Abstract Rationale Recent studies have subdivided acute respiratory distress syndrome (ARDS) into Hypo- and Hyper-inflammatory phenotypes, which have differing outcomes and responses to treatments. Prior work suggests these phenotypes differ in energy metabolism, which may influence response to enteral feeding. Lipids also regulate inflammation and recovery, suggesting potential phenotype-specific effects of lipid supplementation. We therefore evaluated whether ARDS phenotypes respond differently to full versus trophic enteral nutrition and to omega-3 fatty acid supplementation. Methods We performed retrospective analysis of the EDEN and OMEGA trials. EDEN was a randomized, double-blind, multicenter trial of mechanically ventilated ARDS patients assigned to full or trophic enteral feeding for 6 days. A subset also participated in OMEGA, a randomized, double-blind, placebo-controlled trial of twice-daily enteral omega-3 fatty acids, gamma-linolenic acid, and antioxidants. In primary analyses, EDEN showed no benefit, and OMEGA was stopped for futility. ARDS phenotype was determined using a machine learning classifier based on clinical variables. The primary outcome was ventilator-free days (VFDs); secondary outcomes included ICU-free days, organ failure-free days and 60-day mortality. Outcomes were compared between treatment groups within each phenotype using Wilcoxon rank-sum or chi-square tests. We assessed phenotype-treatment interactions for VFDs using competing risk regression to model time-to-successful extubation, with death as a competing risk (Yehya et al., Am J Respir Crit Care Med, 2019). Results A total of 1000 patients were included in EDEN, of which 273 (27%) were classified as Hyperinflammatory. Randomization to full feeding was approximately balanced across phenotypes. Among Hypoinflammatory patients in EDEN, outcomes were similar between treatments. Among Hyperinflammatory patients in EDEN, those randomized to full feeding had numerically greater ventilator- and ICU-free days and decreased mortality, although differences did not meet statistical significance (Table). There was no significant interaction between phenotype and treatment for VFDs (P = 0.052). 272 patients were included in OMEGA, of which 79 (29%) were Hyperinflammatory. A similar proportion of patients within each phenotype were randomized to omega-3 supplementation. Hypoinflammatory patients randomized to omega-3 supplementation had significantly fewer ventilator-, ICU-, and organ failure-free days and higher mortality (Table). No differences were detected in Hyperinflammatory patients. There was no significant interaction between phenotype and treatment for VFDs (P = 0.18). Conclusions In a post-hoc analyses, we found suggestive but not statistically significant evidence that early full enteral feeding may benefit Hyperinflammatory ARDS patients, while omega-3 FA, gamma-linoleic acid, and antioxidant supplementation may be harmful in Hypoinflammatory ARDS patients. Evaluation in prospective trials is warranted. This abstract is funded by: R35HL177135, K23HL173659, 5TL1TR001871-10
Yang et al. (Fri,) studied this question.