Isolate selection. The isolates in our sample were selected in order to represent different years and hospitals from the collection of Enterobacterales isolates that were sent to the National Institute for Antibiotic Resistance & Infection Control in the years 2001–2017. An initial screening for carbapenemase identification was performed using polymerase chain reaction (PCR). We used a multiplex assay to detect blaKPC, blaOXA−48−like, blaNDM, blaIMP, blaVIM, and blaIMI [16] and a simplex PCR to detect blaSME [17]. The PCR results served as the gold standard to which the CARBA-5 assay was compared. Carbapenemase activity was assessed using the qualitative colorimetric β CARBA test (Bio-Rad, Marnes-la-Coquette, France) according to the manufacturer's instructions.
The sample consisted of 197 unrelated, non-duplicate isolates from various sources (sputum n = 51, blood n = 74, urine n = 48, rectal n = 20, other n = 4). There were 19 (19/197, 9.6%) carbapenem-susceptible Enterobacterales isolates (CSE; meropenem MIC < 4 µg/ml) and 178 (178/197, 90.4%) carbapenem-resistant Enterobacterales (CRE; meropenem MIC ≥ 8 µg/ml). Of the 178 CRE isolates, 151 were identified as CPE: 47 harbored blaKPC, 40 blaOXA−48−like, 25 blaNDM, 19 blaVIM, 3 blaIMI, 1 blaIMP, 1 blaSME and 15 isolates harbored two or three carbapenemases. The other 27 CRE isolates (11 K. pneumoniae, 7 E. coli, 4 Enterobacter spp., 1 K. oxytoca, 1 Proteus spp., 1 E. aerogenes and 2 Providencia spp.) had meropenem MIC values of ≥ 8 µg/ml but no carbapenemase activity and none of the carbapenemases genes screened for in this study. They were classified as non-carbapenemase-producing CRE (Non-CP CRE).
All isolates were stored at -80 °C, sub-cultured aerobically at 35 ± 2 °C and transferred twice prior to testing.
OXA-48-like target cross reactivity evaluation. To evaluate the cross reactivity with the OXA-48-like target, we tested 73 carbapenem-resistant A. baumannii (CRAB) isolates (meropenem MIC ≥ 8 µg/ml), all of them Ambler class D β-lactamases producers. A. baumannii isolates were identified by Vitek MS (BioMe´rieux, Marcy-l'Étoile, France). Isolates were tested by PCR for the blaOXA− 24, blaOXA− 23 and intrinsic blaOXA− 51−LIKE genes, as described previously [18]. PCR products were sequenced and OXA-type variants were determined using DNAMAN® software version 7.0 (Lynnon Corporation, Pointe-Claire, Quebec) and the Basic Local Alignment Search Tool (BLAST, http://blast.ncbi.nlm.nih.gov). The CRAB isolates harbored blaOXA− 23 (n = 6), blaOXA− 65 (n = 6), blaOXA− 66 (n = 11), blaOXA− 69 (n = 2), blaOXA− 70 (n = 4), blaOXA− 71 (n = 11), blaOXA− 248 (n = 8) and various OXA-type combinations (n = 25).
The CRAB isolates were stored at -80 °C, sub-cultured aerobically at 35 ± 2 °C and transferred twice prior to testing.
NG-Test CARBA 5 assay. The CARBA 5 assay (NG Biotech, Guipry, France) is based on the reaction of carbapenemases with labelled anti-carbapenemase monoclonal antibodies. The assay was performed using fresh colonies grown on CHROMagar™ mSuperCARBA™ (Hy Laboratories, Rehovot, Israel) or Mueller Hinton agar (Hy Laboratories), as recommended in the CARBA 5 instruction manual. One colony was suspending in five drops (150µL) of extraction buffer. The bacterial suspension was homogenized by vortex and 100 µl was loaded into a nitrocellulose membrane. The suspension migrated through the membrane due to capillary force, and interacted with the corresponding anti-carbapenemase monoclonal antibodies immobilized on the membrane. Results were read by researchers blinded to the PCR results, to ensure unbiased interpretation, following 15 min of incubation time at room temperature. Results were considered positive if a red line appeared on the control region and on one or more of the test regions (KPC, OXA, VIM, IMP or NDM), indicating that the isolate carried one or more carbapenemases. Results were considered negative if a red line appeared only on the control line, indicating that the isolate did not carry any of the five carbapenemases.
Discrepancies between the CARBA 5 results and the PCR results were further investigated by genome sequencing.
Genome sequencing. Isolates requiring investigation by genome sequencing were grown overnight on Brain Heart Infusion broth (Hy Laboratories) and DNA was extracted using the MagAttract HMW DNA Kit (Qiagen, Hilden, Germany). Purified DNA was sequenced using the Rapid Barcoding Sequencing (Ref. No. SQK-RBK004, Oxford Nanopore Technologies, Oxford, UK), on a MinION sequencing device (Oxford Nanopore Technologies). FastQ files were subjected to antibiotic resistance gene search using ResFinder K-mer alignment web interface [19]. Genome sequence was submitted to GenBank under BioProject no. SAMN17121013.
Statistical analysis. Using PCR as the gold standard, we calculated the sensitivity and specificity of CARBA 5. We calculated 95% confidence intervals (CI) for each of these measures using VassarStats (http://vassarstats.net/prop1.html). If CARBA 5 detected the presence of a carbapenemase, but it differed from the carbapenemase detected by PCR, we considered the result a false positive (FP).