Post-hoc mediation analysis of two biomarkers, and survival in acute respiratory distress syndrome.

Previous studies have shown that plasma soluble receptor for advanced glycation end-products (sRAGE) and the radiographic assessment of lung edema (RALE) are associated with the severity of acute respiratory distress syndrome (ARDS) both at baseline and over time. This study aims to explore the causal relationships among sRAGE, the RALE score, their fluctuations, and 90-day survival. Causal mediation analysis was conducted as a secondary analysis of the randomized controlled lung imaging for ventilator setting in ARDS (LIVE) trial, which assessed a mechanical ventilation strategy based on lung morphology. The primary outcome was survival at day 90. We used group-based trajectory modeling to summarize the biomarker patterns over time, followed by mediation analysis with the sRAGE and RALE score as mediators. Out of 400 patients in the LIVE study, 115 were included, resulting in three trajectory groups for both sRAGE and RALE. The mechanical ventilation strategy appeared to influence survival directly and indirectly: one indirect effect was mediated by one RALE score trajectory (aligning with the direct effect), and another by one sRAGE trajectory (opposing the direct effect). Both plasma sRAGE and the RALE score appeared to exhibit a mediation effect on survival for specific patient clusters. Identifying these clusters and using biomarkers as surrogate endpoints warrants further investigation in precision ARDS trials.