Hepatocellular carcinoma (HCC)

Overview

• Epidemiology: ~55,000 hepatobiliary cancers/yr in the US (HCC, cholangiocarcinoma, gallbladder ca). (US 2025: ~42,000 liver ca/yr; HCC ~75 to 85% of primary liver ca.)
• Risk factors: cirrhosis (any cause), HBV (can occur without cirrhosis), HCV, alcohol, MASLD/MASH, aflatoxin, hereditary hemochromatosis, α1-antitrypsin deficiency, Wilson disease. In the US and Western countries, HCV, alcohol, and MASLD/MASH are the leading causes (MASLD rising as HCV declines). HBV vaccination and HCV treatment reduce HCC.
• Surveillance: US ± AFP q6 mo in cirrhosis (any etiology), HBV without cirrhosis (Asian men >40, women >50, family history of HCC), and non-cirrhotic HBV at elevated risk (high viral load is one risk factor; surveillance guided by overall risk: demographics, family history, validated risk scores). AFP-L3 (Lens culinaris agglutinin-reactive fraction of AFP) is more HCC-specific than total AFP but is used for risk stratification, not routine surveillance (US +/- AFP) or diagnosis. AFP sensitivity ~60%; not a stand-alone test, but trends are useful.
• Fibrolamellar HCC: rare, adolescents/young adults (median age ~22 to 23), roughly equal sex distribution; rarely due to infection or cirrhosis; highly resectable even with regional adenopathy; ↑ recurrence risk so transplant is controversial; systemic therapy may be used for advanced/unresectable disease (no standard regimen; e.g., fluorouracil plus interferon, platinum doublets, or trials); 5-yr OS ~32%.
• Traditional HCC: more often men >65 yrs; 5-yr OS historically cited ~7%, contemporary all-stage ~18 to 22% (SEER liver/IBD relative survival 21.9%).

Imaging and diagnosis

• AASLD, EASL, OPTN, and LI-RADS use imaging criteria to diagnose HCC for nodules ≥1 cm; no criteria for nodules <1 cm (difficult to characterize; wait-and-see given high false-positive rate).
• Major imaging features on multiphasic CT/MRI: arterial-phase hyperenhancement, nonperipheral venous/delayed washout, enhancing capsule, and threshold growth. LI-RADS 5 = HCC (diagnose without biopsy in high-risk patients: cirrhosis, current/prior HCC, or chronic HBV).

Child-Pugh and staging

• Child-Pugh mnemonic: A (Albumin), B (Bilirubin), C (Coags/INR), D (Distention/ascites), E (Encephalopathy). Each of the 5 variables scores 1 to 3; sum: class A = 5 to 6, class B = 7 to 9, class C = 10 to 15 (one severely abnormal variable alone can reach class B).
• Child-Pugh A is required for most systemic therapies; Child-Pugh B selectively.
• Other staging systems: Okuda (ascites, albumin <3, TBili >3, tumor size); CLIP (extent of liver replacement, Child-Pugh, AFP, portal-vein thrombosis, tumor morphology); Barcelona Clinic Liver Cancer (BCLC: liver function, symptoms, tumor stage).
• BCLC integrates tumor extent, liver function, and PS, and dictates treatment:
  • Stage 0/A (very early/early): resection, ablation, or transplant.
  • Stage B (intermediate): TACE.
  • Stage C (advanced): systemic therapy.
  • Stage D (terminal): best supportive care.

Localized HCC

• Localized HCC is treated with surgery or transplant (OLT). Only ~25% are resectable, and OLT is limited by organ availability.
• Macrovascular tumor invasion (tumor thrombus) → generally no transplant; resection only in selected cases at expert centers. Bland (non-tumoral) portal vein thrombosis is not an absolute contraindication.
• Liver remnant after resection should be 20% for a healthy liver and 30 to 40% for Child A cirrhosis.
• Systemic therapy is reserved for unresectable/advanced disease; ablation is a curative option for early HCC (BCLC 0/A), including small resectable tumors and can improve OS.
• No established adjuvant therapy after surgery or ablation. STORM showed no RFS or OS benefit of adjuvant sorafenib.

• Optimal candidates: Child-Pugh A, no portal HTN, no significant cirrhosis, single tumor.
• Adjuvant atezolizumab + bevacizumab (IMbrave050, Qin Lancet 2023): initial interim analysis in high-risk resected/ablated Child-Pugh A HCC met RFS (HR 0.72; grade 3/4 AE (all-cause) 41% vs 13%). However, updated 2024 analysis showed loss of RFS benefit (HR 0.90) and concerning OS HR 1.26 (Yopp, ESMO 2024). Not FDA-approved as adjuvant; not recommended outside trials.

• Excellent option for early-stage tumors: therapeutic for both cancer and cirrhosis. Bridging therapy (TACE, ablation) is common while waiting.
• PD-L1/PD-1 therapy is contraindicated in patients with a history of transplant (e.g., relapse after transplant), due to rejection risk.
• Milan criteria (Milan study, 48 cirrhotic patients, 94% HBV/HCV; 4-yr OS 75% post-transplant): 1 tumor ≤5 cm, or ≤3 tumors each ≤3 cm; no macrovascular involvement; no extrahepatic extension. UNOS criteria = Milan plus AFP ≤1000 ng/mL. (UCSF criteria are an expanded alternative.)
• Patients with extrahepatic disease, gross vascular invasion, or tumor (malignant) PVT are not transplant candidates; bland PVT is not an absolute contraindication.
• Downstaging (American Liver Tumor Study Group stage III/IV, TACE to reach Milan): of 76 potential candidates, 18 (23.7%) downstaged to Milan; 17 underwent OLT after successful downstaging; at median follow-up 19.6 mo, 16/17 (94.1%) were alive, and only 1 had recurrent HCC.

• Includes radiofrequency (RFA, superior to ethanol injection; comparable to microwave ablation, no proven superiority), cryoablation, percutaneous alcohol injection, and microwave. Caution near major vessels, bile ducts, diaphragm, and other organs.
• Ablation can be curative (CR) in ~80% of tumors ≤3 cm (only ~50% CR in tumors 3 to 5 cm); equivalent to resection for very small HCC.
• Lesions 3 to 5 cm: arterially directed therapy, or combination with ablation if accessible. No ablation for unresectable lesions >5 cm (use arterially directed therapy, systemic therapy, or EBRT).

• Bland transarterial embolization (TAE), transarterial chemoembolization (TACE), and radioembolization (RE) with Yttrium-90 microspheres.
• TACE: OS benefit in intermediate (BCLC B) HCC. Conventional TACE = doxorubicin-oil emulsion then gelatin sponge; DEB-TACE (drug-eluting bead) ↑ tumor drug delivery and ↓ systemic exposure.
• PRECISION trial (DEB-TACE vs conventional TACE, 212 patients, Child-Pugh A/B, large/multinodular unresectable N0M0 HCC): CR 27% vs 22%, ORR 52% vs 44%, DCR 63% vs 52%; superiority not met (one-sided P = 0.11), but significant response improvement in Child-Pugh B, ECOG 1, bilobar, and recurrent disease (P = 0.038). DEB-TACE has better tolerability and less liver/doxorubicin toxicity.
• Y-90 radioembolization: ORR 72.2% in the LEGACY study; no effect on micrometastasis.
• Relative contraindications to all arterially directed therapy: Child-Pugh C; portal vein thrombosis (may be considered with limited PV invasion); bilirubin >3 mg/dL unless segmental (Y-90 has increased risk of radiation-induced liver disease with bilirubin >2 mg/dL).

• Per NCCN, EBRT is an option for unresectable or medically inoperable disease. SBRT (typically 3 to 5 fractions) is an alternative to ablation/embolization or when these fail or are contraindicated, usually for 1 to 3 tumors, with sufficient uninvolved liver and no or minimal extrahepatic disease.

• EMERALD-1 (Sangro, Lancet 2025, PMID 39798579): durvalumab + bevacizumab + TACE vs TACE for embolization-eligible unresectable HCC (n=616): mPFS 15.0 vs 8.2 mo (HR 0.77); OS immature. NOT FDA-approved (Imfinzi HCC label covers only STRIDE); investigational option for intermediate/BCLC-B HCC amenable to TACE.

Advanced HCC: systemic therapy

• Atezolizumab + bevacizumab: preferred 1L (category 1). IMbrave150 (Finn NEJM 2020 and 2-yr update): mOS 19.2 vs 13.4 mo (sorafenib), HR 0.66; mPFS 6.9 vs 4.3 mo; grade 3 to 4 AE (all-cause) 56.5% vs 55.1%. The trial included macrovascular invasion of the main portal trunk or contralateral branch, bile-duct invasion, or ≥50% hepatic involvement. Pre-treatment EGD required to assess and treat varices before bevacizumab. PVT is not a contraindication to bevacizumab in HCC (usually tumor thrombus). FDA May 29, 2020.
• Durvalumab + tremelimumab (STRIDE): category 1; alternative when bevacizumab best avoided with untreated/high-risk varices or recent significant bleeding (adequately treated varices or stable anticoagulation do not absolutely preclude it). HIMALAYA (Abou-Alfa NEJM Evid 2022): mOS 16.4 vs 13.8 mo sorafenib (HR 0.78); ORR STRIDE 20.1% vs durvalumab 17% vs sorafenib 5.1%; grade 3 to 4 TRAE 25.8% STRIDE (serious TRAE 17.5%). Single priming dose of tremelimumab + durvalumab, then durvalumab q4wk. FDA Oct 21, 2022.
• Nivolumab + ipilimumab: CheckMate 9DW (Galle, ASCO 2024): mOS 23.7 vs 20.6 mo (lenvatinib or sorafenib), HR 0.79; ORR 36% vs 13%. FDA Apr 11, 2025. Higher immune-related AE rate (29% required high-dose steroids; grade 3 to 4 immune hepatitis 15%).
• Lenvatinib: single-agent option when IO is contraindicated. REFLECT (Kudo Lancet 2018, non-inferiority): mOS 13.6 vs 12.3 mo (sorafenib); mPFS 7.3 vs 3.6 mo (HR 0.64); ORR 41% vs 12% per mRECIST (19% vs 7% per RECIST 1.1). Excluded ≥50% liver involvement, bile-duct invasion, or main portal vein invasion; NCCN does not exclude PVT. TRAE: HTN, fatigue, diarrhea, decreased appetite/weight, proteinuria, palmar-plantar erythrodysesthesia, dysphonia, hypothyroidism. FDA Aug 16, 2018.
• Sorafenib: SHARP (Llovet NEJM 2008): mOS 10.7 vs 7.9 mo placebo (8.1 vs 4.9 mo in macrovascular invasion). TRAE: diarrhea, weight loss, hand-foot skin reaction, hypophosphatemia. FDA 2007; largely supplanted by IO regimens.
• LEAP-002 (Llovet, Lancet Oncol 2023, PMID 38039993): lenvatinib + pembrolizumab vs lenvatinib as 1L: mOS 21.2 vs 19.0 mo (HR 0.84), did NOT meet the prespecified boundary; combination not preferred.
• CARES-310 (Qin, Lancet 2023, PMID 37499670): camrelizumab + rivoceranib (apatinib) vs sorafenib: mOS 22.1 vs 15.2 mo (HR 0.62), mPFS 5.6 vs 3.7 mo. Approved in China; FDA complete response letter May 2024 (manufacturing/inspection), not currently FDA-approved.

• Tyrosine kinase inhibitors:
  • Regorafenib (RESORCE, Bruix Lancet 2017): post-sorafenib (must have tolerated sorafenib (>=400 mg/day for >=20 of last 28 days) and progressed on it) mOS 10.6 vs 7.8 mo, mPFS 3.1 vs 1.5 mo, ORR 11% vs 4%. Dose 160 mg daily for 21 of every 28 days. Blocks VEGFR 1 to 3, TIE-2, RAF-1, BRAF, BRAF V600, KIT, RET, PDGFR, FGFR. FDA Apr 2017.
  • Cabozantinib (CELESTIAL, Abou-Alfa NEJM 2018): 2L/3L after sorafenib (Child-Pugh A) mOS 10.2 vs 8.0 mo, mPFS 5.2 vs 1.9 mo. FDA Jan 2019.
  • Lenvatinib: option in 2L if not used in 1L.
  • Ramucirumab (anti-VEGFR2; REACH-2, Zhu Lancet Oncol 2019): only if AFP ≥400 ng/mL after sorafenib; mOS 8.5 vs 7.3 mo (HR 0.71). REACH overall was negative, but the AFP ≥400 subgroup benefited (7.8 vs 4.2 mo). FDA May 2019.
• Immunotherapy:
  • Pembrolizumab (KEYNOTE-394, Qin JCO 2023): 2L mOS 14.6 vs 13.0 mo (placebo). Conditional approval via KEYNOTE-224 (single-arm); KEYNOTE-240 phase III did not meet prespecified significance (OS HR 0.78). FDA Nov 2018 (accelerated); current label restricted to HCC secondary to hepatitis B after prior non-PD-(L)1 systemic therapy (based on KEYNOTE-394).
  • Nivolumab + ipilimumab 2L (CheckMate 040, cohort 4, 49 patients): ORR ~32 to 33%, DOR range 4.6 to 30.5+ mo (not a median; + = ongoing response). FDA Mar 2020 accelerated (post-sorafenib).
  • Nivolumab monotherapy: CheckMate 040 accelerated approval (Sep 2017) was later withdrawn after the confirmatory CheckMate 459 failed to significantly improve OS. Nivolumab monotherapy was negative in CheckMate 459, but pembrolizumab monotherapy met OS in phase III KEYNOTE-394 (14.6 vs 13.0 mo, HR 0.79, P=0.0180).
  • Sequencing: after 1L atezolizumab + bevacizumab, options include lenvatinib, cabozantinib, or regorafenib; after 1L STRIDE, similar TKI options. Optimal sequencing is not well established.

Combined hepatocellular-cholangiocarcinoma (cHCC-ICC)

• Biphenotypic primary liver tumor with features of both HCC and intrahepatic cholangiocarcinoma; clinically and pathologically distinct from HCC. Risk factors: HBV, HCV, alcohol, primary sclerosing cholangitis. 5-yr OS markedly lower than HCC.
• Localized disease: definitive treatment (surgery). No firm standard for advanced disease. A retrospective series of 30 patients using gemcitabine + platinum frontline (mostly GEMOX, some with bevacizumab, some gem/cis): PR 28.6%, SD 50%, PD 21.4%; mPFS 9.0 mo and mOS 16.2 mo.

High-yield HCC pearls

• LI-RADS 5: diagnose HCC by imaging (no biopsy) in cirrhotics, arterial enhancement + venous washout.
• Milan: single ≤5 cm or up to 3 nodules each ≤3 cm for transplant eligibility; UNOS adds AFP ≤1000.
• IMbrave150: atezolizumab + bevacizumab 1L (mOS 19.2 mo); EGD before bevacizumab.
• HIMALAYA (STRIDE): durvalumab + tremelimumab 1L if bevacizumab contraindicated.
• CheckMate 9DW: nivolumab + ipilimumab 1L; FDA Apr 11, 2025; mOS 23.7 vs 20.6 mo.
• IMbrave050: adjuvant atezolizumab + bevacizumab had initial RFS benefit (HR 0.72) but updated 2024 data did not confirm it (RFS HR 0.90, OS HR 1.26); not approved, not recommended.
• AFP ≥400 is the biomarker for ramucirumab response.
• HBV can develop HCC even without cirrhosis; surveillance differs from HCV.
• Aflatoxin → TP53 codon 249 hotspot mutation.
• PD-1/PD-L1 therapy is contraindicated after liver transplant (rejection risk).