The emergence of carbapenem-resistant Acinetobacter baumannii is of particular concern worldwide as infections with this pathogen have high mortality rates and limited treatment options [1^,2^,3^,4]. A. baumannii's growing carbapenem resistance is mainly due to the production of acquired carbapenemases [5]. A rapid and reliable phenotypic method for detecting these carbapenemase producers in clinical samples is indispensable for therapeutic and infection control reasons. Although the number of non-molecular methods described for rapid detection of carbapenemase activity in gram-negative bacteria has increased over the past few years [6^,7], phenotypic screening of carbapenemase in A. baumannii remains a challenge for clinical microbiologists. The OXA-type β-lactamase enzymes harbored by some carbapenem-resistant A. baumannii hydrolyze carbapenem less efficiently than many class A and B carbapenemases, but can still lead to high-level carbapenem resistance when present together with other resistance mechanisms such as porin mutations or efflux [8]. Various procedures have been developed to optimize the method for A. baumannii; however, each has led to undesirable trade-offs between sensitivity and specificity [9].

According to the 26th edition of the CLSI recommendations (2016), the modified Hodge test should be applied to only Enterobacteriaceae, and the Carba NP test should be applied for all Enterobacteriaceae, Pseudomonas aeruginosa, and Acinetobacter spp. [10]. These two phenotypic methods were maintained in the 27th edition with the same procedures described in 2016, but another method, the modified carbapenem inactivation method (mCIM), was introduced for only Enterobacteriaceae [11]. However, the Carba NP test was no longer recommended for Acinetobacter spp. in the 28th edition (2018) [12^,13]. Moreover, the mCIM was further modified to detect carbapenemases in P. aeruginosa but remained inapplicable for Acinetobacter spp. [12^,13]. Furthermore, the modified Hodge test was no longer included in the 28th edition, even for Enterobacteriaceae. In 2017, a research group in Japan described another revised version of the carbapenem inactivation method, CIMTris, for detecting carbapenemase production in Acinetobacter and Pseudomonas spp. [14]; they used Tris-HCl buffer instead of tryptic soy broth (TSB) to extract the carbapenemases, resulting in higher sensitivity but lower specificity than the mCIM [14]. However, to date, no data exist that illustrate the interactions between the matrix and carbapenem stability and/or the effects of matrices on the enzyme-substrate reaction of carbapenem hydrolysis, which may cause the lower specificity for CIMTris [14].

We evaluated the influence of the matrices used in mCIM and CIMTris on the stability of the meropenem (MEM) disk for detecting carbapenemase producers. We also suggest optimal mCIM conditions for detecting carbapenemase-producing A. baumannii.

Of the 73 bacterial isolates tested, 46 (63.0%) were resistant to both IPM and MEM. Of these, 38 isolates (82.6%) were positive for both bla_OXA-23-like and bla_OXA-51-like genes; 37 isolates had ISAba1 upstream of the bla_OXA-23-like gene; and the remaining single isolate (2.1%) was positive for the bla_OXA-51-like gene and had an ISAba1 element upstream of this gene (Table 1). Of the 73 isolates, 27 that showed susceptibility to carbapenem by VITEK 2 were confirmed using the disk diffusion method, with inhibition zones ≥18 mm for MEM and ≥22 mm for IPM.

Procedure A of the mCIM was modified from a screening method initially called CIM [6^,7]. Of the 46 carbapenem-resistant bacterial isolates tested using mCIM procedure A, three (6.5%) were positive and two (4.3%) were indeterminate (data not shown). In addition, 27 carbapenem-susceptible isolates were negative using procedure A. Following procedure B, the mCIM showed higher sensitivity; 30 carbapenem-resistant isolates (65.2%) were carbapenemase producers. Furthermore, CIMTris showed that all 46 (100%) carbapenem-resistant isolates were carbapenemase falsely positive; two (7.4%) and one (3.7%) of the 27 carbapenem-susceptible isolates were positive and indeterminate, respectively (Table 1).

We next tested whether the incubation buffer inactivates MEM activity in the absence of carbapenemase-producing organism. An inhibition zone diameter of 30±1 mm was measured after an overnight incubation and was termed the original inhibition zone (repeated 12 times). The results after incubating the MEM and IPM disk in the matrix showed that the inhibition zone surrounding the incubated disk was smaller than that around the non-incubated disk (repeated five times).

Similar results were observed in both the mCIM and CIMTris. Noticeably, the inhibition zone in CIMTris was significantly smaller (P<0.001) than that in the mCIM and the original inhibition zone mentioned above (Fig. 1B and C). In addition, we evaluated the influence of the matrices on the MEM disk; we measured the inhibition zone diameters dependent on the following three factors: (1) type of matrix, (2) volume of matrix, and (3) duration of incubation (Fig. 1D and E).

When incubated in the same volume for the same time period, the MEM disk maintained higher stability in TSB broth than in Tris-HCl buffer. In addition, when the volume was increased and the incubation time was extended, the diameter of the inhibition zone surrounding the MEM disk incubated in TSB changed slightly, while the change in Tris-HCl was greater. Moreover, the MEM disk incubated in Tris-HCl buffer for four hours could not inhibit the growth of indicator organism E. coli ATCC 25922 and gave false-positive results, with a 6 mm diameter of inhibition zone in both 400 µL and 2 mL (Fig. 1C). No inhibition zone >6 mm in diameter was observed for disks incubated in Tris-HCl (Fig. 1E).

Based on the modification of the mCIM with 400 µL TSB and (i) one 10-µL loopful of bacteria, 36 (78.3%) of the 46 isolates were detected as carbapenemase producers, while the remaining two (4.3%) and eight (17.3%) isolates were identified as indeterminate and negative, respectively (Table 2). When the mCIM with 400 µL TSB and (ii) one and a half 10-µL loopfuls of bacteria was examined, eight more isolates were detected as carbapenemase-producing, resulting in 44 (95.7%) carbapenem-resistant isolates, and the two remaining isolates were identified as indeterminate; with (iii) two 10-µL loopfuls of bacteria, all 46 (100%) isolates were detected as carbapenemase-producing A. baumannii. Our experiment showed that the optimal mCIM with two 10-µL loopfuls of bacteria incubated in 400 µL TSB had markedly high sensitivity and specificity (100% and 100%, respectively).

We determined that (1) incubation in either matrix, even in the absence of the test bacterial isolates, leads to some of the MEM and IPM dissolving into the matrix, resulting in a reduced amount of active MEM in the disk at the end of incubation; (2) both a longer duration of incubation and an increased volume of matrix reduce active MEM in the disk at the end of incubation; (3) a longer incubation causes a more pronounced loss of active carbapenem from the disk when the incubation takes place in Tris-HCl; and (4) incubation in Tris-HCl actually leads to not only dissolution of MEM into the matrix but also its inactivation. We demonstrated that Tris-HCl inactivates MEM activity in the absence of an organism. This might be caused by the effects of the matrix on the MEM disk during incubation. CIMTris differed from mCIM in the following three aspects: (1) type of matrix, (2) volume of matrix, and (3) duration of incubation. These results indicate that an increased volume of matrix and an extended incubation time lead to an elevated amount of dissolved MEM and IPM (Fig. 1D and E).

Our results support a previous report that CIMTris shows a high percentage of false-positive results after four hrs of optimal incubation time [14]. The aforementioned two carbapenem-susceptible isolates that showed positive results in CIMTris are considered false-positive. The false-positive results suggest that MEM might be not only dissolved but also inactivated by Tris-HCl following incubation. Therefore, Tris-HCl should not be used as the buffer when performing mCIM to assess MEM hydrolysis.

Building on the principles learned from this first set of experiments, we then moved forward with TSB as the incubation matrix of choice and performed several variations of the mCIM. Inoculum size and matrix volume were important parameters for adjustment of mCIM.

Our results showed an improvement with increasing inoculum size and reduced volume of the liquid base (Tables 1 and 2). No false-positives were detected for any modification of these two parameters; mCIM with 400 µL TSB had higher sensitivity than mCIM with 2 mL TSB.

The present study had several limitations. The isolates were not comprehensive and were mainly from a single medical center. In addition, the genotypes of the isolates were not determined. Therefore, further testing of various A. baumannii genotypes is required to evaluate the performance of the mCIM. Despite these limitations, the optimal mCIM is an easy-to-perform, inexpensive, and reproducible method using readily available materials and reagents for detecting carbapenemase-producing A. baumannii in clinical microbiology laboratories.

(A) Schematic drawing of remaining potency of MEM disks in two types of buffers (TSB and Tris-HCl) after two or four hours of incubation. Diameter of inhibition zone with (B, C) incubated MEM disk depending on type and volume of matrix and duration of incubation, (D, E) MEM dissolved in TSB or Tris-HCl buffer. Tris-HCl weakened MEM potency, which was not maintained in either the buffer or disk.

Abbreviations: MEM, meropenem; TSB, tryptic soy broth.