Sensitivity to colistin: evaluation of available interpretive criteria of disc diffusion test

Abstract number: 1733_1140

Galani I., Kontopidou F., Rekatzina P., Deliolanis J., Souli M., Antoniadou A., Giamarellou H.

Objectives: Increased antibiotic resistance in Gram-negative bacteria isolated in ICUs of tertiary-care hospitals, has augmented the use of colistin (CL) in recent years, since it is the only available active agent against these isolates. However CLSI and EUCAST does not currently provide interpretive criteria for disc susceptibility as there is very little data on the accuracy of disc testing methods for CL. In this study, we evaluate the accuracy of two available interpretive criteria of disc susceptibility testing for CL, compared to E-test.

Methods: A total of 409 clinical isolates of Acinetobacter baumannii [AB] (178), K. pneumoniae [KP] (185) and Escherichia coli [EC] (46) were included in the study. All strains were isolated from hospitalised patients in 11 different tertiary hospitals. Isolates were tested by E-test for susceptibility to CL, and results were compared with those obtained by disc susceptibility testing method (DD). Two breakpoints were evaluated, Gales et al. (JCM, 2001; 39: 183–90) (≤11 and ≥14 mm) and BSAC 2006 (≤14 and ≥15 mm) and compared to MIC breakpoints established by BSAC (>4 and ≤4 mg/L) for Enterobacteriaceae and Acinetobacter spp.

Results: CL displayed good activity against AB (MIC₉₀ 0.5 mg/L), and EC (MIC₉₀ 0.625 mg/L) but was less active against KP (MIC₉₀ 32 mg/L). Forty-six (24.9%) KP, six (3.4%) AB and two (4.3%) EC strains were resistant (R) to CL according to the E-test method. Totally, 16.4% and 5.5% of CL-resistant isolates were falsely reported as susceptible (S) when evaluated by the two available criteria, respectively. Applying the BSAC breakpoints decreased the very major error (R by E-test but S by DD) to 1.5% from 2.9% compared to Gales et al criteria. On the contrary major discrepancies (S by E-test but R by DD) were found in 45% of strains tested with BSAC recommendations while 11.5% of minor errors (R by E-test and intermediate by DD) were found during application of Gales et al criteria.

Conclusion: Use of the BSAC criteria decreased the very major errors but significantly increased the major errors when compared with the breakpoints proposed by Gales et al. Disc susceptibility testing methods are unreliable at detecting CL resistance. E-test or dilution methods should be the method of choice for susceptibility testing of CL as zone diameter interpretive criteria exhibiting good correlation with MICs seems difficult to establish.