Skip to content

Hepatocellular carcinoma — molecular landscape

TL;DR — HCC is genomically heterogeneous, but recurrent alterations converge on telomere maintenance, Wnt/β-catenin, TP53/cell-cycle control, chromatin remodeling, oxidative stress, and growth-factor signaling (Schulze 2015, PMID 25822088). TERT promoter alteration is often early; CTNNB1 and TP53 define partly opposing biological classes, and etiology leaves mutational signatures. Unlike lung cancer, this map has not produced routine mutation-directed therapy: prospective sequencing found potentially actionable alterations but limited successful matching and no validated genomic treatment algorithm (Harding 2019, PMID 30373752). Wnt/CTNNB1 activation is a compelling immune-exclusion mechanism in models and retrospective human data, but is not yet a clinically validated rule for withholding immunotherapy (Ruiz de Galarreta 2019, PMID 31186238; Pinyol 2019, PMID 30617138). Spatial and single-cell studies show that the immune microenvironment is multicellular and dynamic, making a one-biopsy classifier intrinsically incomplete (Lu 2022, PMID 35933472).

The genomic backbone

Exome sequencing of 243 tumors linked recurrent alterations and mutational signatures to risk factors. Major groups involved TERT promoter, CTNNB1, TP53, AXIN1, chromatin-remodeling genes, and oxidative-stress pathways; approximately 28% of tumors carried an alteration potentially targetable by an FDA-approved drug at the time, but dominant drivers were not directly druggable (Schulze 2015, PMID 25822088).

Pathway Recurrent alteration/examples Biological consequence Clinical status
Telomere maintenance TERT promoter; viral insertion/amplification routes Replicative immortality Common and early; no routine TERT-directed therapy
Wnt/β-catenin CTNNB1, AXIN1 Differentiation, proliferation, immune exclusion Candidate IO-resistance axis, not validated selector
Cell cycle/genome integrity TP53, CDKN2A, RB-pathway changes Genomic instability and aggressive phenotype No routine mutation-directed agent
Chromatin remodeling ARID1A, ARID2 and related genes Transcriptional-state alteration Investigational vulnerabilities
Oxidative stress NFE2L2, KEAP1 Stress adaptation and metabolism Investigational
RAS/MAPK/PI3K growth signaling Low-frequency alterations across pathway Proliferation and survival Not a common clinically dominant addiction
FGF signaling/amplification Selected tumors Growth and angiogenesis Biomarker strategies remain investigational

The important negative finding is structural: many alterations are tumor suppressor losses or transcriptional-state regulators, which are harder to inhibit than an activated kinase.

Etiology leaves molecular traces

Exposure/etiology Reported molecular association Interpretation
HBV Viral integration; TP53-associated patterns in some cohorts Direct viral and inflammatory carcinogenesis
HCV Chronic inflammation and fibrosis; heterogeneous genomic classes No single HCV-specific actionable driver
Aflatoxin B1 Characteristic mutational signature and TP53 pattern Molecular epidemiology can recover exposure history
Alcohol/tobacco Exposure-associated mutational signatures Co-exposure and geography complicate attribution
MASLD/NAFLD TERT promoter mutation and chromosome 8p loss reported Supports partly distinct evolutionary routes (Ki Kim 2016, PMID 27511114)

Associations are probabilistic. A mutation does not reliably disclose one patient's etiology, and tumors from the same etiology occupy multiple expression and immune classes.

Evolution from diseased liver to HCC

Carcinogenesis can be viewed as nested selection:

  1. chronic injury creates inflammation and regenerative pressure;
  2. fibrosis changes matrix, perfusion, and cellular communication;
  3. hepatocyte clones with survival advantage expand;
  4. telomere crisis and TERT reactivation enable persistence;
  5. additional driver and copy-number events establish malignant growth;
  6. immune editing and stromal remodeling permit progression and spread.

This sequence is not universal. HBV integration can contribute before cirrhosis; some MASLD-HCC is diagnosed in non-cirrhotic liver; multifocal tumors may be independent primaries or intrahepatic metastases.

Expression and phenotype classes

Molecular classifications repeatedly separate proliferative/poorly differentiated tumors from more differentiated, often CTNNB1-associated tumors. Immune-rich and immune-excluded subclasses cut across these axes. A review of phenotypic diversity concluded that classifications had not yet formed a unified corpus ready for clinical practice, despite extensive molecular data (Désert 2018, PMID 30386103).

Broad class Typical features Possible implication Boundary
Proliferation class Cell-cycle/growth signaling; poorer differentiation Aggressive behavior; target discovery Multiple overlapping taxonomies
Non-proliferation/differentiated class Hepatocyte-like programs; CTNNB1 enrichment in a subset Immune exclusion in some tumors Not synonymous with indolence
Immune-active class T-cell infiltration and inflammatory signaling Potential checkpoint sensitivity Exhaustion and suppressive cells coexist
Immune-excluded class Wnt/β-catenin signaling; low dendritic/T-cell recruitment Candidate primary IO resistance Requires prospective validation

The same tumor can contain regions with different classes, and prior locoregional or systemic therapy can reshape the phenotype.

Wnt/β-catenin and immune exclusion

In genetically engineered mouse models and human correlates, β-catenin activation impaired antitumor immune surveillance and produced resistance to anti-PD-1 therapy (Ruiz de Galarreta 2019, PMID 31186238). Clinical commentary linked Wnt/CTNNB1 alterations with a “cold” immune-excluded class and proposed them as candidate resistance biomarkers (Pinyol 2019, PMID 30617138).

This evidence supports a hypothesis, not a standard-of-care exclusion test:

  • retrospective treatment cohorts are small and selected;
  • genomic alteration is an imperfect proxy for pathway activation;
  • combination regimens may overcome a mechanism relevant to monotherapy;
  • biopsy timing and spatial heterogeneity matter;
  • prospective interaction testing is required to show predictive—not merely prognostic—value.

The multicellular tumor microenvironment

HCC develops in chronically inflamed, fibrotic tissue containing malignant hepatocytes, endothelial cells, fibroblasts, macrophages, dendritic cells, T cells, B cells, and other immune populations.

A single-cell atlas sampled primary and metastatic HCC across four tissue sites, identified heterogeneous malignant hepatocytes, MMP9-positive tumor-associated macrophages, T-cell states, early tertiary lymphoid structures, and seven microenvironment-based subtypes associated with prognosis (Lu 2022, PMID 35933472).

Compartment Candidate role Translational question
CD8 T cells Cytotoxic antitumor activity and exhaustion Which state predicts reversible checkpoint response?
Dendritic cells Antigen presentation and T-cell recruitment Can Wnt-driven exclusion be reversed?
Macrophages Inflammation, suppression, angiogenesis Which lineage/state is targetable without liver injury?
Endothelium Abnormal vasculature and immune trafficking How does VEGF blockade normalize access?
Fibroblasts Matrix and cytokine signaling Which subtypes promote resistance?
Tertiary lymphoid structures Local adaptive immune organization Location and maturation may determine effect

Sampling a single core at one time cannot represent all compartments or metastases.

Why angiogenesis and immunity intersect

VEGF signaling supports abnormal vasculature and immunosuppression. The clinical success of atezolizumab plus bevacizumab is therefore biologically compatible with simultaneous vascular and immune modulation, although the phase 3 trial established efficacy rather than proving a specific microenvironmental mechanism (Finn 2020, PMID 32402160).

Mechanistic biomarkers must distinguish:

  • prognostic association with aggressive disease;
  • pharmacodynamic change after treatment;
  • predictive interaction with one therapy versus another;
  • causal mediation of clinical benefit.

Precision-oncology experience

Prospective next-generation sequencing in HCC evaluated whether tumor alterations could match patients to targeted or immune therapies. The work demonstrated feasibility and candidate associations, including a Wnt/CTNNB1 resistance signal, but also exposed low matching rates and the gap between “actionable on a report” and an HCC-effective therapy (Harding 2019, PMID 30373752).

Barrier Consequence
Dominant undruggable drivers Few direct genotype-agent pairs
Tumor suppressor loss Restoration is harder than kinase inhibition
Low-frequency targets Large screening effort for small trial cohorts
Cirrhosis and Child-Pugh restrictions Excludes many genomically eligible patients
Tissue scarcity Diagnosis often made without biopsy
Heterogeneity One sample may miss resistant clones
Rapid clinical decline Sequencing turnaround can exceed decision window
Etiology/microenvironment interaction DNA alone incompletely predicts IO response

Liquid and spatial molecular profiling

Circulating tumor DNA, methylation, extracellular vesicles, and proteins could sample multiple tumor sites and support early detection or resistance monitoring. Their tumor fraction is lowest precisely in small early cancers, while cirrhosis creates a noisy background.

Spatial transcriptomics and multiplex imaging preserve neighborhood information lost in dissociated single-cell data. They can test whether immune cells are absent, excluded at the margin, or present but dysfunctional. Clinical translation requires standardized tissue handling, reproducible features, and prospective treatment interaction.

A practical molecular-testing position

Routine broad sequencing is not required to choose standard HCC therapy because no validated genomic algorithm ranks approved regimens (Singal 2023, PMID 37199193). Testing is most defensible when:

  • pathology is atypical or suggests mixed hepatobiliary cancer;
  • a molecularly selected trial is available;
  • rare targetable biology would materially change treatment;
  • adequate tissue can be obtained safely;
  • the result will return within the decision window.

Negative sequencing does not imply absence of therapeutic options; current systemic selection is clinical rather than mutation-driven.

From initiating lesion to branched tumor

Hepatocarcinogenesis is a sequence, not a single genomic event. TERT-promoter mutations and broad chromosome 1/8 changes appear in dysplastic nodules and small HCC and persist as trunk lesions, whereas later branches acquire pathway-specific alterations and copy-number complexity (Torrecilla 2017, PMID 28843658). HBV adds direct routes: integration can cause insertional mutagenesis and genomic instability, while HBx and preS/S proteins alter transcription, unfolded-protein response, proliferation control, and chromatin state (Levrero 2016, PMID 27084040). Whole-genome/exome/transcriptome sequencing further identifies HBV integrations and recurrent TP53, CTNNB1, AXIN1, chromatin-regulator, and cell-cycle lesions, but most are not druggable in routine practice (Jhunjhunwala 2014, PMID 25159915; Lee 2015, PMID 26523267).

Molecular axis Phenotype Clinical implication and limit
CTNNB1/Wnt Well differentiated, cholestatic, microtrabecular/pseudoglandular, relatively non-inflamed In 343 resected tumors CTNNB1 mutation prevalence was 40%; immune exclusion is a hypothesis for resistance, not a validated exclusion biomarker (Calderaro 2017, PMID 28532995)
TP53/cell cycle Poor differentiation, compact/pleomorphic morphology, vascular invasion TP53 mutation prevalence was 21% in the same series and largely mutually exclusive with CTNNB1 (Calderaro 2017, PMID 28532995)
TERT/telomere maintenance Early trunk gatekeeper with later promoter mutation/integration events Biologically common but not yet a routine treatment selector (Torrecilla 2017, PMID 28843658)
Onco-fetal endothelium/macrophages PLVAP/VEGFR2 endothelial and FOLR2 macrophage programs Links VEGF/NOTCH signaling to an immunosuppressive ecosystem and provides a mechanistic rationale for vascular-immune combinations (Sharma 2020, PMID 32976798)
POSTN-positive fibroblasts Spatial barrier to T-cell infiltration Compiled single-cell analysis across 220 samples associates this CAF state with lower immunotherapy efficacy; prospective assay validation is absent (Wang 2024, PMID 39067872)
MVI-associated ecosystem Cycling T cells, LAMP3 dendritic cells, TREM2 macrophages, myofibroblasts and arterial endothelial cells Single-cell analysis of 46,789 cells maps invasion-associated states but does not yet predict occult MVI with clinical-grade accuracy (Li 2024, PMID 37972953)
Epigenetic dysregulation DNA methylation, histone/chromatin remodeling, non-coding RNA changes Reversibility is attractive, yet no epigenetic classifier is a standard treatment companion diagnostic (Nagaraju 2022, PMID 34324953)

Spatial and longitudinal heterogeneity

Single-cell analysis of 57,000 malignant and non-malignant cells from 46 HCC/iCCA biopsies showed that treatment can select and remodel functional tumor-cell states; a baseline biopsy therefore samples a moving system (Ma 2021, PMID 34216724). In a neoadjuvant cabozantinib-nivolumab study, 5 of 15 patients had a pathologic response; spatial transcriptomics linked responding tumors to immune-rich, pro-inflammatory fibroblast contexts and identified distinct resistance/recurrence programs (Zhang 2023, PMID 37723590). NASH-associated HCC spatial proteomics mapped more than 750,000 cells and found immune cells concentrated in adjacent tissue but depleted toward tumor, with exhausted PD-1-positive CD8 cells connected to suppressive myeloid populations (Li 2024, PMID 37733002).

Histology remains an accessible molecular proxy. Macrotrabecular-massive, steatohepatitic, lymphocyte-rich, and cholestatic patterns enrich for different pathways, although no morphology is perfectly specific (Calderaro 2019, PMID 31195064). Lymphoepithelioma-like HCC has dense lymphocytes and fewer Wnt/Notch alterations in a small genomic series, supporting but not proving a tumor-intrinsic route to immune visibility (Chan 2019, PMID 31168847). Combined HCC-cholangiocarcinoma can carry TERT and TP53 alterations across both morphologic components, consistent with a shared clone despite divergent differentiation (Joseph 2019, PMID 30690729).

Controversies

  • Stable subtype versus plastic state. Trunk mutations are stable, but transcriptional and immune states change with location and therapy; both descriptions are true at different biological levels (Torrecilla 2017, PMID 28843658; Ma 2021, PMID 34216724).
  • Wnt as exclusion biomarker. Preclinical and retrospective evidence links Wnt activation to immune exclusion, but prospective clinical genotyping has not established sufficient negative predictive value to withhold checkpoint therapy (Ruiz de Galarreta 2019, PMID 31186238; Harding 2019, PMID 30373752).
  • Sequencing breadth versus actionability. Prospective genotyping detects alterations, yet matching rates and proven HCC-specific targeted options remain low; tissue may be more useful for trial enrollment and mixed-histology diagnosis than standard drug selection (Harding 2019, PMID 30373752; Rebouissou 2020, PMID 31954487).
  • Etiology-specific immunity. NASH-associated immune architecture differs from viral HCC, but etiology is an imperfect surrogate for tumor-level immune state and should not alone determine therapy (Li 2024, PMID 37733002; Tang 2023, PMID 38007237).

Open questions

  • Can prospective randomized data validate CTNNB1/Wnt activation as a treatment-interaction biomarker rather than a prognostic marker (Pinyol 2019, PMID 30617138)?
  • Which synthetic-lethal strategy can exploit TERT, TP53, or chromatin-remodeler alterations?
  • How many regions and time points are required to represent clinically relevant HCC heterogeneity (Lu 2022, PMID 35933472)?
  • Can single-cell or spatial features be reduced to a deployable assay without losing predictive information?
  • Why have nominally actionable alterations rarely produced successful HCC matching (Harding 2019, PMID 30373752)?

References

  1. Schulze K, et al. Exome sequencing of hepatocellular carcinomas identifies new mutational signatures and potential therapeutic targets. Nat Genet. 2015;47:505-511. PMID 25822088
  2. Désert R, Nieto N, Musso O. Dimensions of hepatocellular carcinoma phenotypic diversity. World J Gastroenterol. 2018;24:4536-4547. PMID 30386103
  3. Ruiz de Galarreta M, et al. β-Catenin activation promotes immune escape and resistance to anti-PD-1 therapy in hepatocellular carcinoma. Cancer Discov. 2019;9:1124-1141. PMID 31186238
  4. Pinyol R, Sia D, Llovet JM. Immune exclusion-Wnt/CTNNB1 class predicts resistance to immunotherapies in HCC. Clin Cancer Res. 2019;25:2021-2023. PMID 30617138
  5. Lu Y, et al. A single-cell atlas of the multicellular ecosystem of primary and metastatic hepatocellular carcinoma. Nat Commun. 2022;13:4594. PMID 35933472
  6. Harding JJ, et al. Prospective genotyping of hepatocellular carcinoma: clinical implications of next-generation sequencing for matching patients to targeted and immune therapies. Clin Cancer Res. 2019;25:2116-2126. PMID 30373752
  7. Ki Kim S, et al. TERT promoter mutations and chromosome 8p loss are characteristic of nonalcoholic fatty liver disease-related hepatocellular carcinoma. Int J Cancer. 2016;139:2512-2518. PMID 27511114
  8. Finn RS, et al. Atezolizumab plus bevacizumab in unresectable hepatocellular carcinoma. N Engl J Med. 2020;382:1894-1905. PMID 32402160
  9. Singal AG, et al. AASLD Practice Guidance on prevention, diagnosis, and treatment of hepatocellular carcinoma. Hepatology. 2023;78:1922-1965. PMID 37199193
  10. Lee JS, et al. The mutational landscape of hepatocellular carcinoma. Clin Mol Hepatol. 2015;21:220-229. PMID 26523267
  11. Calderaro J, et al. Molecular and histological correlations in liver cancer. J Hepatol. 2019;71:616-630. PMID 31195064
  12. Calderaro J, et al. Histological subtypes of HCC are related to gene mutations and molecular classification. J Hepatol. 2017;67:727-738. PMID 28532995
  13. Rebouissou S, et al. Advances in molecular classification and precision oncology in HCC. J Hepatol. 2020;72:215-229. PMID 31954487
  14. Torrecilla S, et al. Trunk mutational events present minimal intra- and inter-tumoral heterogeneity in HCC. J Hepatol. 2017;67:1222-1231. PMID 28843658
  15. Jhunjhunwala S, et al. Diverse modes of genomic alteration in hepatocellular carcinoma. Genome Biol. 2014;15:436. PMID 25159915
  16. Ma L, et al. Single-cell atlas of tumor-cell evolution in response to therapy in HCC and iCCA. J Hepatol. 2021;75:1397-1408. PMID 34216724
  17. Li K, et al. Single-cell dissection of the ecosystem and molecular features underlying microvascular invasion in HCC. Hepatology. 2024;79:1293-1309. PMID 37972953
  18. Li M, et al. Spatial proteomics of the immune microenvironment in NASH-associated HCC. Hepatology. 2024;79:560-574. PMID 37733002
  19. Wang H, et al. POSTN-positive cancer-associated fibroblasts determine efficacy of immunotherapy in HCC. J Immunother Cancer. 2024;12. PMID 39067872
  20. Sharma A, et al. Onco-fetal reprogramming of endothelial cells drives immunosuppressive macrophages in HCC. Cell. 2020;183:377-394.e21. PMID 32976798
  21. Zhang S, et al. Spatial transcriptomics of neoadjuvant cabozantinib and nivolumab in HCC identifies mechanisms of resistance and recurrence. Genome Med. 2023;15:72. PMID 37723590
  22. Nagaraju GP, et al. Epigenetics in hepatocellular carcinoma. Semin Cancer Biol. 2022;86:622-632. PMID 34324953
  23. Levrero M, et al. Mechanisms of HBV-induced hepatocellular carcinoma. J Hepatol. 2016;64:S84-S101. PMID 27084040
  24. Chan AW, et al. Genomic landscape of lymphoepithelioma-like hepatocellular carcinoma. J Pathol. 2019;249:166-172. PMID 31168847
  25. Joseph NM, et al. Genomic profiling of combined hepatocellular-cholangiocarcinoma reveals similar genetics to HCC. J Pathol. 2019;248:164-178. PMID 30690729
  26. Tang W, et al. Single-cell RNA-sequencing atlas reveals an FABP1-dependent immunosuppressive environment in HCC. J Immunother Cancer. 2023;11. PMID 38007237