Effect of Resistance Mechanisms on the Inoculum Effect of Carbapenem in Klebsiella pneumoniae Isolates with Borderline Carbapenem Resistance

Carbapenem resistance in Klebsiella pneumoniae is typically caused by carbapenemase, most commonly the KPC, OXA-48, NDM, and VIM enzymes (1). In addition, carbapenem resistance may be caused by a combined mechanism of both a noncarbapenemase β-lactamase (e.g., extended-spectrum β-lactamase) and a permeability defect (2). Compared with carbapenemase-producing K. pneumoniae (CPKP) isolates, carbapenemase-negative K. pneumoniae isolates that are carbapenem resistant (CRCNKP isolates) exhibit lower MIC values of carbapenem and are less likely to lead to clonal outbreaks (3). The MIC values may be relatively low in CPKP isolates of certain clones or with certain enzymes (e.g., OXA-48) (4). Hence, in some cases the carbapenem MIC values may be similar in CPKP and CRCNKP isolates. As both the CLSI and the EUCAST guidelines do not require carbapenemase testing (5), this information may not be included in the routine laboratory report. The goals of this study were to analyze the resistance mechanisms of CRCNKP and CPKP isolates with similar, borderline resistance to carbapenem and to investigate the differences between these groups in other resistance phenotypes.

We initially examined 100 carbapenem-resistant K. pneumoniae isolates from Israel and from the United States. The MICs to ertapenem, meropenem, and imipenem were tested by agar dilution (AD) assay in two repetitions. Of these 100 isolates tested, isolates were included in the study if they exhibited borderline resistance to carbapenem antimicrobials, defined as (i) resistance (≥2 μg/ml) to ertapenem and (ii) a MIC to imipenem and meropenem in the range of 0.25 to 4 μg/ml. Thirty-three isolates fulfilled the inclusion criteria (Table 1 ). The Carba NP test (6) was used to confirm the absence of carbapenemase in CRCNKP isolates. The carbapenemase and bla_ESBL genes were characterized by PCR (7, 8) and sequencing (9). Among the 33 isolates, the resistance mechanisms were OXA-48 (n = 12), KPC (n = 9), VIM (n = 2), NDM (n = 1), and CTX-M (n = 9). The clonal relation between the isolates was tested by BOX-PCR (10) and by PCR for the pilV-I gene (for KPC-producing isolates only) (11). Three of the 9 KPC-producing isolates belonged to the ST-258 clone. Otherwise, the clonal structure was diverse, with only 1 to 2 isolates per BOX-PCR type.

We sequenced the major porin genes associated with membrane permeability defects: ompK35, ompK36, and ompK37 (3). The presence of a nonsense mutation was analyzed using DNAman software version 6 (Lynnon Corp., USA). Nonsense mutations were identified in one gene in 5 of 9 of the CRCNKP isolates, which was higher than the 5 of 24 mutations found in the CPKP isolates (Fisher's exact test P = 0.0894). We then compared the relative expression of the main β-lactamase genes by reverse transcriptase quantitative PCR (qRT-PCR); these assays were done in order to explain the similar MICs of the CRCNKP and CPKP isolates despite the lack of carbapenemase in the CRCNKP isolates. The primers used are detailed in Table 2. As references for expression, we used the rpoB gene and the major porin gene ompK36. The ompK36 gene was chosen with the assumption that its expression is inversely related to the resistance level (2, 12). The efficiency of each qRT-PCR assay was ≥98%. Referenced with the rpoB gene (Fig. 1A), the expression of the carbapenemase genes bla_KPC, bla_VIM, and bla_NDM were in a range similar to that of the bla_CTX-M genes. Referenced to the ompK36 gene (Fig. 1B), the bla_OXA-48, bla_VIM, and bla_NDM genes had expression levels lower than those of the bla_CTX-M gene (with the exception of isolates 4893 and 7524) and the bla_KPC gene. Interestingly, the CRCNKP isolates 8486 and 4667 had the highest bla_CTX-M expression levels, which may explain their resistance to ertapenem despite their intact ompK gene sequences.

The effects of the resistance mechanisms on resistance phenotypes other than the MIC were studied by testing the inoculum effect, the bactericidal activity, and the effect of carbapenem exposure. An inoculum effect was defined as an increase in the MIC of meropenem or imipenem by ≥8-fold using inocula of 10⁵ and 10⁶ compared to the standard inoculum of 10⁴. This was observed in all 24 CPKP isolates, with the exception of one OXA-48-producing isolate (8367), but was observed in only 3 of 9 CRCNKP isolates (Fisher's exact test P = 0.0005). The bactericidal activity of imipenem and meropenem were tested by the minimal bactericidal concentration (MBC) following broth macrodilution (BMD) testing (13). The MBC was defined as a 99.9% decline in the inoculum compared with the initial inoculum of 10⁵ CFU (13). The MBC was ≤2-fold higher than the MIC in all CRCNKP isolates except one (5466) but was ≥4-fold higher in 7 CPKP isolates (P > 0.05). In one KPC-producing isolate (10014), the lack of bactericidal activity of meropenem was observed despite a relatively low MIC.

The effect of previous exposure to carbapenems was tested by measuring the MIC following overnight growth on MacConkey agar with 1 μg/ml of imipenem. A positive effect was defined as an increase in the MIC of imipenem or meropenem of ≥8-fold; no such increase was observed in any of the isolates.

In this study, we have shown that permeability defects were more common in the CRCNKP than in the CPKP isolates, as were the expression levels of their β-lactamase genes (with the exception of bla_KPC). Both factors are important in determining the resistance phenotype, and together they help to explain the similar carbapenem MIC in the CRCNKP and CPKP isolates despite the lack of catalytic activity against carbapenem by the CTX-M enzymes. We have shown that, despite the similar AD MIC values, CPKP isolates had an inoculum effect greater than that of the CRCNKP isolates and in part were more resistant to the bactericidal effect of imipenem or meropenem. These differences reflect the complex interplay of the various factors that determine resistance phenotypes that are not revealed by standard susceptibility testing. These phenotypes are not taken into account in pharmacokinetics/pharmacodynamics calculations and may have therapeutic implications in certain infections, such as those with a high inoculum of bacteria (14) or at a site with poor penetration of antimicrobials. This suggests that MIC measurements alone may not be sufficient in predicting therapeutic efficacy of carbapenems in infections caused by CPKP with borderline resistance. Hence, carbapenemase testing, in addition to its epidemiological importance, may also have therapeutic implications.

We thank Liz Temkin for her assistance in the writing the manuscript.

This work was supported by the European Commission FP7 AIDA project (preserving old antibiotics for the future: assessment of clinical efficacy by a pharmacokinetic/pharmacodynamic approach to optimize effectiveness and reduce resistance for off-patent antibiotics), grant 278348.