|Home | About | Journals | Submit | Contact Us | Français|
We describe the levels of agreement between broth microdilution, Etest, Vitek 2, Sensititre, and MicroScan methods to accurately define the meropenem MIC and categorical interpretation of susceptibility against carbapenemase-producing Klebsiella pneumoniae (KPC). A total of 46 clinical K. pneumoniae isolates with KPC genotypes, all modified Hodge test and blaKPC positive, collected from two hospitals in NY were included. Results obtained by each method were compared with those from broth microdilution (the reference method), and agreement was assessed based on MICs and Clinical Laboratory Standards Institute (CLSI) interpretative criteria using 2010 susceptibility breakpoints. Based on broth microdilution, 0%, 2.2%, and 97.8% of the KPC isolates were classified as susceptible, intermediate, and resistant to meropenem, respectively. Results from MicroScan demonstrated the most agreement with those from broth microdilution, with 95.6% agreement based on the MIC and 2.2% classified as minor errors, and no major or very major errors. Etest demonstrated 82.6% agreement with broth microdilution MICs, a very major error rate of 2.2%, and a minor error rate of 2.2%. Vitek 2 MIC agreement was 30.4%, with a 23.9% very major error rate and a 39.1% minor error rate. Sensititre demonstrated MIC agreement for 26.1% of isolates, with a 3% very major error rate and a 26.1% minor error rate. Application of FDA breakpoints had little effect on minor error rates but increased very major error rates to 58.7% for Vitek 2 and Sensititre. Meropenem MIC results and categorical interpretations for carbapenemase-producing K. pneumoniae differ by methodology. Confirmation of testing results is encouraged when an accurate MIC is required for antibiotic dosing optimization.
Carbapenems are considered first-line therapy for infection with multidrug-resistant Enterobacteriaceae (14). However, the increasing emergence of serine-based carbapenemase-producing Klebsiella pneumoniae (KPC) worldwide is of growing concern. This problem is particularly worrisome due to the fact that this K. pneumoniae is one of the leading causes of hospital-acquired infections in severely ill patients, and few antibiotics retain microbiological activity against isolates that produce blaKPC (15). Additionally, studies have demonstrated increased mortality rates in patients infected with carbapenem-resistant Enterobacteriaceae compared with those infected with susceptible strains (1, 12, 13).
Detection of KPC based strictly on susceptibility testing is challenging due mostly to the heterogeneous expression of β-lactam resistance (15). Many automated systems report KPC as susceptible to meropenem, and while some isolates truly are, the MICs for most KPC are above the Food and Drug Administration (FDA) susceptibility breakpoint (4 μg/ml) (11). To address testing and reporting issues, the Clinical Laboratory Standards Institute (CLSI) Subcommittee on Antimicrobial Susceptibility Testing changed the susceptibility breakpoint for meropenem, imipenem, and doripenem to ≤1 μg/ml against Enterobacteriaceae in January 2010 (8). At the time of writing, the FDA breakpoint remained at ≤4 μg/ml for meropenem.
Given the lack of options for antibiotics that retain susceptibility against pathogens that produce KPC, selection of a dosing regimen that could potentially treat infections caused by these organisms depends on the ability to accurately determine the antibiotic MIC. With respect to KPC, the accurate determination of the meropenem MIC may permit the application of pharmacodynamic principles to dosing regimen optimization by administering higher doses and using prolonged or continuous infusions, as has been accomplished against other resilient bacteria (3, 10, 14).
Herein, we describe the levels of agreement between commonly used testing methods (broth microdilution [BMD], Etest, Vitek 2, Sensititre, and MicroScan) in their abilities to accurately determine the meropenem MIC and further classify categorical susceptibilities of carbapenemase-producing K. pneumoniae isolates based on the 2010 CLSI breakpoints compared with FDA breakpoints.
A total of 46 clinical K. pneumoniae isolates, all modified Hodge test and blaKPC positive, collected from two hospitals in Manhattan, NY, were included. Standard methods were used to perform both the modified Hodge test and the PCR tests at the collective sites (7, 9).
Analytical-grade meropenem provided by International Health Management Associates, Inc. (Schaumburg, IL) was used for all broth microdilution studies. Based on the potency, meropenem powder was weighed in a quantity sufficient to achieve the required concentration and reconstituted immediately prior to use.
MIC testing was conducted by BMD, Etest (bioMérieux Inc., Hazelwood, MO), Vitek 2 (bioMérieux Inc., Hazelwood, MO), Sensititre (Trek Diagnostic Systems, Inc., Cleveland, OH), and MicroScan (Dade Behring, Inc., West Sacramento, CA). MICs were determined in triplicate for each methodology, and the modal MIC for each isolate was reported for all analyses. BMD was performed according to CLSI M07-A8 standards (6) using cation-adjusted Mueller-Hinton broth (BBL-Becton Dickinson, Sparks, MD) at the Center for Anti-Infective Research and Development (Hartford, CT). The BMD MIC test range was 0.06 μg/ml to 64 μg/ml. The prepared microtiter plates were stored at −80°C before use and were read manually to determine the MIC. Meropenem MICs by Etest were performed at the Center for Anti-Infective Research and Development (Hartford, CT) and interpreted according to the manufacturer's procedures. The Etest MIC test range was 0.016 μg/ml to 256 μg/ml. Since BMD did not test MICs higher than 64 μg/ml, all Etest values above 64 μg/ml were defined as ≥64 μg/ml for comparison. For categorical interpretation, Etest MICs between standard dilutions were rounded up to the nearest 2-fold BMD dilution. Pseudomonas aeruginosa ATCC 27853 was used as the quality control strain for all BMD and Etest experiments. Vitek 2 analyses were performed at New York Presbyterian Hospital (New York, NY). The Vitek 2 susceptibility card AST-GN28 containing a meropenem test was employed according to the manufacturer's instructions, and software version WSVT2-R05.02 was used to read the cards. The MIC calling range for meropenem on the AST-GN28 card was ≤0.25 μg/ml to ≥16 μg/ml in doubling dilutions. Sensititre and MicroScan MICs were performed at Mount Sinai Medical Center (New York, NY). Sensititre GNXF panels and MicroScan Neg MIC 38 panels were used according to the manufacturers' instructions. The panels were inoculated and incubated at 37°C for 18 to 24 h before being read manually. The meropenem MIC test ranges for both Sensititre and MicroScan panels were 1 μg/ml to 8 μg/ml in doubling dilutions.
MIC results were classified as susceptible, intermediate, or resistant based on the 2010 CLSI breakpoints (susceptible, ≤1 μg/ml; intermediate, 2 μg/ml; resistant, ≥4 μg/ml) (8). For comparison, current FDA breakpoints were also applied (susceptible, ≤4 μg/ml; intermediate, 8 μg/ml; resistant, ≥16 μg/ml).
The results obtained by the BMD method were used as the reference against which all other results were compared. Agreement was assessed in two manners. If the Etest, Vitek 2, Sensititre, or MicroScan MIC equaled the BMD MIC within a single 2-fold dilution, or if the meropenem MIC result from both tests was ≥16 μg/ml, then the MIC for that isolate was defined as being in agreement. Results were also compared based on CLSI interpretive criteria (5). Category disagreements were classified as follows: minor error, isolate was categorized as intermediate by one testing method but either susceptible or resistant by the comparative test performed; major error, isolate was categorized as being resistant by Etest, Vitek 2, Sensititre, or MicroScan but susceptible by BMD; or very major error, isolate was categorized as susceptible by Etest, Vitek 2, Sensititre, or MicroScan but resistant by BMD (5).
The percentage of isolates classified as susceptible, intermediate, and resistant to meropenem by each testing method are provided in Table Table11 for both the 2010 CLSI breakpoints and the FDA breakpoints. The frequencies of minor, major, and very major errors are presented in Table Table22.
In a comparison of MICs between Etest and BMD, Etest results agreed with BMD results for 82.6% of isolates (38 of 46) but tended to be higher than those for BMD (Fig. (Fig.11 a). Based on categorical assignments, Etest produced a very major error rate of 2.2% and a minor error rate of 2.2% with the 2010 CLSI breakpoints. In a comparison of MICs between Vitek 2 and BMD, results agreed for 30.4% of the isolates (14 of 46). Overall, Vitek 2 results showed lower meropenem MICs than those from BMD (Fig. (Fig.1b),1b), thereby producing a very major error rate of 23.9% and a minor error rate of 39.1%. For Sensititre, MICs for 26.1% of isolates (12 of 46) were in agreement (Fig. (Fig.1c),1c), with 3% very major errors and 26.1% minor errors. MicroScan demonstrated the most agreement with broth microdilution: 95.6% agreement (44 of 46 isolates) (Fig. (Fig.1d)1d) based on MIC, 2.2% minor errors, and no very major errors. There were no major errors between any of the testing systems.
Error rates based on FDA breakpoints are provided in Table Table22 for comparison. Overall, the application of FDA breakpoints had little effect on changing minor error rates for these four systems relative to BMD results. However, the very major error rates for Vitek 2 and Sensititre increased to 58.7%.
Carbapenems are often used as first-line treatment for drug-resistant Gram-negative organisms, but the increasing frequency of carbapenemase-producing K. pneumoniae has decreased the utility of this class of antimicrobials (15). Recent work by our group has demonstrated that pharmacodynamically optimized doses of meropenem were able to produce reductions in bacterial densities against carbapenemase-producing K. pneumoniae in an in vitro pharmacodynamic model and may be a suitable therapeutic option (4). However, in order to produce such a pharmacodynamically optimized regimen, one must have an accurate account of the antibiotic MIC against the pathogen in question. The objective of this study was to evaluate five methods of meropenem MIC and susceptibility testing for carbapenemase-producing K. pneumoniae. We considered the broth microdilution method to be the reference method and tested a variety of automated systems as well as Etest against this standard.
Our studies observed 82.6% agreement between Etest and BMD, with Etest producing MICs that tended to be higher than those from BMD, thus providing a more conservative measure of the meropenem MIC. This result is not surprising, as studies have demonstrated that susceptibility tests of carbapenemase-producing bacteria using Etest are often difficult to interpret (15). The presence of scattered inner colonies along the inhibition zone can lead to perceived increased resistance. The presence of scattered colonies may be due to decreased expression of the porin channel OmpK36, which has been found in isolates expressing blaKPC (11). As is defined in the package insert for Etest, these scattered colonies should be ignored if not near the edge of the strip. Based on both revised CLSI 2010 breakpoints and FDA breakpoints, Etest categorically provided acceptable error rates in interpreting results for these isolates. CLSI maintains that the percentage of very major and major errors between testing methods should be kept to less than 10% and that minor error rates should be kept to less than 40% when evaluating susceptibility methods against a standard dilution technique (5).
Given the busy environment of today's clinical microbiology laboratory, most departments utilize automated systems for identification and susceptibility and MIC testing; furthermore, these systems produced the most intermethod variability in meropenem MICs against these KPC. While manually read MicroScan panels produced MIC and susceptibility results that essentially mimicked those of broth microdilution, Sensititre and Vitek 2 testing systems were less reliable. Sensititre was read manually by a trained investigator and produced a MIC agreement in only 26.1% of isolates but had very major and minor error rates of 3% and 26.1%, respectively, which are in accordance with acceptable ranges. The Vitek 2 AST-GN28 card resulted in a MIC agreement of 30.4% and the highest percentage of very major errors (23.9%). These results may be surprising given the amount of literature that has demonstrated acceptable results with this automated system; however, these studies have used Vitek 2 to detect the presence of a resistance mechanism only, as opposed to accurately defining MICs (16). In addition, Vitek 2 results are influenced by the software and the cards used to analyze and interpret the data. A new card (AST-N142) that has been approved in Europe should be released in the United States shortly, and will most likely ameliorate this problem for KPC. Accordingly, when the new card was used to test the isolates, all isolates demonstrated a meropenem MIC of ≥16 μg/ml, resulting in 89.1% agreement with BMD, a minor error rate of 2%, and no major or very major errors (data not shown). Both Sensititre and Vitek 2 MIC results tended to be several dilutions lower than those of BMD; thus, when FDA breakpoints were applied, the very major error rates jumped to an unacceptable 58.7% for both tests.
A possible reason for the discrepancy in susceptibility results among automated systems might involve the inoculum size. A study with the MicroScan system by Bratu and colleagues demonstrated false susceptibility rates for K. pneumoniae isolates that were proposed to be due to a low inoculum size (2). This problem has also been reported with the Vitek 2 system, leading to the conclusion that low inoculum size has a major influence on the outcomes of these automated systems, with false susceptibilities being reported (1). We read our MicroScan and Sensititre panels manually, and this may have reduced the number of very major errors compared with having the automated system interpret the MIC.
According to the FDA breakpoints in effect at the time these studies were conducted, meropenem susceptibility is defined by a MIC of ≤4 μg/ml. In our study, a wide variation in meropenem MICs (2 to ≥64 μg/ml) was observed among the 46 KPC-producing isolates. Recently, the CLSI approved new breakpoints for meropenem, imipenem, and doripenem that define susceptibility as a MIC of ≤1 μg/ml and resistance as ≥4 μg/ml (8). When the FDA breakpoints were applied to our isolates, slight increases in minor errors for the Etest, Sensititre, and MicroScan methodologies were observed, but as previously noted, the Sensititre panel and Vitek 2 AST-GN28 cards generated greater rates of very major errors. Of note, the results of this study do not imply that a KPC phenotype will be missed in these systems if another carbapenem, ertapenem, is used as a screen for KPC production, which is what CLSI recommends, followed by the Hodge test for confirmation of carbapenemase presence. Nevertheless, the recent lowering of susceptibility breakpoints for the carbapenems still does not completely remove the potential for a resistant KPC isolate to be called susceptible by these systems, particularly for meropenem. Additionally, depending on the card or panel utilized by each hospital, the microbiology laboratory may not be able to test as low as 1 μg/ml for these agents, and thus the ertapenem screening followed by the Hodge test based on FDA breakpoints will still be applicable for identification.
To our knowledge, this is the largest MIC analysis of KPC isolates among differing testing methods conducted to date. Our KPC isolates were collected clinically, and few demonstrated low meropenem MICs via BMD, which is consistent with the frequency of high-level carbapenem resistance among these bacteria in the New York area. These results, therefore, may not apply to KPC from other areas if lower MICs are more common there. A limitation to our study is that the automated systems tested meropenem concentrations up to only 8 μg/ml, with isolates still growing at that concentration being reported as >8 μg/ml or ≥16 μg/ml (i.e., resistant). This precluded us from truly comparing the accuracy of MICs above 8 μg/ml, as can be done between Etest and broth microdilution. Although these isolates are clearly resistant, the opportunity for carbapenem dosage optimization at these higher MICs is of the utmost interest in the research arena (3, 4).
In summary, while the use of automated systems is convenient in hospital microbiology laboratories, the results of these systems vary widely with regard to the meropenem MICs against these carbapenemase-producing K. pneumoniae. When the standard of broth microdilution cannot be performed, manually read MicroScan and Etest provided accurate and conservative assessments of meropenem MICs.
This study was funded internally by the Center for Anti-Infective Research and Development, Hartford Hospital, Hartford, CT.
We graciously acknowledge the assistance of Henry Christensen, Jennifer Hull, Debora Santini, and Lindsay Tuttle in performing broth microdilution and Etest experiments.
Vitek 2 analyses were performed at New York Presbyterian Hospital (New York, NY) by K. A. Fauntleroy and S. G. Jenkins. Sensititre and MicroScan MICs were performed at Mount Sinai Medical Center (New York, NY) by M. Abuali and V. J. LaBombardi.
Published ahead of print on 19 May 2010.