Molecular therapies and precision medicine for hepatocellular carcinoma

Authors

Josep M. Llovet

Robert Montal

Daniela Sia

Richard S. Finn

Doi

10.1038/s41571-018-0073-4

PMID: 24798001 · DOI: 10.1038/s41571-018-0073-4 · Journal: Nature Reviews Clinical Oncology (2018)

TL;DR

Comprehensive review summarising the molecular landscape, classifications, and systemic therapies for hepatocellular carcinoma ([cancer_types/HCC|HCC]) as of mid-2018. Llovet and colleagues synthesise mutational and copy-number drivers from whole-exome sequencing of >1,200 patients, two principal molecular subclasses (proliferation vs non-proliferation) and an immune-status classification, and pivotal phase III trial results that established [drugs/sorafenib|sorafenib] and [drugs/lenvatinib|lenvatinib] as first-line and [drugs/regorafenib|regorafenib], [drugs/cabozantinib|cabozantinib], [drugs/ramucirumab|ramucirumab], and [drugs/nivolumab|nivolumab] as second-line options. The authors argue that only ~25% of HCCs harbour potentially actionable alterations and that — apart from [[drugs/ramucirumab|ramucirumab]] in AFP-high disease (REACH-2) — no biomarker-driven phase III trial has yet succeeded in HCC.

Cohort & data

  • Review of the molecular landscape of [[cancer_types/HCC|HCC]] integrating whole-exome sequencing data from 1,289 patients across multiple cohorts (mutation frequencies in Table 1) and SNP-array copy-number calls from 704 patients.
  • The largest single-cohort genomic analysis cited is the TCGA Research Network’s integrated molecular characterisation of 363 patients with HCC (and 196 with all five platforms: DNA copy number, methylation, mRNA, miRNA, [[methods/rppa|RPPA]]).
  • The page is registered against study_id: lihc_amc_prv (Asan Medical Center, Korea) in the wiki frontmatter; this is the canonical dataset slug the orchestrator has linked to PMID:24798001 (see warnings below).
  • Trial evidence base spans phase I–III studies of [[drugs/sorafenib|sorafenib]] (SHARP), [[drugs/lenvatinib|lenvatinib]] (REFLECT), [[drugs/regorafenib|regorafenib]] (RESORCE), [[drugs/cabozantinib|cabozantinib]] (CELESTIAL), [[drugs/ramucirumab|ramucirumab]] (REACH and REACH-2), and immune-checkpoint inhibitors [[drugs/nivolumab|nivolumab]] (CheckMate 040), [drugs/pembrolizumab|pembrolizumab] (KEYNOTE-224/240), [drugs/durvalumab|durvalumab] + [drugs/tremelimumab|tremelimumab], and [drugs/atezolizumab|atezolizumab] + [drugs/bevacizumab|bevacizumab].

Key findings

  • Mutational landscape (n = 1,289 WES patients): [genes/TERT|TERT] promoter mutations are present in 54% (range 44–59%) of HCCs and are the single most frequent somatic alteration; [genes/CTNNB1|CTNNB1] activating mutations in 29% (23–36%); [genes/TP53|TP53] loss-of-function in 28% (23–31%); [genes/AXIN1|AXIN1] in 7% (5–10%); [genes/ARID1A|ARID1A] in 8% (4–12%); [genes/ARID2|ARID2] in 7% (3–10%); [genes/ARID1B|ARID1B] in 1% (1–3%); [genes/RB1|RB1] LOF in 4%; [genes/CDKN2A|CDKN2A] LOF in 2%; [genes/PIK3CA|PIK3CA] activating in 2%; [genes/KRAS|KRAS]/[genes/NRAS|NRAS] activating in 1% each; [genes/EGFR|EGFR] activating in 1% (0–2%); [genes/NFE2L2|NFE2L2]/[genes/KEAP1|KEAP1] in 4% and 3%, respectively.
  • Copy-number landscape (n = 704 SNP-array): [genes/MYC|MYC] high-level focal amplification in 12% (4–18%); [genes/CCND1|CCND1] in 7%; [genes/FGF19|FGF19] in 6%; [genes/VEGFA|VEGFA] in 5% (1–8%); [[genes/TERT|TERT]] high-level focal amplification in 5%; [[genes/CDKN2A|CDKN2A]] homozygous deletion in 5%; [[genes/RB1|RB1]] homozygous deletion in 5%.
  • Only ~25% of HCCs harbour potentially targetable driver alterations with currently available drugs — most trunk mutations ([[genes/TERT|TERT]] promoter, [[genes/CTNNB1|CTNNB1]], [[genes/TP53|TP53]], [[genes/AXIN1|AXIN1]], [[genes/ARID1A|ARID1A]], [[genes/ARID1B|ARID1B]]) are not clinically actionable.
  • Molecular subclasses: ~50% of HCCs fall into a proliferation class (enriched for [[genes/TP53|TP53]] inactivation, [[genes/FGF19|FGF19]] and/or [[genes/CCND1|CCND1]] amplifications, [genes/EPCAM|EPCAM]/AFP/[genes/IGF2|IGF2] overexpression in the S2 progenitor subgroup, HBV aetiology, poor outcomes); the other ~50% fall into a non-proliferation class (canonical WNT via [[genes/CTNNB1|CTNNB1]], higher [[genes/TERT|TERT]] promoter mutation rate, HCV/alcohol aetiology, better outcomes).
  • Immune classification (Sia et al. 2017): ~30% of HCCs are an “immune class” with high immune infiltrate, [[genes/CD274|PD-L1]]/[[genes/PDCD1|PD-1]] expression and IFNγ signalling; ~25% are an “immune excluded” class characterised by [[genes/CTNNB1|CTNNB1]] mutations, low infiltrate, and overexpression of [genes/PTK2|PTK2]. Within the immune class, an “exhausted” subclass overexpresses TGFβ-regulated genes.
  • First-line phase III trial outcomes: SHARP — [[drugs/sorafenib|sorafenib]] vs placebo median OS 10.7 vs 7.9 months (HR 0.69, 95% CI 0.55–0.87, P<0.001). REFLECT — [[drugs/lenvatinib|lenvatinib]] non-inferior to [[drugs/sorafenib|sorafenib]], median OS 13.6 vs 12.3 months (HR 0.92, 95% CI 0.79–1.06); mRECIST ORR 24.1% (investigator) / 40.6% (masked review) for lenvatinib.
  • Frontline failures vs sorafenib: [drugs/brivanib|brivanib], [drugs/sunitinib|sunitinib], [drugs/linifanib|linifanib], [drugs/erlotinib|erlotinib], and ⁹⁰Y resin microsphere radioembolisation (SARAH and SIRveNIB, median OS 8.0–8.8 vs 9.9–10.0 months for sorafenib; HR 1.12–1.15).
  • Second-line phase III successes (post-sorafenib): RESORCE — [[drugs/regorafenib|regorafenib]] median OS 10.6 vs 7.8 months (HR 0.63, 95% CI 0.50–0.79, P<0.0001); CELESTIAL — [[drugs/cabozantinib|cabozantinib]] median OS 10.2 vs 8.0 months (HR 0.76, 95% CI 0.63–0.92, P=0.0049); REACH-2 — [[drugs/ramucirumab|ramucirumab]] in AFP≥400 ng/ml patients, median OS 8.5 vs 7.3 months (HR 0.71, 95% CI 0.53–0.95, P=0.0199). REACH-2 is the first positive biomarker-enriched phase III trial in HCC.
  • Second-line phase III failures: [drugs/everolimus|everolimus] (mTOR), [drugs/tivantinib|tivantinib] in MET-high HCC (placebo OS in MET-high group was 9.1 months — the longest ever reported), and trials targeting VEGF/FGF or HGF–MET signalling.
  • Immunotherapy results: CheckMate 040 — [[drugs/nivolumab|nivolumab]] investigator-assessed ORR 20% with 9.9-month median duration of response across the 214-patient dose-expansion cohort; OS 15.6 months in second-line patients (FDA accelerated approval based on 154 sorafenib-pretreated patients, ORR 14.3% by RECIST 1.1, 18.2% by mRECIST, DoR 16.6 months). KEYNOTE-224 — [[drugs/pembrolizumab|pembrolizumab]] ORR 16.3% in 104 patients, median OS 12.9 months. [[drugs/durvalumab|durvalumab]] monotherapy ORR 10%; [[drugs/durvalumab|durvalumab]] + [[drugs/tremelimumab|tremelimumab]] combination ORR 15%; [[drugs/atezolizumab|atezolizumab]] + [[drugs/bevacizumab|bevacizumab]] ORR 65% in 23 patients (FDA breakthrough designation).
  • PD-L1 IHC does not predict response to nivolumab or pembrolizumab in HCC, unlike in lung cancer, so alternative biomarkers are needed. MSI-high disease is rare in HCC (~3%).

Genes & alterations

  • [[genes/TERT|TERT]] — promoter activating mutation in 54% (most frequent HCC alteration); focal amplification in 5%; central to telomere maintenance pathway.
  • [[genes/TP53|TP53]] — loss-of-function mutation in 28% of HCCs; defines TCGA iClust3 subtype associated with chromosomal instability and poor prognosis; enriched in the proliferation class.
  • [[genes/CTNNB1|CTNNB1]] — activating mutation in 29% of HCCs; defines canonical-WNT non-proliferation subclass and the “immune-excluded” class; melanoma data suggest β-catenin activation drives T-cell exclusion and immune-checkpoint resistance.
  • [[genes/AXIN1|AXIN1]] — LOF in 7%; component of the WNT–β-catenin axis; trunk driver, not currently actionable.
  • [[genes/ARID1A|ARID1A]] / [[genes/ARID1B|ARID1B]] / [[genes/ARID2|ARID2]] — chromatin-modifier LOF (8%, 1%, 7%); part of the trunk-mutation set called out as non-actionable.
  • [[genes/FGF19|FGF19]] — high-level focal amplification in 5–10% of HCCs; oncogenic driver implicated in [[drugs/sorafenib|sorafenib]] resistance; predictive biomarker (by IHC) for FGFR4 inhibitors [[drugs/blu-554|BLU-554]], [[drugs/h3b-6527|H3B-6527]], and [[drugs/fgf401|FGF401]]. BLU-554 produced 16% response rate in FGFR4-driven (≥1% FGF19 IHC) patients vs 0% in FGFR4-negative.
  • [[genes/FGFR4|FGFR4]] — predominant hepatic FGFR; therapeutic target via selective inhibitors.
  • [[genes/MET|MET]] / [[genes/HGF|HGF]] — overexpression associated with poor prognosis and [[drugs/sorafenib|sorafenib]] resistance; [[drugs/tivantinib|tivantinib]] failed in MET-high HCC (mechanism may be MET-independent); [drugs/capmatinib|capmatinib] and other selective MET inhibitors still under evaluation; [[drugs/cabozantinib|cabozantinib]] contribution attributed in part to MET inhibition.
  • [[genes/VEGFA|VEGFA]] — high-level focal amplification in 5% of HCCs; angiogenic signalling is prominent in all HCC subclasses; central to mechanism of [[drugs/sorafenib|sorafenib]], [[drugs/lenvatinib|lenvatinib]], [[drugs/regorafenib|regorafenib]], [[drugs/cabozantinib|cabozantinib]], [[drugs/ramucirumab|ramucirumab]], and [[drugs/bevacizumab|bevacizumab]].
  • [[genes/KRAS|KRAS]] / [[genes/NRAS|NRAS]] — RAS mutations are rare in HCC (1% each; only 4.4% of 1,318 ctDNA samples in BEAMing screening), but a biomarker-enriched phase II of [drugs/refametinib|refametinib] plus [[drugs/sorafenib|sorafenib]] in 16 RAS-mutant patients reported median OS 12.7 months.
  • [[genes/RB1|RB1]] — LOF in 4%, homozygous deletion in 5%; rationale for CDK4/6 inhibitor trials ([drugs/palbociclib|palbociclib] in RB+ disease).
  • [[genes/CDKN2A|CDKN2A]] — LOF in 2%, homozygous deletion in 5%; epigenetic silencing also reported.
  • [[genes/MYC|MYC]] / [[genes/CCND1|CCND1]] — focal amplifications in 12% and 7%, respectively.
  • [genes/CD274|CD274] (PD-L1) and [genes/PDCD1|PDCD1] (PD-1) — defining markers of the immune-active class; expression not predictive of response in HCC trials of [[drugs/nivolumab|nivolumab]] or [[drugs/pembrolizumab|pembrolizumab]].
  • [genes/CTLA4|CTLA4] — target of [[drugs/tremelimumab|tremelimumab]] (phase II ORR 17.6% in 20 HCC patients) and [drugs/ipilimumab|ipilimumab] in combination regimens.
  • [[genes/PTK2|PTK2]] — overexpressed in the immune-excluded HCC class; oncogenic pathway associated with poor T-cell infiltration.
  • [[genes/EPCAM|EPCAM]] / [[genes/IGF2|IGF2]] — markers of the S2 progenitor-cell proliferation subclass; IGF2 is epigenetically upregulated and proposed as an actionable target.
  • [genes/AXL|AXL] — inhibited by [[drugs/cabozantinib|cabozantinib]] alongside MET and VEGFRs.

Clinical implications

  • First-line standard of care: [[drugs/sorafenib|sorafenib]] remained standard for Child–Pugh A, BCLC stage C (or post-TACE stage B) patients from 2007 through 2017; [[drugs/lenvatinib|lenvatinib]] is an alternative non-inferior to [[drugs/sorafenib|sorafenib]] (except in patients with main portal vein thrombosis or >50% liver involvement).
  • Second-line standards: [[drugs/regorafenib|regorafenib]] (FDA approved) for patients who tolerated and progressed on [[drugs/sorafenib|sorafenib]]; [[drugs/cabozantinib|cabozantinib]] regardless of prior-line count (≤2); [[drugs/ramucirumab|ramucirumab]] specifically in patients with AFP ≥400 ng/ml (the first biomarker-selected drug in HCC); [[drugs/nivolumab|nivolumab]] (FDA accelerated approval) and [[drugs/pembrolizumab|pembrolizumab]] as immune-checkpoint options.
  • Biomarker-driven precision medicine: [[drugs/ramucirumab|ramucirumab]] in serum AFP-high (≥400 ng/ml) HCC is the only proven phase III biomarker-driven indication in HCC. FGFR4 inhibitors ([[drugs/blu-554|BLU-554]], [[drugs/h3b-6527|H3B-6527]], [[drugs/fgf401|FGF401]]) require [[genes/FGF19|FGF19]] expression (typically by IHC) for enrichment.
  • No predictive biomarkers identified for sorafenib or lenvatinib despite extensive companion studies; the SHARP-companion biomarker substudy showed only a non-significant trend for greater benefit in [genes/KIT|KIT]-high, low-plasma-HGF tumours.
  • Combination strategies in development: anti-PD-(L)1 plus anti-CTLA-4 ([[drugs/durvalumab|durvalumab]] + [[drugs/tremelimumab|tremelimumab]], NCT03298451); anti-VEGF/VEGFR plus anti-PD-(L)1 ([[drugs/atezolizumab|atezolizumab]] + [[drugs/bevacizumab|bevacizumab]] — FDA breakthrough, NCT03434379; [[drugs/lenvatinib|lenvatinib]] + [[drugs/pembrolizumab|pembrolizumab]] NCT03006926; [[drugs/regorafenib|regorafenib]] + [[drugs/pembrolizumab|pembrolizumab]] NCT03347292; [[drugs/ramucirumab|ramucirumab]] + [[drugs/durvalumab|durvalumab]] NCT02572687).
  • No adjuvant therapy has demonstrated benefit after curative resection or ablation: the phase III STORM trial of [[drugs/sorafenib|sorafenib]] was negative for recurrence-free survival; a multi-gene signature derived from STORM tumours remains to be prospectively validated.
  • TACE combinations failed: [[drugs/sorafenib|sorafenib]] + TACE (SPACE phase II, TACE 2 phase III) and [[drugs/brivanib|brivanib]] + TACE did not improve TTP or OS.
  • Pembrolizumab is approved (tumour-agnostic) for MSI-H / dMMR HCC, but this molecular subset comprises only ~3% of HCCs.

Limitations & open questions

  • The authors call out that most trunk drivers in HCC ([[genes/TERT|TERT]] promoter, [[genes/CTNNB1|CTNNB1]], [[genes/TP53|TP53]], [[genes/AXIN1|AXIN1]], [[genes/ARID1A|ARID1A]], [[genes/ARID1B|ARID1B]]) are not currently druggable; drugs blocking the effects of [[genes/CTNNB1|CTNNB1]] activation are highlighted as a priority for precision-medicine development.
  • No predictive biomarkers exist for response to [[drugs/sorafenib|sorafenib]], [[drugs/lenvatinib|lenvatinib]], [[drugs/regorafenib|regorafenib]], [[drugs/cabozantinib|cabozantinib]], or HCC immune-checkpoint inhibitors. PD-L1 IHC, in contrast to lung cancer, does not stratify [[drugs/nivolumab|nivolumab]] or [[drugs/pembrolizumab|pembrolizumab]] response in HCC.
  • The interpretation of the failed [[drugs/tivantinib|tivantinib]] phase III in [[genes/MET|MET]]-high HCC is ambiguous: high MET expression may itself be a negative prognostic marker; the IHC assay/cut-off may have been miscalibrated; tivantinib’s mechanism may be MET-independent.
  • Liquid biopsy (ctDNA and circulating-cell RNA) is proposed as the future tissue-acquisition strategy for advanced-stage HCC, but its analytical and clinical validity in HCC remain unestablished.
  • Sequencing of available agents and head-to-head comparisons among second-line therapies have not been performed; treatment selection currently relies on patient-specific factors (first-line drug, duration of response, tolerance, performance status).
  • For the [[cancer_types/HCC|HCC]] immune-excluded class with [[genes/CTNNB1|CTNNB1]] mutations, predicted resistance to immune-checkpoint inhibitors (extrapolated from melanoma) has not been clinically confirmed and remains an open question.

Citations from this paper used in the wiki

  • “an average of 40–60 somatic alterations are detected in protein-coding regions of the genome … only ~25% of HCCs harbour alterations that are potentially targetable with existing drugs” (p. 3).
  • Table 1: WES mutation frequencies in n=1,289 HCCs and SNP-array CNAs in n=704 HCCs — [[genes/TERT|TERT]] promoter 54%, [[genes/CTNNB1|CTNNB1]] 29%, [[genes/TP53|TP53]] 28%, [[genes/FGF19|FGF19]] amp 6%, [[genes/VEGFA|VEGFA]] amp 5%, etc. (pp. 32–33).
  • “the TCGA Research Network … 363 patients with HCC … integration of up to 5 other platforms … for 196 tumours yielded 3 subtypes” iClust1/2/3 (p. 4).
  • “30% of HCCs could be classified into an ‘immune class’ … an ‘immune excluded class’ accounting for ~25% of HCCs was characterized by T cell exclusion from the TME and [[genes/CTNNB1|CTNNB1]] mutations” (p. 4).
  • SHARP trial: “median overall survival 10.7 months versus 7.9 months; HR 0.69, 95% CI 0.55–0.87; P < 0.001” (p. 6).
  • REFLECT trial: “[[drugs/lenvatinib|lenvatinib]] was found to be non-inferior to [[drugs/sorafenib|sorafenib]] … median 13.6 months versus 12.3 months; HR 0.92, 95% CI 0.79–1.06” (p. 7).
  • RESORCE trial: “median overall survival of patients who had HCC progression on [[drugs/sorafenib|sorafenib]] from 7.8 months with placebo to 10.6 months with [[drugs/regorafenib|regorafenib]] (HR 0.63, 95% CI 0.50–0.79; P < 0.0001)” (p. 9).
  • CELESTIAL trial: “median overall survival of 10.2 months in the [[drugs/cabozantinib|cabozantinib]] group versus 8.0 months in the placebo group (HR 0.76, 95% CI 0.63–0.92; P = 0.0049)” (p. 9).
  • REACH-2: “[[drugs/ramucirumab|ramucirumab]] becomes the first agent with a demonstrated clinical benefit for a biomarker-selected population of patients with HCC” — AFP ≥400 ng/ml, median OS 8.5 vs 7.3 months (HR 0.71, 95% CI 0.53–0.95; P = 0.0199) (p. 10).
  • CheckMate 040: “investigator-assessed ORR was 20% … duration of response of 9.9 months … Overall survival for patients in the second-line setting was 15.6 months” (p. 11).
  • FGFR4: “[[drugs/blu-554|BLU-554]] … a response rate of 16% to this agent in an FGFR4-driven group (defined by ≥1% tumour expression of [[genes/FGF19|FGF19]] by IHC) versus 0% in the FGFR4-negative group” (p. 14).
  • Refametinib + sorafenib: “only 59 of 1,318 samples (4.4%) had detectable RAS mutations … led to a median overall survival of 12.7 months in 16 patients” (p. 14).

This page was processed by crosslinker on 2026-05-11.