Hepatol Res. 2026 Jan 26. doi: 10.1111/hepr.70127. Online ahead of print.
ABSTRACT
AIMS: Metabolic dysfunction-associated steatotic liver disease (MASLD) is characterized by the presence of hepatic steatosis and at least one of five cardiometabolic risk factors, leading to the development of cardiovascular-kidney-metabolic syndrome including chronic kidney disease (CKD). However, the impact of heterogeneity of MASLD on new onset of CKD remains unclear. We explored the relationship between subgroups of MASLD divided by using a machine learning (ML) model called supervised clustering and the development of CKD during a 10-year follow-up period.
METHODS: A total of 12,168 Japanese subjects (men/women: 7927/4,241 and mean age: 48 years) who received annual health examinations including abdominal ultrasonography were recruited.
RESULTS: Using the supervised clustering by SHapley Additive exPlanations (SHAP) and uniform manifold approximation and projection (UMAP) for steatotic liver diseases, 10 subclusters including 3 distinctive subgroups of MASLD were detected by a Gaussian mixture model. Kaplan-Meier survival curve analysis showed a significant difference in the cumulative incidence for new onset of CKD among the 3 subgroups of MASLD. Among the MASLD subclusters, an obese subgroup with an atherogenic profile of serum lipids as well as high levels of fatty liver index and uric acid was the worst subcluster for the development of CKD in individuals with MASLD.
CONCLUSIONS: The supervised clustering of MASLD using a SHAP-converted matrix and UMAP reveals phenotypically distinct subpopulations that improved risk stratification for new onset of CKD. An obese subgroup with atherogenic lipid profiles and hyperuricemia in MASLD is associated with an increased risk for the development of CKD.
PMID:41587042 | DOI:10.1111/hepr.70127