Colistin and Polymyxin B Minimal Inhibitory Concentrations Determined by Etest Found Unreliable for Gram-Negative Bacilli

INTRODUCTION

Resistance to polymyxins is being increasingly detected worldwide, and an accurate method of susceptibility determination remains elusive. Colistin (polymyxin E) and polymyxin B diffuse poorly into agar plates, potentially resulting in inaccurate Etest and disk diffusion results.1 In addition, the adsorption of the drug to plastic and glass testing materials and prevalence of heteroresistant subpopulations in organisms like Enterobacter and Acinetobacter spp can cause skipped wells and trailing endpoints in broth microdilution, rendering this reference method sometimes difficult to interpret.2 Currently, our hospital microbiology laboratory performs only the colistin Etest when a colistin minimal inhibitory concentration (MIC) is requested for a clinical bacterial isolate, and the results are reported as “research use only.” Polymyxin B susceptibility testing is not performed. The goal of our in-house quality assessment study was to compare colistin and polymyxin B MICs from Etest with MICs from broth microdilution to determine if colistin MICs are accurately being reported by Etest.

RESULTS

Using broth microdilution, 16/143 (11%) isolates were nonsusceptible to colistin, and 12/143 (8%) were nonsusceptible to polymyxin B. Colistin and polymyxin B MICs by broth microdilution were in agreement (±1 twofold dilution) for 132/143 (92%) of isolates and in categorical agreement (same interpretive category) for 139/143 (97%) of isolates. The 4 isolates not in categorical agreement were all A baumannii, colistin-resistant but polymyxin B-susceptible by broth microdilution.

With Etest, 4/143 (3%) isolates were nonsusceptible to colistin and 7/143 (5%) were nonsusceptible to polymyxin B.

Overall, the rate of discordance between Etest and broth microdilution was high with both colistin and polymyxin B. Essential agreement between methods was unacceptable (<90%) for colistin (59%) and polymyxin B (61%). Categorical agreement performance rates were found to be acceptable (≥90%) between methods when testing P aeruginosa with colistin (95%) and A baumannii with polymyxin B (92%), but all other organism/drug combinations had unacceptable categorical agreements (Table 1).

When isolates were tested with colistin by Etest, very major errors (ie, false susceptibility) were found in 10% (14/143) of isolates (or 88% false-susceptible results among the 16 resistant isolates by broth microdilution). For polymyxin B MICs determined by Etest, the overall very major error rate was 8% (11/143) (or 92% false-susceptible results among the 12 resistant isolates by broth microdilution). No very major errors were detected in the 44 P aeruginosa isolates (Tables 2 and 3).

DISCUSSION

Emerging bacterial resistance to most commonly used antibiotics has necessitated the increased use of polymyxins for treatment of multidrug-resistant gram-negative bacteria. With this resurgence of polymyxin usage, it is important to establish reliable, standardized, and reproducible susceptibility testing methods for these drugs. While broth microdilution and agar dilution have been established by CLSI as reference methods for susceptibility testing of the polymyxins,4 the current standards and breakpoints are under review by CLSI and EUCAST with the Transatlantic Taskforce on Antimicrobial Resistance.7 Broth microdilution has been shown to occasionally produce unreliable and unreadable MICs because of resistant bacterial subpopulations,2 and agar dilution is time-consuming and infrequently used.7 The Etest method of susceptibility determination is widely recognized as unreliable when used with polymyxins, as the large cationic peptides of polymyxins diffuse poorly in agar.2 A number of studies have found that comparison of broth microdilution and Etest methods for polymyxins produces unacceptably high rates of very major errors, with some as high as 39.3%,8 and that concordance between Etest and the reference method of choice is often below the acceptable standard of ≥90%.8,9 Tan and Ng found that most major errors occurred with P aeruginosa and that varying the incubation time by as little as 4 hours can have an impact on Etest MICs with these organisms.9 We speculate that the higher readings we observed with Etest compared to broth microdilution with P aeruginosa could be attributed to the reading of Etest plates during variable times throughout the 16-20 hour window recommended by the manufacturer, as well as the spreading colony morphology that could have been inadvertently interpreted as a slightly higher MIC. Despite the problems associated with broth microdilution and heteroresistance, we found that the broth microdilution method more reliably identified resistant populations in our isolate collection and therefore should be used in lieu of Etest.

CONCLUSION

Our study highlights a high rate of discrepancy between polymyxin susceptibility test methods. Etest very major error rates for colistin were 10% overall (88% false susceptible results among the 16 resistant isolates by broth microdilution) and 8% overall for polymyxin B (92% false susceptible results among the 12 resistant isolates by broth microdilution). These data strongly support other published findings that any colistin or polymyxin B MIC should not be determined by Etest (because of major problems with false susceptibilities) and results should not be reported for clinical use. If polymyxin susceptibility is requested, testing by broth microdilution should be performed.