Preliminary clinical study using a multiplex real-time PCR test for the detection of bacterial and fungal DNA directly in blood

Transcription

1 ORIGINAL ARTICLE Preliminary clinical study using a multiplex real-time PCR test for the detection of bacterial and fungal DNA directly in blood F. Wallet 1, S. Nseir 2, L. Baumann 1, S. Herwegh 1, B. Sendid 1, M. Boulo 2, M. Roussel-Delvallez 1, A. V. Durocher 2 and R. J. Courcol 1 1) Pôle de Microbiologie and 2) Service des Urgences Réanimation Respiratoire Médicale, Centre Hospitalier Régional Universitaire de Lille, Faculté de Médecine de Lille, Lille, France Abstract Early diagnosis of sepsis, rapid identification of the causative pathogen(s) and prompt initiation of appropriate antibiotic treatment have a combined impact on mortality due to sepsis. In this observational study, a new DNA-based system (LightCycler SeptiFast (LC-SF) test; Roche Diagnostics) allowing detection of 16 pathogens at the species level and four groups of pathogens at the genus level has been evaluated and compared with conventional blood cultures (BCs). One hundred BC and LC-SF results were obtained for 72 patients admitted to the intensive-care unit over a 6-month period for suspected sepsis. Microbiological data were compared with other biological parameters and with clinical data. The positivity rate of BCs for bacteraemia/fungaemia was 10%, whereas the LC-SF test allowed detection of DNA in 15% of cases. The LC-SF performance, based on its clinical relevance, was as follows: sensitivity, 78%; specificity, 99%; positive predictive value, 93%; and negative predictive value, 95%. Management was changed for four of eight (50%) of the patients because organisms were detected by the LC-SF test but not by BC. LC-SF results were obtained in 7 15 h, in contrast to the h required for BC. According to the LC-SF results, initial therapy was inadequate in eight patients, and antibiotic treatment was changed. Our results suggest that the LC-SF test may be a valuable complementary tool in the management of patients with clinically suspected sepsis. Keywords: Bacteraemia, blood culture, fungaemia, real-time PCR, SeptiFast Original Submission: 28 July 2008; Revised Submission: 11 February 2009; Accepted: 12 February 2009 Editor: J. L. Mainardi Article published online: 18 August 2009 Clin Microbiol Infect 2010; 16: /j x Corresponding author and reprint requests: R. J. Courcol, Pôle de Microbiologie, Centre Hospitalier Universitaire de Lille, F Lille Cedex, France rcourcol@chru-lille.fr Introduction Rapid detection of bacteraemia and fungaemia is one of the most important functions of the bacteriology laboratory. Microbiological documentation is of paramount importance for the success of the therapeutic strategy during sepsis, as well as for the adequacy of the treatment (delay of <24 48 h) [1]. Microbial documentation has a great impact on mortality and morbidity [2,3]. For instance, rapid and substantiated management of sepsis is recommended for infections due to Pseudomonas aeruginosa [4] and Staphylococcus aureus [5]. Results from traditional blood culture (BC) (and other specimens) are usually not available until h after initial patient presentation or suspected new episodes of infection, necessitating initial use of empirical therapy. During the first 24 h of suspected sepsis, initial antibiotic therapy is reported to be inappropriate in 10 60% of cases (i.e. one or more detected pathogens not covered by, or resistant to, the chosen antibiotics) [2,6]. Inappropriate antibiotic therapy is associated with mortality ranging from 10 45% [1], with emerging antibiotic resistance [2] in intensive-care units (ICUs), and with longer hospital length of stay [6,7]. Targeting antibiotic therapy to causative pathogens is pivotal in reducing resistance; however, greater accuracy of appropriate antibiotic selection requires more rapid and sensitive diagnostic tools for the identification of potential pathogens. The advent of non-culture-based identification of blood-borne pathogens by the detection and identification of bacterial and fungal DNA directly in the blood of patients with suspected early sepsis has the potential to allow Journal Compilation ª2009 European Society of Clinical Microbiology and Infectious Diseases

2 CMI Wallet et al. Detection of bacteraemia and fungaemia with multiplex real-time PCR 775 targeted antibiotic therapy. This may lead to drug cost savings, shorter duration of therapy, shorter hospital length of stay, and reduction in overall resistance rates. A few years ago, a method consisting of a DNA probe matrix directed towards rrna gene targets utilizing seven probe cocktails was proposed [8]. Only positive BC samples were subjected to this technique, giving results within 1 h with very good sensitivity (100%) and specificity (96%). More recently, some studies have reported the use of real-time PCR to detect bacteria either from BC [9] or directly from patient blood [10 13]. The LightCycler SeptiFast (LC-SF) test (Roche Diagnostics GmbH, Mannheim, Germany) was developed to detect directly from blood the 20 most important bacterial and fungal genera or species (Table 1), causing 90% of all bloodstream infections in patients admitted to ICUs [2,7,14]. The present article reports a prospective observational study performed in patients admitted to the ICU with suspected bloodstream infection, using this multiplex real-time PCR kit in comparison with conventional BC in the daily routine. Materials and Methods Study patients and setting This prospective observational cohort study was performed in a 30-bed ICU during a 6-month period. Because it was observational, Institutional Review Board approval was not required according to French law. All adult patients with fever ( 38 C) or hypothermia ( 36 C) were eligible. Clinical data were prospectively collected at ICU admission and throughout the hospitalization period. The patient status was defined according to SIRS criteria [15,16]. Antimicrobial prescription was prospectively recorded on the day of BC sampling. Other microbiological specimens were collected for testing as clinically indicated. BC A single venipuncture was used to collect blood (10 ml for each BC). BC consisted of the following: (i) a pair of bottles (aerobic and anaerobic) incubated for 5 days at 37 C and analysed for bacteria using BacT/Alert (biomérieux, Marcy l Etoile, France); and (ii) one bottle with Mycosis medium incubated for 14 days at 37 C and analysed for fungi with BACTEC 9200 (BD, Pont-de-Claix, France). From the same venipuncture, 5 ml of EDTA whole blood was collected for the LC-SF test. Microorganisms isolated from BC were identified according to standard laboratory methods and good laboratory practice. The LC-SF test The LC-SF test is based on three major processes: (i) specimen preparation with mechanical lysis and purification of DNA; (ii) real-time PCR amplification of target DNA in three parallel reactions (Gram-positive bacteria, Gram-negative bacteria, and fungi) and detection with specific hybridization probes; and (iii) automated identification of species and controls [17]. Target species detected using the LC-SF test are shown in Table 1. The LC-SF test uses proprietary broad-range PCR primers to amplify unique portions of the bacterial or fungal internal transcribed sequences for a particular bacterial or fungal species. A defined volume containing an internal control (IC) is introduced into each specimen along with the lysis solution. The IC contains unique probe-binding regions that allow differentiation of the amplified IC from the target. The volume of whole blood for DNA preparation was 1.5 ml. The specimens were prepared and tested as recommended by the manufacturer [17]. Real-time detection of PCR products was performed using the LightCycler 2.0 instrument. Melting temperatures of specimens and controls were analysed using the LC-SF identification software, and a report was generated. Patient data were valid if the corresponding controls (kit reagent controls and ICs) were positive. The LC-SF test was performed with fresh blood samples that were processed individually or in batches of two or three samples. Instruments and disposables were provided by Roche Molecular Diagnostics (Roche Diagnostics, Meylan, France). All disposables and reagents used within the workflow were labelled M GRADE (DNA-free) and IVD/CE. TABLE 1. Target species of the LightCycler SeptiFast test Gram-negative Gram-positive Fungi Escherichia coli Staphylococcus aureus Candida albicans Klebsiella (pneumoniae/oxytoca) Coagulase-negative staphylococci (Staphylococcus epidermidis, Staphylococcus haemolyticus) Candida tropicalis Candida parapsilosis Serratia marcescens Streptococcus pneumoniae Candida krusei Proteus mirabilis Streptococcus spp. (Streptococcus pyogenes, Candida glabrata Enterobacter (cloacae/aerogenes) Streptococcus agalactiae, Streptococcus mitis) Aspergillus fumigatus Enterococcus faecium Pseudomonas aeruginosa Enterococcus faecalis Acinetobacter baumannii Stenotrophomonas maltophilia

3 776 Clinical Microbiology and Infection, Volume 16 Number 6, June 2010 CMI Statistical methods SPSS software (SPSS for Windows, version 14.1; SPSS, Chicago, IL, USA) was used for data analysis. The Mann Whitney U-test was used to compare demographic data between subjects with positive BC results and subjects with negative BC results. Results of the LC-SF test and BC are presented as positivity rates and by isolate. Results From the 72 patients, a total of 100 sample sets of consecutive BC sets (aerobic, anaerobic and fungi) and LC-SF samples were examined. A total of 14 patients had two sets of blood samples taken, and seven patients had three sets of blood samples taken. No statistically significant differences in demographic data or cause for ICU admission were found between patients with negative BC results and those with positive BC results (p >0.05). The most frequent suspected site of infection was the respiratory tract: 66% (59/90) of samples with negative BC results, and 70% (7/10) of samples with positive BC results. The time from blood draw to delivery of the LC-SF results to the clinician ranged from 7 to 15 h. Table 2 shows the 102 results from the 100 paired samples. In six patients, BCs were positive alone: Escherichia coli was previously isolated from a bursitis; Leuconostoc spp., not included in the panel, were considered by physicians as translocated intestinal bacteria; Staphylococcus epidermidis was considered to be a contaminant from blood drawing. No explanation was found for the recovery of Enterobacter (cloacae/aerogenes). The results of positive cases as determined with the LC-SF test are reported in Table 3. Two yeasts were detected only in BCs but not by the LC-SF test: Candida glabrata/s. aureus (patient 4) and Candida albicans/candida tropicalis (patient 9). In the first case, C. glabrata was recovered from the throat and urine 5 days before the LC-SF test. For the second case, C. albicans was previously isolated from bronchial secretions 15 days before the LC-SF test. In comparing LC-SF results with those from BC, the parameters of analytical performance of the test were as follows: sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were 40%, 88%, 27%, and 93%, respectively. In eight patients, organisms were detected by the LC-SF test but BC results were negative (Table 3). On the basis of LC-SF results and microbiological data, antibiotic management was changed in three patients (patients 2, 3 and 6). In two of these patients (patients 3 and 6), initial therapy was considered to be inadequate. Patient 2, without antibiotics, received antibiotics on day 6 because of the positivity of the LC-SF result (E. coli). In patient 5, although the antibiotic management was considered to be adequate, a catheter was withdrawn following the LC-SF test because the physician considered the catheter to be the focus of infection. Four patients (patients 1, 4, 7 and 8) with positive LC-SF results were already receiving appropriate treatment at the time of the LC-SF test, and no change was made to their antibiotic regimen. Table 3 shows that initial therapy was inadequate in four patients, and management was changed as shown for these patients. From the 11 discrepant cases (Table 3; nine LC-SF-positive/ BC-negative, and two LC-SF-positive/BC-discrepant), ten were LightCycler SeptiFast test result Blood culture result Positive Negative TABLE 2. Test results for all paired blood samples a Positive Serratia marcescens n =3 Escherichia coli n =2 Staphylococcus aureus n =1 Klebsiella pneumoniae n =3 S. aureus n =4 b Pseudomonas aeruginosa n =1 Candida tropicalis n =1 c Total = 4 Total = 11 Negative Escherichia coli n =1 n =81 Enterobacter cloacae n =1 Leuconostoc spp. n =1 d Staphylococcus epidermidis n =1 d Candida glabrata n =1 b Candida albicans n =1 c Total = 6 Total = 81 a One hundred and two microorganisms detected from 100 paired blood cultures and the LightCycler SeptiFast test. b Organisms were detected from the same blood sample (applies to only one of the four blood culture-negative S. aureus results). c Organisms were detected from the same blood sample. d Organisms isolated only by blood culture and not considered to be clinically relevant by the clinician: one case of contamination (S. epidermidis) and one case of possible intestinal translocation (Leuconostoc spp.).

4 CMI Wallet et al. Detection of bacteraemia and fungaemia with multiplex real-time PCR 777 TABLE 3. Description and management of positive LightCycler SeptiFast (LC-SF) cases Patient Bacteria detected with the LC-SF test Bacteria isolated from blood culture Other positive specimen Temperature Cause of ICU admission Leukocytes ( 10 9 /L) CRP (mg/l) PCT (mg/l) Antimicrobial agents at time of blood draw Adequate initial treatment Change in management at time of LC-SF result 1 Escherichia coli Negative NA 34.8 Shock Piperacillin tazobactam + amikacin Yes No change 2 E. coli Negative E. coli in sputum (D ) 6) 38.7 ARDS No No Piperacillin tazobactam amikacin 3 Klebsiella spp. Negative Klebsiella oxytoca in trachea (D ) 0) 38.3 Shock NA Piperacillin tazobactam + amikacin Yes NA Klebsiella spp. Negative Klebsiella oxytoca in trachea (D ) 0) Piperacillin tazobactam No Amikacin re-introduced 4 Staphylococcus aureus Negative S. aureus in blood culture (D ) 3) 39.7 Pneumonia + shock Oxacillin + levofloxacin Yes No change S. aureus Candida glabrata S. aureus in blood culture (D ) 3), C. glabrata in urine (D ) 5) 39.6 Pneumonia + shock Oxacillin + levofloxacin Yes No change 5 S. aureus Negative NA 38.6 Pneumonia + shock Piperacillin tazobactam + teicoplanin + ciprofloxacin 6 S. aureus Negative NA 35.5 Shock Ceftriaxone + gentamicin + metronidazole Yes Catheter withdrawn No Linezolid added 7 Klebsiella spp. Negative Klebsiella oxytoca in urine (D ) 5) 35.8 Pneumonia Piperacillin tazobactam Yes None 8 Pseudomonas aeruginosa Negative P. aeruginosa in trachea (D ) 0) 38.6 ARDS + CF Ceftazidime + amikacin Yes None 9 Candida tropicalis Candida albicans Candida albicans in trachea (D ) 15) 39 Pancolite + shock Vancomycin + metronidazole No Patient died 10 Serratia marcescens Serratia marcescens Catheter: Serratia marcescens (D ) 0) 38.1 Shock + ARDS No No Catheter withdrawn; initiation of imipenem 11 Serratia marcescens Serratia marcescens BAL: Serratia marcescens (D ) 10) 38.4 Pneumonia Fluconazole (no antibiotic) No Initiation of piperacillin tazobactam + amikacin 12 Serratia marcescens Serratia marcescens Trachea: Serratia marcescens (D ) 7) 38.6 ARDS + COPD Amoxycillin + clavulanic acid No Imipenem + ciprofloxacin 13 S. aureus S. aureus S. aureus in trachea (D 2), S. aureus in blood culture (D + 1) 39 Pancreatitis No antibiotic No Initiation of vancomycin gentamicin ARDS, acute respiratory distress syndrome; BAL, bronchoalveolar lavage; CF, cystic fibrosis; COPD, chronic obstructive pulmonary disease; CRP, C-reactive protein; D, day; ICU, intensive-care unit; NA, results not available; PCT, procalcitonin.

5 778 Clinical Microbiology and Infection, Volume 16 Number 6, June 2010 CMI considered to be true-positive results, given the clinical findings and other microbiological results. In addition, two samples with negative LC-SF results but positive BC results were ultimately considered to be not clinically meaningful (Table 2). Thus, the LC-SF performance was adjusted as follows: sensitivity, 78%; specificity, 99%; PPV, 93%; and NPV, 95%. Discussion The LC-SF test is the first DNA-based test to detect microorganisms directly from blood without the need for prior incubation. Such a test has great potential to optimize the management of patients with suspected sepsis. In the present study, the rate of recovery from bacteraemia was apparently better with the LC-SF test than with conventional BC. However, these results raise three questions: (i) what is the clinical relevance of isolates detected only with the LC-SF test; (ii) does the DNAemia found by PCR reflect true infection; and (iii) what is the best design for such a comparison between BC and DNA-based tests? There are no data in the literature on the presence of circulating bacterial DNA in blood, even from patients with subclinical bacterial infections. In contrast, it is well known that the release and circulation of fungal DNA is variable or present in the blood at a very low concentration [18,19]. Sometimes, fungal DNA is detected just within the lower limit of conventional PCR assays [20]. Moreover, the present study had some design limitations, mainly concerning the number of sets of BCs. Indeed, the sensitivity of BCs is increased by culturing greater volumes of blood, whereas the LC-SF test has been compared to a set of only three BCs, collected at the same time. However, the aim of such a molecular test is to reduce the number of samples while trying to achieve the best possible sensitivity. Another aspect is the amplification of circulating DNA of non-viable bacteria and fungi, e.g. that within phagocytic cells or from bacteria inhibited by antimicrobials. PCR-based methods are more sensitive than BC. Indeed, in the 11 cases where bacteria were only detected with the LC-SF test, the results might be considered to be false positive. However, the association of clinical findings and other biological data in these cases suggested that the bacteria recovered were truly responsible for the patients status. Therefore, in the future, BC should no longer be the reference standard for the diagnosis of bloodstream infections. Among the publications on this topic, a recent study underlined that false-positive blood bottles were detected with automated systems, and suggested evaluation of the microbial DNA in these bottles [21]. It has been well established that the time to appropriate antibiotic therapy has a significant impact on mortality [22]. Similarly, it is also well documented that the rate of inappropriate antibiotic use varies and is a strong determinant of outcome [7]. The LC-SF test provides clinicians with the opportunity to fine-tune therapeutic management much earlier than has been previously possible. As this study was preliminary, the number of patients with positive BC and/or LC-SF results was small. However, in a number of cases, patient management was changed in response to the LC-SF results and in accordance with other available clinical information. In particular, a number of these patients had negative BC results at the time and were already receiving antibiotics. Patient management was changed in four of eight patients with positive LC-SF results and negative BC results, although these patients were already receiving antibiotics. In addition, for patients with both BC and LC-SF results positive for the same microorganism, the LC-SF results were available earlier, and patient management could be changed before the BC results were available. The clinical relevance of the LC-SF test was determined by the clinicians, as was also the case for the BC results. With the LC-SF test, inhibition of PCR was observed when leukocytosis was greater than /mm 3 (three cases). This phenomenon, also reported in another study, was probably due to competition between the relatively low level of bacterial DNA and the high level of human genomic DNA, affecting the assay sensitivity [10]. Although the LC-SF test gave a response more quickly than BC, the current version requires considerable handson technician time in the laboratory, except for the PCR step. Consequently, it is recommended that assays be performed in batch format. In the laboratory environment, care had to be taken to avoid contaminants, especially for the detection of Aspergillus fumigatus in blood. Only Mancini et al. [13] have detected A. fumigatus in a patient with the LC-SF test. However, during our study, no exogenous contaminant was recovered, owing to close attention to preparation. Many changes of clothing were made throughout the study. Laboratory surfaces and appliances were also decontaminated frequently. Moreover, PCR amplifications were carried out in the presence of uracil-n-glycosylase. Despite the high degree of precautions and a period of adaptation, laboratory implementation was easy. Assays were routinely performed with good practicability. The likelihood of contamination has significantly decreased, owing to technical improvements such as the use of automated real-time PCR detection. Our results are slightly better than those obtained by Louie et al. in terms of specificity (99% vs. 85%) and sensitiv-

6 CMI Wallet et al. Detection of bacteraemia and fungaemia with multiplex real-time PCR 779 ity (78% vs. 52%), but Lehmann et al., working with spiked blood, found the same specificity as that obtained in the present study [11,12]. However, in the study by Louie et al., cases of contamination were estimated at eight (S. epidermidis), instead of one as in ours. In conclusion, we are witnessing the development of a new paradigm in the field of infectious diseases. Physicians receiving a microbiological result within the day of sampling will be a step ahead in the treatment of infections. At the International Conference on Biodetection Technologies (June 2008), this view was supported by Ritter, who presented the next generation of automated multi-target detection platforms, which are capable of performing 120 discrete analyses simultaneously. The LC-SF test is the first labelled IVD-CE test for the detection of microorganisms that is focused on blood from patients admitted to ICUs, but other methods are currently under development to achieve the same goal [23,24]. The clinical significance of a result from a PCR-based test such as the LC-SF test must be interpreted with consideration of all of the available clinical and other laboratory data, and not alone. The results from this preliminary study suggest that the LC-SF test might be a valuable complementary tool in the optimal management of patients with clinically suspected sepsis. A larger, multicentre, randomized controlled study is in progress to further evaluate the clinical benefit and consequent health economic benefits of using the LC-SF test in addition to traditional culture-based methods of diagnosis of causal pathogens in patients with sepsis. Transparency Declaration Roche Molecular Diagnostics provided materials to perform the study. References 1. Kollef MH. Inadequate antimicrobial treatment: an important determinant of outcome for hospitalized patients. Clin Infect Dis 2000; 31: S131 S Harbarth S, Garbino J, Pugin J, Romand JA, Lew D, Pittet D. Inappropriate initial antimicrobial therapy and its effect on survival in a clinical trial of immunomodulating therapy for severe sepsis. Am J Med 2003; 115: Meehan TP, Fine MJ, Krumholz HM et al. Quality of care, process, and outcomes in elderly patients with pneumonia. JAMA 1997; 278: Kang CI, Kim SH, Kim HB et al. Pseudomonas aeruginosa bacteremia: risk factors for mortality and influence of delayed receipt of effective antimicrobial therapy on clinical outcome. Clin Infect Dis 2003; 37: Lodise TP, McKinnon PS, Swiderski L, Rybak MJ. Outcomes analysis of delayed antibiotic treatment for hospital-acquired Staphylococcus aureus bacteremia. Clin Infect Dis 2003; 36: Ibrahim EH, Sherman G, Ward S, Fraser VJ, Kollef MH. The influence of inadequate antimicrobial treatment of bloodstream infections on patient outcomes in the ICU setting. Chest 2000; 118: Paterson DL, Rice LB. Empirical antibiotic choice for the seriously ill patient: are minimization of selection of resistant organisms and maximization of individual outcome mutually exclusive? Clin Infect Dis 2003; 36: Marlowe EM, Hogan JJ, Hindler JF et al. Application of an rrna probe matrix for rapid identification of bacteria and fungi from routine blood culture. J Clin Microbiol 2003; 41: Thomas LC, Gidding HF, Ginn AN, Olma T, Iredelle J. Development of a real-time Staphylococcus aureus and MRSA (SAM -) PCR for routine blood culture. J Microbiol Methods 2006; 38: Jordan JA, Durso MB. Real-time polymerase chain reaction for detecting bacterial DNA directly from blood of neonates being evaluated for sepsis. J Mol Diagn 2005; 7: Lehmann LE, Hunfeld KP, Emrich T et al. A multiplex real-time PCR assay for rapid detection and differentiation of 25 bacterial and fungal pathogens from whole blood samples. Med Microbiol Immunol 2008; 197: Louie RF, Tang Z, Albertson TE et al. Multiplex polymerase chain reaction detection enhancement of bacteremia and fungemia. Crit Care Med 2008; 36: Mancini N, Clerici D, Diotti R et al. Molecular diagnosis of sepsis in neutropenic patients with haematological malignancies. J Med Microbiol 2008; 57: Böddinghaus B, Brandt C, Brodt HR et al. Evaluation of the PCR-M sepsis kit (Roche Diagnostics), a new real time PCR for the detection of bacterial and fungal DNA in blood. Infection 2005; 33 (suppl 1): Rangel-Frausto MS, Pittet D, Costigan M et al. The natural history of the systemic inflammatory response syndrome (SIRS). A prospective study. JAMA 1995; 273: Headley AS, Tolley E, Meduri GU. Infections and the inflammatory response in acute respiratory distress syndrome. Chest 1997; 111: LightCycler â SeptiFast Test M GRADE. For use with the LightCyclerÒ 2.0 instrument [package insert]. Mannheim, Germany: Roche Diagnostics GmbH, Yeo SF, Wong B. Current status of nonculture methods for diagnosis of invasive fungal infections. Clin Microbiol Rev 2002; 15: Klingspor L, Jalal S. Molecular detection and identification of Candida and Aspergillus spp. from clinical samples using real-time PCR. Clin Microbiol Infect 2006; 12: Quindos G. New microbiological techniques for the diagnosis of invasive mycoses caused by filamentous fungi. Clin Microbiol Infect 2006; 12: Karahan ZC, Mumcuoglu I, Guriz H et al. PCR evaluation of falsepositive signals from two automated blood-culture systems. J Med Microbiol 2006; 55: Kumar A, Roberts D, Wood K et al. Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock. Crit Care Med 2006; 34: Ecker DJ, Sampath R, Blyn LB et al. Rapid identification and straintyping of respiratory pathogens for epidemic surveillance. Proc Natl Acad Sci USA 2005; 102: Domann E, Hong G, Imirzalioglu C et al. Culture-independent identification of pathogenic bacteria and polymicrobial infections in the genitourinary tract of renal transplant recipients. J Clin Microbiol 2003; 41: