Prognostic significance of the endothelial activation and stress index in acute respiratory distress syndrome: a retrospective cohort study

Endothelial cells play a crucial role in the pathogenesis of acute respiratory distress syndrome (ARDS). The Endothelial Activation and Stress Index (EASIX) is regarded as a reliable biomarker of endothelial dysfunction. The aim of this study was to evaluate the prognostic value of baseline and dynamic EASIX trajectories in ARDS patients. The data of this study were sourced from the Medical Information Mart for Intensive Care IV (MIMIC-IV) 2.2 database. The study population was divided into three groups according to the tertiles of the EASIX index. The primary outcome was 28-day ICU mortality. Kaplan-Meier survival analysis, multivariate Cox regression, and restricted cubic spline (RCS) were used to assess the correlation between the baseline EASIX and mortality. Furthermore, Latent Class Mixed Models (LCMM) were employed to identify dynamic EASIX trajectories within the first 72 h. The association between these identified trajectories and clinical prognosis was subsequently evaluated. Finally, the Boruta algorithm was applied to screen for key predictive features, and seven machine learning (ML) algorithms were developed to predict 28-day mortality. According to the established inclusion and exclusion criteria, 1044 ARDS patients were ultimately included in this study. Kaplan-Meier curves showed that patients in the high baseline EASIX group had higher 28-day mortality. Multivariate Cox regression revealed that higher EASIX was associated with increased 28-day mortality (HR = 1.07; 95% CI 1.01-1.14, P = 0.03) and remained significant at 60 and 180 days. The RCS curves indicated a linear relationship (P-non-linear > 0.05). In the dynamic analysis, LCMM identified three distinct trajectories: Trajectory 1 (Persistently Low), Trajectory 2 (Persistently Increasing), and Trajectory 3 (Rise-and-Fall). Notably, Trajectory 2 exhibited the poorest prognosis (HR = 7.56; 95% CI 3.45-16.57,P < 0.001). Feature selection via Boruta algorithm consistently identified EASIX as a key predictor. Among the seven ML models evaluated, the Random Forest algorithm demonstrated superior performance, achieving an AUC of 0.826. Both baseline EASIX and dynamic EASIX trajectories are independent predictors of mortality in ARDS patients, suggesting that EASIX has great potential as a reliable prognostic indicator for ARDS patients.