PCR design

We developed specific real-time PCRs for detection of KPC, VIM, NDM, OXA-48 and IMP. For design of the primers, sequence variations of carbapenemase genes published at http://www.lahey.org/Studies were taken into account, along with synonymous mutations (Additional file 1: Table S1). Detection of CTX-M [9] was included for optional use since it complements OXA-48 for resistance to extended spectrum cephalosporins, a characteristic that is intrinsic to the other carbapenemases. For verification of newly designed PCRs, sequencing was done on larger gene fragments. For NDM sequencing primers 5′-GCGAAAGTCAGGCTGTGTTG-3, and ‘5′-CATTAGCCGCTGCATTGATG-3′, were used, and for IMP sequencing primers 5′-GGCGGAATAGAGTGGCTTAATTCTC-3′, and 5′-CGTACGGTTTAACAAAACAACCACC-3′ For each separate carbapenemase gene, PCR primers and probe concentrations were optimized. Multiplex combinations were compared with single PCRs in presence of the internal control Phocine Herpes Virus (PhHV).

Bacterial strains

PCRs were optimized and validated using the following control strains: a KPC producing Klebsiella pneumonia, a VIM-2 producing Pseudomonas aeruginosa, an IMP-18 producing P. aeruginosa, an IMP-28 producing K. pneumoniae, a NDM-1 positive K. pneumoniae, and an OXA-48 positive K. pneumoniae isolate.

Testing of PCR specificity was carried out on the following strain collection. The test collection of 86 isolates, included 58 carbapenemase producing isolates, and 28 carbapenemase negative controls. The 58 carbapenemase positive isolates consisted of 45 K. pneumoniae, 4 E. coli, 3 Enterobacter species, 2 P. mirabilis, 2 Citrobacter species, and 2 P. aeruginoasa isolates producing the following carbapenemases: 20 KPC-2/3, 4 KPC plus VIM, 21 VIM, 4 NDM-1, 2 IMP, and 7 OXA-48. The 28 carbapenemase negative controls consisted of 10 K. pneumoniae, 2 E. coli and 16 Enterobacter isolates, producing either an ESBL (20 isolates) or an AmpC beta-lactamase (Additional file 2: Table S2). As the reference test for presence of beta-lactamases, PCR and sequencing was used [10].

PCR evaluation

To determine the sensitivity and specificity of PCR, 86 unrelated test isolates were investigated both from agar plate and broth. Two protocols were followed; protocol 1) simultaneous multiplex detection of OXA-48, VIM, IMP, NDM and KPC. In this PCR reaction, the OXA-48 probe was labelled with FAM, and the other carbapenemases with VIC. Protocol 2) 3 multiplex PCRs for identification of respectively OXA-48/CTX-M, VIM/IMP, and NDM/KPC. PhHV was labelled with NED. Fluorescent labels of carbapenemase genes were FAM, and VIC, respectively (Additional file 3: Table S3).

PCR

Strains were grown on MacConkey agar (Oxoid), and in Brain Heart broth, both supplemented with ertapenem (0.125 mg/l). One colony was taken from the plate and 50 μl from the broth. Both were suspended in 100 μl Extraction Solution (SIGMA, E7526). Mixtures were incubated at 95°C for 10 minutes, cooled to room temperature, 100 μl Dilution Buffer (SIGMA, D5688) was added and mixed. PCR reactions were carried out using PCR-ReadyMix™ (SIGMA, E3004). Amplification was performed on ABI 7500 Real-Time PCR system (LifeTech, Glasgow, UK). The temperature profile included initial denaturation of 4 min. at 94°C, followed by 50 cycles (40 cycles, protocol 2) of 94°C for 15 sec., and 60°C for 1 min.

Evaluation of PCR

PCRs on serial dilutions of template showed Cq (Cycle threshold) values ranging from 19.5-41.5 for single PCRs, and Cq values of 20.2-46.3 for multiplex reactions. The mean difference in Cq value between single and multiplex reactions was 0.9, with a tendency towards an increased difference with decreasing template concentrations. Mixing of templates did not affect Cq values. The lower limit of detection was approximately 10 colony forming units. The amplification efficiencies in multiplex format ranged from 85% for VIM to 82% for IMP (Figure 1). We concluded that the PCRs were compatible.

The PCR assay according to protocol 1 was carried out on the test collection of 86 strains (Additional file 2: Table S2). All carbapenemase producing isolates were positive in the PCR with Cq values ranging from 15.4 to 23 for isolates from agar, and Ct values 20.1 to 28 from broth cultures (Figure 2A), corresponding to a sensitivity of 100%. All 26 carbapenemase negative control isolates were negative in the PCR, corresponding to a specificity of 100%.

To evaluate the capacity of the real-time PCR to identify carbapenemase genes, protocol 2 was used. Twenty-five strains were PCR positive for VIM, 2 for IMP, 4 for NDM, 24 for KPC, 7 for OXA-48 and 18 for CTX-M. Four strains were positive for both VIM and KPC and 2 strains both for OXA-48 and CTX-M (Figure 2B). Ct values ranged from 15.1-22.5 (VIM), 13.2-20.6 (KPC), 17.8-20.1 (NDM), 16.8-23.8 (OXA-48) and, 14.7-17.2 (CTX-M). The results were 100% concordant with listed strain characteristics.

Multi-centre comparison

Thus, both inter- and intra-laboratory, PCR sensitivity and specificity corresponds to 100%. Two laboratories also subjected the 20 control strains to Check-points PCR assay (laboratory 3) and Check-MDR Carba Assay (laboratory 6). The results were concordant with our PCR assay, except that In the Check-MDR Carba Assay IMP-28 was missed.