In-process bioburden testing of antibiotics using a membrane filtration device

InfoTrac OneFile (R)
Pharmaceutical Technology Europe , 11/01/2004 16 11
In-process bioburden testing of antibiotics using a membrane filtration device: this study highlights several important factors for consideration when designing a filtration-based bioburden validation study. It is intended as a guide for pharmaceutical manufacturers when developing bioburden test methods in conjunction with current pharmacopoeial guidelines. Lindsay, Jennifer *~|~*Souza, Kathleen *~|~*
COPYRIGHT 2004 Advanstar Communications, Inc.

Bioburden testing antibiotics during their production is an essential element of the manufacturing process. Current good manufacturing practice (cGMP) dictates that microbiological contamination must be managed in the production of both sterile and non-sterile pharmaceuticals, helping to ensure a consistent level of quality. (1,2) Membrane filtration is a sensitive method of monitoring the bioburden levels potentially present during production and verifying a key aspect of process control. For antibiotics and other inhibitory agents, membrane filtration is the method of choice for bioburden testing. (3-5) Antibiotics present some challenges with regard to bioburden testing. Some of the microbial contaminants may be sensitive to the antibiotic, requiring that it be somehow removed or inactivated for these microbes to be recovered. Membrane filtration can enhance the physical separation of the test sample from any microbiological contaminants, although antibiotics may a!
dsorb to some types of membrane filters. (7,9) In addition, some device configurations may limit the ability to rinse the antibiotics sufficiently from the membrane. (7,8) Also, it is possible that rinsing alone will not remove enough of the antibiotic to allow for recovery of some microbes. (10) Each of these factors may lead to reduced recoveries if test methods are not optimized and appropriately verified.
Methods for testing five antimicrobial agents from multiple antibiotic classes were developed and evaluated using two commercial disposable devices (Milliflex and Microfil S; Millipore Corporation, Billerica. Massachusetts, USA) containing a 0.45 [micro]m pore size membrane filter and plastic funnel. In the Milliflex device, the edge of the membrane filter is heat-sealed to the base, whereas the membrane filter is pinched between the funnel and base in the Microfil S device. Two different membrane types were also studied: mixed cellulose esters (MCE) and polyvinylidene fluoride (PVDF). Used for general applications, MCE membranes are composed of cellulose nitrate and cellulose acetate. PVDF membranes tend to have a lower affinity for proteins than MCE and are a good alternative when binding may be an issue. (7,9)
This study addresses several important factors for consideration when designing a filtration-based bioburden validation study. It is intended as a guide for pharmaceutical manufacturers when developing bioburden test methods in conjunction with current pharmacopoeial guidelines. (3-5)
Materials and equipment
Materials. Milliflex filter funnel units with 0.45 [micro]m PVDF membranes (catalogue number MXHVWP124, Milliflex-HV unit), Milliflex filter funnel units with 0.45 [micro]m MCE membranes (catalogue number MXHAWG124, Milliflex-HA units) and Microfil S filtration devices with 0.45 [micro]m PVDF membranes (catalogue number MVHVWPS24, Microfil S-HV devices) are standard devices from Millipore Corporation. Two production lots of each device were used. Cefuroxime, doxycycline, penicillin G, ofloxacin, streptomycin and penicillinase were purchased from Sigma-Aldrich (St Louis, Missouri, USA). Escherichia coli ATCC 8739 (E. coli), Pseudomonas aeruginosa ATCC 9027 (P. aeruginosa) and Staphylococcus aureus ATCC 6538 (S. aureus) were obtained from the American Type Culture Collection (Manassas, Virginia, USA).
Equipment. All devices were run using the Milliflex PLUS vacuum pump (catalogue number MXP-PLUS01; Millipore Corporation). The Milliflex pump head (catalogue number MXPHEAD01) was used with the Milliflex devices and the Microfil S pump head (catalogue number MCCHEAD01) was used with the Microfil S devices.
Methods
Selection of challenge micro-organisms. The micro-organisms used in this study were chosen by ascertaining the spectrum of activity of each antibiotic, (6) and determining which microbes are recommended for antibiotic testing based on US pharmacopoeial guidance. (3) Overnight cultures were prepared by inoculating frozen glycerol stocks into 10 mL of tryptic soy broth ([TSB] Northeast Laboratory Service, Waterville, Maine, USA) and incubating at 32.5 [+ or -] 2.5[degrees]C for 18-24 hours. The overnight cultures were serially diluted in 0.1% peptone water just prior to testing. The inoculum volume (50-100 mL) was adjusted for each microbial species such that less than or equal to 100 colony forming units (cfu) were inoculated into the sample in the filtration device. Each microbial strain selected was tested to verify antibiotic susceptibility prior to the commencement of the study. This susceptibility testing was performed by inoculating less than or equal to 100 cfu into !
10 mL of antibiotic solution in a Milliflex-HV device, filtering and incubating the membrane on tryptic soy agar ([TSA] Hardy Diagnostics, Santa Maria, California, USA) for up to 3 days at 32.5 [+ or -] 2.5[degrees]C. Lack of colony formation after incubation was considered indicative of microbial susceptibility to the antibiotic.
Rinse fluids defined. Two different rinse fluids produced by Millipore Corporation were used in this study. (3) Fluid A consisted of a peptic digest of animal tissue in water (1 g/L). Fluid D was composed of Fluid A with the addition of polysorbate 80 (1 mL/L). A 600 mL septum/screw-cap configuration was used for both rinse fluids.
General experimental design. The general methodology is outlined in Table I. Antibiotic sample concentrations and challenge micro-organisms are presented in Table II. The specific procedure used with each antibiotic was individually determined and optimized prior to each portion of the study. The antibiotic sample concentrations used in this study are those which are typically used in the clinical environment. (6) Samples (10 mL) were used as per pharmacopoeial guidelines, (3-5) except for cefuroxime and penicillin G, which were diluted to 20 mL to allow for the complete dissolution of these concentrated solutions. It is important to note that the antibiotic must be completely dissolved prior to filtration, as any residual particulates in the solution would remain on the membrane and potentially lead to microbial inhibition. The rinsing methods were determined by trial and error whilst taking into account pharmacopoeial guidelines. The prepared antibiotic solutions were fi!
lter sterilized using 0.22 [micro]m Stericup PVDF filter devices (catalogue number SCGVU02RE; Millipore Corporation) prior to use. Membranes were pre-wet with the appropriate rinse fluid prior to contact with antibiotic solutions. Challenge microbes were diluted in 0.1% peptone and added to the final rinse at a target concentration of less than or equal to 100 cfu per filter. Filters were incubated on Tryptic Soy Agar Milliflex Cassettes (Millipore Corporation) at 32.5 [+ or -] 2.5[degrees]C for up to 3 days. Per cent recovery of the sensitive challenge microbes from the antimicrobial agents was determined versus antibiotic-free controls.
Membrane composition comparison. The Milliflex-HV membrane (PVDF) was compared with the Milliflex-HA membrane (MCE) on the basis of per cent recovery of challenge micro-organisms from doxycycline, ofloxacin and streptomycin. The general method was followed for both membrane types and specific methodologies are listed in the first section of Table III. Two device lots per membrane type were tested.
Device comparison. The Milliflex-HV device was compared with the Microfil S-HV device on the basis of per cent recovery of test microbes from doxycycline, ofloxacin and streptomycin. The Milliflex-HV filter funnel unit is a presterilized device that features a larger-than-standard filtration area (19.2 c[m.sup.2]) for quicker filtration and an integral membrane which is heat-sealed to the funnel and base. This one-piece design eliminates membrane handling, thereby decreasing contamination risk. The Microfil S filtration device is presterilized and individually packaged, wherein the membrane is pinched between the funnel and base. This device features an easily-removed funnel and is designed to eliminate prefiltration assembly--reducing the contamination threat and preparation time. The general method was followed with both device types and the specific techniques used are presented in the second part of Table III.
Enzymatic inactivation of [beta]-lactam antibiotics. Per cent recoveries of challenge microbes from penicillin G and from cefuroxime were determined in the presence and absence of penicillinase using Milliflex-HV devices. The general method was followed, except that penicillinase was added to each antibiotic just prior to testing. Refer to the third section of Table III for additional details. Two lots of Milliflex-HV devices were tested.
Data analysis. The data from the membrane comparison and the device comparison were analysed using analysis of variance (ANOVA) by test type interaction F-test using the logarithm of the cfu (p-value <0.001) to determine whether microbial recoveries were equivalent by membrane type and by device type.
Results and discussion
Membrane comparison. A data summary obtained from the comparison of Milliflex devices with a PVDF membrane with those containing an MCE membrane is reported in Table IV. In the case of ofloxacin, 33% of the challenge organisms were recovered using Milliflex-HV devices, whereas only 1% were recovered using Milliflex-HA devices. A less obvious difference was noted with doxycycline; per cent recoveries were not significantly different when comparing PVDF and MCE membranes. However, the colonies recovered with Milliflex-HA devices were considerably smaller (pinpoint-sized) than those with Milliflex-HV devices (3-5 mm) after 24 hours of incubation. No significant difference in recovery existed between PVDF and MCE in the case of streptomycin. The data indicate that some antibiotics may adsorb more to membrane filters composed of MCE than to those made of PVDF, as suggested previously. (7,9)
Device comparison. The device comparison data are reported in Table V. Significantly greater per cent recoveries of challenge microbes were obtained using Milliflex-HV devices than noted using Microfil S-HV devices with ofloxacin and streptomycin. There was no significant difference between the devices in per cent recovery of microbes from doxycycline, though S. aureus recovered with Microfil S-HV devices formed much smaller colonies than those recovered with Milliflex-HV devices. These differences may result from incomplete rinsing of any antibiotic that may have infused into the membrane edges pinched between the funnel and the base in the Microfil S devices. (7,8)
Enzymatic inactivation of [beta]-lactam antibiotics. The per cent recoveries achieved from the enzymatic inactivation of penicillin G and cefuroxime using Milliflex-HV devices are shown in Table VI. The addition of 3 mg penicillinase to penicillin G prior to testing allowed for 113% recovery of S. aureus, whereas the addition of 5 mg of penicillinase to cefuroxime resulted in 84% recovery of E. coli. Challenge microbes were not recovered from either antibiotic in the absence of penicillinase. This demonstrates that additional antibiotic neutralization is beneficial when attempting to recover sensitive microbes from penicillin G and from cefuroxime using membrane filtration.
Conclusion
This study examined several factors that must be considered when designing a membrane filtration-based method for the recovery of microbes from antibiotics. These include membrane composition, filtration device configuration, rinsing method and the potential need for additional inactivating agents. The results demonstrate that microbial recovery from some antibiotics may be improved by using a PVDF membrane, but a membrane composed of MCE may be sufficient for the recovery of microbes from other antibiotics. There was also a tendency toward greater recovery of sensitive microbes using a device with an integral membrane (for example, Milliflex device) than with a device wherein the membrane is pinched between the funnel and the base (for example, Microfil S device). This may result from improved rinsing capabilities with the larger filtration area of the Milliflex device. Another possible reason for the lower recovery of sensitive microbes with the Microfil S device may hav!
e to do with the device configuration itself, which may reduce the ability to sufficiently rinse some antibiotics. In the case of the [beta]-lactam antibiotics--penicillin G and cefuroxime--the challenge microbes were only recoverable when penicillinase was used to inactivate these agents. This information illustrates the need to consider various factors that may affect microbial recovery when devising bioburden test methods, particularly when inhibitory agents such as antibiotics are present.
<pre>
Table I General experimental design for membrane comparisons, device
comparisons and enzymatic inactivation.

Sample Antibiotic Rinses Challenge microbe

Rinses only control (negative) X
Antibiotic solution and rinses
control (negative) X X
Antibiotic solution and microbe
control (negative) X X
Rinses and microbe control
(positive) X X
Test X X X

Table II Antibiotic solution compositions and sensitive challenge
micro-organisms. (6)

Antibiotic Sample concentration/diluent Challenge microbe

Cefuroxime 2.2 g/20 mL ultrapure* water Escherichia coli
and 5 mg [beta]-lactamase ATCC 8739
Doxycycline 10 mg/10 mL Staphylococcus aureus
0.1 N hydrochloric acid ATCC 6538
Ofloxacin 40 mg/10 mL Pseudomonas aeruginosa
0.1 N hydrochloric acid ATCC 9027
Penicillin G 10 million units/20 mL Staphylococcus aureus
ultrapure*
water and 3 mg [beta]-lactamase ATCC 6538
Streptomycin 4 g/10 mL ultrapure* water Escherichia coli
ATCC 8739

* Ultrapure water is defined here as having a resistivity of 18.2
M[ohm].cm at 25[degrees]C. Here, the water was obtained from a Milli-Q
Biocel Ultrapure Water System (Millipore Corporation, Bedford,
Massachusetts, USA).

Table III Optimized methods the membrane comparison, device comparison
and enzymatic inactivation experiments.

Experiment Antibiotic Penicillinase Rinse fluid Number of
per sample 100-mL rinses

Membrane Doxycycline* N/A D 6
comparison Ofloxacin N/A A 3
Streptomycin N/A A 3
Device Doxycycline* N/A D 3
comparison Ofloxacin N/A A 3
Streptomycin N/A A 3
Enzymatic Cefuroxime 5 mg A 6
inactivation Penicillin G 3 mg A 3

*The number of 100-mL rinses used with doxycycline was reduced from six
in the membrane comparison testing to three in the device comparison
experiments after re-optimization of the test method.

Table IV Membrane comparison: Per cent recovery of challenge organisms
on 0.45 [micro]m PVDF versus 0.45 [micro]m MCE.

Per cent recovery*
Doxycycline Ofloxacin Streptomycin
MCE PVDF MCE PVDF MCE PVDF

Average 98% 87% 1% 33% 52% 48%
Lot 1 113% 102% 0% 30% 54% 54%
Lot 2 82% 72% 2% 36% 49% 42%
Significant No** Yes No
difference
in recovery?

* Percent recovery is defined as average cfu on test samples divided by
average cfu of positive control ("rinses and microbe control")
** No statistically significant difference noted; however, colonies on
MCE were much smaller than those on PVDF.

Table V Device comparison: Per cent recovery of challenge organisms on
pinched membrane (Microfil S-HV, "S-HV") versus integral membrane
(Milliflex-HV, "X-HV").

Per cent recovery*
Doxycycline Ofloxacin Streptomycin
S-HV X-HV S-HV X-HV S-HV X-HV

Average 67% 92% 13% 51% 15% 70%
Lot 1 68% 87% 13% 58% 16% 77%
Lot 2 66% 97% 12% 43% 13% 63%
Significant No** Yes Yes
difference
in recovery?

* Per cent recovery is defined as average cfu on test samples divided by
average cfu of positive control ("rinses and microbe control")
** No clear statistically significant difference noted for per cent
recovery; however, colonies appearing on Microfil S were much smaller
than those on Milliflex.

Table VI Per cent recovery of challenge organisms with enzymatic
inactivation of [beta]-lactam antibiotics using Milliflex-HV devices.

Per cent recovery*
Cefuroxime Penicillin G
+Enzyme +Enzyme +Enzyme +Enzyme

Average 0% 84% 0% 113%
Lot 1 0% 71% 0% 117%
Lot 2 0% 96% 0% 108%
Significant Yes Yes
difference
in recovery?

* Per cent recovery is defined as average cfu on test samples divided by
average cfu of positive control ("rinses and microbe control"). </pre>
Acknowledgement
The authors wish to thank Mark Blanchard for performing the statistical analysis.
References
1. US Food and Drug Administration, "Code of Federal Regulations, Title 21, Volume 4, Part 211, Section 211.113" (2004). www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?FR=211.113
2. European Commission, "EudraLex Volume 4, Annex 1" (2003). http://pharmacos.eudra.org/F2/eudralex/vol-4/home.htm
3. The United States Pharmacopeial Convention, Inc., "USP 27-NF 22 through Second Supplement" (2004).
4. European Directorate for the Quality of Medicines, "European Pharmacopoeia, 5th Edition 5.00" (2005).
5. "The Japanese Pharmacopoeia, Fourteenth Edition, Part I" (2001).
6. Physicians' Desk Reference, 58th Edition (Thomson PDR, Montvale, New Jersey, USA, 2004).
7. A. Sugiura et al., "The Removal of Cephalosporins Adsorbed on a Membrane Filter during Sterility Testing," J. Takeda Res. Lab. 47, 97-104 (1988).
8. A.S. Breeze and A.M. Simpson, "Actual and Potential Methods of Testing Antibiotics for Sterility," Society for Applied Bacteriology Technical Series--Antibiotics 18, 339-348 (1983).
9. M.J. Akers, G.E. Wright and K.A. Carlson, "Sterility Testing of Antimicrobial-Containing Injectable Solutions Prepared in the Pharmacy," Am. J. Hosp. Pharm. 48(11), 2414-2418 (1991).
10. P. Casetta and F. Negretti, "Experimental Observations on the Bacteriological Controls of Antibiotics--II. Inactivation of the Antimicrobial Activity of Membranes Employed for the Filtration of Antibiotics," J. Pharm. Biomed. Anal. 7(12), 1867-1870 (1989).
Jennifer Lindsay* is a microbiological scientist I and Kathleen Souza is a microbiological scientist III at Millipore Corporation, 80 Ashby Road, Mail Stop E12C, Bedford, Massachusetts 01730, USA.
Tel. +1 781 533 2550
Fax +1 781 533 3395
Jennifer_Lindsay@Millipore.com
*To whom all correspondence should be addressed.

InfoTrac OneFile (R)