Abstract
BACKGROUND: Relapses in relapsing-remitting multiple sclerosis (RRMS) are acute neuroinflammatory events that shape long-term disease trajectory. Yet, the molecular events during the early recovery period remain poorly characterized, particularly in early, treatment-free disease. Defining these molecular dynamics may clarify mechanisms of repair relevant for understanding early disease progression. OBJECTIVE: Identify serum metabolomic signatures that discriminate early-stage, untreated RRMS by time from last clinical relapse and capture trends in the recovery period. METHODS: Using a cross-sectional cohort study design, we performed untargeted serum metabolomic profiling in 79 RRMS cases (<2 years from onset; treatment-naïve or untreated). Time from a clinical relapse was determined through chart-review. Discriminatory analysis of recent (<4 weeks) versus non-recent (>24 weeks) relapse used supervised machine learning (random forest and partial least squares-discriminant analysis) and rank-based tests. Machine learning and rank-based correlation identified monotonic changes across five relapse proximity stages (<4, 5-8, 9-12, 13-24, and >24 weeks). RESULTS: Multiple metabolites distinguished recent from non-recent relapse (p < 0.05; AUCs 0.72-0.85) and were significantly lower near relapse, including medium-chain acylcarnitines (C6-C10), fatty acid conjugates (hexanoylglycine), microbial-derived aromatics (3-phenylpropionate, 4-vinylcatechol sulfate), and conjugated steroids. Forty-two metabolites had monotonic changes with increasing time from relapse, including 30 decreased and 12 elevated near relapse (p < 0.05). Perturbed pathways involved mitochondrial β-oxidation, tricarboxylic acid cycle flux, membrane lipid remodeling, phase II conjugation, steroid metabolism, and gut-microbial fermentation of polyphenols and aromatic amino acids. CONCLUSIONS: Serum metabolomic profiles in early RRMS reflected coordinated biochemical changes following clinical relapse, implicating mitochondrial dysfunction, membrane remodeling, and microbial co-metabolism.