Evaluation of three variants of the carbapenem inactivation method for carbapenemase detection in Enterobacterales and description of the cCIM, supplemented with cloxacillin Authors: Julia Abels, Jessica Eisfeld, Jennifer Pienkoß, Sören G Gatermann, Niels Pfennigwerth Published: 2026 ABSTRACT AbstractObjectivesTo determine which of the three following phenotypic tests is the most suitable for the detection of carbapenemases in Enterobacterales: the modified carbapenem inactivation method (mCIM), zCIM (carbapenem inactivation method supplemented with zinc) or cCIM (carbapenem inactivation method with cloxacillin).MethodsEach test was evaluated with a set of 199 isolates representing the epidemiological situation in German hospitals. Furthermore, we wanted to establish first data for the cCIM, and evaluate whether the addition of cloxacillin improves the discrimination between carbapenemase-producing and AmpC-overproducing isolates, which have been shown to be problematic for other CIM-based assays.ResultsThe overall sensitivity/specificity was 96.1%/85.4% for the mCIM, 100%/94.8% for the zCIM and 94.2%/99.0% for the cCIM. For AmpC producers without carbapenemases the specificity of the cCIM was 98.6%, for the mCIM 89.0% and for the zCIM 95.9%.ConclusionsThe zCIM provided the best results. The cCIM improved the performance for AmpC producers without carbapenemases. Introduction Infections with carbapenemase-producing Enterobacterales are associated with a significant mortality and high healthcare costs.1–3 Therefore, reliable and efficient detection is of vital importance for patient outcome and infection control.4 Resistance to carbapenems in strains of Enterobacterales can result from carbapenemase production, but also from overproduction of ESBL and AmpC β-lactamases in combination with reduced membrane permeability due to porin loss or overexpression of efflux pumps.5 Furthermore, mutations of PBPs can result in carbapenem resistance.6 There are numerous tests available for carbapenemase detection with different characteristics. One of the most well known is the modified carbapenem inactivation method (mCIM), which is a modification of the carbapenem inactivation method (CIM). It uses trypticase soy broth (TSB) instead of water, to further improve the detection of OXA-48-type carbapenemases.7,8 It shows a high sensitivity and specificity, but can produce false-positive results when an overproduced AmpC β-lactamase is present.7,9,10 The carbapenem inactivation method supplemented with zinc (zCIM) has been designed to improve the MBL detection of the CIM. This is achieved by adding zinc sulphate to the inoculum, and it has been shown to have a higher sensitivity than the CIM.11,12 A modification of the mCIM is the carbapenem inactivation method with cloxacillin (cCIM), which was developed at the Department of Medical Microbiology at the Ruhr-University Bochum. For the cCIM, a meropenem/cloxacillin disc instead of a meropenem disc is used. Cloxacillin has an inhibitory effect on AmpC β-lactamases, and the aim of the test is to reduce the number of false-positive results of CIM-based methods for AmpC-overproducing isolates. Other testing methods are, for example, the colorimetric tests (CarbaNP, Blue-Carba and others), which indicate carbapenemase activity via pH change that results in a colour shift.13,14 MALDI-TOF MS can also be used by monitoring peaks associated with carbapenems and their respective degradation products, which indicate the presence or absence of a carbapenemase.15 However, these tests represent only a small proportion of all possible detection methods and their modifications. The major aim of this study was to compare the mCIM, zCIM and cCIM in terms of their capability to reliably detect carbapenemase-producing clinical isolates of various Enterobacterales. Methods Bacterial isolates The three tests were evaluated on a set of 199 pre-characterized clinical isolates of Enterobacterales from the strain collection of the German National Reference Center for MDR Gram-negative bacteria (NRC), representing the German epidemiology (Table 1). Species identification was performed by MALDI-TOF (Bruker, Bremen, Germany). Table 1.Strains used for this study and their resistance determinants Klebsiella pneumoniae Escherichia coli Enterobacter cloacae complex Klebsiella aerogenes Citrobacter freundii complex Serratia marcescens Klebsiella oxytoca Hafnia alvei OthersaIsolates without carbapenemase1415201898354 AmpC positive54201898153 ESBL positive91121Carbapenemase-producing isolates3521142105106 FRI2 GES121 GIM-11111 GMB1 IMI4 IMP121 KPC4221 KPC, NDM1 KPC, OXA-481 KPC, VIM11 LMB-11 NDM5411 NDM, OXA-48411 NDM, OXA-1811 NMC-A1 OXA-483113123 OXA-16232 OXA-18132 OXA-2041 OXA-23241 OXA-24413 TMB-11 VIM144241 a Klebsiella variicola, Morganella morganii, Proteus mirabilis, Raoultella planticola and Raoultella ornithinolytica. Strains used for this study and their resistance determinants a Klebsiella variicola, Morganella morganii, Proteus mirabilis, Raoultella planticola and Raoultella ornithinolytica. In total, 103 of these isolates were producing a carbapenemase, while 96 isolates were negative for carbapenemases. Among these, 73 were overexpressing an AmpC β-lactamase or an ESBL in combination with porin loss. Thirty-seven isolates harboured a carbapenemase and an AmpC β-lactamase, while only three isolates with a carbapenemase also produced an ESBL. Thirty-five isolates expressed both AmpC β-lactamases and ESBLs. One isolate combined a carbapenemase, AmpC β-lactamase and ESBL. Carbapenemases were pre-characterized as described before.16 Briefly, the carbapenemases were detected by a set of routine PCRs targeting the most common carbapenemase-encoding genes (KPC, VIM, IMP, NDM and OXA-48) with adjacent Sanger sequencing of PCR products. Additionally, a modified Hodge test (MHT) and combined disc tests with EDTA and boronic acid were performed. If phenotypic testing suggested the presence of a carbapenemase not detected by routine PCR, we performed a lysate-based microbiological bioassay analysing carbapenem hydrolysis influenced by EDTA, boronic acid, clavulanic acid and cloxacillin, with additional PCR for less common carbapenemases (e.g. FRI, GES or IMI) or WGS. Isolates potentially overproducing AmpC β-lactamases were analysed by performing an MHT on Mueller–Hinton (MH) agar with and without cloxacillin (200 mg/L) and a synergy assay with phenylboronic acid (PBA) and discs of meropenem on agar with and without cloxacillin.17 Isolates were classified as AmpC overproducers when they showed a positive MHT in combination with a negative result on MH-cloxacillin agar, and when they showed an inhibition effect by PBA for meropenem that was not present on agar supplemented with cloxacillin, indicating inhibition of the AmpC. In advance of testing, the isolates were cultured on blood agar plates and incubated for 18 h at 35 ± 2°C. E. coli ATCC 25922 was used as an indicator strain for all assays. All three tests were performed on the same day using the same inoculum for each strain over a 5 month period by a single operator and using the same lot of antibiotic discs and media. mCIM The mCIM was performed as previously described.7 Due to delivery problems during the study period, meropenem discs (10 µg) from two different manufacturers had to be used (Oxoid, Hampshire, UK or Mast Group, Lancashire, UK). The interpretation criteria for the zone of inhibition diameter were: ≤15 mm = positive, 16–18 mm = indeterminate and ≥19 mm = carbapenemase-negative. Indeterminate results were considered as false negative in case of a carbapenemase producer, and as false positive otherwise. For pinpoint colonies within the zone of inhibition, a diameter ≤18 mm indicated a positive result and ≥19 mm a negative one. zCIM For the zCIM, a 10 µL loopful of the colony material was added to 400 µL of TSB supplemented with 1.5 mM zinc sulphate, with the subsequent steps corresponding to the procedure for the mCIM. A zone of inhibition ≤20 mm was considered a positive result, while inhibition zone diameters >20 mm were considered as carbapenemase-negative. cCIM The cCIM was performed analogously to the mCIM except that instead of a meropenem disc, a meropenem/cloxacillin disc (10 µg + 200 µg) (Liofilchem, Roseto degli Abruzzi, Italy) was used. Statistical methods The sensitivity, specificity and their 95% CIs were calculated using MedCalc version 23.0.8.18 For the ROC analysis, IBM SPSS statistics version 29 (IBM, Armonk, USA) was used. The Youden Index was calculated manually (J = sensitivity + specificity − 1). Results mCIM Performances of each test with sensitivity and specificity are shown in Table 2. The sensitivity of the mCIM was 96.1% with 99/103 correctly positive results. Four isolates showed false-negative results: two isolates of Escherichia coli producing OXA-244, one isolate of Klebsiella pneumoniae producing NDM-1 and one isolate of K. pneumoniae producing GES-5 (Table 3). The specificity of the mCIM was 85.4%, with 82/96 correct negative results. In total, 14 isolates that did not produce a carbapenemase gave false-positive results with 8 of them producing an AmpC or an ESBL, while 5 did not produce any detectable β-lactamase (Table 4). Table 2.Performance of the different CIM-based methods for carbapenemase detectionmCIMzCIMcCIMCorrect resultsSensitivity,a %Specificity,b %Correct resultsSensitivity,a %Specificity,b %Correct resultsSensitivity,a %Specificity,b %Carbapenemase producers99/10396.1 (90.4–98.9)—103/103100, (96.5–100)—97/10394.2 (87.8–97.83)Non–carbapenemase producers82/96—85.4 (76.7–91.8)91/96—94.8 (88.3–98.3)95/96—99.0 (94.3–100)Ambler class A19/2095.0 (75.1–99.9)—20/20100 (83.2–100)—19/2085.0 (62.1–96.8)—Ambler class B37/3897.4 (86.2–99.9)—38/38100 (90.8–100)—35/3892.1 (78.6–98.3)—Ambler class D32/3494.1 (80.3–99.3)—34/34100 (89.7–100)—34/34100 (89.7–100)—Ambler class C65/73—89.0 (79.5–95.2)70/73—95.9 (88.5 -99.1)72/73—98.6 (92.6–99.97)Numbers in parentheses indicate the 95% CI. aSensitivity is given for Ambler class A, B and D. bSpecificity is given for Ambler class C. Performance of the different CIM-based methods for carbapenemase detection Numbers in parentheses indicate the 95% CI. aSensitivity is given for Ambler class A, B and D. bSpecificity is given for Ambler class C. Table 3.False-negative results with species and associated carbapenemase for each testTestSpeciesCarbapenemaseTotalmCIM E. coli OXA-2442 K. pneumoniae GES-51 K. pneumoniae NDM-11zCIM———cCIM K. pneumoniae GES-51 K. pneumoniae NDM-11 C. freundii complexGES-51 S. marcescens VIM-11 E. cloacae complexFRI-31 K. oxytoca GIM-11 False-negative results with species and associated carbapenemase for each test Table 4.False-positive results with number of AmpC producers for each testTestSpeciesTotalamCIM E. coli 5 (1)a E. cloacae complex4 (4) K. aerogenes 2 (2) S. marcescens 1 (1) K. pneumoniae 2 (0)zCIM E. coli 2 (0) E. cloacae complex1 (1) K. aerogenes 1 (1) S. marcescens 1 (1)cCIM S. marcescens 1 (1) aNumbers in parentheses represent the number of AmpC producers among these isolates. False-positive results with number of AmpC producers for each test aNumbers in parentheses represent the number of AmpC producers among these isolates. Eight isolates gave indeterminate results. Two of them were expressing a carbapenemase (NDM-1, GES-5) and were counted as false negative. Six isolates were interpreted as false positive, with one of them harbouring an AmpC β-lactamase and five phenotypically showing an ESBL combined with porin loss. zCIM The zCIM was the only test that detected all carbapenemases dependably resulting in a sensitivity of 100%. Three AmpC producers gave false-positive results (Table 4), respectively one isolate each of Enterobacter cloacae, Klebsiella aerogenes and Serratia marcescens. The specificity of the zCIM was 94.8%. Two isolates of E. coli without carbapenemase production were not detected correctly. cCIM Although the cCIM showed good overall performance, its sensitivity of 94.2% was lower compared with mCIM and zCIM. The cCIM had the most problems in the detection of class A and class B carbapenemases (sensitivity of 85.0% and 92.1%, respectively) with three false-negative isolates each (Table 3). One isolate of FRI-3 (E. cloacae complex) and two GES-5-producing Enterobacterales (K. pneumoniae and Citrobacter freundii complex) were false negative, respectively. The false-negative tests of class B carbapenemases included NDM-1, VIM-1 and GIM-1. With the addition of cloxacillin the specificity of the cCIM was 99.0%. Only one isolate of chromosomally encoded AmpC-producing S. marcescens was tested false positive (Table 4). On seven occasions the cCIM gave a correct negative result with an AmpC-producing isolate when the mCIM tested false positive with the same isolate. These were mostly chromosomally encoded AmpC-producing isolates of E. cloacae complex (n = 4). This constellation also occurred twice with the zCIM. Isolates showing false results in more than one test were an isolate of S. marcescens not producing a carbapenemase that was tested false positive in all tests, while two isolates of K. pneumoniae producing NDM-1 and GES-5, respectively, showed false-negative results in both mCIM and cCIM. Defining cut-off values for the cCIM As a newly developed modification of the CIM, the cCIM yet lacks cut-off values and consequently an ROC analysis was performed. The Youden Index indicated the best cut-off values for the inhibition zones in the range of 6–12 mm for positive and ≥13 mm for negative results (J = 0.932). As previously described, the sensitivity and specificity deriving from these values was 94.2% and 99.9%, respectively. As a modification of the mCIM, a zone for indeterminate results was considered for the cCIM. However, in contrast to the mCIM, the distribution of results for the cCIM was not scattered (Figures 1 and 2), therefore a classification into only positive and negative intervals was chosen. The same was observed for the zCIM (Figure 3), confirming the threshold set by Baeza et al.11 Inhibition zone distribution of the cCIM for the isolates analysed in this study. Cut-off parameter is indicated by the vertical black line. Inhibition zone distribution of the mCIM for the isolates analysed in this study. Cut-off parameters are indicated by the vertical black lines. Inhibition zone distribution of the zCIM for the isolates analysed in this study. Cut-off parameter is indicated by the vertical black line. Even though the cCIM is a screening test, where sensitivity should be favoured, the specificity would be very low (20.8% and 35.4%, respectively) when the sensitivity is maxed out to 100% or 99.9%. We observed two carbapenemase-positive isolates that showed pinpoint colonies within the zone of inhibition. Due to the small amount of these isolates a definition of cut-off values for the appearance of pinpoint colonies was not possible as it was defined for the mCIM.7 With a zone of inhibition of 11 mm, respectively, these two isolates were interpreted as true positives. The results from all three tests for all analysed bacterial isolates are shown in Table S1 (available as Supplementary data at JAC-AMR Online). Discussion This study showed that for our challenging set of clinical Enterobacterales isolates, the zCIM had the best overall performance, with a sensitivity and specificity of 100% and 94.8%, respectively. Looking at the performance for each Ambler class, only the zCIM managed to detect all carbapenemase producers, especially MBL-producing isolates. The solid performance on the detection of carbapenemase-producing Enterobacterales by the zCIM was confirmed by other evaluations in the past, albeit showing similar limitations for AmpC-producing isolates.11,12 In our evaluation, the mCIM had problems with the exclusion of class C AmpC β-lactamases and the detection of the OXA-244 class D carbapenemases and GES-5. OXA-244, which is an OXA-48-like variant, was characterized by Potron et al.19 as showing weak hydrolysis rates for the standard reference carbapenem imipenem. This low hydrolysis suggests a lower enzymatic activity against carbapenems in general, leading to more false-negative results as it did in this evaluation of the mCIM (Table 3). In recent years there has been a rise in OXA-244 detection in Germany, which makes reliable detection of this carbapenemase important.20,21 GES-5 has been described in the literature as showing low hydrolysis rates for imipenem, which was also described for GES enzymes in general.22 This could be an explanation for the false-negative results of the mCIM and cCIM for GES-5-producing isolates in our study. The relatively low specificity of the mCIM (85.4%) that we encountered contrasts with other studies. There have been reported specificities of 97% to 100%; however, these studies contained relatively few AmpC-producing isolates, and it has been shown that false-positive results can occur with the presence of an AmpC β-lactamase.7,10 We intentionally selected a larger number of AmpC-overproducing isolates for our evaluation to reflect the epidemiological situation in Germany, including strains producing chromosomally encoded AmpCs. In particular, K. aerogenes is known to overproduce the intrinsic AmpC in many clinical strains, easily leading to false-positive results in assays for carbapenemase detection, not sparing the mCIM as demonstrated here. With this study we provide the description and first evaluation data for the cCIM as a newly developed CIM variant. Cloxacillin is known as an inhibitor for AmpC β-lactamases and has been used in a variety of modifications of tests for the detection of carbapenemases.17,23 Given that knowledge, we developed the cCIM to improve the mCIM in the field of AmpC exclusion and enhance specificity. Ninety-eight percent of all AmpC producers were correctly identified as carbapenemase-negative by the cCIM, being superior to the mCIM, which only reached 89.0% specificity. This shows that in settings with a high prevalence of AmpC-producing isolates and frequent false-positive results of carbapenemase detection assays, the cCIM could provide a significantly better performance. The cCIM also showed a better performance in the detection of class D carbapenemases than the mCIM. The cCIM detected 100% of class D carbapenemases, whereas the mCIM had a sensitivity of 94.1% for this Ambler class. OXA-48 represents by far the most frequently detected carbapenemase in Germany, and OXA-48-like variants are also among the most common carbapenemases,24 underlining the importance of reliable detection of these enzymes. The lowest sensitivity was detected for class A carbapenemases (85.0%). Like the mCIM, the cCIM showed a false-negative result for the same isolate of K. pneumoniae producing GES-5, and an isolate of C. freundii also producing GES-5. Among the FRI enzymes, FRI-3 is known to have a significantly lower hydrolytic activity against carbapenems than FRI-1.25 This is concordant with a false-negative cCIM result for this enzyme in an isolate of E. cloacae complex. In conclusion, the cCIM might have a weak spot in the detection of some carbapenemases that show a lower hydrolytic activity. All evaluated tests are phenotypic methods that cannot determine the specific carbapenemase subtype. Identifying the subtype is crucial for initiating an effective antibiotic treatment. So, in order to preselect isolates that later undergo PCR testing a reliable, fast and cost-efficient method needs to be available. This underlines the significance of the evaluated tests to determine the one that best meets these criteria for such a method. Regarding the concentration of the meropenem/cloxacillin discs we chose a commercially available product, as the aim was to develop a method that can be easily established in routine laboratories. It is possible that other cloxacillin concentrations may provide better results, and this needs to be analysed in future studies. A major drawback of the tests evaluated in this study is the overnight incubation. Other phenotypic tests like the rapid carbapenem inactivation method (rCIM) can generate results in approximately 3–4 h.26 However, the rCIM has been described as having lower exclusion rates for AmpC producers and gave false-negative results for OXA-244-producing E. coli as well as IMI-producing isolates.27 Muntean et al.26 described the results of the rCIM to be comparable to the unmodified CIM. They also developed a version of the rCIM with the addition of cloxacillin in order to improve the specificity of the rCIM to naturally AmpC-producing isolates, the rCIM-A.23 However, with both rCIM and rCIM-A, IMI-producing isolates gave false-negative results. In our study 4/4 IMI producers were correctly detected in all versions of the CIM. This might show that a longer incubation time improves the detection of such isolates. It needs further evaluation to show whether the rCIM or one of its modifications (TSBrCIM, rCIM-A) could be competitive with the zCIM, mCIM or cCIM.23,27 A limitation of this study is that the strain collection was not subject to comprehensive WGS, due to budget restrictions. Despite 16 years of experience as the National Reference Center, we consequently cannot fully exclude that single strains of the AmpC-producing subset had additional mechanisms not detected with our routine phenotypic assays, e.g. ESBLs that were masked by AmpC production. Another limitation of the study is that we could not avoid using meropenem discs from different manufacturers due to delivery problems during the time the study was performed (2022), which affected the mCIM and zCIM. As it has been shown several times that discs from different manufacturers can show varying results,28 we cannot completely exclude that the results of this study could be biased by the use of discs from different manufacturers, despite the quality controls indicating no variations in disc quality. However, this should be taken into account when implementing such tests in the clinical laboratory, and stringent quality controls should be implemented as, depending on the setting and the respective resources, frequent changes of suppliers can occur. In conclusion, this study showed that the zCIM was the most reliable phenotypic detection test for carbapenemase-producing Gram-negative clinical isolates among the those evaluated. It was demonstrated that the use of cloxacillin in the cCIM can reduce the rate of false-positive results with species that harbour chromosomally encoded AmpC β-lactamases. It might also be worth combining it with the zCIM to form a zcCIM in future follow-up studies to further improve performance, especially since the mzCIM, a further modified version of the zCIM, showed better performance on the detection of OXA-48-like variants and rare carbapenemases such as GES.29 Supplementary Material