Nephrotoxicity of Vancomycin in Combination With Beta-Lactam Agents: Ceftolozane-Tazobactam vs Piperacillin-Tazobactam

Background Vancomycin (VAN)-associated acute kidney injury (AKI) is increased when VAN is combined with certain beta-lactams (BLs) such as piperacillin-tazobactam (TZP) but has not been evaluated with ceftolozane-tazobactam (C/T). Our aim was to investigate the AKI incidence of VAN in combination with C/T (VAN/C/T) compared with VAN in combination to TZP (VAN-TZP). Methods We conducted a multicenter, observational, comparative study across the United States. The primary analysis was a composite outcome of AKI and risk, injury, failure, loss, end stage renal disease; Acute Kidney Injury Network; or VAN-induced nephrotoxicity according to the consensus guidelines. Multivariable logistic regression analysis was conducted to adjust for confounding variables and stratified Kaplan-Meir analysis to assess the time to nephrotoxicity between the 2 groups. Results We included VAN/C/T (n = 90) and VAN-TZP (n = 284) at an enrollment ratio of 3:1. The primary outcome occurred in 12.2% vs 25.0% in the VAN-C/T and VAN-TZP groups, respectively (P = .011). After adjusting for confounding variables, VAN-TZP was associated with increased odds of AKI compared with VAN-C/T; with an adjusted odds ratio of 3.308 (95% confidence interval, 1.560-6.993). Results of the stratified Kaplan-Meir analysis with log-rank time-to-nephrotoxicity analysis indicate that time to AKI was significantly shorter among patients who received VAN-TZP (P = .004). Cox proportional hazards analysis demonstrated that TZP was consistent with the primary analysis (P = .001). Conclusions Collectively, our results suggest that the AKI is not likely to be related to tazobactam but rather to piperacillin, which is a component in VAN-TZP but not in VAN-C/T. Vancomycin in combination with piperacillin-tazobactam was associated with increased odds of acute kidney injury compared with vancomycin in combination with ceftolozane-tazobactam, with an adjusted odds ratio of 3.308 after adjusting for confounding variables in a multivariable logistic regression model.