Prediagnostic Metabolites Linked To Future Liver Cancer Risk In 12 Cohorts

Key Takeaways
- Across 1,485 metabolites, broad associations were observed for future HCC, while no metabolites remained significant for ICC after false discovery rate correction.
- Pathway enrichment differed by cancer type, with HCC signals in methylhistidine metabolism, amino acid metabolism, and bile acid biosynthesis, and ICC signals in valine, leucine, and isoleucine pathways.
- The elastic net model showed stronger discrimination for HCC than for ICC, and the authors interpreted the HCC signal as potentially reflecting liver dysfunction while calling for larger ICC studies.
Researchers evaluated 1,485 prediagnostic metabolites after false discovery rate correction. Among HCC cases, 600 metabolites were associated with risk, with 366 positive associations and 234 negative associations. Both higher and lower metabolite levels were linked to later HCC. No metabolites were significantly associated with ICC in the corresponding analysis. Overall, the association pattern was much broader for HCC than for ICC.
Pathway analyses linked HCC-associated metabolites to methylhistidine metabolism, amino acid metabolism, and bile acid biosynthesis. For ICC, enrichment involved valine, leucine, and isoleucine pathways, indicating a different distribution of pathway signals. Using elastic net regression, investigators selected 38 metabolites for HCC and 8 for ICC. The HCC model yielded an AUC of 0.87 with a 95% CI of 0.83 to 0.91. The ICC model yielded an AUC of 0.64 with a 95% CI of 0.49 to 0.80, showing weaker discrimination.
The authors concluded that the broad prediagnostic HCC metabolite signal may reflect underlying liver dysfunction. They also said the enrichment patterns suggest distinct metabolic perturbations across HCC and ICC. For ICC, they called for larger studies to better characterize metabolic risk factors. They also said additional work is needed to evaluate metabolites as predictors of HCC risk.