Colistin susceptibility testing: time for a review

Sir, Colistin has re-emerged as an important antimicrobial in recent times owing to limited therapeutic options against carbapenem-resistant Gram-negative bacteria.1,2 Current guidelines (BSAC, CLSI and EUCAST) recommend routine colistin susceptibility testing by estimation of MIC because the disc diffusion test does not reliably detect low-level resistance.3–5 Broth microdilution (BMD) is widely used as a method of MIC estimation in Europe and the USA. Colistin exhibits a varying degree of adherence to organic and inorganic materials due to its polycationic nature, resulting in loss during experimental conditions.6 Also, polysorbate 80 (P-80), a surfactant widely used as a dispersing agent in BMD panels, may influence the free drug concentration of colistin and hence MIC results.7 We evaluated the impact of the use of different BMD panels and the presence of P-80 on colistin MIC estimation. A total of 146 clinical isolates collected from a variety of sources and stored at −70°C were evaluated in this study. The isolates included 56 Pseudomonas aeruginosa, 29 Acinetobacter spp. and 61 Enterobacteriaceae. The MIC testing was carried out on two different types of polystyrene microtitre trays (MTTs), namely non-coated V-bottom MTTs (NMTTs; costar 3896; Corning, NY, USA) and tissue-culture-coated round-bottom MTTs (TCMTTs; costar 3799; Corning). The MICs of colistin for the isolates were determined using the CLSI broth dilution method using colistin sulphate. MIC determination was carried out by using an initial bacterial inoculum of 5 × 105 cfu/mL in Mueller–Hinton broth with or without P-80 (final P-80 concentration of 0.002%) on both types of MTT. The experiments were done in triplicate, and quality control was assured by concurrent testing of P. aeruginosa ATCC 27853 as a control, with all results within the range published by the CLSI.4 MICs for the isolates in both types of MTT with or without P-80 are shown in Table 1. The NMTT MICs (mean 0.54 ± 0.58) were significantly lower than the TCMTT MICs (mean 2.84 ± 1.93) (P < 0.0001; 95% CI –2.5 to –2.1). The tissue coating on MTTs, achieved by means of excess negative electric charge, resulted in an overall 5.3-fold increase in MIC value, probably due to decreased free colistin concentration within the microwells. The differences in MIC results were seen among all types/groups of isolates (3.2, 5.5 and 9.4, respectively, for P. aeruginosa, Enterobacteriaceae and Acinetobacter spp.). The addition of P-80 to NMTTs significantly decreased the colistin MIC (mean 0.09 ± 0.09) by 6-fold (P < 0.0001; 95% CI 0.4–0.5). Although there was a relatively smaller decrease (1.24-fold) in the mean MIC determined using TCMTTs with added P-80 (mean 2.3 ± 1.5), this was also statistically significant (P < 0.001; 95% CI −0.31 to −0.75). Comparing the MICs determined using NMTTs and TCMTTs containing P-80, there was an even bigger difference in the MIC result than without P-80. There were 25.6-fold differences in the mean MIC results between NMTTs and TCMTTs containing P-80, compared with just 5-fold differences without P-80 (P < 0.0001; 95% CI –2.35 to –2.1). Colistin MICs for isolates Colistin MICs for isolates In conclusion, colistin MIC results were greatly influenced by the characteristics of the MTTs. Also, the addition of a commonly used surfactant agent such as P-80 not only significantly altered the result in a single type of MTT but also exponentially exacerbated the difference when tested on different types of MTT panel. The effect of the make-up of MTTs and the presence of P-80 on MIC results were similar among all types/groups of isolates (i.e. P. aeruginosa, Acinetobacter spp. and Enterobacteriaceae). A recent study comparing BMD (with or without P-80), the Etest and the agar dilution method against 50 clinical isolates of multidrug-resistant Gram-negative bacilli showed significant variability among colistin MIC results.8 At this present time when the therapeutic use of colistin is on the increase with an anticipated rise in colistin resistance, a review of the methodology for colistin MIC testing is urgently needed.9 This research work received support from ‘The Showering Fund’, Department of Pathology, North Bristol NHS Trust, Bristol, UK. None to declare. This research work was presented as an abstract at the Fifty-second Interscience Conference on Antimicrobial Agents and Chemotherapy, San Francisco, CA, USA, 2012 (Abstract D-793).