Evaluation of a Modified Carbapenem Inactivation Method for Detection of Carbapenemases in Pseudomonas aeruginosa

LETTER

Detection of carbapenemases in Pseudomonas aeruginosa isolates is of utmost importance for preventing nosocomial transmission and detecting outbreaks. The availability of accurate and yet simple and affordable carbapenemase detection tests may offer an incentive for laboratories and hospitals to closely monitor this issue. One such assay is the modified carbapenem inactivation method (mCIM) which has recently been published by CLSI for use in detection of Enterobacteriaceae (1) and evaluated in a multicenter trial, with a resulting mean sensitivity of 97% and specificity of 99% across nine laboratories (2). Standardized guidelines for testing of glucose-nonfermenting Gram-negative bacilli by mCIM have not yet been published.

Here, we evaluated the performance of mCIM for detection of carbapenemases in 100 genotypically characterized, carbapenem-resistant P. aeruginosa clinical isolates. Resistance to carbapenems was confirmed by disk diffusion using CLSI M100-S27 breakpoints (1). Molecular detection of KPC, GES, IMP, VIM, NDM, OXA-48, and NMC/IMI carbapenemases was performed at the National Microbiology Laboratory (Winnipeg, Canada) (3). All carbapenemase PCR-positive organisms available in our collection (n = 31) plus an additional 69 carbapenemase PCR-negative organisms were included to complete a sample set of 100 isolates. mCIM was performed as described in CLSI M100-S27 (1) using a 10-μl inoculum based on preliminary CLSI results (CLSI January 2017 AST Subcommittee Meeting minutes [http://clsi.org/standards/micro/microbiology-files/]) and existing literature (4–10). Inocula were suspended in 2 ml of TSB broth (BD, Franklin Lakes, NJ), and a 10-μg meropenem disk (Oxoid, Hampshire, UK) was added prior to incubation at 35°C for 4 h. The disks were then transferred onto MH agar II (BD, Franklin Lakes, NJ), inoculated with a 0.5 McFarland suspension of Escherichia coli ATCC 25922, and incubated in ambient air at 35°C for 18 to 24 h. Interpretation of results was performed as follows. The presence of a carbapenemase was indicated by an inhibition zone ≤15 mm in diameter or the presence of colonies within the disk zone, and absence was indicated by zones ≥19 mm in diameter. Isolates requiring retesting (i.e., those giving discrepant and indeterminate results) were tested using broth volumes of 2 ml and 400 μl (4–8) and inocula obtained after subculture of isolates in the presence of a meropenem disk. An additional step of 10 to 15 s of vortex mixing was performed immediately after addition of the meropenem disk to facilitate suspension of the inocula (8). Tests using Carba NP and β Carba (Bio-Rad, Hercules, CA) were performed on isolates with persistently discrepant results after mCIM retesting.

A total of 100 P. aeruginosa isolates from unique local patients were included. The isolates were resistant to imipenem (n = 99) or meropenem (n = 92) or both (n = 91), with 31 containing a carbapenemase (GES-5 [n = 18], VIM [n = 10], or NDM [n = 3]). mCIM results are summarized in Table 1. Initial testing yielded 16 indeterminate results (4/16 isolates encoding GES-5; 12/16 carbapenemase PCR-negative isolates) and 6 mCIM/PCR discrepant results. Upon retesting of the isolates with indeterminate or discrepant results (n = 22) using 2 ml of broth, correct identification of 25/31 carbapenemase producers was achieved (sensitivity [Se], 81% [confidence interval {CI}, 67% to 95%]; specificity [Sp], 97% [CI, 93% to 100%]). In comparison, use of 400 μl of broth further improved mCIM performance (Se, 87% [CI, 75% to 99%]; Sp, 97% [CI, 93% to 100%]), correctly identifying 27/31 carbapenemase producers, including 2 additional GES-5-encoding isolates. Discrepant results persisting after retesting with reduced broth volume (Table 2) included 4 mCIM-negative/PCR-positive isolates (GES-5, n = 2; VIM, n = 1; NDM, n = 1) with no/low phenotypic carbapenemase activity and 2 mCIM-positive/PCR-negative isolates with strong carbapenemase activity upon additional phenotypic testing. The test performance remained unchanged (Se, 88% [CI, 72% to 97%]; Sp, 100% [CI, 95% to 100%]) after mCIM-positive/PCR-negative isolates were proven to encode a novel metallo-β-lactamase (unpublished data).

TABLE 1.

Modified carbapenem inactivation method testing results from 100 carbapenem-resistant Pseudomonas aeruginosa clinical isolates (31 carbapenemase producers by PCR) upon initial testing and after retesting of 22 isolates with indeterminate or discrepant results^a

Resultb	No. of isolates at indicated test broth vol
	Upon testing	After retesting
	2 ml	2 ml	400 μl
mCIM
Positive	27	27	29
Indeterminate	16
Negative	57	73	71
Concordant
mCIM+/PCR+	24	25	27
mCIM−/PCR−	54	67	67
Discrepant
mCIM+/PCR−	3	2	2
mCIM−/PCR+	3	6	4

^aPCR-based detection included the following carbapenemases: KPC, GES, IMP, VIM, NDM, OXA-48, and NMC/IMI.

^bmCIM, modified carbapenem inactivation method; indeterminate, no carbapenemase detected (n = 12) or GES-5 detected (n = 4).

TABLE 2.

Additional phenotypic carbapenemase test results for 6 carbapenem-resistant Pseudomonas aeruginosa clinical isolates with discrepant mCIM results after retesting in reduced broth volume^a

Result	Carbapenemase	Carba NP result	β Carba result
mCIM+/PCR−	Novel MBL	+++	+++
	Novel MBL	+++	+++
mCIM−/PCR+	GES-5	+	+++
	GES-5	+	−
	NDM	−	−
	VIM	−	−

^aMetallo-β-lactamase (MBL) phenotype as determined by KPC/MBL in P. aeruginosa/Acinetobacter Confirm kit 98020 (Rosco Diagnostica, Taastrup, Denmark). Carba NP was performed per CLSI M100-S27. Carba NP and β Carba results were graded according to the strength of colorimetric changes as follows: +++, strong/positive test result, <30 min; +, weak but clearly positive test result; −, no change/negative test result.

Our study was the second in the literature to evaluate mCIM for testing of P. aeruginosa and was unique in that it included a high proportion of isolates harboring the emerging Amber class A carbapenemase GES-5. Simner et al. evaluated mCIM alongside 10 other carbapenemase detection phenotypic tests in glucose-nonfermenting organisms, including 67 P. aeruginosa isolates, among which 21% encoded a carbapenemase and only 2 carried a class A carbapenemase (KPC) (9). The P. aeruginosa collection in the current study contained 33% carbapenemase-encoding isolates and a near-50/50 split between class A and B carbapenemases, and the results showed slightly reduced mCIM sensitivity compared to that observed by Simner et al. for testing of P. aeruginosa (Se, 100% [CI, 73% to 100%]; Sp, 98% [CI, 89% to 100%]).

GES-5 has been documented in Canada since 2006 (11) and specifically in P. aeruginosa since 2009 to 2010, when it corresponded to the second-most-frequently detected carbapenemase in this organism species (12). Its presence in P. aeruginosa has been previously documented in the Americas, Africa, Europe, and Asia. Overall, mCIM as here described was able to detect almost 90% of GES-5-encoding isolates. GES-5 is known to have relatively lower carbapenem hydrolytic activity (13), perhaps helping to explain why a 5-fold inoculum concentration or different enzyme substrates utilized in Carba NP (imipenem) and β Carba (proprietary substrate) improved the detection results. Additional help for detection of GES-5 may have been offered by subculture in the presence of meropenem for isolates with indeterminate mCIM results, perhaps through selection of heteroresistant populations (14). Failure to detect a carbapenemase was also seen in Ambler class B isolates encoding VIM (n = 1) or NDM (n = 1), both of which are thought to have relatively higher hydrolytic activity and yet were not detectable by using different substrates. This observation suggests that other isolate-specific factors such as mutations could have affected carbapenemase hydrolytic activity and therefore might have been responsible for these mCIM-negative results.

Our contribution highlights that mCIM offers a viable alternative for carbapenemase detection in P. aeruginosa. While not all carbapenemases could be detected, that limitation was seen in similar proportions for the Ambler class A and B enzymes. Test volume reduction for retesting of isolates as presented may be an acceptable alternative to increase mCIM test sensitivity in regions where carbapenemases with lower hydrolytic activity are present.