Abstract
PURPOSE: Liver fibrosis and cirrhosis represent critical stages in the progression of chronic liver disease, yet their key molecular features remain incompletely understood.</p>
EXPERIMENTAL DESIGN: We performed large-scale Olink-based proteomic profiling in over 40,000 participants from the UK Biobank with a median follow-up of 15.6 years to elucidate disease pathophysiology and identify pre-diagnostic biomarkers. Cross-sectional analysis included 66 prevalent cirrhosis cases, and prospective analysis identified 224 incident cirrhosis cases. Machine learning and Mendelian randomization (MR) were applied. An independent cohort was used for validation.</p>
RESULTS: Distinct dysregulated proteins were observed in compensated cirrhosis (CC) and decompensated cirrhosis (DC). In the prospective analysis, 696 proteins were associated with disease onset. A proteomic panel based on these markers achieved an AUC of 0.832 for predicting incident cirrhosis, outperforming established fibrosis scores including FIB-4, APRI, and NFS, and demonstrated robust performance across CC and DC populations. The protein panel showed predictive value (AUC = 0.743) for disease progression in an independent cohort. MR identified 66 proteins with putative causal roles, including 11 potential therapeutic targets.</p>
CONCLUSIONS AND CLINICAL RELEVANCE: These findings provide novel molecular insights into cirrhosis development and support integrated proteomic biomarkers as a discovery and prioritization framework for early risk stratification.</p>