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

Shiyi Yu, Chunling Chen, Jing Feng, Qinming Li, Weihong Sha, Hao Chen

1. 01

   Five-protein panel: FUOM, ACY1, KRT18, CDHR2 and GGT1

2. 02

   5-year AUC 0.838; 16.6-year AUC 0.756

3. 03

   Long-horizon performance sustained in EPIC-Norfolk (16.6-year AUC 0.710)

4. 04

   External confirmation in the Southern China Inception Cohort (AUC 0.912)

5. 05

   Adding routine clinical data raised performance to 5-year AUC 0.904 and 16.6-year AUC 0.822

6. 06

   Validated across Southern UK, Northern UK, EPIC-Norfolk and Southern China cohorts (>50,000 participants)

Study Design

Cohort Study

Sample Size

50,000 subjects

Study Duration

Up to 16.6 years

Methodology

Plasma proteomic discovery and validation across four independent prospective cohorts; incident MASLD as endpoint; time-stratified AUC analysis with and without integration of routine clinical variables.

Limitations

Observational prediction model only. No trial has tested whether acting on the signature changes hepatic outcomes, and no assay is yet clinically standardised or reimbursed. AUC values are cohort-level discrimination metrics; calibration and net clinical benefit in routine practice are unestablished.