Unmet Clinical Need Liver cancer is the second leading cause of cancer death worldwide, with approximately 700,000 deaths annually. Hepatocellular carcinoma (HCC) is its dominant form, arising from chronic inflammation driven by hepatitis B and C viruses, alcohol use, obesity, and metabolic disease. Despite this enormous burden, sorafenib remains the only approved first-line drug for advanced HCC - a drug with modest survival benefit - and over ten drugs have failed in phase III trials.
The TCGA HCC Study The Cancer Genome Atlas Research Network performed the most comprehensive molecular analysis of HCC to date: whole exome sequencing and copy number profiling in 363 patients, plus DNA methylation, mRNA, microRNA, and proteomics in 196 patients. This multi-platform approach across hundreds of patients provides a definitive landscape of the molecular alterations driving HCC.
Key Findings Preview The study identified new significantly mutated driver genes (LZTR1, EEF1A1, SF3B1, SMARCA4), characterized metabolic reprogramming driven by mutation or epigenetic silencing of liver-specific genes (ALB, APOB, CPS1), defined three molecular subtypes with different prognoses, and identified multiple therapeutic targets including WNT signaling, MDM4, MET, VEGFA, and immune checkpoint proteins.
Most Common Drivers Among 26 statistically significantly mutated genes (SMGs), the most frequent were TERT promoter mutations (44%), TP53 (31%), and CTNNB1/beta-catenin (27%). TERT promoter mutations activate telomere maintenance and were associated with older age, male sex, and HCV positivity. TP53 was predominantly mutated in HBV-positive tumors. AXIN1 (8%) and CTNNB1 mutations together activate WNT/beta-catenin signaling, the most frequently altered pathway in HCC.
Novel Driver Genes Eight previously unrecognized HCC drivers were identified. LZTR1 (3%), an adaptor for CUL3 ubiquitin ligase complexes with germline mutations in schwannomatosis, was found mutated in HCC. EEF1A1 (translation elongation factor) was mutated in 10 tumors. SF3B1 (a splicing factor with driver mutations in blood cancers) and SMARCA4 (a SWI/SNF chromatin modifier) were also newly implicated in HCC.
Mutational Signatures Three independent mutational signatures were identified. Nine samples showed signatures matching aristolochic acid (a plant-derived carcinogen found in herbal medicines) with characteristic A:T-to-T:A transversions. Seven samples showed aflatoxin B1 (AFB1) signature with G:C-to-T:A transversions and TP53-R249S hotspot mutations. HBV-positive tumors showed much higher AFB1 activity, suggesting synergy between AFB1 exposure and HBV in driving HCC-specific mutation patterns.
Widespread Epigenetic Changes Genome-scale methylation profiling revealed extensive hypo- and hypermethylation in HCC tumors. CDKN2A (encoding the tumor suppressor p16) was epigenetically silenced by promoter hypermethylation in 53% of HCCs - far more common than its mutation rate of 4% - identifying DNA methylation as the dominant mechanism for CDKN2A inactivation in this cancer type.
Metabolic Reprogramming via Methylation Several liver-specific metabolic genes were silenced by hypermethylation: CPS1 (a rate-limiting urea cycle enzyme), ALB (albumin), and APOB (apolipoprotein B). Inactivation of these high-energy-cost liver-specific programs appears to divert cellular resources toward cancer-supportive metabolic pathways. For example, CPS1 silencing may shift from the urea cycle toward de novo pyrimidine synthesis via CAD, supporting rapid cell division.
HCC Methylation Subtypes Unsupervised clustering defined four hypermethylation clusters with distinct molecular and clinical associations. Cluster 3 was characterized by IDH1/2 mutations and a distinct DNA hypermethylation profile consistent with D-2-hydroxyglutarate-induced methylation dysregulation. Cluster 4 showed co-occurrence of CDKN2A silencing, TERT promoter mutations, and CTNNB1 mutations, associated with HCV infection.
Multi-Platform Subtyping Integrative clustering combining data from five molecular platforms (mRNA, miRNA, methylation, copy number, protein expression) identified three robust HCC subtypes (iCluster 1, 2, 3). This subtyping was validated in three independent HCC cohorts, confirming biological and clinical relevance beyond the TCGA dataset.
Subtype Characteristics One subtype was characterized by CTNNB1 activation, WNT pathway enrichment, and relatively better prognosis. Another subtype showed high copy number alterations, TP53 mutations, and HBV association. The third subtype, associated with poorest prognosis, was enriched in markers of immune checkpoint expression and p53 pathway dysregulation, and showed a distinct gene expression signature.
p53 Signature Predicts Survival Integrated analyses enabled development of a p53 target gene expression signature that significantly correlated with poor survival across all three independent validation cohorts. This multi-gene signature captures broader p53 pathway dysfunction beyond TP53 mutation status, potentially explaining why TP53 mutation alone is an imperfect prognostic biomarker in HCC.
Druggable Pathway Alterations The study systematically mapped potentially druggable alterations. WNT/beta-catenin signaling was altered in 37% of HCC (CTNNB1 mutations 27%, AXIN1 8%). Copy number amplifications of MET (7q31), VEGFA (6p21), MCL1 (1q21), and CCND1 (11q13) identify kinase and anti-apoptotic targets for which inhibitors exist. MDM4 amplification in tumors without TP53 mutation suggests MDM4 inhibitors could restore p53 function in this subset.
Immune Checkpoint Landscape Comprehensive immunophenotyping revealed that HCC tumors express immune checkpoint proteins. CTLA-4, PD-1, and PD-L1 expression was measured across the cohort, identifying subsets of HCC with high checkpoint expression that may be most responsive to checkpoint inhibitor immunotherapy. This analysis provided an early genomic basis for the clinical trials of nivolumab and other checkpoint inhibitors in HCC.
IDH1 as Emerging Target IDH1/2 mutations, which occur in a subset of HCC (particularly the hypermethylated cluster 3), generate 2-hydroxyglutarate and are being targeted by FDA-approved inhibitors in glioma and leukemia. The IDH-mutant HCC subtype may represent a patient population for IDH inhibitor clinical trials, and the methylation signature could serve as a biomarker.
miR-122 as Tumor Suppressor miR-122 is the most abundant liver-specific microRNA. This TCGA analysis confirmed significant downregulation of miR-122 in HCC, consistent with its role as a liver-specific tumor suppressor. Restoration of miR-122 in experimental models suppresses HCC growth, and this analysis provides genomic context for miR-122's clinical relevance.
Integration of Protein Expression Reverse phase protein array (RPPA) proteomics on a subset of tumors validated genomic findings and revealed pathway activation that was not always predictable from DNA alone. Protein-level analysis of key signaling nodes (AKT, mTOR, EGFR pathway components) complemented the mRNA data and provided functional readouts of pathway activity.
HBV Integration Analysis Viral integration analysis in HBV-positive tumors identified recurrent integration sites near cancer-related genes. HBV integration was more frequent and at different genomic loci compared to HCV, potentially explaining part of the epidemiological difference in HCC development between HBV and HCV-infected patients.
Biomarker-Matched Clinical Trials A critical gap identified by this study is the lack of biomarker-matched trials in HCC. The molecular landscape defined here - with druggable alterations in WNT, MET, VEGFA, IDH1, and immune checkpoints - provides the biological rationale for basket trials and biomarker-stratified studies. Translating genomic findings to matched therapies for HCC patients requires both further preclinical validation and redesigned clinical trial frameworks.
Combination Regimens The complexity of HCC's molecular landscape suggests that combinations of targeted agents and immunotherapy will be needed. The finding that different HCC subtypes have distinct molecular vulnerabilities (WNT-driven vs. TP53-mutant vs. immune checkpoint-high) suggests that optimal combinations will vary by subtype. Development of subtype-specific combination regimens is a high priority.
Applying the Atlas to Emerging Technologies As single-cell sequencing, spatial transcriptomics, and liquid biopsy technologies advance, the TCGA HCC dataset serves as a reference for interpreting tumor heterogeneity, tracking clonal evolution, and developing blood-based biomarkers that reflect the broader molecular landscape identified here.