Epithelial injury and inflammatory phenotypes for personalized ventilation in ARDS: Secondary analysis of the LIVE trial.

BACKGROUND: Guidelines recommend against systematic lung recruitment strategies in acute respiratory distress syndrome (ARDS). sRAGE, an alveolar epithelial injury plasmatic biomarker, and inflammatory phenotypes may guide personalized approaches. RESEARCH QUESTION: Does baseline sRAGE or inflammatory phenotype identify heterogeneity in the treatment effect of lung recruitment on mortality in ARDS? STUDY DESIGN AND METHODS: This analysis included 259 patients with plasma sRAGE measurements from the original LIVE trial that compared lung recruitment with low-PEEP strategies in patients with focal or non-focal ARDS. Baseline sRAGE was dichotomized at 2,440 pg/mL into high and low groups. Inflammatory phenotypes were identified using a validated parsimonious model (IL-8, sTNFr-1, bicarbonate). The primary and secondary outcomes were 90-day mortality and sRAGE trajectory until day 6. Inverse probability weighting and weighted Cox models were used; sRAGE trajectories were assessed using joint models. RESULTS: High plasma sRAGE (r=0.26, p<0.001), but not the inflammatory phenotype, was moderately correlated with non-focal ARDS. Treatment effect varied by baseline sRAGE (p-for-interaction=0.006), but not by inflammatory phenotype (p-for-interaction=0.56); however, the hyperinflammatory phenotype was associated with 90-day mortality (57% vs 26%, p<0.001). Among high-sRAGE patients, lung recruitment was associated with lower 90-day mortality (HR 0.41, 95%CI 0.18-0.93; p=0.033); among low-sRAGE patients it was associated with higher mortality (HR 3.27, 95%CI, 1.06-10.1; p=0.039). Lung recruitment was associated with different sRAGE slopes compared with low PEEP, with effect modification by baseline sRAGE. Lung recruitment was associated with a steeper decline in sRAGE among patients with high baseline sRAGE (p-for-interaction<0.001), and with a slower decline among patients with low baseline sRAGE (p-for-interaction=0.038). Inflammatory phenotypes did not identify heterogeneity of treatment effect. Sensitivity analyses without inverse probability weighting were directionally consistent. INTERPRETATION: In ARDS, epithelial injury assessed by plasma sRAGE may identify heterogeneity of treatment effect with lung recruitment, whereas inflammatory phenotypes remained prognostic for mortality.