Genetic screen of blood proteins flags INHBC as a driver of liver cancer and a drug target
Proteome-wide Mendelian randomization finds 27 plasma proteins linked to liver cancer risk. Blocking the INHBC receptor ACVR2B shrank tumors in mice.
Summary
Liver cancer has few treatable molecular targets. Researchers combined genetic data from 816 primary liver cancer cases and 631,599 controls with two large blood-protein datasets, testing 3,907 proteins using Mendelian randomization. They found 27 circulating proteins with likely causal links to liver cancer risk. The top candidate, INHBC (a TGF-β family ligand), appeared to activate ACVR2B–Smad2 signaling, promoting liver cancer cell growth and invasion in lab experiments. In a mouse xenograft model, Bimagrumab, an ACVR2B-blocking antibody already in clinical development, reduced INHBC-driven tumor volume by about 42%. Genetic overlap with metabolic liver disease was also seen at the NCAN locus, and ACVR2B and C1QA were prioritized as druggable targets. The findings are preclinical and hypothesis-generating.
Detailed Summary
Primary hepatic carcinoma (PHC), which includes hepatocellular carcinoma and intrahepatic cholangiocarcinoma, is the third leading cause of cancer death worldwide, with five-year survival below 20%. Drugs such as sorafenib and lenvatinib help only some patients, and resistance is common. Fewer than 10% of patients qualify for genotype-directed therapy. Known genetic risk loci such as PNPLA3 and TM6SF2 explain little of the heritability, so new causal drivers and drug targets are needed.
The team ran a proteome-wide Mendelian randomization (PW-MR) study. They meta-analyzed two PHC GWAS datasets: UK Biobank (168 cases) and FinnGen (648 cases), for 816 cases and 631,599 controls in total. They paired these with plasma protein quantitative trait loci from deCODE (SomaScan, 35,559 Icelanders) and UK Biobank Pharma Proteomics (Olink, 54,219 participants). Instruments were cis-pQTLs (within ±1 Mb of the gene, P < 5×10⁻⁸, LD-clumped, F > 10). Analyses used Wald ratio or inverse-variance weighted MR, with FDR correction, reverse MR, heterogeneity and pleiotropy tests, MR-PRESSO, and Bayesian colocalization (PPH4 ≥ 0.7). Follow-up work included FUMA annotation, STRING protein-interaction networks and DrugBank druggability profiling. Experiments used recombinant INHBC in LO2 normal liver cells and HepG2 liver cancer cells (CCK-8, EdU, wound-healing, Transwell and qPCR assays, plus Western blotting).
Of 3,907 proteins tested, 27 had significant causal associations with PHC risk at FDR < 0.05. INHBC, a TGF-β superfamily ligand, emerged as the leading pathogenic candidate. Mechanistic experiments indicated that INHBC activates ACVR2B–Smad2 signaling to drive liver cancer cell proliferation and invasion. In a mouse xenograft model, Bimagrumab reduced tumor volume by about 42% versus controls (P = 0.008). Cross-phenotype analyses linked PHC genetically to metabolic liver diseases, with colocalization supporting a shared causal variant at the NCAN locus (PPH4 = 0.782). Druggability profiling prioritized ACVR2B and C1QA as actionable targets.
The work suggests that the INHBC–ACVR2B axis contributes to liver cancer and that existing ACVR2B inhibitors could be repurposed. It also fits with growing evidence that metabolic liver disease underlies much liver cancer risk.
Several caveats apply. The case count is small for a GWAS, and the case definition is broad. The datasets are of largely European ancestry. Genetically proxied plasma levels may not reflect tissue-level or lifelong biology. Experimental validation appears limited to one cancer cell line and a xenograft model. The full results and discussion sections were not available in the provided text, so these details come from the abstract and methods.
Key Findings
- Proteome-wide MR of 3,907 plasma proteins identified 27 with significant causal links to primary liver cancer risk (FDR < 0.05).
- INHBC was the top pathogenic candidate and appeared to activate ACVR2B–Smad2 signaling to promote liver cancer cell proliferation and invasion.
- Bimagrumab, an ACVR2B-blocking antibody, cut INHBC-driven xenograft tumor volume by about 42% versus controls (P = 0.008).
- Colocalization showed a shared causal variant between PHC and metabolic liver disease at the NCAN locus (PPH4 = 0.782).
- Druggability profiling prioritized ACVR2B and C1QA as actionable targets for liver cancer.
Methodology
Two-sample proteome-wide Mendelian randomization used cis-pQTLs from deCODE and UKB-PPP against a meta-analysis of UK Biobank and FinnGen PHC GWAS (816 cases, 631,599 controls), with FDR correction, reverse MR, sensitivity tests and Bayesian colocalization. Prioritized targets were annotated with FUMA, STRING and DrugBank. INHBC was tested in HepG2 and LO2 cells and in a mouse xenograft model with Bimagrumab.
Study Limitations
The PHC GWAS is small (816 cases) and mixes cancer subtypes, which limits power and specificity, and ancestry is mostly European. Genetically predicted plasma protein levels may not capture tissue-level effects, and functional validation used a single liver cancer cell line and a xenograft model. The provided full text was truncated after the methods, so results and discussion details were not independently verified.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.
Enter your email to subscribe: