Detection of Extended-Spectrum β-Lactamase and Klebsiella pneumoniae Carbapenemase Genes Directly from Blood Cultures by Use of a Nucleic Acid Microarray

Abstract

The growing crisis of multidrug-resistant (MDR) Gram-negative bacteria requires that current technologies permit the rapid detection of extended-spectrum β-lactamase (bla_ESBL) and Klebsiella pneumoniae carbapenemase (bla_KPC) genes. In the present study, we assessed the performance characteristics of a commercially available nucleic acid microarray system for the detection of bla_ESBL and bla_KPC genes directly from positive blood cultures. Using blood cultures (BCs) that contained Gram-negative bacilli identified by Gram staining, we isolated bacterial DNA using spin columns (BC-C) and rapid water lysis (BC-W). Twenty ESBL/KPC-positive and 20 ESBL/KPC-negative blood culture samples, as well as 20 non-lactose-fermenting organisms, were tested. The 20 isolates that were ESBL positive by phenotypic testing were also evaluated on solid medium (SM), and the DNA was extracted by use of a spin column (SM-C). The resulting 140 DNA extractions were assessed for DNA quantity and quality using 260/280-nm absorbance ratios, and DNA microarray analysis was performed in a blinded fashion. Microarray and phenotypic results were concordant for 98.3% of BC-W, 90% of BC-C, and 95% of SM-C samples. Compared to phenotypic testing, the sensitivity and specificity for BC-C samples were 88.9% and 100%, respectively, and for BC-W samples, the sensitivity and specificity were 94.4% and 100%, respectively. BC-W samples yielded the highest concordance with phenotypic results. Nucleic acid microarrays offer promise in the identification of bla_ESBL and bla_KPC genes directly from blood cultures, thereby reducing the time to identification of these important pathogens.

INTRODUCTION

The prevalence of extended-spectrum β-lactamase (ESBL)- and Klebsiella pneumoniae carbapenemase (KPC)-producing Gram-negative bacilli is increasing worldwide, with the attendant loss of many previously effective antimicrobial therapeutic agents (1, 5). As a result, appropriate empirical antibiotic therapy becomes even more necessary with regard to KPC-producing Klebsiella pneumoniae (13). Mortality has been shown to be markedly higher when inadequate or ineffective antimicrobial therapy is administered as empirical therapy (or delayed by 72 h) pending the results of phenotype testing (7, 10, 13, 16). The application of rapid molecular diagnostic techniques to the identification and characterization of pathogens promises to improve outcomes as a result of more timely and effective antibiotic therapy.

Important advances have recently come from the use of mass spectrometry (MS) in the clinical microbiology laboratory (11, 12, 15). Rapid and inexpensive identification of pathogens to the genus and species levels using matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) mass spectrometry is becoming more commonly used and is rapidly advancing in scope and applications (14). Using a hydrolysis reaction to identify β-lactamases and carbapenemases, MALDI-TOF is able to accurately report the presence of carbapenemases. PCR followed by electrospray ionization mass spectrometry (PCR/ESI-MS) quickly identifies Gram-negative bacteria and can target identification of common resistance genes such as gyrA, parC, and bla_KPC with >96% sensitivity and specificity (4, 6). Both of these applications are awaiting large-scale trials in the clinical arena.

Among methods that are currently available, nucleic acid microarray technologies are proving to be important in the characterization of resistance genes and molecular epidemiology (2, 3, 8, 9). The clinical experience, to date, shows that this technology correctly identifies the presence of resistance genes with 90 to 100% accuracy when applied directly from single bacterial colonies grown on agar plates. Here we tested the nucleic acid microarray technology to identify the presence of bla_ESBL and bla_KPC genes directly from blood cultures (BCs) positive for Gram-negative bacilli. In order to reduce the total testing time required for reporting of these results (shortening the notification time to providers), we also evaluated two DNA isolation techniques and the impact of DNA quality and quantity on the accuracy of the results.

MATERIALS AND METHODS

Patient samples.

Clinical samples for this feasibility study were obtained from St. John Hospital and Medical Center (St. John Providence Health System), an 800-bed community teaching hospital in Detroit, MI. Blood samples for culture sent to the core microbiology lab were prospectively monitored for the presence of Gram-negative bacilli (Gram's stain) and growth of Gram-negative lactose- and non-lactose-fermenting organisms (MacConkey agar; BBL-BD, Sparks, MD). Identification to the species level and antimicrobial susceptibility testing were performed with the Vitek2 system (bioMérieux, Durham, NC). When the Vitek2 ESBL and/or KPC screen was positive, results were confirmed in accordance with the Clinical and Laboratory Standards Institute (CLSI) M100-S21 method (1a), including a 10-disk test and modified Hodge test, when indicated.

Sixty unique blood culture samples containing Gram-negative bacteria were used for this project. Twenty Gram-negative bacteria were ESBL and/or modified Hodge test (KPC) positive, 20 isolates were ESBL and KPC negative (phenotype negative controls), and 20 isolates were non-lactose fermenters (clinically relevant negative controls). Blood culture samples obtained from patients known to be hepatitis or HIV positive were excluded. This protocol was approved by the St. John Hospital and Medical Center Institutional Review Board (IRB).

Blood culture processing.

Aliquots of 0.5 ml each were withdrawn from BC bottles (BacT/Alert; bioMérieux) and separated by differential centrifugation at 850 × g for 2 min to remove erythrocytes and charcoal, followed by 11,500 × g for 5 min to pellet bacteria. The aliquot samples were taken and processed within 24 h when a positive blood culture was reported by the clinical laboratory to contain Gram-negative bacilli. Two separate aliquots were obtained for each of the 60 blood culture samples. Aliquots were stored at 4°C, and DNA extractions (see below) were performed in batches 1 to 3 times per week. Only a single blood culture bottle was used for each individual patient's organism.

One set of the 60 BC samples underwent DNA extraction by rapid water lysis (BC-W), and the other set underwent DNA extraction using a commercial spin column (BC-C) (QIAamp DNA minikit; Qiagen, Valencia, CA). The 20 known ESBL/KPC-positive blood cultures were streaked on solid medium (SM), and isolated colonies were processed using the spin column DNA extraction method to serve as a reference standard (SM-C). In total, 140 DNA samples were available for microarray analysis.

DNA extraction and assessment. (i) BC-W method.

The bacterial pellet was suspended in 200 μl sterile distilled water (modified from reference 11) and heated to 100°C for 10 min. Debris was removed by centrifugation at 850 × g for 2 min, and the remaining supernatant served as sample DNA.

(ii) BC-C method.

Blood culture-positive samples underwent DNA extraction after differential centrifugation (described above) following the Qiagen manufacturer instructions.

(iii) SM-C method.

Growth of phenotypically ESBL/KPC-positive organisms on solid medium was also extracted using QIAamp DNA minikits, and these samples served as a control group.

(iv) Microarray testing and discrepancy resolution.

DNA quantity and quality were measured for all 140 samples by spectrophotometry (260 nm/280 nm ratio) according to standard procedures. Data on DNA quantity and quality were analyzed using analysis of variance, followed by multiple pairwise comparisons using the Bonferroni correction of the P value.

The quantities of the DNA samples were adjusted (if needed) so that they equaled a range of from 25 to 150 ng/μl for all extractions. Samples were then frozen and analyzed (140 in total) in a blinded fashion using DNA microarray testing (Check-KPC/ESBL kits; Check-Points, Inc., The Netherlands) according to the manufacturer's instructions. Samples initially yielding noninterpretable results (presence of suspected inhibitors) were retested at additional DNA dilutions (1:5, 1:10, or 1:20) to a final concentration at the lower limits of the Check-Points-recommended range (blinding was maintained through these steps).

Upon completion of microarray sample testing, the results were unblinded. Discrepant results among BC-C, BC-W, SM-C, and phenotypic tests were analyzed by repeat microarray testing, PCR, and/or DNA sequencing as needed for discrepancy resolution. The results obtained after removal of blinding were not used to recalculate concordance rates.

Results were analyzed and reported on the basis of phenotypic concordance. The presence of any of the bla_ESBL or bla_KPC genes when the pathogen was phenotypically an ESBL/KPC producer was considered a concordant result. If the phenotypic result would have matched the microarray report, despite missing a mutation (when multiple mutations were present), the results were considered concordant. When microarray analysis failed to identify any bla_ESBL or bla_KPC genes and phenotype results were negative, data were regarded as concordant. Results were then evaluated to assess if the microarray failed to consistently identify all of the bla_ESBL or bla_KPC genes for any particular extraction method or sample (see Table 1).

Table 1.

Nucleic acid microarray results

Organism	Phenotype	bla gene detectiond
		BC-C	BC-W	SM-C
A. baumannii	−	TEM-1	TEM-1	ND
A. baumannii	−	−	−	ND
A. baumannii	−	TEM-1	TEM-1	ND
A. baumannii	−	−	−	ND
A. baumannii	−	−	−	ND
E. coli	−	NE	−	ND
E. coli	−	NE	−	ND
E. coli	−	−	−	ND
E. coli	ESBL	CTX-M-1	CTX-M-1	KPC, CTX-M-1a,b
E. coli	−	TEM-1	TEM-1	ND
E. coli	−	TEM-1	TEM-1	ND
E. coli	ESBL	TEM-1b	CTX-M-1b	SHV-5, TEM-1
E. coli	ESBL	CTX-M-1	CTX-M-1	CTX-M-1
E. coli	−	−	−	ND
E. coli	−	TEM-1	TEM-1	ND
E. coli	ESBL	CTX-M-1	CTX-M-1	CTX-M-1
E. coli	−	−	−	ND
E. coli	ESBL	CTX-M-1	CTX-M-1	CTX-M-1
E. coli	ESBL	CTX-M-1	−b	CTX-M-1
E. coli	−	TEM-1	TEM-1	ND
E. coli	ESBL	CTX-M-1	CTX-M-1	CTX-M-1
E. coli	−	−	−	ND
E. coli	−	−	−	ND
E. coli	−	TEM-1	TEM-1	ND
E. coli	−	−	−	ND
E. cloacae	−	−	−	−
E. cloacae	−	−	−	ND
E. cloacae	−	−	−	ND
K. pneumoniae	−	SHV-1	SHV-1	ND
K. pneumoniae	ESBL	CTX-M-1a	CTX-M-1, SHV-5	CTX-M-1, SHV-5
K. pneumoniae	ESBL and KPC	KPC, SHV-1, SHV-5, TEM-1	KPC, SHV-1, SHV-5, TEM-1	KPC, SHV-1, SHV-5, TEM-1
K. pneumoniae	−	SHV-1	SHV-1	ND
K. pneumoniae	−	NE	SHV-1	ND
K. pneumoniae	−	−	SHV-1	ND
K. pneumoniae	−	−	SHV-1	ND
K. pneumoniae	−	−	−	ND
K. pneumoniae	−	−	−	ND
K. pneumoniae	ESBL and KPC	KPC, SHV-1, E240K, TEM-1c	KPC, SHV-1, SHV-5, TEM-1	KPC, SHV-1, E240K, TEM-1c
K. pneumoniae	ESBL and KPC	KPC, SHV-1, SHV-5, TEM-1	KPC, SHV-1, SHV-5, TEM-1	KPC, SHV-1, SHV-5, TEM-1
K. pneumoniae	KPC	KPC, SHV-1	KPC, SHV-1	KPC, SHV-1
K. pneumoniae	ESBL and KPC	KPC, CTX-M-1, SHV-1, SHV-5, TEM-1	KPC, CTX-M-1, SHV-1, SHV-5, TEM-1	KPC, CTX-M-1, SHV-1, TEM-1c
K. pneumoniae	−	SHV-1	SHV-1	ND
K. pneumoniae	KPC	KPC, SHV-1, TEM-1	KPC, SHV-1, TEM-1	KPC, SHV-1, TEM-1
K. pneumoniae	ESBL and KPC	KPC, CTX-M-1, SHV-1, TEM-1	KPC, CTX-M-1, SHV-1, TEM-1	KPC, CTX-M-1, SHV-1, TEM-1
K. pneumoniae	ESBL and KPC	KPC, SHV-1, SHV-2, TEM-1a	KPC, SHV-1, E240K, TEM-1a	KPC, SHV-1, SHV-5, TEM-1
K. pneumoniae	ESBL and KPC	NE	KPC, SHV-1, SHV-5, TEM-1	KPC, SHV-1, SHV-5, TEM-1
K. pneumoniae	ESBL	SHV-1, SHV-5	SHV-1, SHV-5	SHV-1, SHV-5
K. pneumoniae	ESBL	TEM-1a	SHV-1, SHV-5, TEM-1	SHV-1, SHV-5, TEM-1
M. morganii	−	−	−	ND
P. mirabilis	−	−	−	ND
P. stuartii	−	−	−	ND
P. aeruginosa	−	−	−	ND
P. aeruginosa	−	−	−	ND
P. aeruginosa	−	−	−	ND
P. aeruginosa	−	−	−	ND
P. aeruginosa	−	−	−	ND
Salmonella	−	−	−	ND
Salmonella	−	−	−	ND
S. marcescens	−	−	−	ND
S. rubidaea	−	−	−	ND

^aDiscrepancy resolved by repeat microarray test.

^bDiscrepancy resolved by PCR.

^cDiscrepancy resolved by sequencing.

^dThe designations SHV-2 and SHV-5 indicate the bla genes containing the G238S mutation and the G238S and E240K mutations, respectively. KPC and CTX-M-1 represent the family of bla genes. TEM-1 and SHV-1 indicate no mutations found relative to these sequences for the sites assayed by the microarray. ESBL, extended spectrum β-lactamase; NE, not evaluable; ND, test not done.

RESULTS AND DISCUSSION

The characterization of the 60 identified pathogens used for this study is presented in Table 1. Forty-one ESBL- and KPC-negative and 19 ESBL- and KPC-positive samples were used for final analysis. Phenotypic ESBL-positive isolates included Escherichia coli (n = 7) and K. pneumoniae (n = 3). Nine K. pneumoniae isolates tested positive for the presence of an ESBL and carbapenemase by modified Hodge testing. ESBL-negative isolates included E. coli (n = 13), K. pneumoniae (n = 8), Enterobacter spp. (n = 3), and Serratia rubidaea (n = 1). Other ESBL-negative pathogens included Acinetobacter baumannii (n = 5), Pseudomonas aeruginosa (n = 5), Salmonella spp. (n = 2), Proteus mirabilis (n = 1), Serratia marcescens (n = 1), Providencia stuartii (n = 1), and Morganella morganii (n = 1). As shown below, the pathogens that tested positive by the modified Hodge test were all K. pneumoniae containing bla_KPCs (n = 9). One sample of Enterobacter cloacae was misidentified by the clinical microbiology laboratory as ESBL positive. This error was discovered only upon completion of the study.

The Check-KPC/ESBL kit performed very well in identifying ESBL- and KPC-producing pathogens from positive blood cultures. The overall concordance rate was 95% (133/140 samples) (Table 1). The concordance rates varied by extraction method and were 90% (54/60) for BC-C, 98.3% (59/60) for BC-W, and 95% (19/20) for SM-C. On the basis of a comparison to the identification of ESBLs by phenotypic testing, the blood culture column extraction method, BC-C, demonstrated a sensitivity of 88.9% and a specificity of 100%. Among the 140 samples that were analyzed, 4 samples of DNA (2 E. coli and 2 K. pneumoniae isolates) extracted by the BC-C method failed to produce the necessary quality control spots on microarray testing, and therefore, the results were considered not interpretable. Similarly, the BC-W method performed with a sensitivity of 94.4% and a specificity of 100%. The BC-W and BC-C methods each had a positive predictive value of 100%.

Only three false-negative results were reported in this study. All 3 samples should have been reported to be ESBL positive but were not (Table 1). The microarray failed to detect a bla_SHV-5-like gene in one sample with E. coli DNA (both the BC-C and BC-W samples) and one sample with K. pneumoniae DNA extracted by the BC-C method. This discrepancy was resolved either by repeat microarray testing or by a positive PCR amplification of the SHV gene in these samples. The microarray for a different E. coli isolate failed to initially detect the bla_CTX-M-1-like gene in the BC-W sample. The gene was detected, however, with PCR testing. One false-positive result (KPC) was noted in an E. coli SM-C sample. Repeat microarray testing and a lack of standard PCR amplification of bla_KPC led to discrepancy resolution. All non-lactose-fermenting Gram-negative bacilli tested negative by microarray analysis.

The DNA quantity from BC-W samples was significantly greater than that from BC-C or SM-C samples (417 ± 251.6 ng/μl versus 64.8 ± 24.6 ng/μl or 58.2 ± 21.5 ng/μl, respectively; P < 0.0001). The DNA quality ratios (260/280 nm) also differed significantly. The DNA quality ratio for SM-C was the highest at 1.802 ± 0.330 and was significantly different from that for BC-W (1.362 ± 0.151; P < 0.0001) or BC-C (1.444 ± 0.382; P < 0.0001).

Samples extracted from blood culture often showed interference from inhibitors with microarray testing and required additional dilutions for resolution (prior to unblinding). Dilutions were successful in reducing inhibitors for all samples extracted by the BC-W method, which had higher concentrations of DNA, but remained a problem in 6.7% of the extractions obtained by the BC-C method.

Concluding remarks.

The rising incidence of ESBL- and KPC-producing pathogens presents challenges in the empirical therapy for Gram-negative bacterial infections. Acknowledging the increased mortality rates with ineffective initial therapy (13, 16), the timely recognition of ESBL- and KPC-producing organisms may have a positive impact on clinical outcomes. Nucleic acid microarrays can be important in achieving this end.

In this study, the nucleic acid microarray technology (Check-KPC/ESBL kits) demonstrated accurate and reproducible results with regard to the presence of ESBL and KPC genes within the Enterobacteriaceae (2, 3, 8, 9). Current practice requires 24 h of adequate growth prior to testing on the microarray system. Our study shows that the Check-KPC/ESBL kit could be employed with DNA extracted directly from blood cultures before species identification. This could reduce the notification time by as much as 18 to 20 h. Depending on the method used, the time savings, costs of materials, and technician time can be further reduced. The application of the water lysis process for DNA extraction required about an hour, whereas the spin column method required about 4 h of technician time. Reduced time to reporting of accurate and valid resistance patterns will be paramount as the number and density of drug-resistant pathogens increase in health care.

We recognize that there are limitations to this study. Sixty samples in all were tested in this feasibility study that was designed to demonstrate proof of concept. In addition, we included pathogens not normally used for ESBL/KPC testing (the non-lactose fermenters). We felt that the most commonly encountered, clinically relevant, Gram-negative pathogens should be included in this feasibility study. The non-lactose fermenters were therefore included and served as negative clinical controls.

The collection examined did not possess a full array of Gram-negative bacteria harboring other important bla_ESBL or carbapenemase genes (bla_NDM, bla_VIM, bla_IMP, etc.). We also did not test for AmpC genes because the microarray kits that allow AmpC testing were made available only after completion of our study. In addition, the lower limits of detection were not assessed, and the performance characteristics with regard to mixed or polymicrobial cultures potentially containing aerobic Gram-negative bacilli, anaerobes, fungi, or staphylococci are unknown. Larger prospective studies are required to further define the performance characteristics of this technology when it is applied to blood culture isolates and study the impact on clinical outcomes when antibiotic modification occurs earlier on the basis of provider notification of the presence of resistance genes. Standardization of the laboratory techniques, used with a focus on simple and cost-effective methods, will be necessary as this technology moves into the clinical laboratory for such large studies. Nevertheless, this investigation is an important first step in the expansion of critically needed rapid diagnostic testing to this clinical arena.