Comparison of susceptibility testing methods for determining the activity of colistin against Gram-negative bacilli of clinical origin.

Abstract

Despite being in clinical use for decades, colistin susceptibility testing remains challenging because of its inherent cationic properties. We aimed to compare the performance characteristics of different methods for testing susceptibility to colistin in a series of clinical isolates of Gram-negative bacilli. One hundred and nine clinical isolates of Klebsiella pneumoniae (n=34), Escherichia coli (n=20), Acinetobacter baumannii (n=17) and Pseudomonas aeruginosa (n=38) were studied for colistin susceptibility using broth microdilution (BMID), broth macrodilution (BMAD), agar dilution (AD) as well as disc-diffusion (DD) utilizing two different commercial disc sources. By using BMID as reference method, 88 and 21 isolates were found to be colistin susceptible and resistant, respectively. Overall, acceptable essential agreement (EA) and categorical agreement (CA) were observed between BMAD and reference method (100 %). Whereas the AD method revealed the lowest rate of EA (61.7, 11.7, 5.0 and 5.2 % for K. pneumoniae, A. baumannii, E. coli and P. aeruginosa, respectively), it showed acceptable or near acceptable CA for K. pneumoniae (100 %), E. coli (100 %) and A. baumannii (88.2 %) isolates but not for P. aeruginosa (13.1 %). DD failed to detect resistance in colistin-resistant (colR) P. aeruginosa (n=5, very major errors of 100 %) but successfully identified all high-level colistin-resistant A. baumannii and K. pneumoniae isolates. We found BMAD to be very reliable for colistin MIC determination. Methods AD and DD should not be used for colistin susceptibility testing in P. aeruginosa isolates as these are associated with false-resistant and -susceptible results, respectively.