Perioperative neutrophil-to-lymphocyte ratio trajectories and creatinine-defined acute kidney disease after cardiac surgery: a study using the INSPIRE database.

BACKGROUND: Acute kidney disease (AKD) encompasses persistent or newly developing renal dysfunction beyond 7 days after surgery. Associations between perioperative neutrophil-to-lymphocyte ratio (NLR) changes and AKD after cardiac surgery remain unclear. We identified NLR trajectories and examined their associations with overall, persistent, and delayed AKD. METHODS: This retrospective, hypothesis-generating cohort study used the single-center INSPIRE database. Adults with NLR available in at least four of five perioperative windows were classified using group-based trajectory modeling. The primary outcome was AKD during postoperative days 8-90. Time-to-event, competing-risk, follow-up-weighted, early-AKI-stage, expanded-confounder, and medication-exposure sensitivity analyses were performed. Mortality and postoperative in-hospital hemodialysis/continuous renal replacement therapy (CRRT) records were exploratory secondary outcomes. RESULTS: Of 2,265 patients assigned to a trajectory, 1,975 had day-8-90 creatinine follow-up; 298 developed AKD. Three data-driven trajectories were identified: low-response (T1), intermediate peak-recovery (T2), and high persistent-response (T3). Adjusted odds of overall AKD were higher for T2 vs. T1 (OR: 2.02, 95% CI: 1.35-3.04) and T3 vs. T2 (OR: 1.42, 95% CI: 1.02-1.96), with the clearest gradient for persistent AKD. Kaplan-Meier curves differed for overall and persistent AKD (both log-rank P < 0.001), and Cox and Fine-Gray analyses were concordant. Associations were broadly similar in sensitivity analyses, although adjustment for early AKI attenuated some adjacent contrasts. Ninety-day mortality and postoperative in-hospital hemodialysis/CRRT records were most frequent in T3, but these exploratory endpoints were uncommon and susceptible to residual confounding. CONCLUSIONS: Perioperative NLR trajectories were associated with creatinine-defined AKD after cardiac surgery, most clearly with persistent AKD. Concordant findings for mean perioperative and late postoperative NLR suggest that the magnitude and persistence of NLR elevation were more informative than relative within-patient variability.