Validation of a colistin plasma concentration breakpoint as a predictor of nephrotoxicity in patients treated with colistin methanesulfonate

Abstract Nephrotoxicity limits the effective use of colistin for the treatment of multidrug-resistant Gram-negative bacteria (MDR-GNB) infections. We previously defined a steady-state colistin plasma concentration ( C ss ) of 2.42 mg/L that predicted nephrotoxicity at end of treatment (EOT). The objective of this study was to validate this breakpoint in a prospective cohort. This was a multicentre, prospective, observational study conducted at three hospitals with a cohort of patients treated for MDR-GNB infection with colistin methanesulfonate from September 2011 until January 2015. Nephrotoxicity was evaluated at Day 7 and at EOT using the RIFLE criteria. C ss values were measured and analysed using HPLC. Taking the previously defined breakpoint for colistin concentration as a criterion, patients were divided into two groups ( C ss , ≤2.42 mg/L vs. >2.42 mg/L). Sixty-four patients were included. Seven patients (10.9%) had a C ss  > 2.42 mg/L and were compared with the remaining patients. Bivariate analysis showed that patients with a C ss  > 2.42 mg/L were older and had a significantly higher incidence of nephrotoxicity at Day 7 and EOT. Although not statistically significant, nephrotoxicity occurred earlier in these patients (6.2 days vs. 9.2 days in patients with lower C ss ; P  = 0.091). Multivariate analysis of nephrotoxicity showed that C ss  > 2.42 mg/L was the only predictive factor. Nephrotoxicity was more frequent and occurred earlier in patients with colistin plasma concentrations higher than the previously defined breakpoint (2.42 mg/L). Colistin therapeutic drug monitoring should be routinely considered to avoid reaching this toxicity threshold and potential clinical consequences.