Professional Documents
Culture Documents
ABSTRACT: A leak test performed according to ASTM F2338-09 Standard Test Method for Nondestructive Detection
of Leaks in Packages by Vacuum Decay Method was developed and validated for container-closure integrity
verification of a lyophilized product in a parenteral vial package system. This nondestructive leak test method is
intended for use in manufacturing as an in-process package integrity check, and for testing product stored on stability
in lieu of sterility tests. Method development and optimization challenge studies incorporated artificially defective
packages representing a range of glass vial wall and sealing surface defects, as well as various elastomeric stopper
defects. Method validation required 3 days of random-order replicate testing of a test sample population of
negative-control, no-defect packages and positive-control, with-defect packages. Positive-control packages were
prepared using vials each with a single hole laser-drilled through the glass vial wall. Hole creation and hole size
certification was performed by Lenox Laser. Validation study results successfully demonstrated the vacuum decay
leak test method’s ability to accurately and reliably detect those packages with laser-drilled holes greater than or equal
to approximately 5 m in nominal diameter. All development and validation studies were performed at Whitehouse
Analytical Laboratories in Whitehouse, NJ, under the direction of consultant Dana Guazzo of RxPax, LLC, using a
VeriPac 455 Micro Leak Test System by Packaging Technologies & Inspection (Tuckahoe, NY). Bristol Myers
Squibb (New Brunswick, NJ) fully subsidized all work.
KEYWORDS: ASTM F2338, Container closure, Container-closure integrity, Defects, Glass vial, Glass vial defects,
Leak, Leakage, Leak detection, Leak test method, Lyophilized product, Package, Package defects, Package integrity,
Package integrity method, Vacuum decay, Vacuum decay leak detection, VeriPac
LAY ABSTRACT: A leak test performed according to ASTM F2338-09 Standard Test Method for Nondestructive
Detection of Leaks in Packages by Vacuum Decay Method was developed and validated to detect defects in stoppered
vial packages containing lyophilized product for injection. This nondestructive leak test method is intended for use
in manufacturing as an in-process package integrity check, and for testing product stored on stability in lieu of sterility
tests. Test method validation study results proved the method capable of detecting holes laser-drilled through the glass
vial wall greater than or equal to 5 m in nominal diameter. Total test time is less than 1 min per package. All method
development and validation studies were performed at Whitehouse Analytical Laboratories in Whitehouse, NJ, under
the direction of consultant Dana Guazzo of RxPax, LLC, using a VeriPac 455 Micro Leak Test System by Packaging
Technologies & Inspection (Tuckahoe, NY). Bristol Myers Squibb (New Brunswick, NJ) fully subsidized all work.
(2). Integrity tests are also an important component of lenge the method in regard to sensitivity and reliabil-
packaging line qualification studies and as in-process ity. For instance, leaks located below the product fill
checks of routine manufacturing operations. level might clog the flow of leaking gas, preventing
their detection; therefore, this study used product-
The current research describes the development, opti- filled, with-defect packages to evaluate this possibil-
mization, and validation of a container-closure integ- ity. Large leaks might prove hard to detect, as the
rity test intended for a lyophilized dosage form con- package headspace gas may be lost during a lengthy
tained in a parenteral vial package system. The method test system evacuation, prior to test start. For this
chosen follows ASTM F2338-09 Standard Test reason, larger defect sizes were included, and with-
Method for Nondestructive Detection of Leaks in defect packages were filled with glass beads to simu-
Packages by Vacuum Decay Method (3). This ap- late typical product headspace volumes. Leaks at the
proach is nondestructive, simple to perform, and ap- stopper-vial interface under the cap might require a
propriate for packages of lyophilized dosage forms not longer test time to allow leaking gas to escape from
requiring a headspace of vacuum or nonreactive gases. the package. Therefore, positive controls included
ASTM method precision and bias (P&B) studies sup- packages with channels cut across the glass vial finish
port the sensitivity and reliability of this approach for and with various stopper defects. Finally, vials with
parenteral product packages (3, 4). Leak detection holes laser-drilled through the wall simulating cracks
results reported in the ASTM P&B work demonstrate or chips were also included; each hole was sized for
this method’s ability to detect leaks as small as ap- nominal diameter using a positive air flow measure-
proximately 5m in nominal diameter in rigid, non- ment technique. Holed vials enabled verification of the
porous packages. Published studies have shown opti- test method’s sensitivity limit.
mized vacuum decay methods to be more sensitive and
reliable than compendial and ISO standard dye ingress Once vacuum decay test method parameters were op-
tests (5). timized, final method validation entailed testing a
sample population of negative-control, no-defect
Prior to each day of testing, the suitability of the packages and positive-control packages made using
vacuum decay leak test system was confirmed by vials with laser-drilled holes (size certified). All vali-
introducing a controlled leak into the test system via a dation study packages contained lyophilized product.
National Institute of Standards and Technology The test sample population was tested in random order
(NIST)-certified air flowmeter while conducting leak over three replicate test days.
tests using no-defect master packages. The consistent,
controlled airflow challenge will yield repeatable and All research was financed by Bristol Myers Squibb,
reliable test results given an integral leak test system. Co. (BMS) of New Brunswick, NJ, and was conducted
These system suitability checks used a test method at Whitehouse Analytical Laboratories (WAL) in
with pass/fail criteria established during instrument Whitehouse, NJ, under the direction of Dana M.
performance qualification. This method is not the final Guazzo of RxPax, LLC.
optimized test method for the finished product package
detailed in the current report. A glossary of terms is provided at the conclusion of
this report.
Development and optimization of the final test method
for the finished product package utilized actual pack- Materials and Methods
age test samples representing no-defect product pack-
ages (negative controls) as well as packages with 1. Test System
artificially created defects (positive controls). By test-
ing a population of negative-control, no-defect pack- All vacuum decay leak tests employed a VeriPac 455
ages, final method parameters and acceptance criteria Micro Leak Test System by Packaging Technologies
took into account virtual leakage derived from outgas- & Inspection (PTI) of Tuckahoe, NY. The VeriPac
sing of moisture or other volatiles from package com- 455 system is a non-destructive benchtop leak tester.
ponents and/or release of air trapped under the alumi- The system consists of a 1000 Torr absolute pressure
num cap. Positive control packages representing a transducer (measurements in mbar units) plus a 10
variety of possible defect scenarios were included in Torr differential transducer (measurements in Pa
method development and optimization to fully chal- units), an external vacuum pump, and an external test
Figure 1
A. PTI VeriPac 455 Micro Leak Test System. B. Microcalibrator airflow meter. C. PTI VeriPac 455 Micro Leak
Test System functional diagram.
chamber specifically designed for the container to be test system under constant vacuum while a closed
tested. The design of the leak test system, including valve separates the test chamber from the evacuated
the test chamber, ensures minimal deadspace volume, test system (Maintaining the system under a vacuum
so the smallest leaks will trigger maximal pressure rise between tests provides more consistent, reliable test
readings. (Test chambers designed to closely fit the results by preventing the accumulation of volatiles in
given test package system are required to ensure max- the test system, stabilizing the absolute pressure trans-
imum test method sensitivity.) A microcalibrator by ducer, and shortening the evacuation time segment.)
Furness Controls, Ltd. (Indian Trail, NC) with an Upon test sequence activation, this valve opens allow-
operational range of 0.001 to 2.000 cm3 䡠 min⫺1 ing the rapid establishment of a vacuum in the test
(⫾0.001 cm3 䡠 min⫺1) was used to introduce a con- chamber. When the evacuation time is complete, a
trolled leak into the system as part of suitability tests second valve closes, isolating the vacuum source from
performed prior to each day of operation, and to the fully evacuated test system plus test chamber. The
challenge the leak test system during method devel- absolute pressure transducer measures pressure in the
opment. Photographs of the test system and the mi- test system plus test chamber (system/chamber)
crocalibrator flowmeter are shown in Figures 1A and throughout evacuation.
1B, respectively. A test system functional diagram is
included in Figure 1C.
2.2. Equalization Time (Seconds): An equalization
time segment immediately follows evacuation. This
2. Test Method Overview
time segment allows for system/chamber pressure
The test is performed by first placing a test package equilibration (stabilization). The absolute pressure
inside the test chamber. The test chamber is closed and transducer monitors system/chamber pressure through-
the operator activates the test sequence. The test se- out equalization.
quence consists of three sequentially programmed
time segments discussed below. The length of each 2.3. Test Time (Seconds): A test time segment fol-
time segment is developed specifically for each prod- lows equalization while the absolute pressure trans-
uct-package combination. Two programmed pressure ducer continues to monitor the pressure rise in the
reference limits, also described below, define each test system/chamber. The differential pressure transducer
as either pass (no leak) or fail (leak). These limits are simultaneously tracks the net rise in pressure within
also distinct for each product-package system. the system/chamber from the beginning to end of the
test time. At the completion of this last time segment,
2.1. Evacuation Time (Seconds): Prior to test start, the system/chamber vents allowing test package re-
an external vacuum source maintains the VeriPac 455 moval from the test chamber. Immediately post-vent-
ing, the valve separating the test system from the 3. Test Sample Population
test chamber closes, the valve between the vacuum
source and the test system opens, and the test system 3.1. Product-Package System: The lyophilized prod-
returns to its pre-test evacuated state. The test result uct was packaged in a 20 mL capacity glass vial of
is reported as either pass or fail in accordance with Type I tubing with a 20 mm finish diameter. Each vial
the established pressure reference limits. In addi- was stoppered using a 20 mm lyophilization-style
tion, the final pressure readings from the absolute stopper, and was closed with an aluminum crimp cap
and differential pressure transducers are displayed with a plastic flip top. The headspace in each product-
and recorded. filled package consisted of air at ambient pressure.
2.4. Pressure Reference Limits: Tests pass or fail 3.2. Test Packages: Test packages included master
according to two pressure reference limits unique to packages guaranteed free of leaks, negative-control
each product-package combination. For a test sample test packages with no known defect, and positive-
to pass (i.e., to show no leakage), test results must control packages with known package defects, all
comply with both pressure reference limit specifica- briefly described below. Test packages specified for
tions. A test will fail (i.e., show leakage) if either test system suitability, method development, and method
pressure reference limit specification is not met. validation are identified in Table I. Test package and
component illustrations are included in Figures 2–12.
2.4.1. Absolute Pressure Reference Limit (mbar)
The test will pass if the absolute pressure trans- 3.2.1. Master packages (test sample ID codes 5-1 to
5-5), guaranteed leak-free, were filled to over-
ducer reading monitored from equalization start
flow capacity with silicone sealant, stoppered,
to the end of the test is less than the absolute
and capped. Master packages were used for
pressure reference limit (APRL). The test will
system suitability tests and as part of method
fail (and the cycle will abort) if the absolute
development and method validation. (Note:
pressure is equal to or greater than the APRL.
Masters made of other materials were explored
but were rejected. Vacuum decay results of
2.4.2. Differential Pressure Reference Limit (Pa)
masters made of the same vial/stopper/cap com-
The test will pass if the differential pressure
ponents as actual product packages more
transducer reading monitored throughout the test
closely mirrored results of actual product-filled,
time segment remains less than the differential
no-leak packages.)
pressure reference limit (DPRL). The test will
fail (and the cycle will abort) if the differential
3.2.2. Negative-control packages for method develop-
pressure is equal to or greater than the DPRL. ment included empty vials, stoppered and
capped at WAL (test sample ID codes 2-1 to
2.5. System Flush: The vacuum decay leak test in- 2-10), and product-filled vial packages assem-
strument/chamber itself must be relatively free of bled by BMS (test sample ID codes 13-1 to
virtual and real leaks in order to detect the smallest 13-10). Negative-control packages used for
leakage from test packages. To prevent the accumu- method validation were all product-filled and
lation of volatiles in the test system that might yield assembled by BMS (test sample ID codes N-1
a virtual leak pressure rise, the test system is pro- to N-70).
grammed to maintain vacuum conditions between
tests, as previously noted in Section 2.1. Test cham- 3.2.3. Positive-control packages with defective stop-
ber sealing surfaces are cleaned periodically to pre- pers (test sample ID codes S-1, S-6, and S-7)
vent debris build-up on test chamber sealing sur- used during method development were filled
faces that could cause background leakage. with glass beads to simulate normal product fill
Cleaning is performed by wiping surfaces with a level. Defective stoppers were either pierced
low-lint wipe moistened with distilled or purified using a needle or cut with scissors. These pack-
water. After cleaning, the chamber is closed and the ages were prepared by WAL.
test system/chamber is flushed (i.e., evacuated) in
order to dry gasket surfaces. A 60 s flush time was 3.2.4. Positive control packages used for method de-
selected for this study. velopment included a variety of vial defect
TABLE I
Test Sample Identification
Package
Identification
Code Description Defect Size
System suitability
5-1 to 5-5 Master no-defect packages n/a
Filled with silicone sealant
2-1 to 2-10 Negative-control packages n/a
Empty, assembled at WAL
Method development
5-1 to 5-5 Master no-defect packages n/a
Filled with silicone sealant
2-1 to 2-10 Negative-control packages n/a
Empty, assembled at WAL
13-1 to 13-30 Negative-control packages n/a
Product-filled, supplied by BMS
S-1 Stopper defect, 21gauge needle single pierce 21 gauge puncture
through closure
Glass-bead-filled package
S-6 Stopper defect, single radial cut through flange Closure top radius
across land seal (Not measured)
Glass-bead-filled package
S-7 Stopper defect, land seal flange outer edge removed Flange width
Glass-bead-filled package (Not measured)
V-1 to V-5 Vial defect, land seal surface channel 0.74–89 mm (W) ⫻
Glass-bead-filled packages 0.76–0.96 mm (H)
V-6 to V-8 Vial defect, valve seal surface channel 1.12–1.65 mm (W) ⫻
Glass-bead-filled packages 0.27–1.08 mm (H)
V-9 to V-11 Vial defect, land ⫹ valve seal surfaces channel Valve: 0.13–1.34 mm (W) ⫻
Glass-bead-filled packages 0.41–0.97 mm (H)
Land: 0.88–1.24 mm (W) ⫻
0.16–1.80 mm (H)
27-1-, 27-3- to 27-5- Vial with hole in neck 2.39–3.38 m
27-6- to 27-10- -X: stopper leg blocking hole 12.16–14.14 m
27-11- to 27-15- -O: open hole; no stopper leg blocking hole 4.74–5.58 m
Glass-bead-filled package
Method validation
5-1 to 5-5 Master no-defect packages n/a
Filled with silicone sealant
N-1 to N-70 Negative-control n/a
Product-filled, supplied by BMS
P-1 to P-10 Vial with hole in neck 2.39–3.17 m
Product-filled, hand-assembled at WAL
P-11 to P-20 Vial with hole in neck 4.36–6.02 m
Product-filled, hand-assembled at WAL
P-21 to P-30 Vial with hole in neck 12.08–15.04 m
Product-filled, hand-assembled at WAL
Figure 4
Test package components: glass vial, elastomeric Evacuation time should be just long enough to estab-
stoppers, aluminum caps with plastic flip tops. lish consistent system/chamber target pressure read-
Figure 7
Figure 8
Figure 6
Glass vial defect V-1. Land seal surface channel.
Stopper defect S-6. Single radial cut. 25ⴛ (Magellan V200 Videomicroscope).
Figure 10
Figure 9
Glass vial defect V-9. Land ⴙ valve seal surfaces
channel. 25ⴛ (Magellan V200 Videomicroscope).
Glass vial defect V-6. Valve seal surface channel.
Top view: 25ⴛ; Bottom view: 115ⴛ (Magellan
V200 Videomicroscope).
performed to demonstrate the detection limit of the
leaking packages when tested according to the pro- final vacuum decay leak test method for testing BMS
posed time segment limits (prior to final pressure product-filled packages with and without laser-drilled
reference limits selection, programmed pressure limit hole defects, using the PTI VeriPac 455 Micro Leak
values should be quite high to prevent test cycle Test System. A random-order test sample population
abortion). The final time segment lengths and pressure of 70 negative-control, no-defect packages and 30
reference limits chosen should permit no-leak pack- positive-control, with-defect packages were tested in
ages to pass, while detecting with-leak packages. replicate over 3 days. Method validation acceptance
criteria required all negative-control packages to pass
The current study considered package-to-package and (no leak detected), and all positive-control packages
day-to-day differences when selecting time and pres- made with laser-drilled holes in the vial neck 5 m
sure limits. Varying time gaps between tests were also and larger in nominal diameter to fail (leak detected).
explored. ASTM F2338-09 P&B studies (3) also in-
vestigated variability among instruments and test sites. Results and Discussion
In the present study, research was limited to one
instrument/test chamber and one test site. 1. System Suitability Tests
5. Package Leak Test Method Validation Suitability tests were performed each day of operation
to verify the VeriPac 455 test system’s functionality
After method optimization and pressure limit accep- and to confirm the absence of system/chamber leaks.
tance criteria were defined, a validation study was System suitability checks used a test method with
Figure 11 Figure 12
Glass vial defect V-10. Land ⴙ valve seal surfaces Laser-drilled hole in vial neck. Top image, left to
channel. 25ⴛ (Magellan V200 Videomicroscope). right: Vial 27-3—Hole in “O” closure position,
Hole in “X” closure position. Bottom image: 2.93
m nominal hole diameter, by Lenox Laser, of Glen
pass/fail criteria established during instrument perfor-
Arm, MD. 25X (Magellan V200 Videomicroscope).
mance qualification. This method is not the final op-
timized test method for the finished product package
detailed in the current report. Test acceptance criteria erence limits were set based on data from no-leak
for system suitability were mean absolute and differential pressure readings mea-
sured over multiple days of testing. Table II outlines
● All 10 leak tests of master packages must be the parameter limits of the final system suitability test
passed. method. All current study system suitability tests per-
formed met the acceptance criteria, namely, all master
● All five leak tests of master packages with a NIST
airflow challenge of 0.100 ⫾ 0.005 (cm3 䡠 min⫺1)
TABLE II
introduced into the test system must show failure.
System Suitability Test Method Parameter
Settings
The time parameter limits for system suitability tests
were established using a set of five master packages Parameter Setting
and 10 negative-control, product-filled packages to Evacuation 6.00 s
represent no-leak tests, and using the NIST microcali-
Equalization 20.00 s
brator airflow meter to introduce a controlled leak into
Test 10.00 s
the test system. The test parameter selection procedure
Final absolute pressure reference limit 2.7 mbar
outlined in Materials and Methods, Section 4, was
followed. The absolute and differential pressure ref- Final differential pressure reference limit 30 Pa
tests passed and all with-leak tests (0.100 cm3 䡠 min⫺1) defects (V-2, V-4) and all vials with channels cut
failed. across both valve and land seal surfaces (V-9 to V-11)
gave absolute pressure readings much greater than
2. Package Leak Test Method Development those of the no-defect vials. Therefore, with appropri-
ate APRL settings, packages with such large defects
The final package integrity test method was required should be detectable following a 6 s evacuation time.
to meet the following acceptance criteria. The other vials with either land channels or valve
channels (V-1, V-3, V-5 to V-8) exhibited absolute
● All packages with no known defect must pass (i.e., pressure readings matching the negative control pack-
no leakage detected). ages (0.1 mbar) after 6 s evacuation; these vials either
leaked more slowly or the closures effectively sealed
● All packages with a hole in the vial of greater than the channel paths. During all evacuation optimization
or equal to 5 m in nominal diameter must fail tests the system suitability test method APRL of 2.7
(i.e., leakage detected). mbar was used; the DPRL was irrelevant, as tests did
not proceed beyond initial evacuation.
Method time and pressure parameter limits were op-
timized according to the test parameter selection pro- 2.2. Equalization Time Optimization:Negative-con-
cedure presented in Materials and Methods, Section 4, trol, product-filled BMS packages (13-1 to 13-10)
and described in detail in the following sections. Pack- were tested both with and without the addition of a
ages used for this work included master packages and NIST airflow artificial leak (0.100 cm3 䡠 min⫺1), using
negative-control packages filled with lyophilized a 6 s evacuation period, followed by variable equal-
product. Positive-control packages were used to inves- ization times (Tables IVa, IVb). A 20 s equalization
tigate the method’s ability to find various defect types. time produced consistent absolute pressure readings
Positive-control packages included vials with laser- for negative-control packages. During the equalization
drilled holes, vials with channel defects cut across time segment, noticeable differences could be ob-
finish surfaces, and artificially defective closures (re- served between no-leak-added and with-leak-added
fer to Table I for a list of negative and positive pressure readings.
controls).
Masters and negative-control and positive-control
2.1. Evacuation Time Optimization:Negative-con- packages were subsequently tested using a 20 s equal-
trol, product-filled packages supplied by BMS (13-1 to ization time segment, preceded by either 6 s or 7 s
13-10) were exposed to evacuation times ranging from evacuation (Table IVc). Tests using 6 s evacuation
3 to 7 s (Tables IIIa, IIIb). An evacuation time of 6 s yielded pressure readings 10 fold greater for bigger
was sufficiently long to ensure consistent 0.1 mbar channel defect packages (V-2, V-4, V-9, V-10, V-11),
absolute pressure readings at the end of evacuation and for packages with larger 12 to 14 m holes
time for all negative-control packages tested. Positive- (27-6-O to 27-10-O), as compared to no-defect pack-
control packages with vial channel defects (V-1 to ages (masters, negative controls). Some smaller chan-
V-11) were then tested with an evacuation time of 6 s nel defect packages (V-5 to V-8) and vials with holes
(Table IIIc). Two vials with land surface channel about 5 m in nominal diameter (27-11-O to 27-15-O)
TABLE III
Test Method Evacuation Time Optimization
IIIa. Test Parameter Settings
Parameter Setting
Evacuation Variable
Equalization 0.00 s
Test 0.00 s
Absolute pressure reference limit 2.7 mbar
Differential pressure reference limit N/A1
1
Differential pressure reference limit is irrelevant as test time ⫽ 0.00s.
demonstrated absolute pressure readings about 2 or 3 Extending equalization time from 20 s to 60 s was
fold greater than no-defect packages. There was no attempted to see if very long equalization may produce
difference in pressure readings between negative-con- lower, more consistent test time differential pressure
trol packages and packages with holes smaller than 5 readings for negative-control packages, making it pos-
m. Increasing evacuation time to 7 s lowered abso- sible to detect the smallest vial-hole leaks during the
lute pressure readings for several of the known-leak- final test time segment (Table V). A 20 s test time
ing packages tested, likely due to more package head- segment and a DPRL of 100 Pa was chosen for these
space gas loss from these larger leak packages during tests (Table Va). Positive controls included packages
test cycle evacuation (Table IVc). Therefore, a 6 s having 2.4 to 3.4 m holes blocked by the closure leg
evacuation time rather than 7 s, coupled with 20 s for (27-1-X to 27-5-X). A 60 s equalization time did result
equalization, was preferred to ensure detection of the in lower differential pressure readings, but no signif-
largest package leaks. icant difference in differential pressures was observed
TABLE IV
Test Method Equalization Time Optimization, 0 –20 s
IVa. Test Parameter Settings
Parameter Setting
Evacuation 6.00 s or 7.00 s
Equalization Variable
Test 0.00 s
Absolute pressure reference limit 50 mbar
Differential pressure reference limit N/A1
1
Differential pressure reference limit is irrelevant as test time ⫽ 0.00s.
IVb. Negative Control Tests, with and without NIST Airflow Leak
TABLE V
Test Method Equalization Time Optimization, 20 s versus 60 s
Va. Test Parameter Settings
Parameter Setting
Evacuation 6.00 s or 7.00 s
Equalization Variable
Test 20.00 s
Absolute pressure reference limit 50 mbar
Differential pressure reference limit 100 Pa
Vb. Negative and Positive Control Tests
between negative controls and the smallest-hole pack- results duplicate no-defect packages for all tests at-
ages (Table Vb). Therefore, lengthening equalization tempted).
beyond 20 s proved unbeneficial.
A microcalibrator was used to introduce a NIST trace-
2.3. Test Time Optimization: Test time optimization able airflow leak into the test system/chamber during
trials included test time segments of 10 s, 20 s, and negative-control package leak tests to evaluate a 30 s
30 s (evacuation and equalization times were kept test time leak test method (with 6 s evacuation and
constant at 6 s and 20 s, respectively) (Table VI). 20 s equalization, Table VII). Absolute and differen-
Absolute and differential pressure reference limits tial pressure reference limits were set high (50 mbar
were set high (50 mbar and 100 Pa) to prevent pre- and 100 Pa) to avert test cycle abortion. Tests with a
mature test cycle abortion. 0.050 cm3 䡠 min⫺1 microcalibrator leak yielded similar
results to packages with holes about 3 m in nominal
As summarized in Table VIb, the shortest 10 s time diameter, with pressure readings about twice those of
segment easily permitted reliable detection of vial negative-control packages (refer to Table VIb, 30 s
holes greater than or equal to about 5 m, producing tests). With-leak data on day 1 versus day 2 were
absolute and differential pressure readings well above essentially identical. Thus a test time of 30 s supports
master or no-defect packages. Lengthening the test the required leak size detection limit of 5 m and
time to 20 s or 30 s made it possible to reliably detect larger with a comfortable safety margin.
vials with holes of about 3 m, as differential pressure
readings of with-hole packages were as much as 2 fold 2.4. Test Method Repeatability, and Proposed
those of nonleaking packages. Upright and inverted APRL and DPRL Determination: The proposed leak
packages tested for 30 s demonstrated no difference in test method— 6 s evacuation, 20 s equalization and
results, implying that powdered product, simulated by 30 s test—was used to perform 3 day replicate tests of
glass beads, would have no impact on test results (the a random order population set of positive- and nega-
hole in test package 27-2-O appears to have clogged; tive-control packages in order to verify the method’s
TABLE VI
Test Method Test Time Optimization
VIa. Test Parameter Settings
Parameter Setting
Evacuation 6.00 s
Equalization 20.00 s
Test Variable
Absolute pressure reference limit 50 mbar
Differential pressure reference limit 100 Pa
VIb. Master, Negative and Positive Control Packages Tests
Test 30.00 s
Test 10.00 s Test 20.00 s Test 30.00 s Inverted
Diff Diff Diff Diff
Package Abs P P Abs P P Abs P P Abs P P
ID Description (mbar) (Pa) (mbar) (Pa) (mbar) (Pa) (mbar) (Pa)
5-1 to 5-3 Masters 0.7–0.8 20–22 0.8–1.2 28–30 0.9–1.4 43–45 Not Not
tested tested
2-1 to 2-10 Negative controls, 1.0–1.6 15–19 1.4–1.8 24–30 1.5–1.9 36–40 1.5–2.0 40–46
empty packages
27-1-O Vial 3.38 m 2.1 37 2.8 66 3.3 97 3.7 97
hole, open
27-3-O Vial 2.93 m 1.7 29 2.0 44 2.9 55 2.6 66
hole, open
27-4-O Vial 2.39 m 1.4 25 2.1 51 2.7 72 2.9 75
hole, open
27-5-O Vial 2.67 m 1.7 32 2.5 57 3.1 84 3.4 80
hole, open
27-11-O Vial 4.92 m 4.1 73 Not Not Not Not Not Not
hole, open tested tested tested tested tested tested
27-12-O Vial 5.09 m 4.1 78 Not Not Not Not Not Not
hole, open tested tested tested tested tested tested
27-13-O Vial 4.92 m 3.8 67 Not Not Not Not Not Not
hole, open tested tested tested tested tested tested
27-14-O Vial 4.74 m 3.4 61 Not Not Not Not Not Not
hole, open tested tested tested tested tested tested
27-15-O Vial 5.58 m 5.5 98 Not Not Not Not Not Not
hole, open tested tested tested tested tested tested
day-to-day repeatability. For these tests, the absolute The test sample population included 40 negative-con-
and differential pressure reference limits were set in- trol packages, five negative-control packages with
tentionally high at the time of testing in order to NIST airflow leak added, and 24 positive-control,
prevent test cycle abortion, enabling end-of-test pres- with-defect packages. Thirty negative-control pack-
sure readings for all but the largest leaks (APRL ⫽ 50 ages included BMS-manufactured product packages
mbar, DPRL ⫽ 100 Pa). Absolute and differential and 10 WAL hand-assembled empty packages. Posi-
pressure readings were necessary so that appropriate tive-control packages included vials with laser-drilled
APRL and DPRL values could be calculated for the holes ranging in size from 2.39 to 14.14 m, all with
final test method, taking into account normal day-to- stoppers positioned so that the inserted stopper leg
day and sample-to-sample pressure reading variations. blocked the vial neck hole. Other positive controls
TABLE VII
Test Method Sensitivity Verification Using NIST Airflow Leaks
VIIa. Test Parameter Settings
Parameter Setting
Evacuation 6.00 s
Equalization 20.00 s
Test 30.00 s
Absolute pressure reference limit 50 mbar
Differential pressure reference limit 100 Pa
VIIb. Negative Control Tests, with and without NIST Airflow Leaks
included stoppers with a single needle puncture (S-1), channel along both valve and land surfaces (V-9 to
a radius cut (S-6), and a missing flange edge (S-7). V-11).
Vials with channel defects included those with a chan-
nel at the land surface only (V-2, V-4), with a channel Test results from the no-defect packages tested ac-
along the valve or neck only (V-6 to V-8), and with a cording to the parameter settings listed in Table VIIIa
TABLE VIII
Proposed Final Method Repeatability Tests
VIIIa. Test Parameter Settings
Parameter Setting
Evacuation 6.00 s
Equalization 20.00 s
Test 30.00 s
VIIIb. Negative-Control Packages1 Statistical Analysis
are summarized in Table VIIIb. Final test method 䡩 all packages with vial valve land seal channels
APRL and DPRL were calculated by adding six stan- only (V-6, V-7, V-8)
dard deviations to the no-defect package absolute and
differential pressure reading means, respectively. The Based on these results, the final test method with
calculated final APRL was 3.7 mbar, and the DPRL proposed pressure reference limits met the required
was 67 Pa. With these limits as leak test pass/fail package integrity leak test acceptance criteria.
criteria, the 3 day replicate test results were re-exam- Namely, all packages with no known defect passed,
ined, and the following observations summarized in and all packages with a hole in the vial greater than
Table VIIIc were made. or equal to 5 m in nominal diameter failed. The
additional positive control tests demonstrated the
● Packages that failed (i.e., leakage was detected) method’s ability to detect various defect types and
included sizes. For instance, the method detected an air leak
of 0.050 cm 3 䡠 min ⫺1 introduced into the test sys-
䡩 all packages with vial laser-drilled holes greater tem/chamber. The method detected vial hole defects
than or equal to 3.38 m in nominal diameter as small as 3.38 m in nominal diameter. The four
hole defects missed were 2.39 –2.93 m in diameter,
䡩 all negative control packages with a NIST 0.050
with differential pressure measurements of 54 – 61
cm3 䡠 min⫺1airflow leak
Pa, closely approaching the 67 Pa DPRL limit. The
method also found vial channel defects that permit
䡩 all packages with vial land seal surface channels
package headspace gas leakage at the stopper/vial
only (V-2, V-4)
land seal surface interface. Vial neck defects not
䡩 all packages with vial land plus valve seal sur- affecting the land sealing surface did not cause
face channels (V-10, V-11) package leakage. Stopper defects not detected in-
cluded a single 21-gauge needle puncture, a single
● Packages that passed (i.e., no leakage detected) radius cut, and a missing closure flange edge. The
included method was unable to identify these defects likely
due to the viscoelastic nature of the closure formu-
䡩 all negative-control, no-leak packages lation making it possible for the closure to re-close
when punctured or cleanly cut, and to conform to
䡩 all packages with vial laser-drilled holes less vial surfaces when compressed and capped into
than 3.38 m in nominal diameter place. The inability to find certain stopper defects
by vacuum decay, and the lack of leakage from vials
䡩 all stopper defect packages (S-1, S-6, S-7) having only neck defects, both illustrate the point
that no leak test method exists that can prevent or and 15 m (range, 12.08 –15.04 m). Validation study
detect all package defect types. Other means are test samples are listed in Table I.
necessary to preclude defects such as these from the
package assembly line. To avoid risk of false-positive results due to test
chamber gasket debris, test chamber sealing surfaces
3. Package Leak Test Method Validation were wiped clean followed by a 60 s system flush at
regular intervals throughout validation. Nevertheless,
3.1. Purpose:A validation study was performed to the protocol included the following procedure to ad-
demonstrate the detection limit of the final vacuum dress possible false-positive test results.
decay leak test method for testing BMS product-filled
packages with and without laser-drilled hole defects, ● If a test package result was questionable (e.g., a
using the PTI VeriPac 455 Micro Leak Test System. no-defect package fails), the test package was re-
moved from the test chamber, the chamber sealing
3.2. Test System Suitability:Prior to each test day, surfaces were cleaned, and a 60 s system flush was
the VeriPac 455 test system was shown suitable for performed. A master was tested. Once the master
use by successfully meeting the system suitability test passed, confirming test system performance
criteria (refer to Table II for system suitability test and test chamber integrity, the questionable pack-
method parameters). Specifically, five master pack- age was retested.
ages were leak-tested using a test cycle consisting of
6 s evacuation, 20 s equalization, and 10 s test, with ● If the test package retest result was pass, then the
APRL and DPRL of 2.7 mbar and 30 Pa, respectively. initial result was considered invalid due to debris
Each master package was tested once with the NIST- on the test chamber sealing surfaces or operator
traceable microcalibrator closed (no leak). Then each error (e.g., the test chamber was improperly
master was tested again with the microcalibrator open closed). The repeat test result was the final result.
to introduce an airflow rate of 0.100 cm3 䡠 min⫺1 into The test package passed.
the test system/chamber. Finally, each master was
tested again with the microcalibrator closed. The test ● If the test package retest result showed failure,
system was suitable for use if all 10 tests with micro- both the original and the repeat tests were consid-
calibrator closed passed (no leak detected), and all five ered final. The test package failed.
tests with microcalibrator open failed (leak detected).
3.4. Acceptance Criteria:The following acceptance
3.3. Validation Protocol:The optimized leak test criteria were applied to test results on each day of
method was used to test a random-order test sample validation.
population of negative- and positive-control packages
on each of three replicate test days. The final method ● All 70 negative-control packages must pass.
consisted of 6 s evacuation, 20 s equalization, and 30 s
test time, with an APRL and DPRL of 3.7 mbar and 67 ● All 10 positive-control packages with holes ap-
Pa, respectively. The refrigerated test samples were proximately 5 m in nominal diameter must fail.
tested after coming to ambient room temperature to
avoid false-positive results caused by package surface ● All 10 positive-control packages with holes ap-
condensation. proximately 15 m in nominal diameter must fail.
Test samples included 70 negative-control, no-defect Test results obtained with the 10 positive-control
lyophilized product packages, filled and assembled by packages with holes approximately 3 m in nominal
BMS. Test samples also included 30 positive-control, diameter were included in the validation study for
with-defect packages prepared at WAL by transferring information only.
lyophilized product from BMS-supplied packages to
vials with a single laser-drilled hole in the vial neck, 3.5. Results:
hand-stoppered and capped using a hand-crimping
tool. Three groups of 10 holed vials were included, 3.5.1. Test System Suitability Tests
representing nominal hole diameters of about 3 m Test system suitability criteria were success-
(range, 2.39 –3.17 m), 5 m (range, 4.36 – 6.02 m), fully met each day of validation. All 10 tests
TABLE IX
Final Method Validation Tests
IXa. Test Parameter Settings
Parameter Setting
Evacuation 6.00 s
Equalization 20.00 s
Test 30.00s
Absolute pressure reference limit 3.7 mbar
Differential pressure reference limit 67 Pa
IXb. Positive and Negative Control Tests Summary
with microcalibrator closed passed (no leak de- holes were smaller, ranging in size from 2.39 to
tected), and all five tests with microcalibrator 2.67 m in nominal diameter. Creating and
open to deliver a 0.100 cm3 䡠 min⫺1 leak failed sizing such tiny holes is challenging. These
(leak detected). holes are prone to clogging with environmental
debris, making their viability questionable. In
3.5.2. Package Leak Test Method Validation Tests addition, variability in pressure rise readings
Test method results are summarized in Table observed with negative control packages, com-
IX. In short, test method validation criteria bined with test system degassing, make reliable
were successfully met on each of day of vali- detection of the minuscule pressure rise from
dation. All 70 negative control packages (N-1 leaks much smaller than 5 m difficult.
to N-70) passed (leak detected). All 10 positive
control packages with holes approximately 5 Conclusion
m in nominal diameter (P-11 to P-20) failed
(leak detected). All 10 positive-control pack- A vacuum decay leak test method was successfully
ages with holes approximately 15 m in nom- developed and validated for nondestructive container
inal diameter (P-21 to P-30) also failed. closure integrity testing a BMS lyophilized product
packaged in a glass vial, stoppered, and closed with an
The 10 positive-control packages with holes aluminum seal. This nondestructive leak test method is
approximately 3 m in nominal diameter (P-1 intended for use in manufacturing as an in-process
to P-10) were included in the validation study package integrity check, and for testing product stored
for information only. Three of 10 packages in on stability in lieu of sterility tests.
this hole size group failed (were detected),
seven of 10 passed (were not detected). Inter- The method is performed according to ASTM
estingly, each of the three holes detected on all F2338-09 Standard Test Method for Nondestructive
three days were 3.17 m in nominal diameter Detection of Leaks in Packages by Vacuum Decay
(P-2, P-9, P-10). The remaining undetected Method. The instrument used for all work was the PTI
VeriPac 455 Micro Leak Test System. The time re- through which the flow occurs. Fluid denotes any
quired for performing the test is less than 1 min per liquid or gas that can flow. Factors that affect leakage
package. The leak test method was evaluated for de- rate include the fluid itself, the leak path dimensions,
tection limit as defined by USP General Chapter and the pressure difference across the leak path. Gas
具1225典 Validation of Compendial Procedures. The de- mass flow leakage rate is expressed in SI units of
tection limit of the VeriPac 455 leak test method was Pascal cubic meters per second (Pa 䡠 m3 䡠 s⫺1). The
proven greater than or equal to 5 m in nominal hole current study reports air flow measured in units of
size. Development studies investigated the ability of cubic centimeters per minute (cm3 䡠 min⫺1), taken at
the method to detect other package defect types, in- ambient (not standard) temperature and pressure con-
cluding vial channel defects and stopper defects. ditions.
1. Department of Health and Human Services. Con- 4. Wolf, H.; Stauffer, T.; Chen, S.-C. Y.; Lee, Y.;
tainer Closure Systems for Packaging Human Forster, R.; Ludzinski, M.; Kamat, M.; Godorov,
Drugs and Biologics, CMC Documentation, Guid- P.; Guazzo, D. M. Vacuum decay container/closure
ance for Industry; U.S. Food and Drug Adminis- integrity testing technology. Part 1. ASTM
tration, Center for Drug Evaluation and Research F2338-09 precision and bias studies. PDA
and Center for Biologics Evaluation and Research: J. Pharm. Sci. Technol. 2009, 63 (5), 472– 488.
Rockville, MD, 1999.
5. Wolf, H.; Stauffer, T.; Chen, S.-C. Y.; Lee, Y.;
2. Department of Health and Human Services. Con- Forster, R.; Ludzinski, M.; Kamat, M.; Mulhall, B.;
tainer and Closure System Integrity Testing in Lieu Guazzo, D. M. Vacuum decay container/closure
of Sterility Testing as a Component of the Stability integrity testing technology. Part 2. Comparison to
Protocol for Sterile Products. Guidance for Indus- dye ingress tests. PDA J. Pharm. Sci. Technol.
try; U.S. Food and Drug Administration, Center for 2009, 63 (5), 489 – 498.