A P P L I C A T I O N N O T E

Validation and Qualification

for Durapore® Sterilizing-Grade (0.22 µm) Membrane VMF4

Introduction – A Technical and A hydrophilic bubble point LRV hydrophilic membrane has a
correlation was not developed water-wet bubble point value, the
Historical Perspective
because hydrophilization does not hydrophobic bubble point is the
This document contains Millipore’s most
impact retention. The hydrophobic integrity test value most closely linked
current Validation and Qualification
structure of the Durapore membrane to intrinsic particle-retention perfor-
information for Durapore‚ Sterilizing-
is the primary determinant of this mance. This rational is explained in
Grade (0.22 µm) Membrane VMF4.
membrane’s retention characteristics. further detail below, along with the
This membrane is a hydrophobic
Hydrophilization, which is the second data that support this correlation.
polyvinylidene fluoride (PVDF) back-
step in manufacturing, simply makes
bone that is rendered hydrophilic by
the initially cast membrane hydrophilic. Abstract
applying a surface modification of A correlation between bubble point
It does not change the pore size,
cross-linked hydrophilic polymer. and LRV was established experimen-
membrane structure, methanol bubble
In 1995, consultants and regulators tally on hydrophobic PVDF membrane.
point or microbial retention character-
began to discuss the need to use From this correlation, the bubble point
istic for sterilizing-grade Durapore.
worst case or minimum bubble point of sterilizing-grade hydrophobic
The Durapore membrane structure
membranes to validate the capability membrane can be predicted using
can be changed to yield a series of
of a membrane to sterilize a drug statistical methods of analysis. Several
membranes with different pore sizes
solution as part of an aseptic process. statistical analysis methods were used
and corresponding bubble points. By
As a result of this discussion, PDA in this study. The method chosen as
testing this series of membranes using
Technical Report 26, “Sterilizing most appropriate for the data was a
the standard ASTM test, one can
Filtration of Liquids,” was published. method that combined Poisson
demonstrate that through a range of
The Technical Report stated that “a regression and the negative binomial.
bubble points, membrane retention
physical integrity test is meaningful This statistical method shows that the
performance undergoes a transition
only when it can be related to specific rapid transition from non-retentive to
from incomplete to complete bacterial
filter-retention characteristics.” This retentive membrane occurs over the
retention. This demonstration fulfills the
statement focuses on the need to range of membranes spanning 14 to
regulatory requirement for a demon-
support any membrane designated as 16 psi methanol bubble point.
strable correlation between integrity
sterilizing-grade with a documented Additionally, membranes with the
test value and bacterial retention
correlation between bacterial retention manufacturing release specification
performance.
and an integrity test value. of 18 psi methanol bubble point are
Although the hydrophobic structure
Millipore’s approach to providing shown to be outside of this transition
provides the bulk of particle retention,
this correlation between bacterial region, and well above the bubble
most filtration applications require a
retention performance and integrity test point range in which the statistical
hydrophilic membrane. Because the
value is the hydrophobic membrane method provides a meaningful
hydrophobic base membrane cannot
structure of Durapore. estimate of the probability that a disk
be easily water wetted, this membrane
is surface-modified with a hydrophilic with a given bubble point value will
cross-linked polymer. Although this yield a sterile filtrate.

Â
Statistical Analyses Table 1. Summary of Experimental Data
The bubble point and LRV data for
hydrophobic membrane were Trial Initial LRV Measured Trial Initial LRV Measured
Number Bubble Thickness Number Bubble Thickness
analyzed using linear regression, Point (µm) Point (µm)
survival analysis and Poisson (MeOH) (MeOH)
regression.
Linear regression (Weisberg) is 47 10.1 3.72 115 55 13.6 7.45 108
a statistical tool in which data are 46 10.3 3.77 116 43 13.6 7.55 104
analyzed to determine the equation 38 9.7 3.85 117 8 13.4 7.60 105
for a line that best represents the 37 10.0 3.94 116 45 13.4 7.64 101
relationship between LRV and bubble
39 10.8 3.95 115 26 13.4 7.66 106
point. Using this technique, only data
from non-fully-retentive disks can be 19 10.2 4.01 114 58 14.1 7.75 91
analyzed, because fully-retentive disks 51 11.0 4.11 90 27 13.6 7.76 109
do not have a real number LRV 50 11.0 4.13 89 9 13.1 7.99 102
associated with them. Typically, this 2 10.2 4.17 118 59 14.0 >8.22 87
technique has been used to derive a 49 10.9 4.17 94 60 14.9 >8.22 115
linear relationship between LRV and
48 11.0 4.21 95 61 15.0 >8.22 114
bubble point over the range of several
bubble point units. Importantly, since 3 10.0 4.22 116 62 16.2 >8.22 94
this method cannot include data of 1 10.2 4.29 117 63 16.0 8.22 96
fully retentive disks, extrapolation of 21 10.2 4.30 118 7 13.6 >8.29 106
the linear relationship is required to 20 10.3 4.35 113 10 14.1 >8.29 99
arrive at a value for bubble point that 54 12.6 6.22 108 11 13.8 >8.29 93
represents a “bubble point cutoff” for
22 12.4 6.25 104 12 13.6 >8.29 89
retentive membrane.
40 12.6 6.33 106 13 14.9 >8.29 114
Results 53 12.8 6.35 106 14 15.0 >8.29 116
Four trials were executed in this study. 42 12.8 6.38 104 15 14.5 >8.29 115
Each trial was done on a different day.
6 12.8 6.50 104 16 16.0 >8.29 98
Microbial Challenge 41 12.8 6.50 104 17 15.9 >8.29 98
Suspensions 52 12.6 6.50 108 18 15.6 >8.29 92
In this study, a sterilizing-grade filter 23 12.6 6.53 106 28 14.1 >8.36 85
is defined as one that yields a sterile 4 12.4 6.59 108 29 14.0 >8.36 88
filtrate when challenged with at least
5 12.5 6.75 105 30 14.0 >8.36 90
107 organisms per square centimeter,
which is 1.38 x 108 CFU/disk when 24 12.9 6.88 110 31 14.8 >8.36 113
using 47 mm disks. A filter that retains 56 13.4 6.98 114 32 15.2 >8.36 112
all organisms at this challenge level is 57 13.5 7.08 105 33 14.8 >8.36 116
reported to have an LRV of greater 44 14.2 7.29 104 34 15.6 >8.36 92
than 8.14. In the trials presented, 25 13.4 7.41 103 35 15.8 >8.36 95
every effort was made to keep the
36 16.0 >8.36 102
challenge level the same so that this
variable could be considered a
constant in the analysis of the data.
Subsequently, all challenge suspen-
sions were between 1.67 x 108 and
2.30 x 108 CFU/disk and retentive
filters had LRVs reported as “>8.22”
to “>8.36”. The data from these
experiments are found in Table 1.
Retention Correlation Table 2. Bubble Points of Sterilizing-Grade Hydrophobic PVDF Membrane as
The data in Table 1 were analyzed Calculated by Two Statistical Methods
by linear regression, survival analysis Statistical BubblePoint1 Confidence Interval2
and Poisson regression. Method LRV=8.14
The data fit the following linear
regression equation: 99.5% 99% 97.5% 95%
Linear Regression Analysis (retentive data excluded)
LRV = 1.02811*bubble point - 6.5255 + error Linear
regression 14.3 15.3 15.2 15.0 14.9
The coefficient of regression, R2, is Log linear
0.932. This curve intersects an LRV of regression 14.0 15.0 14.9 14.7 14.6
8.14 (the value representing sterilizing
membranes) at a bubble point of Survival Analysis (retentive data included)
14.3 psi. Linear
Using the mean square error analysis 14.0 15.0 14.9 14.7 14.6
(sigma), for the “error” value, the Log linear
following equation is derived: analysis 13. 8 14.8 14.7 14.5 14.4
1
Bubble point at which the curve predicted from a given statistical model intersects an LRV of 8.14
bubble point=(LRV-b-sigma)/m 2
Bubble points that give the indicated level of confidence that a filter disk with this bubble point will yield
a sterile filtrate
The bubble point of a membrane
predicted to yield a sterile effluent can
be calculated for any confidence level
by multiplying sigma by a z factor. Table 3. Poisson Regression Analysis: Prediction of the Ability of Hydrophobic
The practical interpretation is that if the PVDF Membrane to Yield a Sterile Filtrate
entire experiment were repeated 100
Visual Methanol Bubblepoint P(0cfu)*
times, 99% of the time the prediction
line (regression line) would intersect 15.0 95.48
LRV=8.14 at a bubble point of 14.9 16.0 99.59
or lower. The same data can be used 17.0 99.96
to calculate predicted “sterilizing” 18.0 >99.99
bubble points for any confidence 19.0 >99.99
level, as is shown in Table 2. 20.0 >99.99
The same data were analyzed by 21.0 >99.99
Poisson regression. A summary of this
* Probability that an individual disk with the designated visual methanol bubble point will yield a sterile
analysis is provided in Table 5. effluent in the standard microbial retention test
Poisson regression gives statistically
derived probabilities that an individual
disk with a given bubble point will
yield a sterile effluent. For example,
according to the data in Table 3, a
disk with a 17.0 psi bubble point will
have a 99.96% chance of yielding a
sterile effluent when tested according
to the methods used in this report.
Conclusion and Discussion From a practical perspective, the Correlation of the bubble point of
In this study, microbial LRV was most relevant use of the data and the Durapore to microbial retention fulfills a
compared to membrane bubble point, statistical modeling of the data is to regulatory and industry requirement of
and a correlation between the two compare them to the current manufac- sterilizing-grade filters. This report will
variables was established. The data turing release specification for assist in validating the use of Durapore
were analyzed by five different hydrophobic Durapore, which is filters in manufacturing processes.
statistical methods. 18 psi. The statistical model derived
The Poisson regression was chosen from the data verifies that the 18 psi
as the primary statistical method. The specification is well beyond the region
primary value of the Poisson regres- of transition to fully-retentive filters, and
sion method is in describing the is well into the region of fully-retentive
relationship between bubble point and filters. Therefore, the 18 psi specifica-
microbial retention of filters that are in tion remains solid, and this work has
the narrow region of transition from requalified hydrophobic Durapore
not-fully-retentive to fully-retentive with a bubble point of 18 psi or
filters. Therefore, the transition region greater as a sterilizing-grade
can be very well characterized. membrane.

To Place an Order or Receive Millipore Worldwide

Technical Assistance
For additional information call your AUSTRALIA CZECH REPUBLIC IRELAND POLAND
nearest Millipore office: Tel. 1 800 222 111 Tel. 02-2051 3841 Life Sciences and Tel. 22-669 12 25
or (02) 9888 8999 Fax 02-2051 4298 Laboratory Water Divisions: 22-663 70 31
In the U.S. and Canada, Fax (02) 9878 0788 Tel. +44 1923 816375 Fax 22-663 70 33
DENMARK
call toll-free 1-800-MILLIPORE AUSTRIA Tel. 70 10 00 23
Fax +44 1923 818297
PUERTO RICO
BioPharmaceutical Division:
(1-800-645-5476) Tel. (01) 877 89 26 Fax 70 10 13 14
Tel. (021) 883 666
Tel. (787) 273-8495
Fax (01) 877 16 54 Fax (787) 747-6553
In the U.S., Canada and Puerto EASTERN EUROPE, Fax (021) 883 048
Rico, fax orders to 1-800-MILLIFX BALTIC COUNTRIES C.I.S., AFRICA, MIDDLE
EAST AND GULF
ITALY
SINGAPORE
Tel. +358 9 804 5110 Tel. (65) 842 1822
(1-800-645-5439) Fax +358 9 256 5660 Life Sciences Division: Vimodrone (Milano):
Fax (65) 842 4988
Tel. (02) 25.07.81
Internet: www.millipore.com BELGIUM AND
Tel. +33 3 88 38 9536
Fax (02) 26.50.324 SPAIN AND PORTUGAL
Fax +33 3 88 38 9539
Tech Service: LUXEMBOURG BioPharmaceutical Division: Roma: Tel. +34 917 283 960
Tel. +32 2 726 88 40 Tel. (06) 52.03.600 Fax +34 917 292 909
www.millipore.com/techservice Fax +32 2 726 98 84
Tel. +43 1 877-8926
Fax (06) 52.95.735
Fax +43 1 877-1654 SWEDEN
BRAZIL Laboratory Water Division: JAPAN Tel. 08-628 6960
Tel. (011) 5548-7011 Tel. +33 1 30 12 7000 Tel. (03) 5442-9711 Fax 08-628 6457
Fax (011) 5548-7923 Fax +33 1 30 12 7180 Fax (03) 5442-
9736 Life Sciences SWITZERLAND
CANADA FINLAND Tel. (01) 908-30-60
Tel. (09) 804 5110 9737 BioPharm.
Tel. 1-800-645-5476 9734 Lab Water Fax (01) 908-30-80
Fax 1-800-645-5439 Fax (09) 256 5660
KOREA TAIWAN
CHINA, PEOPLE'S FRANCE Tel. 886-2-2792-9333
Tel. (822) 551-0990
REPUBLIC OF Tel. 01 30 12 7000 Fax 886-2-2792-6555
Fax (822) 551-0228
Beijing: Fax 01 30 12 7180
Tel. (8610) 8519 1250 MALAYSIA U.K.
GERMANY Tel. 01923 816375
(8610) 6518 1058 Tel. (06196) 494-0 Tel. 603-757 1322
Fax (8610) 8519 1255 Fax 603-757 1711 Fax 01923 818297
Fax (06196) 43901
Guangzhou: U.S.A.
MEXICO
Tel. (8620) 8755 4049 HUNGARY Tel. (781) 533-6000
Tel. (525) 576-9688
Fax (8620) 8752 0172 Tel. 01 381 0433 Fax (781) 533-3110
Fax (525) 576-8706
Hong Kong: 01 381 0434
Tel. (852) 2803 9111 01 209 3232 THE NETHERLANDS IN ALL OTHER
COUNTRIES

 Fax (852) 2513 0313

Shanghai:
Tel. (8621) 5306 9100
Fax (8621) 5306 0838
Fax 01
INDIA
209 0295

Tel. (91) 80-839 46 57

Fax (91) 80-839 63 45
Tel. 076-5022000
Fax 076-5022436
NORWAY
Tel. 22 67 82 53
Millipore Intertech (U.S.A.)
Tel. +1 (781) 533-8622
Fax +1 (781) 533-8630

Fax 22 66 04 60
Millipore and Durapore are registered trademarks
of Millipore Corporation.
ASTM is a trademark of American Society for
Testing and Materials.
Lit. No. AN1525EN00 Rev. A 10/01 01-268
Printed in U.S.A.
© 2001 Millipore Corporation, Bedford, MA.
All rights reserved.