Polymyxin B Etest ^® compared with gold-standard broth microdilution in carbapenem-resistant Enterobacteriaceae exhibiting a wide range of polymyxin B MICs

Objectives: Polymyxins have been revitalized to combat carbapenem-resistant Enterobacteriaceae (CRE). However, evaluating the activity of these agents by traditional broth dilution methods is not practical for busy clinical laboratories. We compared polymyxin B activity using two quantitative susceptibility testing methods, Etest ^® and broth microdilution (BMD), against CRE isolates from patients at an academic medical centre. Methods: Polymyxin B activity against 70 CRE clinical isolates was determined by Etest ^® according to the manufacturer and by BMD according to CLSI guidelines. Pseudomonas aeruginosa ATCC ^® 27853 and Escherichia coli NCTC 13846 served as quality control strains. The EUCAST colistin susceptibility breakpoint of Enterobacteriaceae (≤2 mg/L) was used. Essential agreement was isolates with an MIC within 1 log ₂ dilution over total isolates. Categorical agreement was number of isolates in the same susceptibility category (susceptible or resistant) over total isolates. Major and very major error rates were calculated using number of susceptible and number of resistant isolates, respectively, as the denominator. McNemar's test was used for determining a difference in susceptibility between methods. Results: The CRE isolates were primarily Klebsiella spp. (49%) and Enterobacter spp. (36%). Polymyxin B susceptibility was significantly higher by Etest ^® compared with BMD (97% versus 77%; p 0.0001). Categorical agreement was 80%, but essential agreement was low (10%). False non-susceptibility was never observed by Etest ^® (BMD reference), but the very major errors were high (88%). Conclusions: Etest ^® reporting of false susceptibility may result in inappropriate antibiotic use and treatment failure clinically. We do not recommend using Etest ^® for polymyxin B susceptibility testing for routine patient care.