein
assays were obtained
from
Pierce (Rockford,
IL). Avicel PH 101 was obtained
from FMC. Mannitol
was obtained
from ICI. Trehalose dihydrate and all other chemicals were obtained
from Sigma (St. Louis, MO).
2.2. Spray drying
A 60 mg/ml
solution of P-galactosidase
was prepared by dispersing
the enzyme, as supplied by the
manufacturer,
in deionized water, centrifuging
and dialyzing the supernatant
against deionized
water. Trehalose, at a concentration
of 50 mg/ml, was added to
this solution to prevent activity loss during drying and
storage. Spray drying was carried out using a Biichi 190
mini spray dryer at an inlet temperature
of 140 C and
an outlet temperature
of approx. 95 C. The solution
was fed to the dryer at a rate of 2 ml/min.
An air flow
rate of 600 l/h and aspirator vacuum of 35 mbar were
used. The moisture content of the spray dried material,
determined
by Karl Fischer analysis, was 3.7%.
2.3. In vitro inhalation behavior
The spray dried material was evaluated both in the
presence and absence of a carrier. The carrier formulation was prepared by mixing equal weights of the spray
dried powder and micro crystalline
cellulose (Avicel
PH 101). Avicel was used simply as a model carrier
material, since lactose is a substrate for P-galactosidase
and therefore could not be used. The geometric mean
size of the Avicel 101 was 24.3 pm as determined
by
image analysis. Hard gelatin capsules were filled with
either 20 mg of the spray dried material or 40 mg of
the carrier blend. This represents
approx. 11-12 mg of
protein per capsule. Four inhalation
devices were evaluated (Spinhaler,
Rotahaler,
Berotec and ISF). The
deposition
properties
were evaluated using a twin impinger (Copley Instruments)
and a cascade impactor

J. Broadhead et al. /Pharmaceutics

(Andersen
8 stage 1 ACFM Non Viable Particle Size
Sampler Mark II). In both cases five capsules were
discharged
into the apparatus
per determination
and
each determination
was carried out five times. An air
flow rate of 60 l/min
and an inhalation
time of 10 s
was used throughout.
Deionized water was used as the
impingement
fluid for the twin impinger. The volumes
in stage 1 and 2 were 5 ml and 20 ml, respectively.
After each determination
the throat and two stages of
the twin impinger were rinsed with deionized
water,
and the washings analyzed for protein content. For the
cascade impactor determinations,
the eight glass plates
of the impactor were coated with a thin layer of silicone grease. This prevents particles bouncing
off the
plates and becoming
re-entrained
in the air stream
(Allen, 1990). A preseparator
was attached to the top
of the cascade impactor to prevent large particles or
aggregates
reaching
the stages. The twin impinger
throat piece was attached to the top of the preseparator. After each determination
the material
on each
plate of the impactor
was collected
by rinsing with
deionized
water. The material which deposited
in the
throat piece, the preseparator
and on the metal stages
was also determined.
The aerodynamic
diameters
collected with 50% efficiency
at each stage of the impactor are 7.39, 4.71, 3.21, 2.21, 1.43, 0.73, 0.46 and
0.32 at an air flow rate of 60 l/min (calibration
data
from Andersen).
The cut off aerodynamic
diameter
between stage 1 and 2 of the twin impinger is approx.
6.4 pm.
The protein
content
of the various
samples was
determined
using the BCA protein assay (concentrations of 20-300 pgg/ml) or micro BCA protein assay
(concentrations
less than 20 pg/mlI
(Smith et al.,
1985). The percentage
of the protein
filled into the
capsules which was recovered
in each region of the

Table 1
The effect of formulation
Formulation

and device on twin impinger

Region

of deposition

No carrier

Avicel blend

a Results

expressed

as percentage

and inhaler

and inhaler

of protein

121

impinger
or impactor was thus calculated.
The mass
median aerodynamic
diameter of the samples was determined
from the cascade impactor
data. The Students t-test (two-sided) was used to evaluate the significance of differences
in deposition
results at the 95%
confidence
level.

2.4. Particle size determination

The geometric mean size and size distribution
of the
spray dried material
were determined
using image
analysis (Magiscan,
Joyce Loebl). A dispersion
of approx. 50 mg of the sample in 15 ml of silicone fluid was
prepared
and sonicated
for one minute.
Four slides
were prepared
from this dispersion
and four fields of
view were analyzed per slide. In total, over 500 particles were counted.
The particle size of the spray dried material (without
a carrier) was also evaluated
using the Aerosizer
system (Amherst Process Instruments).
This is a relatively
new piece of equipment
which measures particle size
based on aerodynamic
time of flight. The pulse jet
disperser attachment
was used to determine
the size of
dry powder samples. This uses a jet of compressed
air
to create an airborne
suspension
of the powder. The
particles then pass through a high shear air flow region
which breaks up agglomerates
prior to measurement.
The AeroBreather
attachment
was used to evaluate
the spray dried powder in the inhaler devices. This
allows MDIs or DPIs to be inhaled
directly by a
system that mimics the action of taking a breath. The
Aerosizer system requires that the density of the material be determined.
This was measured
using an AutoPycnometer
1320 (Micromeritics)
and was determined to be 1.42 g/cm3.

deposition
Device a
ISF

Capsules
Throat
Stage 1
Stage 2
Capsules
Throat
Stage 1
Stage 2

Acta Hebetiae 70 (1995) 125-131

32.55
19.99
21.06
17.80
21.72
13.50
23.99
24.74

(13.15)
(3.86)
(2.72)
(6.10)
(7.74)
(1.58)
(3.82)
(4.06)

filled into capsules

reaching

Berotec

Rotahaler

Spinhaler

56.78
20.18
6.40
10.49
49.55
11.76
9.05
19.33

69.25
9.78
15.16
1.96
67.42
5.57
12.19
5.89

78.56
4.26
14.28
0.74
70.26
4.19
10.96
5.72

respective

(7.84)
(3.19)
(0.59)
(2.48)
(8.94)
(1.92)
(2.38)
(3.94)
regions.

(10.01)
(2.84)
(5.16)
(0.48)
(3.66)
(1.18)
(0.81)
(0.93)

n = 5; standard

deviations

(14.28)
(2.49)
(9.36)
(1.13)
(11.30)
(1.01)
(3.39)
(1.46)

in parentheses.

J. Broadhead et al. /Pharmaceutics

128
Table 2
The effect of formulation

impactor

deposition

Device a

Formulation

No carrier
Stages 1-8
Stages 3-8
Ar?cel blend
Stages l-8
Stages 3-8
a Results

and device on cascade

Acta Hebetiae 70 (19951 125-131

expressed

ISF

Berotec

Rotahaler

Spinhaler

17.52 (2.06)
9.20 (1.27)

10.58 (1.92)
6.63 (1.34)

3.51 (0.66)
1.84 (0.52)

1 SO (0.84)
0.99 (0.56)

22.48 (2.34)
9.90 (0.98)

16.58 (3.71)
8.01 (1.83)

8.03 (0.45)
2.78 (0.44)

3.19 (0.46)
1.69 (0.26)

as percentage

of protein

in capsules

reaching

respective

properties

Table 1 shows the distribution

of protein between
the regions of the twin impinger. The material which
reaches stage 2 is considered
to be respirable.
Table 2
shows the percentage
of the protein fill recovered from
stages 1-8 and 3-8 of the cascade impactor. The stage
1-8 deposition
was considered
to be respirable.
This
represents
material less than 7.39 pm (as compared to
a theoretical
cut-off of 6.4 pm for the twin impinger).
Material collected on stages 3 through 8 should be less
than the theoretical
cut-off size of 3.21 ,um. This gives
an indication
of the amount of material with potential
to reach the alveolar region of the lung. There were no
significant
differences
between the respirable
percentages obtained from the two analytical devices when the
Berotec or ISF inhalers were tested. The Rotahaler

Particle
Fig. 1. The particle

size distribution

n = 5; standard

deviations

in parentheses.

gave significantIy
different estimates for both formulations, and the Spinhaler
gave a significantly
different
estimate for the Avicel formulation.
In the case of the
Rotahaler,
the cascade impactor gave higher estimates,
whereas with the Spinhaler
the twin impinger gave a
higher estimate for the Avicel formulation.

3. Results
3.1. In vitro inhalation

regions.

3.2. Particle size determinations

The particle size distribution
of the material determined by image analysis is shown in Fig. 1. The Feret
diameter was used as the measure of particle size. This
is the distance between two tangents on opposite sides
of the particle
(Allen,
1990). The geometric
mean
particle size was 2.67 pm, determined
as a number
average.
Table 3 shows the mass median aerodynamic
diameter (MMAD)
of the spray dried material
calculated
from the cascade impactor data. In all cases there was

Size

(microns)

of the spray dried P-galactosidase

formulation

determined

using image analysis.

J. Broadhead et al. /Pharmaceutics

Table 3
Effect of formuation
Formulation

and device on MMAD

MMAD

Acta Hebetiae 70 (1995) 125-131

determinations

(pm)

ISF

Berotec

Rotahaler

Spinhaler

No carrier
Avicel blend

2.86 (0.17)
3.40 (0.14)

2.36 (0.11)
3.00 (0.25)

2.68 (0.44)
4.40 (0.60)

2.06 (0.49)
2.76 (0.21)

n = 5; standard

deviations

in parentheses.

an apparent increase in MMAD when blended with

avicel, although this was only statistically significant for
the Rotahaler and the ISF device.
Table 4 shows the Aerosizer results. The data was
analyzed both as a number distribution based on geometric mean size and as a volume distribution based on
aerodynamic size. The size of the dry powder was
determined using the pulse jet disperser. The other
determinations were made using the AeroBreather system. A serious drawback encountered with the AeroBreather device is that it is very easily blocked by
powder aggregates or gelatin fragments. This makes it
extremely difficult to obtain accurate and reproducible
measurements using capsule devices. This was particularly problematic with the ISF device because it causes
extensive capsule fragmentation, resulting in inhalation of gelatin fragments into the analytical system, be
it the cascade impactor, twin impinger or Aerosizer.
This repeatedly caused blockages in the Aerosizer system, making determinations impossible.

4. Discussion
4.1. Effect of formulation and inhaler type

In terms of respirable percentages, the effect of

blending with a carrier was dependent on the particular inhaler being evaluated. The twin impinger showed
significant differences between the respirable percentages achieved with and without a carrier for all except
the ISF device. The cascade impactor showed significant differences between the formulations in terms of
stage l-8 deposition for all the inhalers. However in

Table 4
Aerosizer

particle

terms of stage 3-8 deposition, only the Spinhaler and

Rotahaler gave significant differences. With the ISF
device which achieved the highest respirable fractions,
the presence of a carrier was less critical to performance than with the other devices. The Berotec device
was able to achieve respirable percentages close to
those obtained with the ISF device when a carrier was
included in the formulation, but without a carrier the
respirable fraction dropped to around 10%. A similar
pattern was observed with the Spinhaler and Rotahaler, in that in the absence of a carrier respirable
amounts were significantly reduced. When mannitol
was used as a carrier we achieved twin impinger results
which corresponded very closely with those obtained
with the Avicel formulation (Mean respirable percentages of 22.8%, 19.8%, 4.3% and 5.5% for ISF, Berotec,
Spinhaler and Rotahaler respectively). This is despite
the fact that mannitol has very different properties
from avicel. Therefore the nature of the carrier does
not appear to be critical, merely the presence of a
coarser material to reduce the cohesive forces between
the spray dried particles and thereby facilitate dispersion. VidgrCn et al. (1988) have previously attributed
the superior in vivo performance of the Berotec and
ISF devices to the narrower diameter of the air channels in the inhalers, which they claim is more likely to
result in turbulent air flow and thus more effective
powder dispersion. Presumably with the ISF inhaler,
this effect is sufficient to enable the cohesive interparticulate forces to be overcome without the aid of a
carrier.
The ISF device achieved slightly better respirable
percentages than the Berotec device for the Avicel
formulation. However, there was also much greater
deposition in stage 1 of the twin impinger than with the
Berotec inhaler (23.99% versus 9.05%). Thus if respirable percentage were calculated as the amount of
material reaching stage 2 as a percentage of the total
recovered from the throat and both stages (as is typically the case when calculating respirable fractions for
MDIs), the Berotec device would actually be superior.
Since substantial oropharyngeal deposition is not desir-

size determinations

Device

Mean aerodynamic

Dry powder
Rotahaler
Berotec
Spinhaler

4.29
6.02
4.64
4.42

n = 3; standard

129

deviations

(0.49)
(0.40)
(0.60)
(0.50)

in parentheses.

diameter

(pm) (by

volume)

Mean geometric
1.54 (0.16)
2.14 (0.07)
1.90(0.11)
1.60 (0.32)

diameter

(pm)

(by number)

130

J. Broadhead et al. /Pharmaceutics

able, there may be advantages to choosing the Berotec

device over the ISF device. Another
problem associated with the ISF device is that it causes extensive
capsule fragmentation,
resulting
in gelatin fragments
reaching the throat and first stage of the impinger.
This would result in a similar effect in an in vivo
situation
which is obviously
undesirable.
The main
drawback associated with the Berotec device is retention of a substantial
amount of material in the capsules
and inhaler.
4.2. Particle sizing analysis methods
The aerodynamic
diameter
may be the most relevant sizing parameter
for inhalation
products, since it
reflects most closely their ultimate mode of use. The
aerodynamic
diameter of a particle is equivalent
to the
diameter
of a spherical particle of unit density which
will fall through free air at the same velocity as the
particle
in question.
Thus, since the density of the
spray dried material is 1.42, the aerodynamic
diameter
will obviously be larger than the true diameter i.e. the
diameter
which would be measured
optically.
The
MMAD calculated
from cascade impaction
data is a
weight average mean. Similarly the aerodynamic
diameter determined
by the Aerosizer
is generally
expressed as a volume average. Both these means will
be significantly
skewed by the presence of a few large
particles.
This is because, for example, a tenfold increase in diameter
will result in a thousand
fold increase in volume or weight. In contrast the geometric
mean determined
by either image analysis or the Aerosizer and expressed
as a number
average will not be
skewed by the presence
of a few large particles
or
aggregates.
Thus the latter expressions
of mean size
would be expected to be somewhat
smaller than the
former.
In fact the geometric
mean determined
by image
analysis was very close to the MMADs calculated from
the cascade impactor data. These in turn were somewhat lower than the aerodynamic
diameters
determined by the Aerosizer,
although
the two methods
gave the same rank order for the three inhalers tested
using
both (Rotahaler
> Berotec > Spinhaler).
The
smallest estimates of particle size were the geometric
means generated
by the Aerosizer.
They were considerably smaller than the aerodynamic
means with considerably
less variation
between
the values obtained
using different inhalers.
These results demonstrate
the variation
between
size determinations
when different analytical methods
are used, and the difficulty
in trying to obtain
an

Acta Helvetiae 70 (1995) 125-131

absolute value for particle size. Whilst image analysis is

an optical method and so should, in theory, provide an
absolute value of mean diameter, it too can be subjective. This is because it is highly dependent
on slide
preparation.
It is extremely difficult to avoid the presence of aggregates even on skillfully prepared
slides.
Furthermore
it is considerably
more time consuming
than the Aerosizer, and bears little resemblance
to the
ultimate mode of use of a dry powder for inhalation.
The main advantage
of the Aerosizer
is the rapidity
with which measurements
can be made. We also found
that the results obtained using the dry powder sampler
were very reproducible.
In addition it can more closely
simulate the end use of a DPI formulation.
4.3. In vitro deposition methods
In our experiments,
both the twin impinger
and
cascade impactor were able to discriminate
adequately
between the different
formulations
and different
inhalers. Furthermore
the two devices gave very close
estimates for respirable fractions when operated under
identical conditions.
The twin impinger has previously
been reported to over-estimate
respirable
fractions of
metered
dose inhalers when compared
with cascade
impactors
(Phillips et al., 1990). However, for MDIs,
the Andersen
impactor is usually operated
at a flow
rate of 28.30 liters/minute,
which presumably
accounts
for some of the difference.
Dry powder inhalers are
designed to be inspired quickly, however, which makes
a flow rate of 60 liters/minute
more representative
of
real life conditions.
This is also the air flow rate at
which the twin impinger is operated,
making comparisons between the two devices simpler.
The limitation
of the twin impinger lies in that it
divides the sample into only two components.
This
means that the MMAD cannot be calculated.
In addition, if we are considering
using the pulmonary
route
for systemic delivery, it is desirable to obtain estimates
of material
in the l-3 pm range. It has also been
reported
that, theoretically,
the twin impinger cannot
discriminate
among distributions
with certain combinations of size and width (Miller et al., 1992) Nevertheless we have found the twin impinger
to be useful,
reproducible,
and certainly
sufficiently
discriminating
for routine evaluations.
Its main advantage
is that a
determination
can be performed
in approximately
half
the time required using a cascade impactor.
The Aerosizer
is a useful method
for obtaining
rapid size evaluations
of dry powders. It is considerably
less tedious than either image analysis or cascade impaction, so is an extremely useful method for screening

.I. Broadhead et al. /Pharmaceutics

studies during development or routine end product

testing. However, it is difficult to see any way in which
the use of the Aerosizer could obviate the need for
impinger/impactor
testing, since it simply determines
particle size which cannot necessarily be correlated
with in vivo performance. The results we obtained
using the different inhalers highlight this fact. The
Aerosizer gave similar particle size readings for all the
devices, despite the fact that their performance differed markedly in terms of respirable fraction. This is
because the Aerosizer only requires a tiny amount of
material to be inhaled from the capsules in order to
determine size. Furthermore, because it has an in built
disperser system, it gives little indication of the inherent dispersibility of a powder formulation, or of the
efficiency on an inhalation device. Nevertheless it is
extremely useful for obtaining rapid size analyses during the development of a micronization method, or
during routine quality control testing.
The AeroBreather system did not work well for our
purposes due to frequent blockages in the nozzle caused
by both powder aggregates and gelatin fragments.
However, these problems should not occur with multidose devices, such as the Turbuhaler. The main advantage of the AeroBreather is that multidose DPI devices
can be evaluated without removing the powder from
the device. This may be useful in end product and
stability testing. For development purposes, however, it
is easier to simply test the powder directly using the
pulse jet disperser.
In conclusion, it is likely that impaction/impingement methods will remain the mainstay of DPI testing
for the foreseeable future. Whilst image analysis and
the Aerosizer have a role to play in size analysis, they
do not provide the crucial in vitro deposition data
which the development of inhalation products requires.
We have found both the twin impinger and cascade
impactor to be suitable for testing DPIs. For most
situations the twin impinger is probably sufficiently

Acta Helvetiae 70 (1995) 125-131

131

discriminating. Whether to use this device or the cascade impactor would require an assessment as to
whether the additional data which the cascade impactor can provide merits the additional time required
to carry out the analysis.

Acknowledgements

The support of SmithKline Beecham Pharmaceuticals is gratefully acknowledged.

References
Allen, T. (1990) Particle Size Measurement,
4th Edn., Chapman and
Hall, New York, 78-80.
Bell, J.H., Hartley P.S. and Cox J.S. (1971) Dry powder aerosols I: a
new powder inhalation
device. J. Pharm. Sci. 60, 1559-1564.
Broadhead
J., Edmond Rouan S.K., Hau I. and Rhodes C.T. (1994)
The effect of process and formulation
variables on the properties
of spray dried /3-galactosidase.
J. Pharm. Pharmacol.
46, 458-467.
Ganderton
D. and Kassem N.M. (1991) Dry powder inhalers. Adv.
Pharm. Sci. 6, 165-191.
Gupta P.K. and Hickey A.J. (1991) Contemporary
approaches
in
aerosolized drug delivery to the lung. J. Control. Rel. 17, 129-148.
Hallworth
G.W. and Westmoreland
D.G. (1987) The twin impinger:
a simple device for assessing the delivery of drugs from metered
dose pressurized
aerosol inhalers. J. Pharm. Pharmacol.
39, 966972.
Miller N.C., Marple
V.A., Schultz R.K. and Poon W.S. (1992)
Assessment
of the twin impinger for size measurement
of metered dose inhaler sprays. Pharm. Res. 9, 1123-1127.
Pharmacopeia
Forum (1992) U.S. Pharmacopeia
18, 3164-3174.
Phillips E.M., Byron P.R., Fults K. and Hickey A.J. (1990) Optimized
inhalation
aerosols II. Inertial testing methods for particle size
analysis of pressurized
inhalers. Pharm. Res. 7, 1228-1233.
Smith P.K., Krohn RI., Hermanson
G.T., Mallia A.K., Gratner F.H.,
Provenzano
M.D., Fujimoto E.K., Goeke M.B., Olson B.J. and
Klenck D.C. (1985) Measurement
of protein using bicinchoninic
acid. Anal. Biochem. 150, 76-85.
Vidgren M., Kiirkkainen
A., Karjalainen
P., Paronen P. and Nuutinen J. (1988) Effect of powder inhaler design on drug deposition
in the respiratory
tract. Int. J. Pharm. 42, 211-216.