RESUMO
INTRODUCTION: This study aims to assess the association of piperacillin/tazobactam and meropenem minimum inhibitory concentration (MIC) and beta-lactam resistance genes with mortality in the MERINO trial. METHODS: Blood culture isolates from enrolled patients were tested by broth microdilution and whole genome sequencing at a central laboratory. Multivariate logistic regression was performed to account for confounders. Absolute risk increase for 30-day mortality between treatment groups was calculated for the primary analysis (PA) and the microbiologic assessable (MA) populations. RESULTS: In total, 320 isolates from 379 enrolled patients were available with susceptibility to piperacillin/tazobactam 94% and meropenem 100%. The piperacillin/tazobactam nonsusceptible breakpoint (MIC >16 mg/L) best predicted 30-day mortality after accounting for confounders (odds ratio 14.9, 95% confidence interval [CI] 2.8-87.2). The absolute risk increase for 30-day mortality for patients treated with piperacillin/tazobactam compared with meropenem was 9% (95% CI 3%-15%) and 8% (95% CI 2%-15%) for the original PA population and the post hoc MA populations, which reduced to 5% (95% CI -1% to 10%) after excluding strains with piperacillin/tazobactam MIC values >16 mg/L. Isolates coharboring extended spectrum ß-lactamase (ESBL) and OXA-1 genes were associated with elevated piperacillin/tazobactam MICs and the highest risk increase in 30-day mortality of 14% (95% CI 2%-28%). CONCLUSIONS: After excluding nonsusceptible strains, the 30-day mortality difference from the MERINO trial was less pronounced for piperacillin/tazobactam. Poor reliability in susceptibility testing performance for piperacillin/tazobactam and the high prevalence of OXA coharboring ESBLs suggests that meropenem remains the preferred choice for definitive treatment of ceftriaxone nonsusceptible Escherichia coli and Klebsiella.