Regression of metabolic dysfunction–associated steatotic liver disease in older adults: Short- and long-term prediction models in the NHIS-senior and UK Biobank cohorts

초록

Background and aim: We developed and validated prediction models for regression of metabolic dysfunction-associated steatotic liver disease (MASLD) in older adults. Methods and results: Using data from the Korean National Health Insurance Service–Senior cohort, we included older patients with MASLD, defined fatty-liver-index≥30 and at least one cardiometabolic risk factors (CMRFs) at baseline. MASLD regression was assessed at short-term (3-year) and long-term (6-year) follow-ups, defined the absence of steatotic liver disease or any CMRFs. Logistic regression (LR) and decision tree (DT) models were developed to predict MASLD regression, and their performance was evaluated using area under the receiver-operating-characteristic-curve (auROC). The developed models were externally validated in the UK-Biobank cohort. Among 168,198 older adults with MASLD, regression occurred in 38,687 (23.0%) within the short-term period and 30,204 (18.0%) in the long-term period. Internally, auROCs were 0.787/0.760 (LR/DT) for short-term prediction and 0.754/0.720 (LR/DT) for long-term prediction, respectively. Externally, auROCs for short-term predictions were 0.825/0.813 (LR/DT) and 0.784/0.752 (LR/DT) for long-term prediction in the UK-Biobank. Conclusions: We developed an interpretable prediction score for short- and long-term regression of MASLD with good performance. This tool may enable personalized and proactive management by identifying individuals likely to experience regression, thereby informing targeted lifestyle or pharmacologic interventions. © 2026 The Italian Diabetes Society, the Italian Society for the Study of Atherosclerosis, the Italian Society of Human Nutrition and the Department of Clinical Medicine and Surgery, Federico II University

키워드