Plasma proteomics framework predicts metabolic dysfunction-associated steatotic liver disease up to 16 years before...

Metabolic dysfunction-associated steatotic liver disease (MASLD) is a global health challenge, yet preclinical identification remains difficult owing to a lack of reliable predictive tools. Here we show that a panel of five plasma proteins, FUOM, ACY1, KRT18, CDHR2 and GGT1, identified and validated across over 50,000 participants from the Southern UK, Northern UK, EPIC-Norfolk and Southern China cohorts, serves as a predictive signature for incident MASLD. Our five-protein model achieves predictive accuracy of 0.838 (5 year area under the curve (AUC)) and 0.756 (16.6 year AUC), with performance sustained longitudinally in the EPIC-Norfolk cohort (16.6 year AUC = 0.710) and confirmed in the Southern China Inception Cohort (AUC = 0.912). Integrating these biomarkers with routine clinical data further enhances performance (5 year AUC = 0.904; 16.6 year AUC = 0.822). These findings establish a scalable proteomic framework for ultra-early risk stratification and targeted intervention in MASLD up to 16 years before clinical onset.