Identification of Growth Differentiation Factor-15 as An Early Predictive Biomarker for Metabolic Dysfunction-Associated Steatohepatitis: A Nested Case-control Study of UK Biobank Proteomic Data

Abstract

Background & AimsThe lack of non-invasive biomarkers for the early prediction of patients with metabolic dysfunction-associated steatohepatitis (MASH) is a major challenge for timely intervention. This study aims to determine the predictive capability for MASH long before its diagnosis by using six previously identified diagnostic biomarkers for metabolic dysfunction-associated steatotic liver disease (MASLD) with proteomic data from the UK Biobank. MethodsA nested case-control study comprising of a MASH group and three age- and sex-matched controls groups (metabolic dysfunction-associated steatosis, viral hepatitis, and normal liver controls) were conducted. Olink proteomics, anthropometric and biochemical data at baseline levels were obtained from the UK Biobank. The baseline levels of CDCP1, FABP4, FGF21, GDF15, IL-6 and THBS2 were analyzed prospectively to determine their predictive accuracy for subsequent diagnosis with a mean lag time of over 10 years. ResultsAt baseline, GDF15 demonstrated the best performance for predicting MASH occurrence at 5 and 10 years later, with an AUC of 0.90 at 5 years and 0.86 at 10 years. A predictive model based on four biomarkers (GDF15, FGF21, IL-6, and THBS2) showed AUCs of 0.88 at both 5 and 10 years. Furthermore, a protein-clinical model that included these four circulating protein biomarkers along with three clinical factors (BMI, ALT and TC) yielded AUCs of 0.92 at 5 years and 0.89 at 10 years. ConclusionGDF15 at baseline levels outperformed other individual circulating protein biomarkers for the early prediction of MASH. Our data suggest that GDF15 and the GDF15-based model may be used as easy-to-implement tools to identify patients with high risk of developing MASH at a mean lag time of over 10 years.