Evaluation of different methods for in vitro susceptibility testing of colistin in carbapenem resistant Gram-negative bacilli

Abstract

Introduction. The increasing antibiotic resistance like the advent of carbapenem resistant Enterobactarales (CRE), Carbapenem Resistant Acinetobacter baumanii (CRAB), and Carbapenem Resistant Pseudomonas aeruginosa (CRPA) has led to to the use of toxic and older drugs like colistin for these organisms. But worldwide there is an increase in resistance even to colistin mediated both by chromosomes and plasmids. This necessitates accurate detection of resistance. This is impeded by the unavailability of a user-friendly phenotypic methods for use in routine clinical microbiology practice. The present study attempts to evaluate two different methods – colistin broth disc elution and MIC detection by Vitek two in comparison to CLSI approved broth microdilution (BMD) for colistin for Enterobactarales, Pseudomonas aeruginosa , and Acinetobacter baumanii clinical isolates. Methods. Colistin susceptibility of 6013 carbapenem resistant isolates was determined by BMD, Colistin Broth Disc Elution (CBDE), and Vitek two methods and was interpreted as per CLSI guidelines. The MIC results of CBDE, Vitek two were compared with that of BMD and essential agreement (EA), categorical agreement (CA), sensitivity, specificity, very major error (VME), major error (ME) and Cohen’s kappa (CK) was calculated. The presence of any plasmid-mediated colistin resistance (mcr-1, 2, 3, 4 and 5) was evaluated in all colistin-resistant isolates by conventional polymerase chain reaction. Results. Colistin resistance was found in 778 (12.9 %) strains among the carbapenem resistant isolates. Klebsiella pneumoniae had the highest (18.9 %) colistin resistance by the BMD method. MIC of Vitek two had sensitivity ranging from 78.2–84.8% and specificity of >92 %. There were 171 VMEs and 323 MEs by Vitek two method, much more than CLSI acceptable range. The highest percentage of errors was committed for Acinetobacter baumanii (27.8 % of VME and 7.9 % ME). On the other hand, the CBDE method performed well with EA, CA, VME and ME within acceptable range for all the organisms. The sensitivity of the CBDE method compared to gold standard BMD varied from 97.5–98.8 % for different strains with a specificity of more than 97.6 %. None of the isolated colistin resistant organisms harboured mcr plasmids. Conclusion. As BMD has many technical complexities, CBDE is the best viable alternative available for countries like India. A sensitive MIC reported by Vitek two needs to be carefully considered due high propensity for VMEs particularly for Klebsiella spp.