Clinical clustering identifies MASLD subtypes with distinct longitudinal cardiovascular risks and lipidomic profiles

Metabolic dysfunction-associated steatotic liver disease (MASLD) exhibits substantial heterogeneity in progression and cardiovascular outcomes. We aimed to identify reproducible MASLD subtypes in Chinese cohorts, characterize their distinct lipidomic profiles, and evaluate their associations with long-term cardiovascular outcomes.

We investigated the heterogeneity of MASLD using k-means clustering based on six simple clinical variables in a cohort of 5,329 individuals. The identified clusters were applied in an independent cohort of 1,432 participants. Cardiovascular outcomes were compared across clusters using Kaplan-Meier curves and log-rank tests. Untargeted lipidomic profiling was performed with partial least squares discriminant analysis and least absolute shrinkage and selection operator regression to identify discriminative lipids for XGBoost model. Feature importance was evaluated using SHapley Additive exPlanations.

Three distinct MASLD subtypes were identified. Cluster A (Metabolic-Obesity subtype), characterized by the highest BMI and triglycerides with the lowest risk of cardiovascular outcomes. Cluster B (Dysglycemic subtype) distinguished by increased glycated hemoglobin. Cluster C (Aging-Hypertensive subtype) was primarily associated with age, blood pressure, and fibrosis markers, leading to higher risk of cardiovascular outcomes. Lipidomics identified 1061 metabolites, and ten lipids distinguishing the Cluster A and C were selected for modeling, yielding an area under the receiver operating characteristic curve of 0.900. Pathway enrichment analysis further revealed the Cluster C involved in sphingolipid metabolism.

Three distinct MASLD subtypes with varying metabolic features and cardiovascular risks were identified in Chinese cohorts. The Aging-Hypertensive subtype showed the highest cardiovascular risk and was associated with dysregulated sphingolipid metabolism. These findings underscore the clinical heterogeneity of MASLD and highlight the need for risk classification and personalized intervention strategies.