Metabolic Signatures for Liver Cancer Diagnosis and Mechanistic Insights: A Large‐Scale, Multicenter Study

Abstract

ABSTRACT The poor prognosis of liver cancer (LC) highlights the urgent need for more effective strategies for early detection. We conducted a large, multicenter study to develop and validate a serum metabolic signature for LC diagnosis. Using high‐throughput nanoparticle‐enhanced laser desorption/ionization mass spectrometry, we profiled serum metabolites from 1,924 participants in discovery cohort recruited across 11 clinical centers and 225 participants in independent external validation cohort from two additional centers. A nine‐metabolite signature combined with alpha‐fetoprotein achieved excellent discriminatory performance, with an AUC of 0.92 in the discovery cohort and 0.93 in the external validation cohort. The combined model also showed good sensitivity for early‐stage LC, reaching 0.85 and 0.78 in the two cohorts, respectively, and performed particularly well for hepatocellular carcinoma, with AUCs of 0.94 and 0.93. Exploratory analyses suggested potential utility for intrahepatic cholangiocarcinoma diagnosis. Mendelian randomization analysis supported a potential causal association between nicotinamide (NAM) levels and LC risk. Functional studies showed that NAM promotes LC cell proliferation, migration, and invasion, through an NAD+/SIRT1‐dependent mechanism that increases HIF1α stability and activates glycolytic and MAPK signaling. Overall, we developed a promising metabolic diagnostic signature for LC and identified NAM‐associated metabolic signaling as a potential contributor to LC progression.