JOURNAL OF CLINICAL MICROBIOLOGY, Oct. 2010, p. 3475–3481 Vol. 48, No. 10 0095-1137/10/$12.00 doi:10.1128/JCM.00542-10 Copyright ? 2010, American Society for Microbiology. All Rights Reserved.
The Pan Genera Detection Immunoassay: a Novel Point-of-Issue
Method for Detection of Bacterial Contamination
in Platelet Concentrates
Tanja Vollmer, Dennis Hinse, Knut Kleesiek, and Jens Dreier*
Institut fu?r Laboratoriums- und Transfusionsmedizin, Herz- und Diabeteszentrum Nordrhein-Westfalen,
Universita?tsklinik der Ruhr-Universita?t Bochum, Bad Oeynhausen, Germany
Received 12 March 2010/Returned for modi？cation 21 April 2010/Accepted 2 August 2010
Bacterial contamination of platelet concentrates (PCs) still represents an ongoing risk in transfusion-
transmitted sepsis. Recently the Pan Genera Detection (PGD) system was developed and FDA licensed for screening of bacterial contamination of PCs directly prior to transfusion. The test principle is based on the
immunological detection of lipopolysaccharide (for Gram-negative bacteria) or lipoteichoic acid (for Gram-
positive bacteria). In the present study we analyzed the applicability of this method with regard to detection limit, practicability, implementation, and performance. PCs were spiked with Staphylococcus aureus, Bacillus
subtilis, and ；ve different Klebsiella pneumoniae strains, as well as eight different Escherichia coli strains. The
presence of bacteria was assessed by the PGD immunoassay, and bacteria were enumerated by plating cultures. Application of the PGD immunoassay showed that it is a rapid test with a short hands-on time for sample
processing and no demand for special technical equipment and instrument operation. The lower detection
limits of the assay for Gram-positive bacteria showed a good agreement with the manufacturer’s speci；cations 34(8.2 10to 5.5 10CFU/ml). For some strains of K. pneumoniae and E. coli, the PGD test showed analytical 64sensitivities (>10CFU/ml) that were divergent from the designated values (K. pneumoniae, 2.0 10CFU/ml; 4E. coli, 2.8 10CFU/ml). Result interpretation is sometimes dif；cult due to very faint bands. In conclusion, our study demonstrates that the PGD immunoassay is an easy-to-perform bedside test for the detection of
bacterial contamination in PCs. However, to date there are some shortcomings in the interpretation of results
and in the detection limits for some strains of Gram-negative bacteria.
Bacterial contamination of platelet concentrates (PCs) still sampling strategy, carrying a high risk of sampling errors. The represents an ongoing risk in transfusion-transmitted sepsis. In initial levels of most skin-based bacteria in PC units are usually 2004, sterility testing of PCs was recommended by the Amer- remarkably low, and it has been demonstrated that 57% of ican Association of Blood Banks, and the detection of bacterial cultures are false negative at low contamination levels of 10 to contamination in PCs has been implemented in several blood 100 CFU per PC unit (22). Application of rapid detection centers and transfusion services as routine quality control test- methods combined with an early sampling strategy will also ing (11). However, transfusion-transmitted bacterial sepsis has invariably miss bacterial contamination as a result of the sam- still not been completely eliminated, with septic complications pling error. Therefore, a prolonged time frame between sam- observed particularly with older PCs (6, 8, 11, 12, 20, 26, 32, pling and testing will increase the probability that most con- 34). At present, the detection of microbiological contamina- taminated PCs will be identi？ed. With cultural approaches, tion in PCs can be divided into two major methodological results are not available in time to avoid transfusion of the concepts (19): (i) incubation or cultivation methods and (ii) contaminated PC. rapid detection methods, such as nucleic acid ampli？cation Hence, substantial interest focuses on rapid detection meth- techniques (NAT) (9, 16, 28), ，uorescence-activated cell sort- ods for bacterial screening combined with a late sampling strat- ing (FACS) (10, 13, 17, 30, 31), or immunological detection egy. In this context, a sensitive, speci？c, cost-effective, rapid, methods (Pan Genera Detection [PGD] system) (24, 25). and easy-to-perform point-of-issue bacterial detection test im- Incubation or cultivation methods are currently the most mediately before transfusion (24, 25), requiring only a small sensitive detection methods and are utilized predominantly for test sample volume, is considered optimal. Recently, the Pan sterility testing of PCs (7). Nevertheless, culture-based meth- Genera Detection (PGD) system was developed and FDA ods require 24 h prior to sampling and at least 18 h to 24 h of licensed [501(k) clearance] for the screening of bacterial con- incubation to obtain a positive result (5, 15, 23, 24). Therefore, tamination directly prior to transfusion. Experience regarding culture-based methods have to be combined with an early the implementation and performance of this technology has seldom been reported (24, 25). A ？rst study using the PGD test for screening of 7,733 whole-blood derived PCs has been pub- * Corresponding author. Mailing address: Institut fu?r Laboratori- lished (38). The test principle is based on the immunological ums- und Transfusionsmedizin, Herz- und Diabeteszentrum Nord- detection of the conserved bacterial antigens lipopolysaccha- rhein-Westfalen, Universita?tsklinik der Ruhr-Universita?t Bochum, ride (LPS) (for Gram-negative bacteria) or lipoteichoic acid Georgstrasse 11, 32545 Bad Oeynhausen, Germany. Phone: 49-5731- (LTA) (for Gram-positive bacteria) by lateral-，ow immuno- 97-1391. Fax: 49-5731-97-2307. E-mail: email@example.com. precipitation. In a previous study, we developed a novel rapid Published ahead of print on 11 August 2010.
3476 VOLLMER ET AL. J. CLIN. MICROBIOL.
different E. coli and ？ve different K. pneumoniae strains and incubated at 22?C screening method based on ，ow cytometric detection (10), with agitation for 48 h. Samples were diluted in a 10-fold dilution series with which was compared with the PGD test, among others. Pre- sterile PCs, three negative controls were added randomly, and all samples were liminary results revealed that the PGD test detected Gram- analyzed by PGD. Test devices were read by ？ve independent experimenters in positive bacteria in the given range, but Gram-negative bacte- a blinded trial. Each individual result was scored (positive, 1 points; arguable, 0.5 rial species such as Klebsiella pneumoniae were detected with point; negative, 0 points), and total results were evaluated as follows: (i) 0 to 1, negative; (ii) 1.5 to 2.5, arguable, and (iii) 3 to 5, positive. considerably divergent detection limits, as speci？ed by the RAPD PCR analysis and serotyping. Bacteria were harvested from cultures manufacturer. Based on these data, we have evaluated this grown overnight, and bacterial DNA was extracted using the QIAamp DNA effect systematically and in detail in the present study. blood kit (protocol D; Qiagen, Hilden, Germany) according to the manufactur- er’s instructions. Nucleic acids were eluted with 200 l elution buffer (Qiagen). Randomly ampli？ed polymorphic DNA (RAPD) PCR analysis was performed MATERIALS AND METHODS using arbitrary primers (ERIC-1, 5 -ATGTAAGCTCCTGGGGATTCAC-3 ; ERIC-2, 5 -AAGTAAGTGACTGGGGTGAGCG-3 ) (36). DNA ampli？cation Bacterial strains and culture conditions. The two strains Escherichia coli was carried out in 0.2-ml tubes containing 45 l reaction mix and 5 l DNA ATCC 35218 and K. pneumoniae ATCC 13882 were purchased from the Amer- extract. The reaction mixture consisted of 1 AmpliTaq buffer, including 1.5 mM ican Type Culture Collection (ATCC) (LGC Promochem GmbH, Wesel, Ger- MgSO(Applied Biosystems, Foster City, CA), 200 M each deoxynucleoside 4many). Strains K. pneumoniae PEI-B-08-08 and Staphylococcus aureus PEI-B- triphosphate, 2000 nM each primer, and 5 U of AmpliTaq DNA polymerase 23-04 were obtained from the Paul-Ehrlich-Institute (PEI) (Langen, Germany). (Applied Biosystems). DNA ampli？cation was carried out with the following E. coli strain L01207081 and K. pneumoniae strain L01204084 were provided by thermal cycling pro？le: preliminary denaturation at 95?C for 5 min, followed by Verax Biomedical Inc. (Worcester, MA). Bacterial spore suspensions with de- 45 cycles of denaturation at 95?C for 30 s, annealing at 35?C for 60 s, and ？ned titers of Bacillus subtilis ATCC 35031 (SGM Biotech Inc., Bozeman, MT) extension at 72?C for 120 s. DNA fragments were separated by gel electrophore- were cultured in Trypticase soy broth (bioMe?rieux, Nu?rtingen, Germany) under sis in a 1.5% agarose gel (Carl Roth, Karlsruhe, Germany) in 1 UltraPure aerobic conditions at 37?C for 24 to 48 h. All other K. pneumoniae and E. coli Tris-borate-ETA (TBE) buffer (pH 8.0) (Gibco Technologies, Paisley, Scotland) strains were previously isolated from patient specimens at our hospital and were containing 500 ng/ml ethidium bromide. After addition of 2 l of 6 loading dye characterized in our microbiological laboratory by standard methods. (Fermentas, St. Leon-Rot, Germany), 10 l of PCR product was loaded onto the PC collection. Apheresis-derived single-donor PCs were obtained from the gel. The pUC mix 8 DNA ladder (Fermentas) was used as a molecular size transfusion service Uni.Blutspendedienst OWL (Bad Oeynhausen, Ger- marker. Electrophoresis was carried out at room temperature and at a constant many). PCs were prepared using the Haemonetics MCS (Haemonetics voltage of 6 V/cm in 0.5 TBE buffer. GmbH, Munich, Germany) from healthy blood donors and stored in gas- Serotyping of O and H antigens of E. coli strains was performed by the Robert permeable containers (LN994CF-CPP; Haemonetics GmbH) at 20 to 24?C Koch Institute (RKI) (Wernigerode, Germany). with agitation. Predonation sampling was performed after donor arm disin- fection using a single-swab method with 70% isopropyl alcohol. The ？nal PC volume was approximately 235 ml. PGD testing. The PGD test (Verax Biomedical Inc.) was performed accord- RESULTS ing to the manufacturer’s instructions. Brie，y, 500 l of PC sample was mixed with 8 drops of reagent 1 (water, methanol, surfactants, and preservative Performance and implementation. The PGD test is very [ProClin300]) by inverting the tube three times, followed by centrifugation at easy to perform, with a short hands-on time. Altogether, the 11,000 g for 5 min. Subsequently, the supernatant was removed and 8 drops of processing of one sample at a time took 8 min. Reading of the reagent 2 (water, sodium hydroxide, surfactants, and preservative [sodium test device took a maximum of an additional 6 min (1 min per azide]) were added to the cell pellet. Sample preparations were not vortexed before the addition of reagent 3 (Tricine buffer with surfactants, anticoagulants, result interpretation, six different times in case of negative protein stabilizers [bovine, mouse, and rabbit], and preservatives [ProClin300 results). The PGD test procedure took 77 min before nega- and sodium azide]). In a change from the manufacturer’s instructions, the pellet tively tested PCs were accessible for transfusion release, was loosened carefully from the bottom of the tube with a disposable pipette but whereas the earliest positive results could be obtained after 37 was not divided into three or four fragments. The maximum residence time of reagent 2 did not exceed 2 min. Afterwards, 4 drops of reagent 3 were added, the min. Increasing processing to up to six samples at a time did pellet was completely resuspended by vortexing, and the total sample volume was not considerably in，uence the hands-on time or time to result. transferred to the test device. Test performance and interpretation of results Detection of bacterial proliferation in PCs. PCs were ana- were implemented as described by the manufacturer. lyzed after inoculation with 0.06 CFU/ml K. pneumoniae PEI- Detection of bacterial proliferation in PCs. Before bacterial inoculation, all B-08-08, 0.19 CFU/ml E. coli ATCC 35218, 0.07 CFU/ml B. PCs used for spiking experiments were sampled to ensure baseline sterility of the original apheresis bags. Five milliliters of sample was inoculated into both the subtilis ATCC 35031, or 0.13 CFU/ml S. aureus PEI-B-23-04 at aerobic (BacT/Alert SA; bioMe?rieux, Nu?rtingen, Germany) and anaerobic eight different times during storage (Fig. 1), under conditions (BacT/Alert SN; bioMe?rieux) culture bottles and incubated for up to 7 days. The which might be encountered in practice. The BacT/Alert au- time periods until detection of bacterial contamination using the PGD test were tomated culture system was used to con？rm successful con- compared during PC storage at 20 to 24?C after inoculation with 1 CFU/ml of K. pneumoniae PEI-B-08-08, E. coli ATCC 35218, S. aureus PEI-B-23-04, and B. tamination. For each inoculated PC, bacteria were detected in subtilis ATCC 35031. Bacterial titers of 1 CFU/ml were achieved by 10-fold a minimum of one BacT/Alert culture bottle at the time of serial dilution of stationary-grown overnight cultures in phosphate-buffered sa- inoculation, whereas bacterial contamination was detected for line (PBS), followed by inoculation of 1 ml of the respective dilution (K. pneu- all four strains under both aerobic and anaerobic conditions 7875moniae, 10; E. coli, 10; S. aureus, 10; and B. subtilis, 10). To control after 24 h (data not shown). Subsequently, the applicability of bacterial inoculation of a PC, samples were taken immediately after inoculation (0 h), as well as after 24 h, and analyzed with the BacT/Alert 3D continuous- the PGD test was analyzed with regard to sensitivity and the monitoring system as described above. Sampling of appropriate sample volumes time period until detection of bacterial contamination. Con- was performed by sterile welding of satellite bags at 0, 6, 12, 24, 30, 48, 54, and tamination with S. aureus was initially detected after 48 h, with 72 h after inoculation directly prior to analysis. Subsequently, 500 l of PC was 6a corresponding titer of 4.67 10CFU/ml (Fig. 1A). The analyzed using the PGD test as described above. 3To monitor bacterial growth kinetics, 100- l aliquots of 10-fold serial dilutions bacterial titer of 8.4 10CFU/ml at the previous sampling 3of PC samples were plated in triplicate onto tryptone soy agar (TS) and incu- point was very close to the speci？ed detection limit of 8.2 10 bated at 37?C for at least 48 h. After incubation, the number of colonies was CFU/ml, resulting in a statistical contingency to obtain a pos- counted and the initial bacterial concentration was calculated. itive or negative result. The PGD test revealed an initially Detection limit for Gram-negative strains. To analyze the detection limit of positive result for B. subtilis after 72 h, with a corresponding the PGD test for different Gram-negative strains, PCs were spiked with eight
VOL. 48, 2010 PGD STERILITY SCREENING OF PCs 3477
FIG. 1. Detection of bacterial proliferation by the PGD immunoassay. One single apheresis-derived PC unit was spiked with 1 CFU/ml of the indicated bacterium and stored at 22?C with agitation. Samples were taken immediately (0 h) and at 6, 12, 24, 48, 54, and 72 h after inoculation. The PGD test was performed as described in Materials and Methods. Bacteria were enumerated by colony-forming assay. The dotted horizontal line represents the lower detection limit of the PGD assay for the respective strain.
4ml. Additionally, we observed dif？culties in obtaining valid titer of 4.63 10CFU/ml (Fig. 1B). At the previous sampling
point, the bacterial titer of 667 CFU/ml was too low to obtain procedural control windows when analyzing high titers of bac- a positive result. In summary, the results obtained for the teria, especially for K. pneumoniae. According to our consul-
Gram-positive strains correlated well with the analytical sensi- tation with the manufacturer, valid procedural control win- 3tivities reported by the manufacturer (S. aureus, 8.2 10 dows are not necessary in the case of a clearly positive result 4CFU/ml; B. cereus [representative for B. subtilis], 1.2 10 (personal communication).
CFU/ml). Detection limits for different E. coli and K. pneumoniae
In contrast, PGD testing of PCs inoculated with Gram-neg- strains. To analyze the detection limits of the PGD test for K.
ative bacteria demonstrated analytical sensitivities that were pneumoniae and E. coli in detail, PCs were spiked with eight considerably divergent from the speci？ed lower detection lim- different E. coli and ？ve different K. pneumoniae strains and
its, followed by a delayed detection of bacteria. The earliest incubated at 22?C with agitation for 48 h. The two strains E.
positive result for the PC unit inoculated with E. coli was coli L01207081 and K. pneumoniae L01204084, which were 6used for the FDA validation studies, were used as reference obtained after 54 h, with a bacterial titer of 9.20 10CFU/ml
strains. Samples were diluted in a 10-fold dilution series with (Fig. 1C), whereas the detection limit reported by the manu- 4sterile PCs and analyzed by PGD. Test devices were read by facturer was 2.8 10CFU/ml. For the PC inoculated with K.
pneumoniae, the PGD test revealed an arguably positive result, ？ve independent experimenters in a blinded trial (Table 1). To with very faint bands after 30 h of incubation, with a corre- allow for the subjectivity of reports by each individual experi- 7sponding bacterial titer of 2.13 10CFU/ml (Fig. 1D). The menter and to obtain an objective qualitative result, we scored lower detection limit was indicated by the manufacturer to be each individual result and calculated the total results as de- 42.0 10CFU/ml; therefore, the PGD test should already scribed in Materials and Methods. In general, very faint bands provide a positive result after 24 h, as well as at this sampling complicated the result interpretation, which is clearly compre- point. De？nite positivity was not achieved until sampling after hensible on the basis of the test device interpretation pre- 8sented in Table 1. The reference strain E. coli L01207081, as 48 h, with a corresponding bacterial titer of 8.97 10CFU/
3478 VOLLMER ET AL. J. CLIN. MICROBIOL.
TABLE 1. Strain characteristics and detection limits for different E. coli and K. pneumoniae strains
Strain characteristics Results of PGD testing
Species and Titer Score given by experimenter: baisolate RAPD cluster Serovar Total score(CFU/ml) 2 3 4 1 5 E. coli 2.85E 07 EC-I O6:H31 5.0 (positive) Isolate MS88 2.85E 06 0.0 (negative) O102:H Isolate 079 EC-II 1.75E 08 5.0 (positive) 1.75E 07 2.0 (arguable) 1.75E 06 1.0 (negative) Isolate 008 EC-III Ont:H4 3.40E 08 5.0 (positive) / 3.40E 07 2.5 (arguable) 3.40E 06 0.0 (negative) ATCC 35218 EC-IV O6:H1 3.55E 08 5.0 (positive) 3.55E 07 3.0 (positive) 3.55E 06 2.0 (arguable) Ont:H35 Isolate MS82 EC-V 3.20E 08 5.0 (positive) / 3.20E 07 0.5 (negative) 3.20E 06 0.0 (negative) Isolate A001S EC-VI Ont:H 1.45E 04 5.0 (positive) / 1.45E 03 0.5 (negative) 1.45E 02 0.0 (negative) L01207081 EC-VI Ont:H 4.28E 04 5.0 (positive) 4.28E 03 0.0 (negative) O1:H7 Isolate 835 EC-VII 7.00E 06 5.0 (positive) / 7.00E 05 0.5 (negative) 7.00E 04 0.0 (negative) K. pneumoniae c 5.10E 08 5.0 (positive) KP-I NA PEI-B-08-08 / / 5.10E 07 1.0 (negative) 5.10E 06 0.0 (negative) ATCC 13882 KP-II NA 1.06E 08 5.0 (positive) / / 1.06E 07 1.0 (negative) 1.06E 06 0.0 (negative) Isolate 13734 KP-III NA 5.10E 08 5.0 (positive) / / 5.10E 07 1.0 (negative) 5.10E 06 0.0 (negative) L01204084 KP-IV NA 7.95E 03 5.0 (positive) 7.95E 02 0.0 (negative) Isolate 01662 KP-V NA 2.45E 08 5.0 (positive) / / 2.45E 07 4.0 (positive) 2.45E 06 0.0 (negative) Isolate 7724 KP-VI NA 3.26E 08 5.0 (positive) / / / 3.26E 07 1.5 (arguable) 3.26E 06 0.0 (negative) a Scores were evaluated as described in Materials and Methods. b Strain clustering is based on identical banding pattern for both primers used in RAPD-PCR analysis. c NA, not applicable.
well as E. coli isolate A001S, were detected in the given range CFU/ml (score, 5.0). Only isolate 01662 (score, 4.0) was also 74of 2.8 10CFU/ml with a maximum score of 5.0 (positive, detectable with a bacterial titer of 10CFU/ml.
concordant results by all experimenters). In contrast, E. coli RAPD-PCR analysis and serotyping. To study the molecular isolates MS82 and MS88 were detected (score, 5.0) only with relatedness of K. pneumoniae and E. coli strains and to infer a 87bacterial titers of 10CFU/ml and 10CFU/ml, respectively. E. potential connection to the differences in detection limit, coli isolates 079 and 008 were de？nitely detected (score, 5.0) RAPD-PCR analysis was performed using two single-oligonu- 8with a bacterial titer of 10CFU/ml, whereas arguable results cleotide primers with arbitrary sequences. To determine 7were obtained for titers of 10CFU/ml (score, 2.0 or 2.5). Only whether the composition of E. coli LPS in，uences the immu- E. coli strain ATCC 35218 (arguable; score, 2.0) and isolate nologic detection, strains were serotyped regarding their O and
H side chains (Table 1). Primer ERIC-1 yielded six different 835 (positive; score, 5.0) were detectable with bacterial titers of 6banding patterns for the different K. pneumoniae strains and 10CFU/ml. Accordingly, the K. pneumoniae reference strain
seven different patterns for E. coli, whereas primer ERIC-2 L01204084 was detected (score, 5.0) with a bacterial titer of 3yielded ？ve different patterns for K. pneumoniae and six dif- 7.95 10CFU/ml, which is closely below the given range of 4ferent banding patterns for E. coli (data not shown). Regarding 2.0 10CFU/ml. However, similarly to the detection of E.
both primers, a molecular relatedness was detectable only for coli, K. pneumoniae strains PEI-B-08-08 and ATCC 13882 and 8the two E. coli strains A001S and L01204081 (cluster EC-VI) isolate 13734 were positive only at a concentration of 10
VOL. 48, 2010 PGD STERILITY SCREENING OF PCs 3479
(Table 1). Both strains AS001 and L01204081 had no H anti- PGD test also occur using nonimmunological rapid detection gen, whereas the O antigens are not typeable so far. Interest- principles, PCs were analyzed in parallel using a recently pub- ingly, these two E. coli strains were detected by the PGD assay lished ，ow cytometric assay (BactiFlow) (10). Strain-depen- 4dent detection limits were not observed using this ，ow cyto- at as low as 10CFU/ml, in accordance with the detection limit
reported by the manufacturer. All other strains demonstrated metric detection principle (data not shown). Molecular
relatedness between RAPD-PCR patterns, as well as sero- no molecular relatedness based on RAPD-PCR pro？le and
types, of all investigated isolates was observed only for the E. serotyping results (Table 1).
coli reference strain L01204081 and isolate A001S. These two
strains were the only strains which were detected in the spec- DISCUSSION i？ed detection limit of the PGD assay. A relationship between Different strategies to optimize the detection of bacterial RAPD-PCR patterns or serotypes of the other strains and the contamination in PCs have been discussed (8): (i) early testing detection limit of the PGD test could not be demonstrated. after preincubation for 24 h using culture methods with PC These results reinforce the observation that the PGD test release and negative-to-date status, (b) early testing after pre- showed differences in the ef？ciency of detection of different incubation for 24 h using rapid methods (NAT and FACS), strains. The LPS and LTA components of the bacterial cell are and (iii) late testing in combination with rapid detection meth- attractive targets for the detection of bacterial contamination
in blood products. However, the use of these two peptides has ods (NAT, FACS, and PGD) (8). The consensus of several
studies is a late-sampling strategy in combination with rapid to ensure that all relevant bacterial species are detected with detection methods (9, 18), offering the possibility of sample equal sensitivity (3), and it seems that the PGD test could not drawing at a later stage to overcome the risk of sampling error detect the LPS components of different E. coli and K. pneu-
moniae strains with equal sensitivity. Potential factors interfer- due to initially low bacterial count or slow-growing bacterial
ing with the test antigens were the capsule structure and ex- species. Decisions regarding bacterial testing strategies always
have to balance the requirements for rapid detection and assay opolysaccharides which disguise the LPS antigen structure. In sensitivity (27). Considering the logistic complications involved particular, Klebsiella spp. express a large capsule (K antigen) in testing immediately before transfusion, the ideal rapid playing a signi？cant role in pathogenicity (29). Bacteria also
express two different classes of LPS: smooth LPS, which is screening method for bacterial contamination of PCs should be
easy to perform, should not be laborious, and should have a composed of O antigen, complete core oligosaccharides and quick turnaround time (24). The optimal solution comprises a the lipid A hydrophobic domain, and rough LPS, which lacks direct, bedside-like bacterial detection method applied directly the O antigen but possesses lipid A and progressively shorter prior to transfusion. The PGD test ful？lls most of these re- core oligosaccharides (35). The epitopes of the utilized anti- quirements. Specialized personnel quali？cations and technical gens strongly affect the detection ef？ciency, and the PGD test
equipment are not required for the application of the PGD might bene？t from an extended antigen detection panel. No- test, and instrument operation does not considerably interrupt tably, three E. coli strains tested did not possess the H antigen, the work ，ow in the PC-providing facilities. The simplicity of including the two strains that met manufacturer speci？cations
performance is a promising aspect that should promote usage for the lower detection limit. This raises the question of of the PGD test as a bedside-like test. The test requires a whether the absence of the H antigen might in，uence the
minimum time to result of approximately 30 min in the case of detection ef？ciency. However, antibodies used for immunolog- positive results; the time to result increased up to 1.5 h in the ical detection by PGD target the bacterial lipopolysaccharide,
whereas the ，agellum is not involved. This is further reinforced case of negative results, which is also comparatively maintain-
able for a bedside-like test. Therefore, the greatest advantages by the fact that E. coli isolate 079, which, in contrast, is de- 7of the PGD test are its rapidity and easy performance. tected only in the 10CFU/ml range, did not possess the H The analytical sensitivity of the PGD test method is speci？ed antigen either. 353to be 10to 10CFU/ml for Gram-negative bacteria and 10to Previous studies have also shown that the storage time for 4contaminated PCs signi？cantly differs between Gram-negative 10CFU/ml for Gram-positive bacteria. Detection of bacterial
contamination under conditions which might be encountered and Gram-positive organisms (median storage time, 2.5 versus in practice showed a good agreement with the given detection 5 days; P 0.0001) (14)). Consequently, the spectrum of limits for Gram-positive bacteria. The PGD test showed ana- bacteria accounting for the majority of transfusion-associated
septic complications and fatal casualties varies. In all fatal lytical sensitivities for some Gram-negative strains, e.g., E. coli
cases, endotoxin was present in the posttested products (14). and K. pneumoniae strains, that were considerably divergent 4Despite the fact that Gram-positive organisms were detected from the designated values (K. pneumoniae, 2.0 10CFU/ml; 4most frequently in PCs, Gram-negative organisms accounted E. coli, 2.8 10CFU/ml), resulting in a delayed detection of
for the majority of transfusion fatalities (59.7%) (4). Kuehnert bacteria during PC storage or false-negative test results. The
et al. also observed that the majority of transfusion fatalities systematic analysis of different Gram-negative K. pneumoniae
were associated with Gram-negative organisms and PC trans- and E. coli strains con？rmed these results. The detection limit 67fusion within 3 days of storage, whereas PCs associated with was mostly determined in the 10to 10range, with the excep-
nonfatal transfusion complications were related to Gram-pos- tion of the reference strains and E. coli isolate A001S. We also
itive organisms and PC transfusion after 5 days of storage observed this problem with some strains of other Gram-nega-
(BaCon study) (14). The presence of endotoxin in PCs con- tive bacteria, such as Serratia marcescens, Pseudomonas aerugi-
taminated with Gram-negative organisms most likely explains nosa, and Klebsiella oxytoca (data not shown). To analyze
the correlation between transfusion fatalities with Gram-neg- whether the strain-dependent detection limits observed by the
3480 VOLLMER ET AL. J. CLIN. MICROBIOL.
ative organisms, short storage time, and death (1). In contrast, present study. The average rate of contamination of PCs varies
from 0.08 to 0.7%, depending on technology, testing protocols, Reading and Brecher observed that fatalities show a tendency
and additional intervention methods (34, 37). Indeed, the in- to be equally divided between Gram-positive and Gram-nega-
cidence is approximately 1:2,000 to 1:3,000 in PCs (2, 7). tive organisms (27). Remarkably, E. coli and K. pneumoniae
were responsible for 5.8% and 17.3% of all PC transfusion Therefore, the chance of missing a contaminated PC using a
PGD test resulting in very faint bands is considerable, regard- fatalities in the United States from 1976 to 1998 (27). Klebsiella
ing the high proportion of negative tests. Of course, PCs with spp. were the most commonly reported Gram-negative organ-
isms (21). As mentioned above, we observed that the PGD an at least arguable result have to be discounted immediately, immunoassay revealed some strain-dependent differences, es- but the general shortage of blood products, the high costs pecially regarding the ef？ciencies of detection of different K. (particularly for apheresis products), and the ethical concerns pneumoniae and E. coli strains. All of these strains were capa- regarding the donor demand a clear and nonambiguous result. ble of growing in PCs and therefore had the potential to induce In this context, an improvement of the PGD test in regard to septic transfusion reactions. The predominant proportion of result interpretation, potentially by an automated reader, Gram-negative organisms involved in transfusion fatalities em- would considerably enhance test performance and practicabil- phasizes the requirement that the divergent intra- and inter- ity. An interlaboratory comparison of pro？ciency testing with
species detection ef？ciencies of the PGD test had to be different blood centers and transfusion facilities needs to be adapted for an effective improvement of PC safety. Recently, a performed to compare experiences using the PGD immunoas- ？rst study using the PGD test for screening of 7,733 whole- say, especially for strains which are dif？cult to detect within the
blood-derived PCs was published (38). Two PCs were positive determined detection limits.
for coagulase-negative staphylococci and group B streptococci, In summary, application of the PGD assay as a novel point- respectively, whereas 12 PCs returned false-positive results. of-issue detection method for bacterial contamination in PCs Gram-negative bacteria were not detected. However, only PCs has revealed some shortcomings demanding further improve- showing a reactive signal after application of the PGD test ment in order to effectively increase the safety of PCs: (i) were tested using automated culture systems. Therefore, the Gram-positive bacteria were detected in the given range, but frequency of PCs contaminated with bacteria which were we observed a strain-dependent detection limit for different E.
missed by the PGD test is unknown. The authors also state that coli and K. pneumoniae strains that, in some cases, is consid- the frequency of bacterial contamination observed in this study erably lower than the given range, and (ii) result interpretation is at the lower end of that reported in the literature, which is often complicated due to very faint bands, and the detection might be explained by a lower sensitivity of the PGD test of positive PGD results for different E. coli and K. pneumoniae
compared to the culture-based test (38). strains is strongly dependent on the experimenter performing The critical factor for evaluation of any detection method is the reading. Without a doubt, the PGD test is a rapid and the de？nition of the number of bacteria representing clinically easy-to-handle immunoassay for the detection of bacterial con- signi？cant contamination (24). In addition to the previously tamination in PCs which does not require special precautions mentioned disadvantages, incubation or cultivation methods or stringent work ，ow conditions.
were, to date, the only methods for sterility testing of PCs,
recognizing a minimum contamination of at least one viable ACKNOWLEDGMENT microorganism. The lower detection limits of all available We thank Sarah L. Kirkby for her linguistic advice. rapid detection methods, including NAT, FACS, and PGD, made them inadequate for sterility testing of PCs due to their REFERENCES relative lack of sensitivity for the detection of a low-level bac- 1. Arduino, M. J., L. A. Bland, M. A. Tipple, S. M. Aguero, M. S. Favero, and teremia. Nowadays, these methods need to be considered only W. R. Jarvis. 1989. Growth and endotoxin production of Yersinia enteroco- litica and Enterobacter agglomerans in packed erythrocytes. J. Clin. Microbiol. as screening methods for bacterial contamination, not for ste- 27:1483–1485. rility testing. The clinical signi？cance of the bacterial titer, and 2. Blajchman, M. A. 2002. Incidence and signi？cance of the bacterial contam- therefore the minimum detection limit required for a rapid ination of blood components. Dev. Biol. (Basel) 108:59–67. 3. Blajchman, M. A., M. Goldman, and F. Baeza. 2004. Improving the bacte- screening method, is still questionable. However, it is com- riological safety of platelet transfusions. Transfus. Med. Rev. 18:11–24. monly accepted that initial levels of bacteria need to increase 4. Brecher, M. E., and S. N. Hay. 2005. Bacterial contamination of blood around 10,000-fold to have a clinical impact. Low levels of components. Clin. Microbiol. Rev. 18:195–204. 5. Brecher, M. E., S. N. Hay, A. D. Rose, and S. J. Rothenberg. 2005. Evaluation bacteria may have no clinical signi？cance, but transfusion re- of BacT/ALERT plastic culture bottles for use in testing pooled whole 23actions may also occur with as few as 10to 10CFU/ml, even blood-derived leukoreduced platelet-rich plasma platelets with a single con- taminated unit. Transfusion 45:1512–1517. with organisms usually regarded as nonpathogenic (e.g., S. 6. de Korte, D., J. Curvers, W. L. de Kort, T. Hoekstra, C. L. van der Poel, E. A. epidermidis) (24). Therefore, the sensitivities of screening Beckers, and J. H. Marcelis. 2006. Effects of skin disinfection method, methods had to be increased as much as possible (e.g., BactiFlow deviation bag, and bacterial screening on clinical safety of platelet transfu- sions in the Netherlands. Transfusion 46:476–485. ，ow cytometry, 150 CFU/ml ; NAT techniques, 22 to 29 7. Dreier, J., M. Sto?rmer, and K. Kleesiek. 2007. Real-time polymerase chain 4CFU/ml ), but sensitivities of up to 10CFU/ml, as speci- reaction in transfusion medicine: applications for detection of bacterial con- ？ed for the PGD test, are commonly accepted. tamination in blood products. Transfus. Med. Rev. 21:237–254. 8. Dreier, J., M. Sto?rmer, L. Pichl, V. Schottstedt, A. Grolle, J. Bux, and K. A second shortcoming of the PGD assay is the interpretation Kleesiek. 2008. Sterility screening of platelet concentrates: questioning the of results, which is sometimes dif？cult due to very faint bands optimal test strategy. Vox Sang. 95:181–188. 9. Dreier, J., M. Sto?rmer, L. Pichl, V. Schottstedt, A. Grolle, J. Bux, K. and a strong dependence on the experimenter. In addition, the Kleesiek. 2008. Bacterial screening of platelet concentrates: results of steril- expectations of the experimenter in routine clinical use are not ity testing in two German blood services. Vox Sang. 95:181–188. comparable to the situation with spiking experiments in the 10. Dreier, J., T. Vollmer, and K. Kleesiek. 2009. Novel ，ow cytometry-based
VOL. 48, 2010 PGD STERILITY SCREENING OF PCs 3481
screening for bacterial contamination of donor platelet preparations com- 24. Palavecino, E. L., R. A. Yomtovian, and M. R. Jacobs. 2006. Detecting pared with other rapid screening methods. Clin. Chem. 55:1492–1502. bacterial contamination in platelet products. Clin. Lab. 52:443–456. 11. Eder, A. F., J. M. Kennedy, B. A. Dy, E. P. Notari, J. W. Weiss, C. T. Fang, 25. Prowse, C. 2007. Zero tolerance. Transfusion 47:1106–1109. S. Wagner, R. Y. Dodd, and R. J. Benjamin. 2007. Bacterial screening of 26. Ramirez-Arcos, S., C. Jenkins, J. Dion, F. Bernier, G. Delage, and M. apheresis platelets and the residual risk of septic transfusion reactions: the Goldman. 2007. Canadian experience with detection of bacterial contami- American Red Cross experience (2004–2006). Transfusion 47:1134–1142. nation in apheresis platelets. Transfusion 47:421–429. 12. Fang, C. T., L. A. Chambers, J. Kennedy, A. Strupp, M. C. Fucci, J. A. Janas, 27. Reading, F. C., and M. E. Brecher. 2001. Transfusion-related bacterial sep- Y. Tang, C. A. Hapip, T. B. Lawrence, and R. Y. Dodd. 2005. Detection of sis. Curr. Opin. Hematol. 8:380–386. bacterial contamination in apheresis platelet products: American Red Cross 28. Rood, I. G., M. H. Koppelman, A. Pettersson, and P. H. Savelkoul. 2008. experience, 2004. Transfusion 45:1845–1852. Development of an internally controlled PCR assay for broad range detec- 13. Karo, O., A. Wahl, S. B. Nicol, J. Brachert, B. Lambrecht, H. P. Spengler, F. tion of bacteria in platelet concentrates. J. Microbiol. Methods 75:64–69. Nauwelaers, M. Schmidt, C. K. Schneider, T. H. Muller, and T. Montag. 29. Rukavina, T., B. Ticac, M. Susa, N. Jendrike, S. Jonjic, P. Lucin, R. Marre, 2008. Bacteria detection by ，ow cytometry. Clin. Chem. Lab Med. 46:947– M. Doric, and M. Trautmann. 1997. Protective effect of antilipopolysaccha- 953. ride monoclonal antibody in experimental Klebsiella infection. Infect. Im- 14. Kuehnert, M. J., V. R. Roth, N. R. Haley, K. R. Gregory, K. V. Elder, G. B. mun. 65:1754–1760. Schreiber, M. J. Arduino, S. C. Holt, L. A. Carson, S. N. Banerjee, and W. R. 30. Schmidt, M., M. K. Hourfar, S. B. Nicol, H. P. Spengler, T. Montag, and E. Jarvis. 2001. Transfusion-transmitted bacterial infection in the United Seifried. 2006. FACS technology used in a new rapid bacterial detection States, 1998 through 2000. Transfusion 41:1493–1499. method. Transfus. Med. 16:355–361. 15. McDonald, C. P., A. Rogers, M. Cox, R. Smith, A. Roy, S. Robbins, S. 31. Schmidt, M., A. Karakassopoulos, J. Burkhart, R. Deitenbeck, J. Asmus, Hartley, J. A. Barbara, S. Rothenberg, L. Stutzman, and G. Widders. 2002. T. H. Muller, F. Weinauer, E. Seifried, and G. Walther-Wenke. 2007. Com- Evaluation of the 3D BacT/ALERT automated culture system for the de- parison of three bacterial detection methods under routine conditions. Vox tection of microbial contamination of platelet concentrates. Transfus. Med. Sang. 92:15–21. 12:303–309. 32. Schrezenmeier, H., G. Walther-Wenke, T. H. Muller, F. Weinauer, A. You- nis, T. Holland-Letz, G. Geis, J. Asmus, U. Bauerfeind, J. Burkhart, R. 16. Mohammadi, T., P. H. Savelkoul, R. N. Pietersz, and H. W. Reesink. 2006. Deitenbeck, E. Forstemann, W. Gebauer, B. Hochsmann, A. Karakassopou- Applications of real-time PCR in the screening of platelet concentrates for los, U. M. Liebscher, W. Sanger, M. Schmidt, F. Schunter, W. Sireis, and E. bacterial contamination. Expert Rev. Mol. Diagn. 6:865–872. Seifried. 2007. Bacterial contamination of platelet concentrates: results of a 17. Mohr, H., B. Lambrecht, A. Bayer, H. P. Spengler, S. B. Nicol, T. Montag, and T. H. Muller. 2006. Basics of ，ow cytometry-based sterility testing of prospective multicenter study comparing pooled whole blood-derived plate- platelet concentrates. Transfusion 46:41–49. lets and apheresis platelets. Transfusion 47:644–652. 18. Montag, T. 2008. Strategies of bacteria screening in cellular blood compo- 33. Sto?rmer, M., K. Kleesiek, and J. Dreier. 2007. High-volume extraction of nents. Clin. Chem. Lab Med. 46:926–932. nucleic acids by magnetic bead technology for ultrasensitive detection of 19. Mu?ller, T. H., H. Mohr, and T. Montag. 2008. Methods for the detection of bacteria in blood components. Clin. Chem. 53:104–110. bacterial contamination in blood products. Clin. Chem. Lab Med. 46:933– 34. te Boekhorst, P. A., E. A. Beckers, M. C. Vos, H. Vermeij, and D. J. van 946. Rhenen. 2005. Clinical signi？cance of bacteriologic screening in platelet 20. Munksgaard, L., L. Albjerg, S. T. Lillevang, B. Gahrn-Hansen, and J. Georg- concentrates. Transfusion 45:514–519. sen. 2004. Detection of bacterial contamination of platelet components: six 35. Trianta；lou, M., K. Trianta；lou, and N. Fernandez. 2000. Rough and years’ experience with the BacT/ALERT system. Transfusion 44:1166–1173. smooth forms of ，uorescein-labelled bacterial endotoxin exhibit CD14/LBP 21. Niu, M. T., M. Knippen, L. Simmons, and L. G. Holness. 2006. Transfusion- dependent and independent binding that is in，uencedby endotoxin concen- transmitted Klebsiella pneumoniae fatalities, 1995 to 2004. Transfus. Med. tration. Eur. J. Biochem. 267:2218–2226. 36. Versalovic, J., T. Koeuth, and J. R. Lupski. 1991. Distribution of repetitive Rev. 20:149–157. 22. Nussbaumer, W., D. Allersdorfer, C. Grabmer, M. Rheinschmidt, L. Lin, D. DNA sequences in eubacteria and application to ？ngerprinting of bacterial Schonitzer, and C. Lass-Florl. 2007. Prevention of transfusion of platelet genomes. Nucleic Acids Res. 19:6823–6831. 37. Walther-Wenke, G. 2006. Bacterial contamination of blood components— components contaminated with low levels of bacteria: a comparison of bac- teria culture and pathogen inactivation methods. Transfusion 47:1125–1133. incidence and signi？cance for homologous and autologous transfusion. 23. Ortolano, G. A., L. F. Freundlich, S. Holme, R. L. Russell, M. A. Cortus, K. J. Lab Med. 30:66–73. Wilkins, H. Nomura, C. Chong, R. Carmen, A. Capetandes, and B. Wenz. 38. Yazer, M. H., D. Stapor, and D. J. Triulzi. 2010. Use of the RQI test for 2003. Detection of bacteria in WBC-reduced PLT concentrates using per- bacterial screening of whole blood platelets. Am. J. Clin. Pathol. 133:564– cent oxygen as a marker for bacteria growth. Transfusion 43:1276–1285. 568.