Common lipidomic signatures across distinct acute brain injuries in patient outcome prediction

Abstract

Lipidomic alterations have been associated with various neurological diseases. Examining temporal changes in serum lipidomic profiles, irrespective of injury type, reveals promising prognostic indicators. In this longitudinal prospective observational study, serum samples were collected early (46 +/- 24 h) and late (142 +/- 52 h) post- injury from 70 patients with ischemic stroke, aneurysmal subarachnoid hemorrhage, and traumatic brain injury that had outcomes dichotomized as favorable (modified Rankin Scores (mRS) 0-3) and unfavorable (mRS 4-6) three months post-injury. Lipidomic profiling of 1153 lipids, analyzed using statistical and machine learning methods, identified 153 lipids with late-stage significant outcome differences. Supervised machine learning pinpointed 12 key lipids, forming a combinatory prognostic equation with high discriminatory power (AUC 94.7 %, sensitivity 89 %, specificity 92 %; p < 0.0001). Enriched functions of the identified lipids were related to sphingolipid signaling, glycerophospholipid metabolism, and necroptosis (p < 0.05, FDR-corrected). The study underscores the dynamic nature of lipidomic profiles in acute brain injuries, emphasizing late-stage distinctions and proposing lipids as significant prognostic markers, transcending injury types. These findings advocate further exploration of lipidomic changes for a comprehensive understanding of pathobiological roles and enhanced prediction for recovery trajectories.