Serial lactate–procalcitonin interaction identifies a high-risk phenotype in 24-h conditional survivors of post-cardiac arrest syndrome: a CART-based analysis

Objective To investigate the time-dependent interplay between early metabolic failure and delayed systemic inflammatory response in 24-h conditional survivors of post-cardiac arrest syndrome (PCAS). We aimed to develop a non-linear risk stratification tool using Classification and Regression Tree (CART) analysis based on serial lactate and procalcitonin (PCT) kinetics. Materials and methods This retrospective cohort study included 158 24-h conditional survivors of PCAS. Arterial lactate levels were monitored at admission (T0), 6 h, 12 h, 24 h, and 48 h, while PCT levels were recorded at T0 and T24. Missing data were handled using multiple imputation by chained equations (MICE). The primary endpoint was in-hospital mortality. A landmark analysis was performed at 24 h to address immortal time bias. Results The in-hospital mortality rate was 70.9%. Non-survivors exhibited significantly higher rates of non-shockable rhythms, longer CPR durations, and higher APACHE II scores. In the multivariable logistic regression model adjusted for Targeted Temperature Management (TTM), persistent hyperlactatemia at 48 h (T48) remained a significant independent predictor of mortality (OR: 1.92; 95% CI: 1.22–3.01, p = 0.003). Lactate burden (AUC) demonstrated superior prognostic performance compared to baseline T0 measurements (Bootstrap-validated DeLong test, p = 0.019). CART analysis identified a high-risk phenotype (T0 lactate >5 mmol/L AND T24 PCT > 5.5 ng/mL) associated with a 92% mortality risk (PPV: 91.8%). The CART model showed excellent calibration (Brier score: 0.14) and comparable discrimination to logistic regression (Logistic Regression Brier score: 0.12 vs. CART Brier score: 0.14; AUC: 0.830 vs. 0.842, p = 0.412). Conclusion In 24-h conditional survivors of PCAS, persistent metabolic debt at 48 h is a potent independent biochemical indicator of poor outcome. Integrating serial lactate and PCT kinetics through non-linear CART analysis identifies a high-risk “metabolic-inflammatory failure” phenotype. This model may facilitate early clinical decision-making and guide the escalation to advanced circulatory or extracorporeal support strategies in high-risk patients.