Abstract Background Sepsis is a heterogeneous, time-dependent syndrome in which static biomarkers may inadequately capture evolving host responses. We investigated whether baseline and early serial serum NLR family pyrin domain-containing 3 (NLRP3) measurements were associated with disease severity and 28-day mortality in sepsis. Methods Adult ICU patients with sepsis were retrospectively enrolled. Serum NLRP3 was measured on ICU days 1 and 3. Early change was defined as ΔNLRP3 (day 3 minus day 1). Analyses involving ΔNLRP3 used a prespecified day-3 landmark design. The primary complete-case logistic model adjusted for age, sex, admission SOFA score, and log-transformed day-1 lactate. Multiple imputation addressed missing day-1 lactate values, and bootstrap resampling was used for internal validation. Results The day-3 landmark cohort included 161 patients; 33 died by day 28. Baseline NLRP3 correlated weakly with admission SOFA score (Spearman’s ρ = 0.25, P = 0.0013) and showed limited standalone discrimination for post-landmark 28-day mortality (AUC, 0.605; 95% CI, 0.496–0.713). In the complete-case cohort ( n = 147), higher ΔNLRP3 was independently associated with mortality, with an adjusted odds ratio (aOR) of 2.03 per 1-ng/mL increase (95% CI, 1.49–2.93; P < 0.001). This association remained consistent after multiple imputation (pooled aOR, 2.15; 95% CI, 1.53–3.02; P < 0.001). Adding ΔNLRP3 to the baseline clinical model improved the AUC from 0.669 to 0.828, with an optimism-corrected AUC of 0.802 after bootstrap validation. Day-3 NLRP3 had similar standalone discrimination to ΔNLRP3 (AUC, 0.788 vs. 0.796; DeLong P = 0.873). Conclusions Early serial serum NLRP3 assessment was associated with post-landmark 28-day mortality and provided incremental prognostic information beyond admission clinical severity. The comparable exploratory performance of day-3 NLRP3 suggests that a single day-3 reassessment warrants further validation as a pragmatic risk-updating approach.
Lai et al. (2026) studied this question.