A gene expression-based approach for the precision use of hydrocortisone in septic shock patients; a secondary analysis of the ADRENAL trial.

Background: Small observational studies suggest the effect of corticosteroids in patients with vasodilatory shock vary depending on endotypes determined by gene expression. We sought to replicate these findings in a larger cohort from a randomised clinical trial. Methods: In a cross-sectional substudy of the Adjunctive Glucocorticoid Therapy In Septic Shock (ADRENAL) trial, patients were classified as one of two immune endotypes using predefined gene expression signatures: immune adaptive-prevalent (IA-P) or immune innate-prevalent (IN-P). We compared the outcomes of the two endotypes using a Bayesian analysis. The primary outcome was Day-28 mortality. Findings: Of 540 patients, 267 (49.4%) were classified as IA-P and 273 (50.6%) as IN-P. In a Bayesian analysis using noninformative priors, there was no difference in the effect of hydrocortisone on 28-day mortality (odds ratio [OR] 1.43, 95% credible intervals [CrI] 0.72-2.87) and OR 1.39, 95% CrI 0.74-2.61, between the IA-P and IN-P groups, respectively. In the subgroup of patients with more severe shock (n = 215/540, 40%), the corresponding figures for IA-P and IN-P were 1.21, 95% CrI (0.31-4.74) and OR 0.72 (95% CrI 0.30-1.67), respectively. In the subgroup of patients with pulmonary sepsis (232/540, 43%), IA-P patients treated with hydrocortisone had increased mortality (OR 5.55, 95% CrI 1.81-21.2). Interpretation: Gene expression data from patients with septic shock reveal distinct immune endotypes. There was no evidence of a heterogeneity of treatment effect of hydrocortisone on mortality in the 2 endotypes or in the subgroup with severe shock. Patients with the IA-P endotype and pulmonary sepsis appear to be harmed by corticosteroids.