Weight-based polymyxin B dosing for mortality and nephrotoxicity: an exploratory dose-response study

The optimal weight-based maintenance dose of polymyxin B is not established. We evaluated dose-response relationships for 28-day mortality and nephrotoxicity and explored a dose threshold.

In this single-centre retrospective cohort, hospitalized adults received a first intravenous polymyxin B course (June 2025 to February 2026); the maintenance dose (mg/kg/day) was the exposure. Main outcomes were 28-day mortality and nephrotoxicity, defined by a creatinine-clearance criterion (≥50% fall, or  ≥ 25% with pre-existing dysfunction); renal injury was also graded by the risk, injury, failure, loss and end-stage kidney disease (RIFLE) classification. We used clinically selected dose categories (2.5, 2.75 and 3.0 mg/kg/day), exploratory classification and regression tree (CART) analysis, multivariable logistic regression and a desirability of outcome ranking.

The mortality cohort comprised 208 patients (28-day mortality 41.3%) and the nephrotoxicity cohort 163 (nephrotoxicity 49.7%). Mortality declined across dose categories in univariable analysis (trend P = 0.05) but was not independently associated with dose after adjustment. Nephrotoxicity increased with dose, reaching 73% at  ≥3.0 mg/kg/day; CART identified a threshold at 2.74 mg/kg/day, above which the RIFLE grade was worse (62.3% of comparisons; P = 0.006). Multivariable analysis confirmed this (adjusted odds ratio 2.60, 95% confidence interval 1.25-5.43). The desirability of outcome ranking did not differ across dose groups.

A higher polymyxin B maintenance dose was associated with increased nephrotoxicity, particularly above approximately 2.75 mg/kg/day. The mortality estimates were imprecise and did not exclude a clinically important effect. This single-centre threshold is hypothesis-generating and requires validation before it could inform a clinical dose ceiling.