Background & aims: Although metabolic dysfunction-associated steatotic liver disease (MASLD) is associated with high multimorbidity and mortality, existing classification systems and risk prediction models largely ignore the heterogeneity of MASLD. Improved subtype definition could improve prediction of outcomes and inform new precision treatment strategies.
Methods: We analyzed individuals with MASLD from population-based electronic health record resource from UK Biobank (n = 125,197) and Health examinee dataset of Nanfang Hospital (n = 995). We identified subtypes with K-means clustering method. The Cox proportional hazard regression model analyzed the relationship between variables and MASLD-related complications.
Results: After identifying five clusters across seven clinical indicators which were age, body mass index, monocyte/lymphocyte ratio, aspartate aminotransferase, waist-hip ratio, low-density lipoprotein-cholesterol, and cholesterol, we labelled MASLD subtypes: (1) Metabolic-Dyslipidemia, (2) Younger, (3) Obesity, (4) Inflammatory, and (5) Hepatotoxic. Metabolic outcomes differed across these five subtypes. Hepatotoxic MASLD showed an increased risk of severe liver diseases compared to Metabolic-Dyslipidemia MASLD [HR = 13.9, 95% CI 10.7-18.1]. The extrahepatic complications were highest in Inflammatory MASLD. These two groups were defined as the high-risk group with higher health burden than other three groups, which classified as low-risk group. Differential single-nucleotide polymorphisms were concentrated on chromosomes 1 and 19 when comparing the high- and low-risk groups, and the annotated genes enrichment pathways were primarily related to lipid metabolism and transport.
Conclusions: Patient subtypes derived by clinical indicators are a valuable addition to existing MASLD classification systems, which could provide a valuable tool to aid in selecting specific treatment approaches.
Keywords: Clinical indicator; Cluster; Complications; Genome-wide association study (GWAS); MASLD.
© 2025. The Author(s).