ORIGINAL ARTICLE Sterility testing of minimally manipulated cord blood products: validation of growth-based automated culture systems Salem Akel, Joan Lorenz, and Donna Regan BACKGROUND: The St Louis Cord Blood Bank sub- mitted a biologics license application for cord blood (CB) products processed by PrepaCyte-CB (BioE), sup- ported with a validation study of a microbial detection system for product sterility testing (BACTEC-FX, Becton Dickinson). This article provides the validation approach followed to fulfill Food and Drug Administration require- ments pertinent to sterility testing method. STUDY DESIGN AND METHODS: System qualifica- tion, culture media quality verification, and validation of CB processing by-product (CB-BP) sample as surrogate to final product for sterility testing were followed by studies evaluating method sensitivity, specificity, repro- ducibility, ruggedness or robustness, and bacteriostatic or fungistatic effect of CB-BP sample. CB-BP cultures and control samples were formulated using BACTEC Plus Aerobic/F, Plus Anaerobic/F, and Myco F/Lytic media. Samples were seeded with selected test organ- isms (n = 13 at 10-100 colony-forming units [CFUs] per vial) and cultured for 14 days (bacterial) and 30 days (fungal). RESULTS: Under testing conditions, no stasis effect of test sample on microbial growth and no false-positive or false-negative results were reported. Although a 7-day culture was sufficient to detect all validation test organ- isms seeded at 26 CFUs/vial, growth in actual product sterility testing practice may require a 10- to 14-day culture. Assay reproducibility was uncertain at very low bioburden (<10 CFUs/vial). Growth time to detection neither varied between different media lots nor pro- longed in culture vials with loading delay (6-8 hr at room temperature). CONCLUSION: BACTEC-FX culture and detection system and BACTEC media formulae have high detec- tion capability and can be effectively validated for steril- ity testing of CB products. A lthough widely used, commercially available automated sterility testing methods are not US Food and Drug Administration (FDA) approved for testing of cellular therapy products. 1-4 The FDA has given the flexibility to take advantage of modern sterility test methods (either culture-based or non– culture-based) provided that they meet certain criteria. For cell therapy products regulated under Section 351 of the Public Health Service Act such as cord blood (CB), sterility test method must be appropriate to the material being tested such that the material does not interfere with or otherwise hinder the test (21 CFR 610.12). In a recent amendment, 5,6 the FDA sterility testing require- ments for biologic products do not specify testing method, media formulae, culture conditions (incubation time and temperature), or test sample (type and volume). The FDA justifiably addressed that product’s manufac- turer must conduct a validation study to demonstrate that the method used is capable of consistently detecting the presence of product-contaminating microorganisms at low bioburden. The validation activity must consider all validation principles, such as limit of detection (sensitivity), specificity, reproducibility, ruggedness, and robustness. Although the FDA may decide to encourage ABBREVIATIONS: BLA = Biological License Application; BP(s) = by-product(s); CB = cord blood; CBB(s) = cord blood bank(s); FP = final product; RMM(s) = rapid microbiologic method(s); SLCBB = St Louis Cord Blood Bank; TTD(s) = time(s) to detection. From the St Louis Cord Blood Bank and Cellular Therapy Laboratory, SSM Cardinal Glennon Children’s Medical Center, St Louis, Missouri. Address reprint requests to: Salem Akel, PhD, St Louis Cord Blood Bank & Cellular Therapy Laboratory @ SSM Cardinal Glennon Children’s Medical Center, 3662 Park Avenue, St Louis, MO 63110; e-mail:[email protected]. Received for publication October 16, 2012; revision received January 21, 2013, and accepted January 29, 2013. doi: 10.1111/trf.12229 TRANSFUSION **;**:**-**. Volume **, ** ** TRANSFUSION 1
Transcript
O R I G I N A L A R T I C L E
Sterility testing of minimally manipulated cord blood products:validation of growth-based automated culture systems
Salem Akel, Joan Lorenz, and Donna Regan
BACKGROUND: The St Louis Cord Blood Bank sub-mitted a biologics license application for cord blood(CB) products processed by PrepaCyte-CB (BioE), sup-ported with a validation study of a microbial detectionsystem for product sterility testing (BACTEC-FX, BectonDickinson). This article provides the validation approachfollowed to fulfill Food and Drug Administration require-ments pertinent to sterility testing method.STUDY DESIGN AND METHODS: System qualifica-tion, culture media quality verification, and validation ofCB processing by-product (CB-BP) sample as surrogateto final product for sterility testing were followed bystudies evaluating method sensitivity, specificity, repro-ducibility, ruggedness or robustness, and bacteriostaticor fungistatic effect of CB-BP sample. CB-BP culturesand control samples were formulated using BACTECPlus Aerobic/F, Plus Anaerobic/F, and Myco F/Lyticmedia. Samples were seeded with selected test organ-isms (n = 13 at 10-100 colony-forming units [CFUs] pervial) and cultured for 14 days (bacterial) and 30 days(fungal).RESULTS: Under testing conditions, no stasis effect oftest sample on microbial growth and no false-positive orfalse-negative results were reported. Although a 7-dayculture was sufficient to detect all validation test organ-isms seeded at �26 CFUs/vial, growth in actual productsterility testing practice may require a 10- to 14-dayculture. Assay reproducibility was uncertain at very lowbioburden (<10 CFUs/vial). Growth time to detectionneither varied between different media lots nor pro-longed in culture vials with loading delay (6-8 hr atroom temperature).CONCLUSION: BACTEC-FX culture and detectionsystem and BACTEC media formulae have high detec-tion capability and can be effectively validated for steril-ity testing of CB products.
Although widely used, commercially availableautomated sterility testing methods are not USFood and Drug Administration (FDA) approvedfor testing of cellular therapy products.1-4 The
FDA has given the flexibility to take advantage of modernsterility test methods (either culture-based or non–culture-based) provided that they meet certain criteria.For cell therapy products regulated under Section 351 ofthe Public Health Service Act such as cord blood (CB),sterility test method must be appropriate to the materialbeing tested such that the material does not interferewith or otherwise hinder the test (21 CFR 610.12). In arecent amendment,5,6 the FDA sterility testing require-ments for biologic products do not specify testingmethod, media formulae, culture conditions (incubationtime and temperature), or test sample (type and volume).The FDA justifiably addressed that product’s manufac-turer must conduct a validation study to demonstratethat the method used is capable of consistently detectingthe presence of product-contaminating microorganismsat low bioburden. The validation activity must considerall validation principles, such as limit of detection(sensitivity), specificity, reproducibility, ruggedness, androbustness. Although the FDA may decide to encourage
ABBREVIATIONS: BLA = Biological License Application;
Received for publication October 16, 2012; revision
received January 21, 2013, and accepted January 29, 2013.
doi: 10.1111/trf.12229
TRANSFUSION **;**:**-**.
Volume **, ** ** TRANSFUSION 1
the use of the specified compendial sterility testingmethod described in the CFR 610.12/USP <71> as abenchmark for validation of culture-based methods, aone-sided validation protocol can be accepted withoutcomparison to a “referee” test method.1,2,5,6 Becausecertain cell therapy products, such as CB, may containantimicrobial agents including antibiotics, sterility testsamples must be evaluated for bacteriostasis and fung-istasis according to the USP <71>.2
Most CB banks determine product sterility usinggrowth-based rapid (5-7 days) microbiologic methods(RMMs) such as BacT/ALERT (bioMérieux, Durham, NC)and BACTEC (Becton Dickinson, Sparks, MD).3,7-9 Duringthe process of preparing the Biological License Applica-tion (BLA) for CB products manufactured at the St LouisCord Blood Bank (SLCBB), a validation plan of theBACTEC microbial testing method was submitted to theCenter for Biologics Evaluation and Research and modi-fied in response to current FDA thinking. Validation com-ponents and results are presented here to assist CBmanufacturers in planning a site-specific method valida-tion and to provide experience with the performance ofthe BACTEC-FX system and BACTEC culture media forthis intended use.
MATERIALS AND METHODS
CB collection, production, and sterilitytesting samplesAseptically, umbilical CB is drawn directly into a closed-bag system containing 35 mL of CPD (Pall Corp., PortWashington, NY). Collections are processed using areagent kit (PrepaCyte-CB, BioE, St Paul, MN).10,11 The kit isa functionally closed three-bag system in which 150 mL ofPrepaCyte-CB is used as a red blood cell (RBC) sediment-ing reagent. After RBC sedimentation out of suspension,centrifugation of suspension rich with white blood cells(WBCs) at 400 ¥ g for 10 minutes is performed to pellet theWBCs and to remove plasma. The WBC pellet is resus-pended in residual plasma, transferred into a freezing bag,and mixed with a sterile freezing solution (1:1 volumemixture of dimethyl sulfoxide [DMSO] and 40% dextran;WAK-Chemie, Steinbach, Germany) to yield a small(about 25 mL) final product (FP) ready for cryopreserva-tion. The processing by-products (BPs; plasma and RBCs)are combined in one bag. Two potential sample types forsterility testing (0.5 mL of FB and 10 mL of BP) werechosen for investigation. These samples are representativeof all reagents, additives, and biologic componentsinvolved in CB production. The BP volume is approxi-mately 65% PrepaCyte-CB reagent, 25% CPD plasma, and10% RBCs while FP volume is mainly nucleated cells andRBCs (40%), PrepaCyte-CB reagent (25%), CPD plasma(15%), and DMSO-dextran (20%).
Selection of culture system and culture mediaEarlier studies concerning sterility testing of cellulartherapy products have demonstrated the overall superior-ity in organism recovery of both BacT/ALERT and BACTECto the CFR/USP method.7-9,12,13 In both systems, auto-mated readings are taken and recorded (every 10 min) forfluorimetric detection of CO2 as a product of microbialmetabolism. Culture media used with automated systemssupport the growth of aerobic and anaerobic bacteria andfungus or yeast. Certain media are formulated to neutral-ize antibiotics in sterility testing samples to reduce the riskof false-negative results.14-17
BACTEC media were shown to support the recovery ofPropionibacterium acnes, a not uncommonly found slow-growing contaminant of CB products, and were reportedto grow organisms in samples containing antibiotics withfewest failures, compared to other media.7,17 For clinicaluse, the BACTEC-FX sterility testing system (Becton Dick-inson) is approved by FDA as 510(k) instrument andmedia formulations. The clinical microbiology laboratoryof the corporate parent of the SLCBB (SSM Healthcare)relies on the BACTEC-FX for blood cultures, which can beutilized as a backup system for testing. Given these obser-vations, the BACTEC-FX system was chosen for sterilitytesting at our facility. Media selected for use included BDPlus Aerobic/F, Plus Anaerobic/F, and Myco F/Lytic media.The Plus medium contains resins that neutralize antibiot-ics not infrequently present in CB collections. MycoF/Lytic medium has no resins but contains lysing agentwhich lyses blood cells and improves recovery of yeast andfungi.18
Organism preparation, seeding of samples, andinoculation or incubation of culture vialsOrganisms were prepared to be tested at a target celldensity of 10 to 100 colony-forming units (CFUs)/vial. A0.5 McFarland suspension of the challenge organism intrypticase soy broth (Becton Dickinson) was made, withthe approximate cell density of 108 CFUs/mL for bacteriaand 106 CFUs/mL for yeast. Suspensions were seriallydiluted to achieve the target cell densities. Routine colonycounts were carried out to determine inoculum density byplating 0.1 mL of the 102 CFUs/mL suspensions.
Throughout all phases of the validation, test andcontrol samples were inoculated, in triplicate, into culturevials and incubated in the BACTEC until flagged positiveor up to 14 and 30 days (for bacterial and fungal growth,respectively). Flagged positive vials were confirmed, bymedia subculture or Gram stain, for growth of inoculated(test) organism (true positive) and absence of cross-contamination. Unflagged vials were confirmed as truenegative. Instrument time to detection (TTD) of microbialgrowth was calculated as the time from loading of vialuntil vial is flagged positive. Results are expressed as themean TTD of triplicate cultures.
AKEL ET AL.
2 TRANSFUSION Volume **, ** **
Phase I: initial validation
System qualificationBACTEC-FX instrument installation and operation quali-fication was performed by the supplier.
Media lot release testingAlthough all received media lots have been certified by themanufacturer for growth promotion quality and sterility,media growth promotion ability and sterility were verifiedon site, using CFR/USP strains and/or those suggestedby the manufacturer (Table S1, available as supportinginformation in the online version of this paper). The targetwas to inoculate vials with not more than 100 CFUs oftest organism and to have evidence of growth within 3days, with the absence of cross-contamination. In caseswhere growth is observed later, within 3 to 7 days, and/orif inoculum count is found to be 100 to 200 CFUs, mediagrowth promotion quality was considered satisfactory andreleased for use. Sterility is shown when uninoculatedvials are confirmed negative after 14 days of incubation.
Sterility testing samples
Sample assessment for potential interference withsystem detection capability. Since CB sterility testingsamples contain production-related components notfound in clinical specimens, it was decided to assess thepotential for individual sample components to causefalse-positive signals in the absence of microbial contami-nation. To this effect, cryopreservation solution (10 mL),PrepaCyte-CB reagent (10 mL), CPD solution (10 mL),RBCs (10 mL), and various densities of CB nucleated cells(20 to 60 million in 1 mL) were added to three types ofblood culture media and loaded to the BACTEC for theentire incubation time.
Sample adequacy and validity for sterility testing.Because the CB FP is with limited volume and cell content,our intent was to use sparingly a minimal volume (0.5 mL)of the FP for sterility testing and/or to use the BP sampleas a surrogate for the FP. The use of a small sample fromthe FP and/or the optimal sample volume (10 mL foraerobic or anaerobic and 5 mL for fungal culture vials)from the BP would be acceptable for sterility testing if thesample retains, by evidence, detectable levels of CB con-taminants present at low bioburden. In this validationphase, microbial recovery was assessed in various CB pro-duction fractions (samples) including FP, BP, postprocess-ing plasma, and RBCs. CB unit (mean volume, 100 mL) isinitially mixed with 150 mL of PrepaCyte-CB reagent toyield a 250-mL preprocessing volume. The test microor-ganism (three representative microbes used includingEscherichia coli, Bacteroides fragilis, and Candida albi-cans) was added to the preprocessing volume at approxi-mately 1 CFU/mL. Units were processed, and sterility
testing was performed on various production fractions.The RBC fraction was tested at 1.0 or 0.5 mL representingthe amount of RBCs in the BP and at higher volume (5 mL)to confirm whether this fraction retains the contaminat-ing microbe. Control samples from the preprocessedvolume were used to confirm assay sensitivity at theinoculated microbial dose. Culture results would demon-strate which postprocessing fraction of CB retains thecontaminating microbe and would also determine sampleadequacy for sterility testing.
Phase II: validation
Sensitivity, specificity, and reproducibilityThe BACTEC method was evaluated for its ability to detectpositive cultures from test samples inoculated with verylow density of selected microorganisms and to decide theoptimal incubation time for microbial recovery. Thirteenmicrobial isolates were selected, representing a variety ofaerobic and anaerobic bacteria, yeast, and fungus thatare observed as common environmental contaminants,common isolates from past CB product cultures, and/orCFR/USP recommended1-4,7 (Table 1). For each test organ-ism the following samples were prepared, inoculated intoorganism-optimal culture media (at 10-100 CFUs/vial),and cultured up to 14 days for bacterial (aerobic andanaerobic) or 30 days for fungal vials:
1. Test: CB-BP sample seeded with the test organism.2. Negative control: CB-BP sample free of test organism.3. Stasis: Sterile normal saline sample seeded with test
organism. TTDs for stasis cultures were comparedwith those of corresponding test cultures to assess thepotential bacteriostatic or fungistatic effect of CB-BPsample.
In multiple initial culture runs, organism growth (10organisms) was tested using organism-optimal andorganism-nonoptimal culture media type(s). Results ofthese runs are mentioned in this report.
Phase III: validation
Evaluation of ruggedness and robustness of theBACTEC system for CB sterility testingThis phase was designed to determine whether variationsin operational or environmental variables might exhibitany significant effect on assay sensitivity and specificity.3
The variables studied, relevant to manufacturing at theSLCBB, were potential slight formula variations betweendifferent media lots and immediate versus delayed time(6-8 hr at room temperature) from inoculation to loadingof the vials. Six test organisms were used: coagulase-negative Staphylococcus and Bacillus subtilis (aerobic),B. fragilis and P. acnes (anaerobic), C. albicans (yeast), and
VALIDATION OF CB STERILITY TESTING
Volume **, ** ** TRANSFUSION 3
Aspergillus niger (fungus), representing organisms with avariety of characteristics and requirements, including twoin-house isolates, four ATCC strains, and one slow-growing organism (P. acnes). Results of this comparisonstudy would determine whether slightly dissimilar testingconditions would cause any significant difference inmicrobial recovery and TTD by the BACTEC.
RESULTS
Phase I validation: sterility testing samples
Sample assessment for potential interference withsystem detection capabilityNegative results were reported in all cultures inoculatedwith various test sample components ensuring that suchcomponents do not generate reactions with media con-stituents that might trigger fluorescence signals and false-positive results.
Sample adequacy and validity for sterility testingAs shown in Table 2, growth varied among samples used.All cultures from samples representing the processing BPwere flagged positive, and subcultures confirmed thepresence of the inoculated microbe. TTDs for cultures ofBP were comparable to those determined for the corre-sponding control cultures. None of the other testedsample types were ideal for capturing all tested microbes.Collectively, the results support the adequacy of the BPsample, in terms of type and volume, as a surrogatesample to FP for sterility testing. Therefore, validationusing the PrepaCyte-CB processing BP as testing sampletype was pursued.
Phase II validationCertain culture results were not informative and wereexcluded from the validation, such as cases where nega-
tive control vials were flagged positive and the positivitywas due to contaminants already present in the CBsamples and/or introduced during sample preparation.
SpecificityUnder testing conditions where 13 test organisms wereseeded into CB-BP samples and inoculated intoorganism-optimal media, 100% recovery of test organismswas achieved, and no false-positive or false-negativeresults were reported in all informative cases. The growthof test organisms (n = 10) in media not optimal for theirgrowth was also monitored during the initial trials of thevalidation process (Table 3). Due to the facultative natureof certain organisms, the growth of these organisms wassupported by nonoptimal media type(s). For example,aerobic organisms have grown in anaerobic or fungalmedium, and yeast or fungal growth was supported by theaerobic media. In no case did anaerobic test organismsgrow in aerobic vials or yeast or fungi grow in anaerobicvials.
SensitivityThrough different trials, it was possible to inoculateorganisms at low CFUs/vial (<100 CFUs). The results indi-cated the sensitivity of the assay to be at or below26 CFUs/vial for all tested organisms inoculated togetherwith CB-BP samples (Table 4). The mean TTD for each testorganism and the TTD for any individual culture waswithin 7 days. Of paramount importance, frozen in-houseisolates were used in initial trials of the validation, someof which were difficult to restore to active growth whensubcultured, indicating their possible attenuation and/orfastidious nature.19 Specifically, isolates of Actinomyces tu-ricensis, P. acnes, and Peptostreptococcus asaccharolyticuseventually showed delayed growth (flagged positive onlywithin 7-14 days) even when inoculated at adequatecounts (>200 CFUs/vial, data not shown). When ATCC
TABLE 1. Organisms selected for microbial assay validation of minimally manipulated CB productsType of organism 13 selected organisms* Source
Yeast/fungal C. albicans*† ATCC 10231A. brasiliensis (niger)* ATCC 16404
* CFR/USP recommended.† Common product contaminant.‡ Slow doubling time.
AKEL ET AL.
4 TRANSFUSION Volume **, ** **
TABLE 2. TTD of tested microbes in various samples after 14 days of culture in BACTEC system: analysis ofmicrobial presence in various product-related fractions and verification of sample adequacy for sterility testing
found and samples adequacyBottle 1 Bottle 2 Bottle 3 Mean
E. coli isolatePlus Aerobic/F0.9 CFUs/mL
Control 10 mL 0;09:59 0;10:09 0;10:18 0;10:09 All fractions.All sample types are adequate but FP at
0.5 mL might be suboptimal.1 mL RBCs 0;14:26 0;13:57 0;14:47 0;14:235 mL RBCs 0;12:56 0;12:04 0;12:34 0;12:319 mL plasma 0;12:05 0;12:25 0;11:44 0;12:0510 mL BP 0;10:54 0;11:04 0;10:54 0;10:570.5 mL FP 0;15:47 0;15:17 NG
B. fragilis isolatePlus Anaerobic/F0.87 CFUs/mL
Control 10 mL 1;09:02 1;07:23 1;10:03 1;08:49 All fractions, but retained better insamples with plasma.
Only samples containing plasma areadequate, including plasma and BP.
1 mL RBCs 1;16:56 NG 1;16:165 mL RBCs NG 2;01:28 NG9 mL plasma 1;16:44 1;14:14 1;13:34 1;14:4410 mL BP 1;10:54 1;11:23 1;09:13 1;10:300.5 mL FP 1;16:32 NG NG
C. albicansATCC 10231Myco F/Lytic1 CFU/mL
Control 5 mL 1;00:20 1;01:22 1;00:19 1;00:40 All fractions, but retained better in the BPand FP
Only samples representing BP and FP areadequate.
0.5 mL RBC NG NG NG NG5 mL RBC 1;05:11 1;04:11 1;03:10 1;04:114.5 mL plasma NG NG 1;06:085 mL BP 1;04:06 1;05:06 1;06:05 1;05:060.5 mL FP 1;02:23 1;03:23 1;03:24 1;03:03
* Control sample represents CB collection mixed with PrepaCyte-CB reagent (pre-processing) inoculated with the microbe at low microbialdose. Other samples are post-processing samples representing various product-related fractions including RBCs, PrepaCyte-CB plasma,BP (9 volumes PrepaCyte-CB plasma + 1 volume RBCs), and FB.
NG: no growth.
TABLE 3. Growth in nonoptimal medium: two representative results for each tested microbe (10 microbes)
Microorganism CFUs/vial
Mean TTD (days;hr:min)
Aerobic medium Anaerobic medium Fungal medium*
Aerobic bacteriaS. aureus 170 0;23:17 1;11:38† NT
61 0;19:49 NG† NTCoagulase-negative staphylococci 53 1;00:42 0;17:48† NT
24 1;00:43 1;22:06† NTGroup B streptococci 110 0;10:01 0;08:44† NT
5 0;10:47 0;09:40† NTB. subtilis 170 0;09:52 1;05:52† NT
2 0;12:47 0;21:40† NTAnaerobic bacteria
C. subterminale 79 NG† 0;16:26 NT7 NG† 0;22:20 NT
B. fragilis 270 NG† 1;00:27 NT13 NG† 1;13:41 NT
Fungus or yeastC. albicans 333 3;15:24† NG† 0;22:11
17 4;36:47† NG† 1;03:13A. niger 32 4;04:58† NT 3;15:08
26 4;07:35† NT 4;02:06
* Volume of BP with organism suspension inoculated into aerobic and anaerobic bottles was 10 mL. Volume of same BP with organism sus-pension inoculated into Myco F/Lytic bottles was 5 mL; therefore, the CFUs/vial is half that of the number for aerobic or anaerobic vials in aspecific run.
† Nonoptimal medium.NG = no growth; NT = not tested.
VALIDATION OF CB STERILITY TESTING
Volume **, ** ** TRANSFUSION 5
strains were substituted for those isolates, unsurprisingly,the ATCC strains had shorter TTD (within 7 days). Suchresults would necessitate incubation for 14 days tocapture all possible CB contaminants.
ReproducibilityTesting for each organism (<100 CFUs/vial) was repeatedat least twice and the mean TTD was consistently below 7days (Table 4). In general, TTD among triplicate vialsvaried slightly but was consistently below 7 days and com-parable (data not shown). In three exceptions, cultures ofStaphylococcus aureus, P. asaccharolyticus, and Clostridi-um subterminale with the very low inoculated organismcounts of 2, 3, and 7 CFUs/vial, respectively, only 2 of 3vials were flagged as positive. Therefore, the assay repro-
ducibility was not certain when contaminating microbeswere present at very low bioburden (<10 CFUs).
Bacteriostatic and fungistatic effect of CB sample inBACTEC culturesThe potential bacteriostatic or fungistatic effect of the ste-rility testing sample (CB-BP) on microbial growth, underculture conditions, was evaluated. Parallel cultures of testorganisms seeded in CB-BP and in sterile normal saline inplace of CB-BP were monitored for TTD. The mean TTD ofall test organisms was shorter in CB-BP compared tosaline (37 hr vs. 45 hr; Table S2, available as supportinginformation in the online version of this paper). Theshorter TTD demonstrated, in most cases, with the organ-ism in CB-BP, would suggest no inhibitory effect of CB-PB
TABLE 4. Informative results of detection of challenge organisms seeded in the CB-BP samples at low dosesand inoculated in optimal blood culture media
Type of organism and medium Selected organism and source CFUs/vial testedMean TTD of triplicatecultures (days;hr:min)
AerobicPlus Aerobic/F
S. aureusATCC 6538
194 0;18:52170 0;23:1761 0;19:493* 0;19:38*
Coagulase-negative StaphylococcusIn-house isolate
53 1;00:4227 1;00:5225 0;23:3024* 1;00:43*
Group B StreptococcusIn-house isolate
110 0;10:0127 0;08:4725 0;10:5617* 0;10:30*
P. aeruginosaATCC 9027
134 0;13:5319 0;15:5315 0;16:3813* 0;15:55*
E. coliIn-house isolate
98 0;08:5914 0;10:068 0;11:065* 0;10:47*
B. subtilisATCC 6633
170 0;09:5283 0;21:0120 0;14:552* 0;12:47*
Facultative anaerobicPlus Anaerobic/F
A. pyogenesATCC 19411
87 1;16:1975 1;11:392* 1;22:59*
AnaerobicPlus Anaerobic/F
C. subterminaleIn-house isolate
23 0;22:385* 0;23:08*
B. fragilisIn-house isolate
13 1;13:417* 2;00:37*
P. acnesATCC 11827
137 4;06:5694 4;19:407* 5;05:40*
P. asaccharolyticusATCC 14963
85 3;01:383* 6;13:34*
FungalMyco F/Lytic
C. albicansATCC 10231
167 0;22:1111 1;01:4711 1;03:269* 1;03:13*
A. brasiliensis (niger)ATCC 16404
13* 4;02:06*16 3;15:08
* Lowest tested CFUs/vial counts, indicating those cultures performed in this validation on which method sensitivity was based.
AKEL ET AL.
6 TRANSFUSION Volume **, ** **
on microbial growth. In the rare cases where the TTD fororganism with CB-BP was longer, the difference in TTDwas less than 2 hours, showing that sample used in steril-ity testing does not have a significant bacteriostatic orfungistatic effect under culture conditions.
Phase III validation
Comparison of two different media lotsThe mean TTD of all test organisms was comparablebetween parallel cultures using the two different medialots (difference <1 hr). The difference in the organism TTDusing the two different media lots ranged from 3 minutes(as in the case of fast-growing coagulase-negative Staphy-lococcus) to 10 hours (as in the case of slow-growingmicrobes P. acnes; Table S3, available as supporting infor-mation in the online version of this paper). Such resultsindicate that slight variations in media characteristics donot have a significant effect on test performance.
Effect of delayed loading of inoculated vials on TTDUsing vials from the same lot, one triplicate set of vialsinoculated with test organism was loaded directly to theinstrument, and an identical set was allowed to remain atroom temperature for a delay period of 6 to 8 hours beforebeing loaded. The mean instrument TTD (from time ofloading) and growth kinetic curve were comparedbetween both sets of cultures. Growth curves were evalu-ated for length of lag phase and exponential phase. Duringthe lag phase of the bacterial growth cycle, bacteria adaptthemselves to growth conditions: individual bacteria arematuring and are not yet able to divide, and synthesis ofRNA, enzymes, and other molecules occurs. Exponentialphase (sometimes called the log phase or the logarithmicphase) is a period characterized by cell doubling anddepends on the doubling rate.20
Incubation of cultures at room temperature beforeloading resulted in a slightly lower instrument TTD
(Table 5), indicating that this incubation actually allowedmicrobial adjustment (adaptation) to the culture mediaenvironment and improved microbial growth behavior.This was apparent when the lag period and exponentialgrowth period were compared between cultures. Ingeneral, incubation at room temperature was associatedwith shorter lag period and with no significant effect onexponential growth period (Table 5), indicating that incu-bation at room temperature had no significant effect onthe doubling rate of tested microbes. In the case ofA. niger, although incubation shortened the lag period,TTD was slightly longer, suggesting that such conditionsmay decrease the doubling rate of A. niger.
DISCUSSION
CB products are used as investigational new drug orlicensed cell therapy to treat various diseases.21,22 Themajority of CBBs employ growth-based RMMs for sterilitytesting in place of the method recommended, describedand specified in 21 CFR 610.12. RMMs have not been com-prehensively validated by most CBBs for this intended useand the noncompliance with FDA regulations is likelyattributed to 1) the stringent validation requirements,sometimes difficult to apply to the unique small products,primarily the requirements for sample type or volume; 2)the lack of clear constituents of the validation study forsterility testing of CB products; 3) the fact that CBBs usevarious CB processing techniques yielding products withdifferent components and thus necessitating the develop-ment of a site- and product type–specific validation;10,11,23
and 4) the laborious nature of the validation process itself,especially when applying a parallel validation of theRMMs against the CFR/USP method.
Taking advantage of the recent FDA amendmentsconcerning the validation requirements for CB sterilitytesting method,5,6 along with the support of the Center forBiologics Evaluation and Research office, the SLCBB sub-mitted to the FDA a plan and validation results specific to
TABLE 5. Ruggedness and robustness: effect of delayed time to loading at room temperature on microbialgrowth (TTD and growth lag period/exponential growth period comparison)
* For P. acnes and A. niger, the end of exponential growth, necessary for calculation of the time of growth phase, was not determined (ND)since the vials were flagged as positive and were removed from the BACTEC instrument for subculture while still in the growth phase,effectively discontinuing the graphs.
VALIDATION OF CB STERILITY TESTING
Volume **, ** ** TRANSFUSION 7
the BACTEC-FX system and PrepaCyte-CB products.Results are discussed below to help CBBs developing areasonable site-specific validation, supporting their BLA.
CBBs use sterile containers, aseptic techniques, andcontrolled environmental conditions to minimize con-tamination, yet risk of CB product contamination remainssignificant (up to 8%) due to the fact that 1) microbes canbe introduced during the collection process and 2) a com-pletely closed processing system is not currently avail-able.24 In principle, each CB product is considered as anindependent lot and must be tested for microbial sterility.Practically, absolute product sterility cannot be guaran-teed employing any of the current sterility testingmethods, in part due to sampling limitations; therefore,our goal was to define and implement an appropriatelyvalidated method to support product sterility under site-specific production conditions. The targets were to inves-tigate the CB processing BP as the surrogate testingsample for FP and to provide evidence that the developedBACTEC method has detection sensitivity of less than100 CFUs/vial for all test organisms within the shortestpossible time between 7 and 14 days, with minimal false-positive results.
False-negative results of culture-based sterilitytesting methods are probable when CB samples containmultiple antibiotics of maternal origin. Penicillinase in theCFR/USP broth and proprietary antibiotic-binding sub-stances in some BacT/ALERT and BACTEC culture mediaare not always fully effective in the neutralization ofresidual amounts of the broad spectrum of antibioticsretained in some CB sterility testing samples.14-17 Previouscomparative studies showed the highest microbial recov-ery rate with BACTEC Plus media when used in cultures ofclinical and cellular therapy samples with antibiotics.9,17
Flayhart and colleagues17 found BACTEC Plus mediumsuperior to BacT/ALERT FA medium in recovering bacte-rial pathogens (95.1% vs. 43.1%) in the presence of thera-peutic levels of frequently used antibiotics. Since morethan 30% of CB products are expected to harbor residualantibiotics, the Plus medium was selected for sterilitytesting. In initial trials, the performance of PlusAnaerobic/F was compared with another BACTECmedium (Lytic/10 Anaerobic/F). It was clear that the useof the Plus medium with antibiotic-binding resins is abso-lutely essential to the sterility testing of CB product toavoid growth failures (data not shown). Our prospectivevalidation did not include studies to evaluate the effec-tiveness of the Plus medium to support microbial growthin CB samples traced to mothers with history of antibiotictreatment. However, a retrospective analysis of dataobtained from different phases of the validation hasserved this purpose. As shown in Tables 2, 3, and 4, all CBsamples (of which presumably >30% would containresidual antibiotics), challenged with test organisms andcultured in Plus medium (aerobic and anaerobic), have
shown growth of the predicted organism with no inci-dence of false-negative results. Such results suggestadequate neutralization of residual antibiotics by Plusmedium resins under production and sterility testing con-ditions. While it is very unlikely that fungal or yeast iso-lates which may contaminate CB products would besensitive to the types of antibiotics routinely given todelivering mothers, it is of note that the culture mediaavailable for fungal or yeast culture do not have antibiotic-binding substances. Interestingly, our test organism panelincluded C. albicans and A. niger, which were seeded inCB samples (of which >30% would presumably containresidual antibiotics) and inoculated into Myco F/Lyticvials, and in all cases (Tables 2-4), yeast or fungal growthof the challenge organism was evident, suggestingadequacy of Myco F/Lytic medium to support microbialgrowth under production and sterility testing conditions.Also of interest is that the aerobic Plus medium systemwith antibiotic-binding resins can support fungal or yeastgrowth, as confirmed in this validation using C. albicansand A. niger. Recently, Rosa and colleagues25 showed agrowth advantage for A. niger in BACTEC Plus Aerobic/Fmedium in cultures from patients undergoing antifungaltreatment.
Another factor that might contribute to false-negativeresults would be a microbial stasis effect of the test samplematrix. While planning for the validation, a pilot feasibilitystudy was conducted to screen individual product com-ponents and materials used in manufacturing for poten-tial microbial stasis effect (data not shown). The maingradients were screened at their relevant levels in samplematrix, including 10% DMSO (1 mL), Dextran 40 (1 mL),CPD (2 mL), PrepaCyte-CB reagent (2 mL), RBCs (5 mL),and WBCs (60 million), in the presence of all test organ-isms (A. niger not tested). Microbial growth in test cultureswas comparable and sometimes faster than that found inthe corresponding positive control culture inoculatedwith test organism in normal saline. CPD exerted differ-ential inhibitory effects on microbial growth especiallywhen included at high volumes (>5 mL). The bactericidalactivity of citrate has been previously described.26,27 Insupport of the above data, Phase II validation results con-firmed that the matrix of the sterility testing sample (CB-BP) has no bacteriostatic or fungistatic effect and ratherexerted stimulatory effects on growth of most test organ-isms. Although the consistent absence of false-negativegrowth using the BACTEC developed method was demon-strated, under no circumstances can absolute productsterility be guaranteed.
Next, the risk of false-positive results was assessed.None of tested sample matrix components interferedwith the BACTEC system detection platform by generat-ing enough CO2 to trigger a false-positive signal.Throughout the entire validation, no incidence of false-positive results was reported among more than 200 vials
AKEL ET AL.
8 TRANSFUSION Volume **, ** **
spiked with CB-BP and challenge organisms. Addition-ally, more than 100 negative control vials containingCB-BP (presumably sterile) alone did not generate anyfalse-positive results. In agreement with our findings,Khuu and coworkers8 tested a wide range of cellulartherapy products and reported false-positive rates of 0.0,0.8, and 7.3% using BACTEC, BacT/ALERT, and CFRmethods, respectively. We intend to continue to monitorthe rate of false-positive results to assure method perfor-mance over time.
Although the most appropriate sterility test materialis the FP, the FDA may allow the use of other material asappropriate and as approved in the BLA application. CB isprocessed using a variety of technologies, most of whichinclude RBC sedimentation and centrifugation steps toyield RBC and/or plasma-reduced products.10,11,23 Manylaboratories perform sterility testing on plasma discardedduring processing. However, plasma may not be a goodsurrogate for detection of microbial contamination in thefinal CB product.9 In a recent study, certain microbes con-taminating CB units at low bioburden were detectableonly in RBCs and FP after unit centrifugation.28 Accord-ingly, we designed a study to determine which processingfraction of PrepaCyte-CB–processed CB units may retainthe contaminants and to determine the most appropriatealternative testing sample. Microbes, inoculated at lessthan 1 CFU/mL of preprocessing volume, were tracedafter processing in various sample types (fractions). Of thesamples tested, the BP has had the best microbial recoveryrate; therefore, we have recommended utilizing the BPsample for method validation and future product sterilitytesting. Since CB processing may vary between sites, it isadvised that each CB production site should validatesample appropriateness in terms of type and volumeduring the initial phase of their validation.
The validation protocol evaluated method specificity,sensitivity, reproducibility, ruggedness, and robustness,using 13 selected isolates. Evaluation of the results of allvariables agreed with the internal acceptance criteria.Although validation results suggest that a 7-day cultureperiod is sufficient to detect test organisms, a 14-dayculture was ultimately selected at the encouragement ofthe FDA and to optimize the recovery of fastidious and/orslow-growing contaminants, especially when present atvery low numbers. Since beginning the 14-day BACTECculture protocol for sterility testing of CB products, three of19 positive cultures were detected between 7 and 10 days ofculture (data not shown). Overall, validation data sup-ported method detection at very low inoculum densities(below 26 CFUs/vial) for all tested organisms. Reproduc-ibility was demonstrated consistently at inoculae higherthan 10 CFUs, where all vials of the same triplicate setshowed positive growth with no significant difference inTTD. For specificity, detection of all tested strains by thedeveloped BACTEC method in optimal test media was con-
firmed. The absence of microbial growth in all negativecultures was also established. Method ruggedness androbustness was challenged with the six organisms (listed inTable S3). This panel of selected organisms will be used forperiodic method and system performance verification andfor performance comparability studies between theprimary and any backup system. Method ruggedness isreflected by degree of result reproducibility under a varietyof normal test conditions such as different analysts, differ-ent instruments, and different reagent lots. Inoculating thesame test organism suspension into two different medialots for each type of BACTEC medium used, the mean TTDof test microbes was comparable between paired cultures,indicating method ruggedness under test conditions(Table S3). For method robustness, one can examine thecapacity of test method to remain unaffected, in terms ofrecovery andTTD, by small deliberate variations in methodvariables such as changes in reagent concentration andincubation temperature. The impact of incubation at roomtemperature and delayed loading (6-8 hr) of culture vialson test results was studied. In agreement with Chapin andLaunerdale,29 results comparing mean instrumentTTD in apaired series of cultured vials have provided evidence formethod robustness (Table 5).
The validation data indicate that the BACTEC auto-mated blood culture system and media formulae havehigh detection capability and can be effectively validatedfor sterility testing of CB products. Caution must be takenduring initial phases of validation with regard to mediaand system selection and evaluation of various types oftest samples for appropriateness. So far, culture-basedmethods may not always be optimal for rapid (<10-14days) sterility testing of CB products. Based on currentCFRs, additional verification activities are warranted todemonstrate that the selected sterility test method contin-ues to be capable of consistently detecting viable con-taminating microorganisms. To this effect, we arecommitted to 1) verify growth promotion properties of themedia when the medium is initially received and atregular intervals over the shelf-life of the medium lot, 2)perform periodic validation verification activities usingrelevant media and test organisms, and 3) revalidate orverify method validation as appropriate whenever thereare changes in test method or CB production method thatcould potentially inhibit detection of viable contaminat-ing microbes.
CONFLICT OF INTEREST
There are no conflicts of interest.
REFERENCES
1. US Food and Drug Administration. General Biologics Stan-
dards (21 CFR parts 600, 610 and 680) and Sterility Testing
(section 610. 12). 2008.
VALIDATION OF CB STERILITY TESTING
Volume **, ** ** TRANSFUSION 9
2. United States Pharmacopeia. United States Pharmacopeia
28 and National Formulary 23 (USP 28 NF 23) 2005; Steril-
ity Tests, General Chapter 71, 1-13.
3. US Food and Drug Administration. Draft guidance for
industry. Validation of growth-based rapid microbiological
methods for sterility testing of cellular and gene therapy
products. Washington, DC: US Department of Health and
Human Services; 2008.
4. United States Pharmacopeia. General Information Chapter
<1223>, Validation of alternative microbiological methods.
USP 32-NF27, U.S. Pharmacopeia Convention. Rockville,
Maryland; 2009.
5. Food and Drug Administration. Amendments to Sterility
Test Requirements for Biological Products. Proposed Rule.
21 CFR Parts 600, 610, and 680 [Docket No. FDA–2011–N–
0080] 2011; 76 36019-27.
6. Food and Drug Administration. Amendments to Sterility
Test Requirements for Biological Products Final Rule. 21
CFR Parts 600, 610, and 680 [Docket No. FDA–2011–N–
0080] 2012; 77:26162-75.
7. Khuu HM, Stock F, McGann M, Carter CS, Atkins JW,
Murray PR, Read EJ. Comparison of automated culture
systems with a CFR/USP-compliant method for sterility
testing of cell-therapy products. Cytotherapy 2004;6:183-
95.
8. Khuu HM, Patel N, Carter CS, Murray PR, Read EJ. Sterility
testing of cell therapy products: parallel comparison of
automated methods with a CFR-compliant method. Trans-
29. Chapin K, Lauderdale TL. Comparison of Bactec 9240 and
Difco ESP blood culture systems for detection of organ-
isms from vials whose entry was delayed. J Clin Microbiol
1996;34:543-9.
SUPPORTING INFORMATION
Additional Supporting Information may be found in theonline version of this article at the publisher’s web-site:
Table S1. Test organisms and vials used for BACTEC mediagrowth promotion quality assayTable S2. Assessment of potential bacteriostatic or fungi-static effect of CB-BP in BACTEC cultures. Mean TTD inorganism optimal mediaTable S3. Ruggedness and robustness, lot comparison
