Determination of Water Vapor Transmission Rate for High

Density Polyethylene Screw-cap Bottles

Executive Summary
The Product Quality Research Institute (PQRI) Container Closure Working Group
undertook testing using four different laboratory sites to find ways to improve USP
General Chapter <671> Containers Permeation. Water Vapor Transmission Rate
(WVTR) testing was performed to construct weight gain profiles due to moisture ingress
for 60 mL high-density polyethylene (HDPE) bottles with intact or broached induction
seals when stored at 23C and 40C and 75% RH. The profiles were compared to
determine the time needed to reach steady state, the time, duration and number of data
points needed for WVTR calculation, the variability of the data due to sampling and lab
to lab variation, and to determine if empty bottles (i.e. controls) are needed for accurate
WVTR determination. This report summarizes the data collected from the four testing
laboratories Abbott, Merck Sharp & Dhome, Pfizer, and sanofi-aventis.
The recommendations are as follows:
USP Chapter <671> should be rewritten or replaced.
For the purpose of moisture-barrier comparison, WVTR testing should be
conducted at 40oC/75% RH
The test duration should be 35 days, and containers weighed at 7 day intervals
WVTR should be calculated by regression analysis over the period from day 7 to
day 35 (5 data points).
Blanks (Controls) are not necessary for bottles with induction-sealed closures, and
should not be included in the test design
If calcium chloride is used as the desiccant, it should be pre-dried at 210oC, rather
than 110oC as specified in USP <671>
Introduction and Objectives
The purpose of this study was to assess the opportunity to improve USP <671>. Weight
gain profiles, due to moisture ingress, of closed 60 mL high density polyethylene (HDPE)
screw-cap bottles (Multiple Unit Container Systems) were obtained during storage at
23oC/75% RH and 40oC/75% RH. The profiles were used to determine the time needed
to reach steady state, the duration of the test and number of data points needed for
calculation of water vapor transmission rate (WVTR), the variability in data due to
samples and testing labs, and to determine if controls (blanks) are needed for accurate
WVTR determination. The controls were bottles containing glass beads in place of
desiccant. The origins of this study also lie, in part, in a previous publication1 that
1 Barry

J., J. Bergum, Y. Chen., R. Chern , R.Hollander, D.Klein, H. Lockhart, D. Malinowski, R.

McManus, C. Moreton, A. Mueller, L. Nottingham, C. Okeke, D. OReilly, K. Rinesmith and S. Shorts
(PQRI Container-Closure Working Group), Basis for Using Moisture Vapor Transmission Rate per Unit

2
provided a theoretical basis for using WVTR per unit dose as a means to compare
different container-closure systems.
Materials and Methods
The bottles used were 60mL HDPE with 28mm double lead semi interrupted threads.
The HDPE resin was Phillips Marlex 5502 BN. The bottles were manufactured by
Rexam. The closures used were 28mm diameter polypropylene (PP) with Squeeze and
Turn child-resistant feature. The PP resin was Phillips-Sumika HLN 120-01. The closures
were manufactured by Rexam. The cap liners were supplied by Selig with FS 1-7 innerseal
and C25P liner. The materials of construction for the inner-seal and liner were
0.0015 medium density polyethylene (MDPE)/0.001 Al foil/Heat Tac Wax/0.005
Paper/0.030 Polyolefin Foam.

The bottles were uniquely identified, then filled with an appropriate amount of anhydrous
calcium chloride desiccant and induction-sealed in accordance with USP <671>. See
Appendix A for details about selection of desiccant. The containers were all filled and
sealed at the same time at one site (Pfizer). For details of sample preparation, see the
protocol in Appendix B. The seals of a designated number of these bottles were then
broached and the bottles re-capped according to the procedure prescribed in USP <671>.
On receipt at the testing laboratory, the seals of a designated number of these bottles were
then broached and the bottles re-capped according to the procedure prescribed in USP
<671>. The broached seals were prepared as follows: The cap of the induction sealed
bottle was removed, and the seal was cut around the inner circumference of the mouth
with a sharp razor blade. The cut-out material was removed, taking care to leave a flat,
unbroken area of inner-seal on the land of the bottle finish. The caps were then reapplied
to the bottles in accordance with the procedure in USP <671>. The details of the
broaching procedure are given in Appendix C.
Subsequently, both the intact and broached bottles were stored in environmental
chambers controlled at 402C/755%RH or 232C/753%RH at the four sites:
Abbott (AB), Merck Sharp & Dhome (MK), Pfizer (PF) and sanofi-aventis (SA).
A testing protocol was developed that required each laboratory to perform WVTR tests as
follows:
1) 15 Broached bottles with desiccant, 10 control samples without desiccant stored at
402C/755%RH and weighed at 7 day intervals for 49 days.
2) 15 Broached bottles with desiccant, 10 control samples without desiccant stored at
232C/753%RH and weighed at 7 day intervals for 49 days.
Product in the Evaluation of Moisture-Barrier Equivalence of Primary Packages for Solid Oral Dosage
Forms, Pharmacopeial Forum (2005) 31(1) Jan-Feb, 226-269

3
3) 15 Intact bottles with desiccant, 10 control samples without desiccant stored at
402C/755%RH and weighed at 14 day intervals for 98 days.
4) 15 Intact bottles with desiccant, 10 control samples without desiccant stored at
232C/753%RH and weighed at 14 day intervals for 98 days.
The weight gain of each sample was measured and recorded according to the procedure
given in the protocol. The average rate of water vapor permeation (WVTR) in mg per
day per bottle was calculated according to the protocol.
Results/Discussion
Weight Gain Profiles
Figures 1 and 2 show the mean weight gain (mg/bottle) profile (average of 15 bottles
with desiccant, or 10 control bottles) for broached bottles over the testing period for each
laboratory. As can be seen in the profile plots, there is no apparent induction period, and
the weight gain profiles are linear over the testing interval. The broached bottles
containing desiccant showed an increase in weight over time, with a higher rate for the
higher temperature condition, which was expected. The controls show very little change
over time.
Broached with Desiccant
Figure 1
40oC/75%RH
4
Figure 2
25oC/75%RH
Figures 3 and 4 show the mean weight gain (mg/bottle) profile (average of 15 bottles
with desiccant or 10 control bottles) for intact bottles over the testing period for each lab.

As can be seen in the profile plots, there is no apparent induction period and the weight
gain profiles are linear over the testing interval. The intact bottles containing desiccant
show an increase in weight over time at the higher temperature, but very little increase at
the lower temperature. The controls show very little change over time.
5
Intact with Desiccant
Figure 3
40oC/75%RH
Figure 4
25oC/75%RH
Slopes compared for value and variability
A linear regression was fitted to the results for each container. The slopes (WVTRs) are
plotted in Figures 5 and 6 for broached and intact bottles, respectively. The four labs
show similar patterns within each temperature. . The containers having intact seals have
smaller slopes and less variation than those having broached seals. Slopes for broached
6
and intact seals are plotted to the same scale in Figure 7, emphasizing the difference in
variability between broached and intact seals.
Figure 5
Slopes (mg/day)
Broached
0
5
10
15
Slope, mg/day
Broached 23C/75%RH - AB
Broached 23C/75%RH - MK
Broached 23C/75%RH - Pf
Broached 23C/75%RH - SA
Broached 40C/75%RH - AB
Broached 40C/75%RH - MK
Broached 40C/75%RH - Pf
Broached 40C/75%RH - SA
Seal/Conditions/Site

7
Figure 6
Slopes (mg/day)
Intact
0
0.5
1
1.5
Slope, mg/day
Intact - 23C/
75%RH - AB
Intact - 23C/
75%RH - MK
Intact - 23C/

75%RH - Pf
Intact - 23C/
75%RH - SA
Intact - 40C/
75%RH - AB
Intact - 40C/
75%RH - MK
Intact - 40C/
75%RH - Pf
Intact - 40C/
75%RH - SA
Seal/Conditions/Site

8
Figure 7
Slopes (mg/day)
Broached and Intact
Slope Summary
Tables 1 and 2 summarize the slopes for the broached and intact bottles. Included in
these tables is the number of containers (N) in the analysis and the minimum and
maximum slope across the N containers. The mean, standard deviation, and RSD
(relative standard deviation) can be used to evaluate differences between sites as well as
variability of the slopes. The p-value indicates whether or not the slope is significantly
different from zero, i.e. a statistically significant result (p < 0.05) indicates that the slope
was significantly different from zero. All slopes were significantly different from zero
except one (controls at Pf at the higher temperature condition).
For the broached bottles (Table 1), note that the mean rate at 40/75 is similar across all
laboratories (about 7 to 9.5 mg/day). The variability of the rates is much less at the lower
temperature with respect to standard deviation but about the same with respect to RSD.
See discussion of outliers below. Adjusting for the controls has little effect on the rate
since the control slopes were so close to zero.
9
Table 1
Slope Summary
Broached
Condition Site N
Mean
Slope
(mg/day)
Std Dev
(mg/day) RSD(%) P-value
Min
Slope
(mg/day)
Max
Slope
(mg/day)
40C/75%RH MK 15 7.22 0.733 10.162 < 0.0001 5.523 8.290
Pf 15 7.17 0.668 9.314 < 0.0001 5.982 8.679
SA 15 8.28 1.549 18.701 < 0.0001 5.215 10.985
AB 15 9.40 1.530 16.270 < 0.0001 4.934 11.445
23C/75%RH MK 15 2.65 0.809 30.496 < 0.0001 1.519 5.219
Pf 15 1.92 0.262 13.664 < 0.0001 1.508 2.446
SA 15 2.40 0.326 13.571 < 0.0001 1.853 2.949
With
Desiccant
AB 15 2.74 0.338 12.332 < 0.0001 2.248 3.604

40C/75%RH MK 10 0.04 0.007 < 0.0001 0.028 0.047

Pf 10 0.00 0.011 0.93 -0.019 0.016
SA 10 0.05 0.007 < 0.0001 0.038 0.060
AB 10 0.11 0.002 < 0.0001 0.109 0.116
23C/75%RH MK 10 0.03 0.009 < 0.0001 0.007 0.036
Pf 10 0.03 0.005 < 0.0001 0.019 0.036
SA 10 0.02 0.002 < 0.0001 0.019 0.025
Control
AB 10 0.06 0.003 < 0.0001 0.056 0.066

For the intact bottles (Table 2), the mean rate is similar across all laboratories within each
temperature. The variability of the rates is much less at the lower temperature with
respect to standard deviation but more similar with respect to RSD. Since the slopes are
much smaller for the intact bottles with desiccant than the broached bottles, adjusting for
the controls has more of an effect.
10
Table 2
Slope Summary
Intact
Condition Site N
Mean
Slope
(mg/day)
Std Dev
(mg/day) RSD(%)
Test
Slope = 0
P-value
Min
Slope
(mg/day)
Max
Slope
(mg/day)
40C/75%RH MK 15 1.18 0.026 2.170 < 0.0001 1.154 1.235
Pf 15 1.19 0.022 1.820 < 0.0001 1.158 1.245
SA 15 1.25 0.062 5.002 < 0.0001 1.210 1.460
AB 15 1.23 0.018 1.470 < 0.0001 1.210 1.285
23C/75%RH MK 15 0.29 0.006 2.122 < 0.0001 0.276 0.297
Pf 15 0.27 0.007 2.752 < 0.0001 0.257 0.283
SA 15 0.28 0.006 2.122 < 0.0001 0.273 0.298
With
Desiccant
AB 15 0.27 0.005 1.660 < 0.0001 0.264 0.281
40C/75%RH MK 10 -0.01 0.002 < 0.0001 -0.012 -0.007
Pf 10 -0.02 0.005 < 0.0001 -0.031 -0.016
SA 10 0.02 0.004 < 0.0001 0.018 0.028
AB 10 0.02 0.005 < 0.0001 0.010 0.030
23C/75%RH MK 10 0.02 0.001 < 0.0001 0.023 0.025
Pf 10 0.02 0.002 < 0.0001 0.015 0.023
SA 10 0.01 0.004 < 0.0001 0.006 0.021
Control
AB 10 0.01 0.002 < 0.0001 0.010 0.016

Outliers
In three cases, the data appeared to be influenced by outliers. A discussion of those three
situations follows:
From Table 1 and Figure 5:
Broached MK at 23/75, 30% RSD result: Container 4 had a slope of 5.22 mg/day. The

other 14 slopes ranged from 1.52 to 3.07 mg/day. The Dixon outlier test indicates that
the 5.22 is an extreme value (p-value < 0.001) assuming that the slopes are normally
distributed. By leaving this point out, the MK mean, std dev, and RSD are 2.47, 0.40,
and 16.3, respectively. The RSD calculated by omitting the outlier is in good agreement
with the other site RSDs.
From Table 1 and Figure 5:
Broached AB at 40/75, 16% RSD result: Container 2 had a slope of 4.93 mg/day. The
other 14 slopes ranged from 8.38 to 11.44 mg/day. The Dixon outlier test indicates that
the 4.93 is an extreme value (p-value < 0.005) assuming that the slopes are normally
distributed. By leaving this point out, the AB mean, std dev, and RSD are 9.72, 0.94, and
9.62, respectively.
From Table 2 and Figure 6:
Intact SA at 40/75, 5% RSD result: Container 6 had a slope of 1.46 mg/day. The other
14 slopes ranged from 1.21 to 1.28 mg/day. The Dixon outlier test indicates that the 1.46
11
is an extreme value (p-value < 0.001) assuming that the slopes are normally distributed.
By leaving this point out, the SA mean, std dev, and RSD are 1.23, 0.02, and 1.72,
respectively. The RSD calculated by omitting the outlier is in good agreement with the
other site RSDs.
Nine Day Study
Since there was no apparent delay to reach steady state, Pfizer performed a 9 day study in
both broached and intact bottles as well as controls. They used 15 test bottles and 10
controls, treated in the same way as the main study. The mean weights are given in
figures 8-9 & 10-11 for broached and intact containers, respectively. For broached seals
stored at 40oC/75%RH, there is no apparent induction period in the 9 day study. There is
a short period of high slope during the first time period (day 0 to day 1) for the broached
seals stored at 23oC/75%RH. For the intact seals, there is an extensive period of irregular
behavior for the first six days of the 9 day study. Figures 10 and 11 show discontinuities
and wide slope variation in the data for the whole six day period. For this nine day
period, the performance is widely different among the four tests and irregular within two
of them. This appears to be an induction period that was not evident in the main study.
Figure 8 Figure 9
Broached Broached
402C/755%RH 232C/753%RH
46.46
46.47
46.48
46.49
46.5
46.51
46.52
Mean(Weight)
-1 0 1 2 3 4 5 6 7 8 9
Time
45.99
46
46.01
Mean(Weight)
-1 0 1 2 3 4 5 6 7 8 9
Time

12
Figure 10 Figure 11
Intact Intact
402C/755%RH 232C/753%RH
45.815
45.8175
45.82
45.8225
45.825
45.8275
45.83

Mean(Weight)
-1 0 1 2 3 4 5 6 7 8 9
Time

45.982
45.983
45.984
45.985
45.986
45.987
45.988
Mean(Weight)
-1 0 1 2 3 4 5 6 7 8 9
Time

Impact of Controls on Slopes

Tables 3 and 4 compare the mean slopes (WVTRs) for broached and intact containers at
the two test conditions. In both tables, the overall slopes adjusted for controls at both
conditions are 94 % or better of the slopes for desiccant filled bottles WVTRs obtained
for desiccant filled bottles alone are slightly higher than WVTRs obtained when controls
are subtracted as specified in USP <671>.
Table 3
Mean Slopes (mg/day) for Broached Seal
40C/75%RH 23C/75%RH
Site Desiccant
Adjusted
for
Control
Ratio
(%) Desiccant
Adjusted
for
Control
Ratio
(%)
MK 7.216 7.176 99.4 2.653 2.626 99.0
Pf 7.173 7.173 100.0 1.917 1.892 98.7
SA 8.283 8.233 99.4 2.399 2.377 99.1
AB 9.403 9.291 98.8 2.739 2.679 97.8
Overall 99.4 Overall 98.6

* Ratio = Adjusted for Control slope as a percent of Desiccant slope

13
Table 4
Mean Slopes (mg/day) for Intact Seal
Intact 40C/75%RH 23C/75%RH
Site Desiccant
Adjusted
for
Control Ratio* Desiccant
Adjusted
for
Control Ratio*
MK 1.179 1.188 100.8 0.286 0.262 91.6
Pf 1.186 1.208 101.9 0.271 0.252 93.0
SA 1.246 1.223 98.2 0.280 0.269 96.1
AB 1.230 1.208 98.2 0.275 0.263 95.6
Overall 99.7 Overall 94.1

* Ratio = Adjusted for Control slope as a percent of Desiccant slope

Comparison of RSDs; Effect of Temperature on Variation
Table 5 compares the RSDs for broached and intact containers with desiccant at the two
storage conditions. Considering the effect of outliers discussed above, the variation for
broached containers at 40oC/75%RH (mean =15.9) is slightly higher than at
23oC/75%RH (mean=13.9). Otherwise, the variations are similar, and there is little

difference in RSDs between the two conditions.

Table 5
RSD (%)*
40C/75%RH 23C/75%RH
Site Broached Intact Broached Intact
MK 10.16 2.17 30.50
(16.33)
2.12
Pf 9.31 1.82 13.66 2.75
SA 18.70 5.00
(1.72)
13.57 2.12
AB 16.27
(9.62)
1.47 12.33 1.66

* RSD without outlier in parentheses

Variation Between Sites, Runs and Containers
Table 6 is an analysis of the RSDs. It breaks the overall RSD into two components; the
variation associated with Between Sites and Runs and that associated with Between
Containers. The least variation occurs when intact and broached containers are tested at
40oC/75%RH, using regression analysis over all five data points. Using initial point to 14
days shows similar mean WVTRs, but in most analyses the variation is larger for the
initial to day 14 calculation than using regression. The RSDs for WVTR calculated using
the data from initial to day 14 at 23oC/75%RH are very large .
14
Table 6
Comparison of Type, Condition, and Slope Method Using RSD
Type Condition Slope Method
Mean
Slope
(mg/day)
RSD(%)
Between
Sites&Runs
RSD(%)
Between
Containers
RSD(%)
Total
Broached 23C/75%RH Initial - Day 14 2.44 16.44 23.92 29.02
Regression 1
(all) 2.43 14.31 19.99 24.58
Regression 2
(wo Initial) 2.42 14.06 19.40 23.96
Regression 3 2.42 13.50 19.93 24.07
40C/75%RH Initial - Day 14 7.12 17.56 17.43 24.74
Regression 1
(all) 8.02 12.59 14.92 19.52
Regression 2
(wo Initial) 8.18 11.96 14.52 18.82
Regression 3 8.14 10.89 14.99 18.53
Intact 23C/75%RH Initial - Day 14 0.36 44.36 5.97 44.76
Regression 1
(all) 0.28 2.35 2.18 3.20
Regression 2
(wo Initial) 0.27 4.15 2.34 4.77
40C/75%RH Initial - Day 14 1.28 15.59 3.47 15.97

Regression 1
(all) 1.21 2.63 3.02 4.00
Regression 2
(wo Initial) 1.21 2.72 3.39 4.34

Notes: (1) Initial-Day 14 is day point 0 and day point 14. (2) Regression 1 is regression
over days 0 to 42 or 49. (3) Regression 2 is regression over days 7 to 42 or 49. (3)
Regression 3 is regression on days 7, 14, 21, 28, and 35; not much different from using
42 or 49 days.
Conclusions
There was no apparent induction period in the main study; steady state appears to
have occurred by the 7 day time point.
The 9 day study showed irregularity and variable behavior during the first 6 days,
suggesting instability during this period which did not appear in the test as
specified by USP Chapter <671>.
Weight gain was linear over time at steady state; WVTR is constant.
Smaller variability occurred when WVTR was calculated by regression rather
than using the two- point calculation of USP <671>.
Mean WVTR rates and container variability were similar between labs within
each of the testing conditions.
The higher temperature resulted in higher WVTR rates. RSDs were similar
between the two temperatures.
Controls (blanks) had statistically significant but small positive slopes. In this
study, the controls had little effect on the WVTR rates for the broached bottles.
15
Controls had a greater effect on the intact bottle because of the smaller intact
WVTR rates.
Controls are not necessary (see Table 3).
Anhydrous calcium chloride should be pre-dried at 210o C, not 110o C as is
currently specified in USP <671> (see Appendix).
Recommendations
USP Chapter <671> should be rewritten or replaced.
For the purpose of moisture-barrier comparison, WVTR testing should be
conducted at 40oC/75% RH. This is an ICH accelerated condition.
The test duration should be 35 days, and containers weighed at 7 day intervals
WVTR should be calculated by regression analysis over the period from day 7 to
day 35.
Blanks (Controls) are not necessary, so they should not be included in the test
design
If calcium chloride is used as the desiccant, it should be pre-dried at 210oC

Appendix A
Selection of desiccant for bottle WVTR study
Anhydrous calcium chloride is used as a desiccant for measuring the WVTR of HDPE
bottles. This desiccant is specified in USP<671> with a requirement that the desiccant be
activated at 110C prior to use. The amount of the desiccant to use is also specified in
USP<671>, depending on container sizes. The purpose of using desiccant is to provide a
sink condition; hence a constant driving force of water vapor pressure over the course of
study at constant temperature and external humidity. For the current study, a sink
condition is defined as an internal relative humidity that does not exceed 5% over the
time course of study. To assure this criterion is met, the following procedures were

carried out to determine the suitable amount of the desiccant to use:

1. A literature search was conducted for information on the properties of anhydrous
chloride, including moisture sorption isotherms
2. The cumulative amount of water vapor that will permeate into the container
during the study was estimated.
3. The desiccant isotherm was linked with the estimated bottle WVTR to verify that
the criterion of not more than 5% RH is met.
16
Data in the literature shows that the relative humidity in air dried by anhydrous calcium
chloride is 1.6% at 25Ci. Literature also indicates that calcium chloride can exist in
multiple hydrates, and that the hexahydrate loses water at 200Cii. This information on
hydrates shows that the activation temperature must be higher than 200C if the desiccant
has been previous exposed to humidity and formed hexahydrate. The history of the
desiccant must be considered because during shipping, storage, and use the anhydrous
calcium chloride may be exposed to moisture at some points, hence care must be taken to
activate the material properly before use.
A preliminary experiment was conducted by a member laboratory to select the activation
temperature for anhydrous calcium chloride after the material was equilibrated at
6.4%RH at room temperature. Activation was conducted by heating at 110C and 210C.
Results showed that loss on drying was 0.5% and 4.5% by activation at 110C and
210C, respectively. These results showed that activation of anhydrous calcium chloride
at 210C is necessary to fully activate the desiccant even if the material has been exposed
to an environment of extremely low humidity.
To estimate the quantity of anhydrous calcium chloride needed to maintain a sink
condition for study, the desiccant moisture capacity was linked to the estimated quantity
of moisture ingress during the study. The moisture capacity of the desiccant activated at
210C is about 2% at 5%RH. The WVTR of the intact 60 mL bottle was estimated to be
0.2mg/day at 23C/75%RH, and 0.9mg/day at 40C/75%RH. Total ingression of
moisture over 98 days is 19.6mg and 88mg under the two study conditions. It can be
calculated that about 5g of fully activated anhydrous calcium chloride is needed to absorb
88 mg of the moisture penetrating the container to result in an internal relative humidity
of about 5%. In the current study, 30 g of fully activated anhydrous calcium chloride was
used for the 60 mL bottle. This amount of desiccant ensured that a sink condition of
<5%RH was maintained throughout the study. This quantity also provides a sink
condition for the broached bottle over a 49 day study (Table A1).
Table A1. Estimation of relative humidity in the headspace of 60 mL HDPE bottles
40C/75%RH
Actual
WVTR
mg/day
Test
Duration
Days
Total Water
Ingressed,
mg
Amount of
Desiccant,
g
Water/Desiccant,
% (w/w)
Headspace
RH iii

%
Intact 1.25 98 122.5 30 0.4% <5%
Broached 9.4 49 460.6 30 1.5% <5%
23C/75%RH
Intact 0.29 98 28.42 30 0.1% <5%
Broached 2.74 49 134.26 30 0.4% <5%
i. G.P. Baxter, H.W. Starkweather, 1916. The efficacy of calcium chloride, sodium hydroxide and
potassium hydroxide as drying agents. J. Am. Chem. Soc., 38, 2038-2041.

17
ii. S. Budavari, M.J. ONeil, A. Smith, P.E. Hekelman. The Merck Index, 11th Ed. Merck & Co., Inc.,
Rahway, N.J., 1989. p 252.
iii. Y. Chen, Y. Li, Determination of water vapor transmission rate (WVTR) of HDPE bottles for
pharmaceutical products. Int. J. Pharm., 358 (2008) 137-143.

Appendix B
Sample Preparation and Handling
Excerpted from PQRI Bottle WVTR Study Protocol, 5 September, 2006
Sample Preparation at the Manufacturing Plant
Desiccant
o Desiccant should be taken from the original container and opened just prior to
use. Place bulk desiccant in shallow container, taking care to exclude any fine
powder. Dry at 210 degree Celsius for a minimum of one hour and cool in a glass
desiccator for at least 30 minutes. The desiccator should be kept closed during
packaging of the bottles. No desiccant should be exposed to the ambient
environment unprotected for more than 10 min
Bottles with Desiccant
o To each of >240 bottles, add 30+-3.0 g of the desiccant. Immediately close the
bottles with the S&T CR cap, manually without significant torque.
o Using a calibrated instrument, close the bottles with the cap in accordance with
the procedure described in USP <671>. Activate the seal with the induction
sealer. The induction seal must pass ASTM D3078-02 at 15 inches mercury (or
equivalent) vacuum-dye leak test.
o Mark each bottle sequentially 1-240
Empty Bottles
o Follow the same procedure as described above and prepare >160 induction-sealed
bottles without any desiccant.
o Instead of desiccant add glass glass beads per USP<671> to the same volume as
the desiccant
o Mark each bottle sequentially 1-160
Shipping of Sample to Testing Labs
o Place at least 60 (e.g. 70) bottles containing desiccant and at least 40 (e.g. 50)
empty bottles (with caps) in an appropriate container and FedEx them to the 4
testing labs within 3 days.
18
Sample Preparation and Weight Measurement at the Testing Labs
Intact Bottles
o Separate 60 bottles-with-desiccant into two equal parts, A & B mark them with
non-erasable ink (e.g. A1, B2, etc.)
o Measure and record the initial weight to the 4th decimal point (0.1 mg) of each
bottle in group A (30 in total). Place 15 group-A bottles each in the 40C and
23C chambers, respectively. Measure and record the weight of each bottle every
14 days for 98 days, to the 4th (0.1 mg). Before each weighing, the bottles are
removed from the chamber and allowed to equilibrate for ~30 min in the room

(preferably with T/RH control to 75F (24C)/40%RH or below) where the

weighing takes place. Minimize the time-out-of-chamber to below 2 hrs.
o Mark the 40 empty bottles with non-erasable ink as C1, C2, etc.
o Place 10 empty bottles with intact seal in each of the environmental chambers at
the same time when the desiccated bottles are placed (20 bottles in total).
o Measure and record the weight of each of the empty bottles as for the desiccated
bottles.
Breached Bottles
o Breach the seal of group-B bottles (30 in total) according to the attached
procedure. Close these bottles with the cap according to the prescribed procedure
(USP <671>).
o Weigh and record the weight of each of the group-B bottles with breached seal to
the 4th decimal point (0.1 mg). Store 15 bottles each in the 23 and 40 chambers,
respectively, as with the intact bottles. Measure and record the weight of each
bottle to the 4th decimal point (0.1 mg) every 7 days for 49 days. Before each
weighing, the bottles are removed from the chamber and allowed to equilibrate
for ~30 min in a room (preferably with T/RH control to 75F (24C)/40%RH or
below) where the weighing takes place.
o Breach the seal of the remaining 20 empty bottles according to the prescribed
procedure. Close these bottles with the cap according to the prescribed procedure
(USP <671>).
o Place 10 empty bottles with breached seal in each of the environmental chambers
at the same time when the desiccated bottles are placed.
o Measure and record the weight of each of the empty bottles as for the desiccated
bottles.
19