Bacterial isolates: A total of 260 isolates resistant to imipenem and meropenem isolated from blood and respiratory specimens were included in this study prospectively from all age groups. Of these 57, 85, 60 and 58 were carbapenem resistant E.coli, Klebsiella spp., Pseudomonas aeruginosa and Acinetobacter baumannii respectively. All the study isolates were identified up to species level as per standard protocols which includes, mannitol motility, triple sugar iron, citrate utilization and indole production test for Enterobacteriaceae; Oxidase, catalase, nitrate reduction, sugar fermentation such as lysine, ornithine and arginine tests for Pseudomonas aeruginosa and Acinetobacter baumannii [16]. They were sourced from various clinical specimens collected between 2012 and 2014 at Christian Medical College, Vellore, South India. The screening for the carbapenem resistance for the study isolates were carried out using Kirby Bauer disc diffusion for imipenem (10μg) and meropenem (10μg) along with the routinely tested antimicrobial agents such as cephalosporins (cefotaxime, ceftazidime); β-lactam/β-lactamase inhibitors (Piperacillin/tazo bactam, cefoperazone/sulbactam); carbapenems (imipenem, meropenem); fluoroquinolones (ciprofloxacin, levofloxacin); PB300 units and tigecycline (For Klebsiella spp. and Acinetobacter spp.) and interpreted according to CLSI guidelines M100-S22, M100-S23, M100-S24.

Modified Hodge Test: MHT was performed only for E.coli and Klebsiella spp. as per CLSI recommendations M100-S21 [8]. Lawn culture of 1 in 10 diluted, 0.5 McFarland turbidity of susceptible ATCC 25922 E.coli was made and ertapenem disc of 10 μg/ml was placed in the centre. Test isolate and controls were streaked onto the lawn culture from centre to periphery. Following 24 hours incubation, presence of clover leaf like indentation was taken as positive for the carbapenemase production and absence of indentation was taken as negative. K.pneumoniae ATCC BAA 1705 and K. pneumoniae ATCC BAA 1706 were used as positive and negative controls for all the assays respectively. Presence of clover leaf like indentation along the streak line of the test isolate indicating the growth of the carbapenem susceptible strain (E.coliATCC 25922) was taken as a positive for carbapenemase production. While, absence of indentation was taken as a negative result for carbapenemase production.

Carba NP test: For all carbapenem resistant E.coli, Klebsiella spp. and P. aeruginosa, Carba NP test was done and a modified protocol of CarbAcineto NP was used for A. baumannii as recommended by Nordmann and Poirel [6–8]. Since Class D Oxacillinases are more prevalent in Acinetobacter spp., a modified protocol using 5 M NaCl was used instead of B-PER II lysis buffer, to avoid any buffer effect, as Oxa enzymes possess weak carbapenemase activity. All the study isolates subjected to CarbaNP and CarbAcineto NP were grown on Mueller Hinton agar plates for 24 hours and the isolated colonies were used for testing. K. pneumoniae ATCC BAA 1705 and K. pneumoniae ATCC BAA 1706 were used as positive and negative controls in all the assays, respectively. CarbaNP test was performed in two 1.5ml Eppendorf tubes labeled as control and test. 0.1 ml of B-PER II Lysis buffer (Thermo Scientific Cat.no. 78260) was taken in both the tubes. 10 μl loopfull of colonies to be tested was inoculated in both the tubes and vortexed briefly. Following that, 0.1ml of phenol red solution with 6μg/ml imipenem was added in the tube labeled as test and 0.1 ml of phenol red solution without imipenem was added in the tube labeled as control. The tubes were incubated at 37^oC for upto two hours. Every 15 minutes once, the tubes were checked for the colour change. Change of the phenol red indicator from red to yellow was taken as a positive for carbapenemase producers. Absence of colour change from red to yellow was taken as negative result. The observation of colour change was taken by two independent readers. Imipenem+cloxacillin test: Combination disc test of imipenem and imipenem plus cloxacillin was used to demonstrate the synergy of cloxacillin with imipenem in order to differentiate between carbapenemase and non-carbapenemase producers (only for P.aeruginosa) [17]. Cloxacillin of 4000μg/ml was used alone and in combination with imipenem (10 μg/ml). Zone diameter of >5mm difference between imipenem alone and in combination of imipenem plus cloxacillin was taken as a positive result for hyper AmpC producers.

Multiplex PCR for Detection of Carbapenemase genes

All the test isolates were grown on blood agar overnight and whole genomic DNA was extracted by boiling lysis method [18]. Conventional multiplex PCR was done for the detection of carbapenemase genes such as bla_SPM, bla_IMP, bla_VIM, bla_NDM, bla_KPC and bla_{OXA-48 like}. In addition, bla_OXA-51, bla_OXA-23 and bla_OXA-24 genes were screened for A. baumannii isolates. Further, presence of bla_OXA-51 was carried out to confirm that all studied isolates are A.baumannii [19–25]. The amplicons were visualized in 2% agarose gel with staining of ethidium bromide. Known positive controls for appropriate genes were used (Courtesy: IHMA, Inc., USA). Primers and cycling conditions are mentioned in [Table/Fig-1]. Furthermore, a total of 30 carbapenem susceptible isolates of E.coli and 30 carbapenem susceptible Klebsiella spp. were included in the assay as a comparator.

Statistical analysis of the categorical data was done using SPSS Version 16.0. The sensitivity and specificity of CarbaNP test were calculated and compared with PCR results as a gold standard, for the individual carbapenemase encoding genes respectively.

CarbaNP was found to be positive in 88% (n=50/57) of E.coli, 81% (n=69/85) of Klebsiella spp., 38% (n=23/60) of P. aeruginosa and CarbAcineto NP was positive in 81% (n=47/58) of A. baumannii. MHT was found to be positive in 89% (n=51/57) of E.coli and 81% (n=69/85) of Klebsiella spp. [Table/Fig-2,3].

The multiplex PCR for the detection of carbapenemase genes revealed the presence of 57 bla_OXA-48like producers, 53 bla_NDM producers, 11 bla_VIM producers and few co-producers of bla_NDM+bla_OXA-48like, bla_NDM+bla_VIM and bla_NDM+bla_OXA-23 as listed in [Table/Fig-3,4]. Synergism testing of imipenem plus cloxacillin for Carba NP negative isolates of P. aeruginosa revealed that, out of 37 isolates, 24 were positive for AmpC plus porin loss detection.

The overall sensitivity and specificity of CarbaNP were found to be 84 % and 100% respectively. The sensitivity and specificity for MHT was found to be 96 % and 100 % for bla_{OXA-48 like} producers and 90% and 100% for bla_NDM producers. While, sensitivity and specificity of CarbaNP shows 77% and 100% for bla_{OXA-48 like} producers and 94% and 100% for bla_NDM producers. On the other hand, CarbAcineto NP showed 81% sensitivity and 100% specificity rates [Table/Fig-4].

In comparison with the CLSI recommendations and the study finding, there is a significant variation observed in the sensitivity and specificity. In case of MHT, CLSI states that there is >90% sensitivity and specificity for bla_KPC carbapenemase and 11% for metallo β-lactamases. In this study, we however found a sensitivity of 96% and specificity of 100% for non metallo β-lactamases. In addition, we found 90% sensitivity for bla_NDM like producers in contrast to 11% reported by CLSI.

Likewise, for CarbaNP test, CLSI states the sensitivity and specificity of >90% for metallo β-lactamases and up to 11% for bla_OXA-48 like producers. Although we found 94 % sensitivity for metallo β-lactamases and 77% sensitivity for bla_OXA-48 like producers unlike 11% reported by CLSI.

Upon testing the carbaNP negative P.aeruginosa for porin loss plus AmpC hyper producers, using imipenem with cloxacillin showed 65% (n=24/37) were positive, while 35 % of (n=13/37) isolates were still negative, which may be due to efflux pumps or loss of porins. These results are in accordance with previously reported studies, in which carbapenem resistance due to carbapenemases is found to be less in non-carbapenemase producers beacuse of porin loss and ampC hyper-production [26]

Interestingly, CLSI does not mention CarbAcineto NP. This gives an even better sensitivity and specificity of 81% and 100% respectively than CarbaNP test. The information derived from this study varies extensively when compared to the CLSI observations, indicating the influence of the type of carbapenemase prevalent in a geographic region and the suitable methodology used for enhanced detection of carbapenemase.

Remarkably, EUCAST neither recommends MHT or CarbaNP, but rather recommends specific enzyme based detection methods for the detection of resistance mechanisms such as using aminophenyl boronic acid, cloxacillin and EDTA for different classes of carbapenemases [27]. It is worthwhile to evaluate and compare these methodologies to with CLSI recommendations for any advantage for Indian settings as mentioned in [Table/Fig-5].

Moreover, studies have evaluated the performance of MHT in detecting metallo β-lactamases and found the sensitivity and specificity rates ranging from 58 to 78% and 57% to 100% respectively [28,29]. This is much higher than that mentioned by CLSI. Furthermore, the addition of Zinc sulfate to the test medium increases the sensitivity of the bla_NDM detection from 50% to 85.7%, but lacks specificity [11]. In addition, Neo-Sensitabs has also been evaluated and shown to be more efficient in detecting carbapenemases [30].

In spite of CarbaNP being simple and easy, its downside includes false positive and false negatives results due to the change in the pH of the test solution, the inoculum size, time of incubation, test isolate growth medium where the inoculum was taken, mucoid phenotypes and presence of class bla_{OXA-48 like} enzymes with low carbapenemase activity [31,32]. In addition, the colour change of red to orange-yellow must be detected to interpret the test and thus it is dependent on the observer. These conditions should be taken into consideration when interpreting the results. Finally in terms of time, the CarbaNP takes two hours, while the MHT requires the 24 hours time period for the carbapenemase detection.

This study reports 90% sensitivity and specificity for bla_{oxa-48 like} enzymes by MHT and >90% sensitivity and specificity for bla_NDM producers by CarbaNP method. Even though MHT was estimated by CLSI to give a sensitivity of 11% for bla_NDM producers, much higher rates were found in Indian settings as in this study where it was around 90% sensitive and 100% specific. Similarly, bla_{oxa-48 like} enzymes detection by CarbaNP was >70% sensitivity and specificity while CLSI observation was only 11%. Although, CLSI recommends CarbaNP for A.baumannii, it is well known that CarbAcinetoNP has a higher sensitivity and specificity for A.baumannii [14].

Other carbapenemase encoding genes were not looked for the isolates that are resistance to carbapenem but negative for the carbapenemase genes tested in PCR.

Detection of carbapenem resistant organism (CRO) using different methodologies demonstrates variations in the sensitivity and specificity across different geographical regions. Overall, it was evident that MHT is comparatively better for the detection of class D Oxa enzymes but not for metallo β lactamases. On the contrary, CarbaNP is excellent rapid test in detecting metallo β lactamases but not for class D Oxa enzymes. Henceforth, it is advisable to screen the isolates through CarbaNP for the rapid detection, followed by MHT for any region where class D Oxa-enzymes are more prevalent. Eventually, combined testing of both Carba NP and MHT will certainly help in detecting the most of the carbapenemases. Moreover, this study signifies the importance of knowing the predominant carbapenemase enzymes seen in a specified geographical region before deciding upon a suitable phenotypic screening technique. Further work should investigate the EUCAST proposed carbapenemase detection algorithm to establish whether this better suits the Indian setting.