Kokkorakis · Metabolism: clinical and experimental 2025 · prospective cohort study · n=502,359

GDF-15 improves the predictive capacity of steatotic liver disease non-invasive tests for incident morbidity and mortality risk for cardio-renal-metabolic diseases and malignancies.

Cited 9 times in the scientific literature.

Level 3 - non-randomized controlled study

Prospective cohort study evaluating prognostic biomarker performance

PubMed 39396641 · doi:10.1016/j.metabol.2024.156047 · record verified 2026-08-26

What was done

Researchers evaluated 502,359 UK Biobank participants who were free of the study outcomes at baseline and followed for a median of 14 years. Participants were analyzed across three categories: the general population, a cohort with potential metabolic dysfunction-associated steatotic liver disease (MASLD), and individuals with type 2 diabetes. The study assessed whether adding growth differentiation factor 15 (GDF-15) to established non-invasive liver tests (APRI, FIB-4, FLI, HSI, LAP, and MAF-5) improved prediction of incident cardio-renal-metabolic morbidity, malignancies, cause-specific mortality, and all-cause mortality.

What was found

Adding GDF-15 significantly improved discrimination across almost all evaluated outcomes compared to traditional non-invasive tests, achieving C-indices between 0.601 and 0.808 (representing an overall C-index improvement > 0.2). Screening with GDF-15-enhanced tests required up to more than seven times fewer individuals to identify more incident adverse outcomes compared to traditional tests. Cumulative incidence across all outcomes increased exponentially in the upper quintile of the enhanced scores.

Why it matters

Supplementing routine non-invasive steatotic liver disease scores with GDF-15 substantially enhances their prognostic accuracy, potentially expanding their utility into multipurpose tools for predicting systemic metabolic, cardiovascular, renal, and cancer mortality.

Limits

The UK Biobank cohort is predominantly of European descent and subject to healthy volunteer selection bias. The abstract does not report calibration metrics, specific hazard ratios, or external validation in an independent clinical population.

Cited by