Detection of carbapenemase activity in Enterobacteriaceae: comparison of the carbapenem inactivation method versus the Carba NP test

Sir,

Carbapenem resistance mediated by plasmid-encoded carbapenemases has emerged worldwide and become a major concern.¹ The detection of the activity of carbapenemases has a strong impact on hospital infection control, because the detection of their presence can initiate measures to avoid potential outbreaks and lateral spread of the resistance. Although molecular detection by PCR is considered the gold standard for carbapenemase gene identification, some limitations are clearly recognized. Between them, false-negative results (the presence of a carbapenemase gene not tested in the PCR reaction, or mutations affecting annealing of primers) or the detection of inactive genes (i.e. no carbapenemase expression) can delay infection-control measures or, oppositely, initiate them when they are not required. A fast and accurate phenotypic method, the Carba NP test (CNPt), was developed; it detects carbapenemase activity with very high sensitivity and specificity and lower costs compared with those of PCR.² This method is now being recommended in the CLSI guidelines for carbapenemase activity detection.³ However, recent studies have shown that this test has lower sensitivity particularly against isolates expressing β-lactamases with low carbapenemase activity, such as OXA-48-like, or expressing mucoid colonies.^4,5 Another new test, the carbapenem inactivation method (CIM), has shown very promising results based on its sensitivity, specificity, low cost and easy interpretation.⁶ Briefly, the CIM consists of two steps: (i) incubation of a meropenem disc with the isolate to be tested; and (ii) incubation of this meropenem disc with the Escherichia coli ATCC strain. After this second incubation step, the presence of carbapenemase activity can be easily detected: the absence of an inhibition zone indicates enzymatic hydrolysis of meropenem during the first incubation step, whereas a ‘clear inhibition zone’ appears when the tested isolate does not express carbapenemase activity.⁶ The goal of this study was to compare the performance of the CIM and the CNPt against a panel of well-characterized enterobacteria.

A total of 182 Enterobacteriaceae were tested (Table 1). Of these, 82 were negative controls with variable carbapenem susceptibilities, which were determined by Etest and their results interpreted using the CLSI guidelines.³ These 82 isolates had PCR results that were negative for carbapenemase genes, and none displayed carbapenemase activity by phenotypic tests (modified Hodge test, CNPt and KPC/MBL Confirm Kit, Rosco Diagnostica).⁴ We also included 100 carbapenemase-producing isolates (including KPC, NDM, VIM, IMP, OXA-48-like, NMC/IMI and SME producers), all confirmed by PCR results and sequence analysis (Table 1). Both the CNPt and the CIM were performed in triplicate for each isolate as described previously.^3,6

All non-carbapenemase producers were negative by both phenotypic methods (Table 1). The fact that some of these negative controls had high carbapenem MICs (≥32 mg/L) suggests the presence of other mechanisms of resistance, such as AmpC and/or ESBL production plus impermeability. These results confirm the specificity and a positive predictive value of 100% for both methods. For carbapenemase producers, the CNPt had a few false-negative results, most of them related to OXA-48-like producers (two OXA-48, two OXA-181, two OXA-232 and one OXA-244), but also with two NDM-1 producers (one of which showed mucoid colonies) and one KPC producer (mucoid). Both methods failed to detect a swarming-producing Proteus mirabilis positive for IMP-27. However, the carbapenemase activities of this enzyme as well as the NDM-1 from the mucoid Providencia rettgeri were identified by both methods in transformant or transconjugant E. coli strains, as described previously.⁴ These results gave a sensitivity (90.1%) and negative predictive value (88.2%) for the CNPt that were similar to those previously reported.^4,5 However, considerably higher sensitivity (98.8%) and negative predictive value (99%) were obtained by the CIM.

The CIM proves to be an accurate method for detection of carbapenemase activity. In our hands, its main disadvantage was that, in practice, it required an overnight incubation of the plates to obtain results (in contrast to the 8 h incubation period claimed in the original report⁶) versus a maximum incubation of 2 h using the CNPt. On the other hand, the CIM had multiple advantages: (i) it is an easier-to-perform test; (ii) only water and a 10 μg meropenem susceptibility-testing disc per isolate were necessary to perform the test; (iii) four isolates plus positive and negative controls can be tested on the same Mueller–Hinton agar plate (standard size) inoculated with E. coli ATCC 25922 (these three first points highlight the low costs of the test); and (iv) the interpretation of results is easy (no inhibition zones versus ≥20 mm for negative carbapenemase activity; in our tests, with the exception of one false-negative result, we did not detect a carbapenemase producer exhibiting meropenem inhibition zones >6 mm).

In summary, both methods proved to be very efficient in the detection of carbapenemase activity, with pros and cons for their implementation in microbiological laboratories. Although the CNPt showed an excellent specificity and a short turnaround time for carbapenemase detection (from 10 min to 2 h), it was unable to detect around 10% of the carbapenemase (mainly OXA-48-like) producers. The CIM, on the other hand, showed excellent specificity and sensitivity and low costs (no specialized reagents, equipment or skills are necessary), but it requires at least 8 h (overnight in our hands) of incubation before the results may be read. In both cases, molecular tests are needed for identification of the carbapenemase-producing genes in the clinical isolate.

Funding

This work was supported by Public Health Ontario Laboratory internal funding.

Transparency declarations

None to declare.

References