Journal: EMBO molecular medicine
This study addresses the critical need for improved early detection of gastric cancer (GC) by developing a non-invasive diagnostic tool that combines untargeted metabolomics with machine learning.
Researchers analyzed plasma and tissue samples from 597 patients using UPLC-MS and identified a six-metabolite panel with high diagnostic accuracy (AUC 0.920–0.982) and sensitivity (0.900–0.940), outperforming traditional biomarkers.
Key findings include:
- Isovalerylcarnitine (C5) was consistently downregulated in GC.
- Mendelian randomization linked C5 causally to GC risk.
- Proteomic and functional analyses showed that C5 inhibits GC cell migration and invasion by affecting cadherin and MMP pathways.
These results support the metabolite panel as a promising diagnostic tool and highlight isovalerylcarnitine (C5) as a potential therapeutic target for GC.