Nine-Metabolite Blood Test Detects Liver Cancer Early With 93% Accuracy
A large multicenter study identifies a serum metabolic signature that detects liver cancer early, outperforming AFP alone — and links nicotinamide to tumor progression.
Summary
Researchers across 13 Chinese medical centers profiled serum metabolites from over 2,100 participants using high-throughput nanoparticle-enhanced mass spectrometry. They identified nine metabolites that, combined with the standard AFP blood marker, detected liver cancer with an AUC of 0.93 in external validation — well above AFP alone. The panel was especially strong for early-stage disease (85% sensitivity) and hepatocellular carcinoma specifically. One standout metabolite, nicotinamide (a form of vitamin B3 and NAD+ precursor), was shown through Mendelian randomization to have a potential causal role in liver cancer risk. Lab experiments revealed nicotinamide drives cancer cell proliferation and invasion by boosting NAD+, activating SIRT1, stabilizing HIF1α, and switching on glycolysis and MAPK signaling pathways.
Detailed Summary
Liver cancer carries a devastating five-year survival rate of only 15.1%, largely because most cases are caught late. The current standard — AFP blood testing plus abdominal ultrasound — misses more than half of early-stage hepatocellular carcinomas (HCC), with AFP sensitivity around 45.5%. This multicenter study set out to find a better blood-based diagnostic by mapping the metabolic fingerprints of liver cancer across a geographically diverse Chinese population spanning 13 clinical centers.
The discovery cohort enrolled 1,924 participants — 1,013 liver cancer patients and 911 non-cancer controls — from 11 centers across five provinces. An independent external validation cohort of 225 subjects (100 liver cancer, 125 controls) came from two additional centers. Serum metabolomics was performed using nanoparticle-enhanced laser desorption/ionization mass spectrometry (NPELDI-MS), a microarray-based high-throughput platform that circumvents many of the throughput and reproducibility limitations of traditional liquid chromatography-MS. Feature selection via LASSO regression and logistic modeling identified nine candidate metabolites from hundreds of detected signals.
The nine-metabolite panel alone achieved an AUC of 0.88 in the discovery cohort. When combined with AFP, the model reached an AUC of 0.92 in discovery and 0.93 in the external validation cohort — significantly outperforming AFP alone (AUC ~0.79). For early-stage (Stage I–II) liver cancer, the combined model achieved sensitivity of 0.85 in the discovery cohort and 0.78 in the external validation cohort, compared to substantially lower sensitivity for AFP alone. For HCC specifically, the combined model produced AUCs of 0.94 (discovery) and 0.93 (validation). Exploratory analyses also suggested potential utility for intrahepatic cholangiocarcinoma (ICC), a subtype that currently lacks reliable early biomarkers.
Among the nine metabolites, nicotinamide (NAM) — a precursor to NAD+ and a form of vitamin B3 — stood out as both a strong diagnostic contributor and a mechanistically interesting oncometabolite. Mendelian randomization analysis using genome-wide association data supported a potential causal relationship between elevated circulating NAM levels and increased liver cancer risk, adding evidence beyond simple correlation. In vitro functional experiments with HCC cell lines showed that exogenous NAM treatment promoted cell proliferation, migration, and invasion in a dose-dependent manner. Mechanistically, NAM enhanced NAD+ biosynthesis, which activated the deacetylase SIRT1. SIRT1 in turn deacetylated HIF1α, protecting it from proteasomal degradation and increasing its stability. Stabilized HIF1α then transcriptionally upregulated glycolytic enzymes and activated MAPK signaling, creating a pro-tumorigenic metabolic environment.
The study's broad geographic sampling across China strengthens generalizability within high-risk HBV-endemic populations, and the use of an entirely independent external cohort from different centers is a notable methodological strength. However, several caveats apply: the cohort is exclusively Chinese, limiting direct applicability to Western populations with different etiologies (e.g., NAFLD-driven HCC). The retrospective design and the mechanistic NAM findings — derived solely from cell lines — require prospective clinical trials and in vivo validation before clinical translation. Nonetheless, this work represents one of the largest serum metabolomics studies in liver cancer to date and offers both a near-term diagnostic tool and a potential therapeutic target.
Key Findings
- Combined 9-metabolite + AFP model achieved AUC of 0.92 in the 1,924-person discovery cohort and 0.93 in the 225-person independent external validation cohort
- Early-stage (Stage I–II) liver cancer sensitivity reached 0.85 in discovery and 0.78 in external validation, substantially exceeding AFP alone (~45.5% sensitivity)
- For hepatocellular carcinoma specifically, the combined model yielded AUCs of 0.94 (discovery) and 0.93 (validation)
- Mendelian randomization analysis supported a potential causal link between elevated circulating nicotinamide (NAM) levels and liver cancer risk
- In vitro experiments showed NAM promotes HCC cell proliferation, migration, and invasion through NAD+/SIRT1-mediated stabilization of HIF1α
- Stabilized HIF1α activated both glycolytic gene expression and MAPK signaling pathways downstream of the NAM/NAD+/SIRT1 axis
- The 13-center study covered 5 provinces in China (n=2,149 total), representing one of the largest multicenter serum metabolomics datasets in liver cancer research
Methodology
This multicenter, retrospective study enrolled 1,924 participants (discovery) and 225 participants (external validation) from 13 clinical centers across China. Serum metabolomics was performed using NPELDI-MS on a microarray platform for high-throughput, reproducible profiling. A nine-metabolite diagnostic signature was selected via LASSO regression and combined with AFP in a logistic regression model; performance was assessed by AUC, sensitivity, and specificity. Mendelian randomization used genome-wide genetic instruments for NAM to test causality, and mechanistic in vitro work used HCC cell lines with NAM supplementation, SIRT1 inhibition, and HIF1α knockdown experiments.
Study Limitations
The study population is exclusively Chinese, predominantly HBV-positive, limiting generalizability to Western populations where NAFLD/NASH-driven HCC predominates. Mechanistic findings for NAM are based solely on cell-line experiments and Mendelian randomization, lacking in vivo animal models or prospective clinical validation. The retrospective design introduces potential selection bias, and the authors note that prospective multicenter trials are needed before clinical implementation.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.
Enter your email to subscribe: