New Predictive Equation for the Estimation of Plasma Concentrations of Formed Colistin in Patients Treated With Colistimethate Sodium for Multidrug-Resistant Gram-Negative Bacterial Infections

Background: The clinical use of colistin methanesulphonate (CMS) is limited by potential nephrotoxicity. The selection of an efficient and safe CMS dose for individual patients is complicated by the narrow therapeutic window and high interpatient pharmacokinetic variability. In this study, a simple predictive equation for estimating the plasma concentration of formed colistin in patients with multidrug and extremely drug-resistant gram-negative bacterial infections was developed. Methods: The equation was derived from the largest clinical cohort of patients undergoing therapeutic drug monitoring (TDM) of colistin for over 8 years in a tertiary Spanish hospital. All variables associated with Css,avg were selected in a multiple linear regression model that was validated in a second cohort of 40 patients. Measured Css,avg values were compared with those predicted by our model and a previous published algorithm for critically ill patients. Results: In total, 276 patients were enrolled [the mean age was 67.2 (13.7) years, 203 (73.6%)] were male, and the mean (SD) Css,avg was 1.12 (0.98) mg/L. Age, gender, estimated glomerular filtration rate, CMS dose and frequency, and concomitant drugs were included in the model. In the external validation, the previous algorithm appeared to yield more optimized colistin plasma concentrations when all types of Css,avg values (high and low) were considered, while our equation yielded a more optimized prediction in the subgroup of patients with low colistin plasma concentrations (Css,avg <1.5 mg/L). Conclusions: The proposed equation may help clinicians to better use CMS among a wide variety of patients, to maximize efficacy and prevent nephrotoxicity. A further prospective PK study is warranted to externally validate this algorithm.