Circulating pre-diagnostic metabolites and risk of hepatocellular carcinoma and intrahepatic cholangiocarcinoma: a population-based study of 12 cohorts
作者:Cody Z. Watling, Eleanor L. Watts, Erikka Loftfield, Barry I. Graubard, Hyokyoung G Hong, Xuehong Zhang, Matthew J Barnett, Julie E. Buring, Mollie E. Barnard, Yu Chen, A Heather Eliassen, J Michael Gaziano, Jonathan N Hofmann, Wen‐Yi Huang, Jae H Kang, Jill Koshiol, I‐Min Lee, Steven C Moore, Lorelei A. Mucci, Marian L Neuhouser, Christina C. Newton, Mark P. Purdue, Howard D Sesso, Martha Shrubsole, Rashmi Sinha, Lesley Tinker, Triplette Matthew, Caroline Y. Um, Kala Visvanathan, Jean Wactawski‐Wende, Walter C Willett, Fen Wu, Wei Zheng, Peter T. Campbell, DINESH BARUPAL, Jessica L. Petrick, Katherine A. McGlynn · 发表于:JNCI Journal of the National Cancer Institute · 年份:2026 · DOI:10.1093/jnci/djag227 · 研究领域:Cholangiocarcinoma and Gallbladder Cancer Studies、Hepatocellular Carcinoma Treatment and Prognosis、Metabolomics and Mass Spectrometry Studies
BACKGROUND: Underlying metabolic perturbations may offer insight into etiological factors related to liver cancer development. The objective of this study was to explore associations between pre-diagnostic metabolites and risk of hepatocellular carcinoma (HCC) and intrahepatic cholangiocarcinoma (ICC). METHODS: We conducted a nested case-control study within 12 prospective U.S. cohort studies, including 727 HCC cases and 126 ICC cases with matched controls. Metabolomic profiling was performed on pre-diagnostic blood samples using the Metabolon platform. Multivariable conditional logistic regression estimated adjusted odds ratios and 95% confidence intervals for associations between metabolites (per doubling in levels) and liver cancer risk. Elastic net regression selected metabolites predictive of HCC and ICC, with predictive performance assessed by area under the curve (AUC). RESULTS: Median time from blood collection to diagnosis was 11.6 years. Of 1,485 pre-diagnostic metabolites, 600 were associated with HCC risk after false discovery rate correction (366 positively, 234 negatively); no metabolites were significantly associated with ICC. Enriched pathways among HCC cases included methylhistidine metabolism, amino acid metabolism, and bile acid biosynthesis; ICC showed enrichment in valine, leucine, and isoleucine pathways. Elastic net models for HCC selected 38 metabolites with strong predictive performance (AUC = 0.87; 95% CI: 0.83 to 0.91), while eight metabolites were ...