Skip to content
Bending the Curve Trial by Trial
?

Liver (digestive system, hepatobiliary)

Hepatocellular carcinoma

How hepatitis control, surveillance, ablation, transplantation, embolisation and systemic therapy shaped care of hepatocellular carcinoma.

Liver and intrahepatic bile duct cancer, not hepatocellular carcinoma alone · 5-year relative survival · United States, SEER 8 registries

2.63%→22.81%

1975 (observed value; first year in the table) to 2018 (observed value; last observed year in the table). Read these with the caveats under “Then and now”.Source

25 milestones, by decade

  • In the SEER Cancer Stat Facts series for liver and intrahepatic bile duct cancer (a category that includes bile duct cancers inside the liver as well as hepatocellular carcinoma), observed five-year relative survival in the SEER 8 registries was 2.63% in 1975 and 22.81% in 2018, with a single-year peak of 27.34% in 2015. The US age-adjusted death rate (a separate series covering 1975 to 2024) was 2.80 per 100,000 in 1975 and 6.64 in 2024, after reaching 6.74 in 2016 and 2018. The two series use different populations (people with a recorded cancer, and the whole population) and the page does not apportion either trend between stage, cause of liver disease, case mix or treatment, so the figures are shown together rather than as one improving trend.
  • The FDA approved sorafenib for unresectable hepatocellular carcinoma on 16 November 2007. The placebo-controlled phase 3 trial of sorafenib (SHARP) enrolled 602 people with advanced disease and no prior systemic treatment and was stopped at its second planned interim analysis: median overall survival was 10.7 months with sorafenib and 7.9 months with placebo (hazard ratio 0.69, 95% CI 0.55 to 0.87). A five-year follow-up of a later phase 3 trial, HIMALAYA, reported in an exploratory analysis that estimated overall survival at 60 months was 19.6% with tremelimumab plus durvalumab and 9.4% with sorafenib (hazard ratio 0.76, 95% CI 0.65 to 0.89). The two trials differ in design (placebo control versus an active sorafenib control), era and follow-up, and the figures are not a cross-trial comparison.
  • Single-agent nivolumab received FDA accelerated approval on 22 September 2017 for hepatocellular carcinoma previously treated with sorafenib; the FDA list of withdrawn cancer accelerated approvals shows the indication withdrawn on 23 July 2021 and gives no reason. In the later phase 3 trial CheckMate 459 of first-line nivolumab versus sorafenib (743 people), median overall survival was 16.4 versus 14.7 months (hazard ratio 0.85, 95% CI 0.72 to 1.02; p=0.075), and the protocol-defined significance level of p=0.0419 was not reached. The documents read do not link the withdrawal to that trial.

Evidence current to October 2026 · Checked against primary sources to the extent possible

A guided tour of 11 milestones. You can pause or step through it yourself.

Filter milestonesshowing all 25
Treatment type
Setting

Molecular origins

  1. 1981 · Understanding · Understanding⚑ needs reviewProspective cohort ties hepatitis B carriage to HCC riskIn 1981, researchers reported following more than 22,000 men in Taiwan. Men who carried the hepatitis B virus developed liver cancer far more often than men who did not. Because infection was measured before the cancer appeared, the study strengthened the case that the virus contributes to liver cancer and that preventing infection could prevent cancer.Prospective cohort of 22,707 men in Taiwan. Figures below are quoted from a 2018 appraisal, not from the 1981 paper: 3,454 (15.2%) were HBsAg-positive at entry; average follow-up 3.3 years; HCC incidence 1158 per 100,000 person-years in HBsAg-positive men versus 5 in HBsAg-negative men (relative risk 223, 95% CI 28 to 1479); 41 HCC deaths in total.
  2. 1992 · Understanding · UnderstandingUrinary aflatoxin markers and HCC risk in a Shanghai cohortResearchers measured aflatoxin, a toxin made by moulds on stored food, in urine collected from men in Shanghai before some of them developed liver cancer. Those who later developed it were more likely to have had aflatoxin traces, and the authors reported a strong interaction with markers of hepatitis B infection.Nested case-control study within a Shanghai cohort of 18,244 men (35,299 person-years; 22 liver cancer cases, five or ten matched controls each). Any detectable urinary aflatoxin metabolite: relative risk 2.4 (95% CI 1.0 to 5.9); aflatoxin P1 6.2 (1.8 to 21.5); 3.8 (1.2 to 12.2) after adjustment including HBsAg. The authors report a strong interaction with HBsAg. A 1994 follow-up had 55 HCC cases.
  3. 1996 · Treatment · Cell therapy & transplantMilan criteria: transplantation for small HCC in cirrhosisIn 1996 a team in Milan reported 48 people with cirrhosis and small liver tumours that could not be removed by surgery, who had transplants under selection limits set in advance. At four years about 85 in 100 of those within the limits were alive, compared with about 50 in 100 of those beyond them.Series of 48 patients with cirrhosis and small unresectable HCC who underwent transplantation (eligibility: one tumour of 5 cm or less, or up to 3 nodules of 3 cm or less each); median follow-up 26 months. Four-year actuarial survival 75%, recurrence-free survival 83%. At pathological review of the explanted liver, 35 patients met the predetermined criteria (4-year survival 85%, recurrence-free survival 92%) and 13 did not (50% and 59%; P=0.01 and P=0.002).

Targeted therapy

  1. 1999 · Understanding · UnderstandingBCLC: a staging classification linking stage to treatmentIn 1999 a Barcelona research group proposed a staging system for liver cancer with four stages, each tied to the treatments it considered best for those patients, from curative treatment for small early tumours to symptomatic care for the most advanced disease. The Barcelona Clinic Liver Cancer (BCLC) system has been updated since, including in 2022.Review proposing the Barcelona Clinic Liver Cancer (BCLC) classification with four stages: A, asymptomatic early tumours suitable for resection, transplantation or percutaneous treatment; B, asymptomatic multinodular HCC; C, symptomatic tumours and/or vascular invasion or extrahepatic spread; D, Okuda stage III or PST 3-4, symptomatic treatment only. Stages B and C: palliative treatments or new agents within trials. A 2022 update follows an official update in 2018.
  2. 2002 · Treatment · SurgeryTwo randomised trials of chemoembolisation for unresectable HCCIn 2002 two randomised trials compared repeated chemoembolisation (chemotherapy into the tumour's artery, which is then blocked) with conservative care in selected people whose liver cancer could not be removed. At two years about 6 in 10 versus 3 in 10 were alive in one trial, and about 3 in 10 versus 1 in 10 in the other.Barcelona RCT (N=112 of 903 assessed; Child-Pugh A/B, Okuda I-II; embolisation, chemoembolisation or conservative treatment), stopped at the ninth sequential inspection: chemoembolisation vs conservative treatment HR of death 0.47 (95% CI 0.25-0.91; p=0.025); 2-year survival 63% vs 27%. Hong Kong RCT (N=80, Asian patients; chemoembolisation vs symptomatic treatment): 3-year survival 26% vs 3% (P=0.002).
  3. 2003 · Detection · Detection & preventionTwo Chinese randomised surveillance trials, two answersTwo randomised trials in China tested six-monthly liver cancer screening in people at high risk. In Shanghai, a blood test plus ultrasound was followed by fewer liver cancer deaths. In Qidong, a blood test alone found cancers earlier but did not significantly change deaths.Shanghai (18,816 people aged 35-59 with hepatitis B or chronic hepatitis; AFP plus ultrasound every 6 months; 58.2% of offered screens completed): HCC mortality rate ratio 0.63 (95% CI 0.41-0.98). Qidong (5,581 HBsAg-positive men; AFP alone): stage I 29.6% versus 6.0%; Poisson-model rate ratio of death 0.83 (95% CI 0.68-1.03), not significant. Different tests, populations and follow-up.
  4. 2004 · Treatment · Detection & preventionLamivudine in advanced hepatitis B: fewer HCC, a borderline findingIn a placebo-controlled trial of 651 people with advanced hepatitis B, the antiviral lamivudine slowed disease progression. Liver cancer was less frequent with lamivudine, about 4 in 100 versus 7 in 100, but that difference was borderline and the virus often developed resistance.Randomised placebo-controlled trial (n=651; 98% Asian, 85% male) of lamivudine 100 mg daily versus placebo (2:1) in hepatitis B with cirrhosis or advanced fibrosis; stopped early at a median treatment duration of 32.4 months. Disease progression (composite including HCC): 7.8% versus 17.7% (HR 0.45, P=0.001). HCC: 3.9% versus 7.4% (HR 0.49, P=0.047). YMDD resistance mutations arose in 49% of lamivudine-treated patients.
  5. 2006 · Treatment · Surgery⚑ needs reviewResection versus ablation for small HCC: conflicting trialsFor small liver tumours, three randomised trials compared surgery to remove the tumour with ablation (destroying it with local treatment such as radiofrequency heat). One found similar survival, one found better survival with surgery, and the most recent found no clear advantage for surgery. The trials differed in size limits and ablation methods.Chen 2006 (N=180; solitary HCC of 5 cm or less; percutaneous ablation vs resection): 4-year OS 67.9% vs 64.0%, no significant difference. Huang 2010 (N=230; Milan criteria; RFA vs resection): 5-year OS 54.78% vs 75.65% (P=0.001). SURF (N=302; 3 cm or less, up to 3 nodules; surgery vs RFA): 5-year OS 74.6% vs 70.4% (HR 0.96; adjusted P=0.84); 5-year RFS 42.9% vs 42.7%.
  6. 2006 · Understanding · UnderstandingREVEAL-HBV: HCC risk rises with serum hepatitis B virus DNA levelA Taiwanese cohort of about 3,650 people with chronic hepatitis B found that liver cancer became more common as the amount of virus in the blood rose. Over an average of about 11 years of follow-up, roughly 1 in 75 people with the lowest virus levels and about 1 in 7 with the highest developed liver cancer.Prospective community cohort of 3,653 HBsAg-positive, anti-HCV-negative adults aged 30-65 in Taiwan (recruited 1991-1992); 164 incident HCC over a mean 11.4 years. Incidence rose with baseline HBV DNA from 108 per 100,000 person-years below 300 copies/mL to 1152 at 1 million copies/mL or more; cumulative incidence 1.3% versus 14.9% (P<.001 for the gradient after adjustment).
  7. 2007 · Treatment · TargetedSorafenib: SHARP and Asia-Pacific trials in advanced HCCSorafenib, a tablet that blocks several growth signals, was tested against placebo in people with advanced liver cancer who had not had drug treatment. In the main trial, median survival was about 11 months with sorafenib and about 8 months with placebo. A separate Asia-Pacific trial pointed the same way (6.5 versus 4.2 months).SHARP (double-blind phase 3, N=602, no prior systemic therapy): sorafenib 400 mg twice daily vs placebo, median OS 10.7 vs 7.9 months, HR 0.69 (95% CI 0.55-0.87; P<0.001), at the second interim analysis (321 deaths). Separate Asia-Pacific phase 3 (226 randomised, Child-Pugh A): median OS 6.5 vs 4.2 months, HR 0.68 (95% CI 0.50-0.93; P=0.014); no predefined primary endpoint.

Immunotherapy

  1. 2012 · Understanding · Detection & preventionHepatitis C clearance and HCC: interferon-era SVR, then DAA debateA long-term study of people with hepatitis C and advanced liver scarring found less liver cancer over ten years in those whose virus was cleared by treatment. After newer antiviral tablets arrived, a small study of people with previously treated liver cancer reported frequent early return, while larger cohorts of new cancers reported lower risk with viral clearance or treatment.Four observational cohorts, each analysed on its own terms. Interferon era (n=530, advanced fibrosis): 10-year HCC incidence 5.1% with SVR versus 21.8% without. DAA era: Reig reported radiologic recurrence in 27.6% of 58 patients with treated HCC (median follow-up 5.7 months, no control group); US Veterans cohort adjusted HR 0.28 for HCC with SVR; French Hepather adjusted HR 0.66 for HCC with DAA exposure (unadjusted 2.77).
  2. 2015 · Reversal · TargetedSTORM: adjuvant sorafenib did not improve recurrence-free survivalAfter surgery or local ablation for liver cancer, 1,114 people were randomly assigned to sorafenib tablets or a placebo for up to four years to see whether it delayed recurrence. Median recurrence-free survival was about 33 months in both groups, and the authors concluded that sorafenib was not effective in this setting.Phase 3, double-blind, placebo-controlled trial, N=1114 with a complete radiological response after resection or ablation: sorafenib 400 mg twice daily vs placebo for up to 4 years. Final analysis (464 events): median RFS 33.3 vs 33.7 months (HR 0.940; 95% CI 0.780-1.134; one-sided p=0.26). Grade 3 or 4 hand-foot skin reaction: 28% vs <1%.
  3. 2016 · Treatment · TargetedRESORCE: regorafenib after sorafenib progressionRegorafenib was compared with placebo in 573 people with liver cancer whose disease had progressed on sorafenib and who had tolerated it. Median survival was about 11 months with regorafenib and about 8 months with placebo. The FDA approved regorafenib for this use in April 2017.RESORCE (double-blind phase 3, 2:1, ITT N=573): regorafenib plus best supportive care (n=379) vs placebo (n=194) in patients who had tolerated sorafenib, progressed on it, and had Child-Pugh A liver function. Median OS 10.6 vs 7.8 months, HR 0.63 (95% CI 0.50-0.79; one-sided P<0.0001).
  4. 2017 · Treatment · ImmunotherapyNivolumab after sorafenib: CheckMate 040 and accelerated approvalIn an early-phase trial of nivolumab in advanced liver cancer, about one in five people in the dose-expansion phase (20%) had a tumour response. In September 2017 the FDA gave accelerated approval for people previously treated with sorafenib, on condition that a randomised trial showed better survival.CheckMate 040 (open-label, non-comparative phase 1/2; 262 treated patients: 48 dose escalation, 214 dose expansion): objective response rate 20% (95% CI 15-26) in dose expansion (nivolumab 3 mg/kg every 2 weeks) and 15% (95% CI 6-28) in dose escalation. FDA accelerated approval 22 September 2017; postmarketing requirement 3270-1 calls for a randomised overall survival trial. FDA label: response rate 14.3% (95% CI 9.2-20.8) in a 154-patient sorafenib-treated subgroup.
  5. 2017 · Reversal · RadiationSARAH and SIRveNIB: no survival gain over sorafenibTwo randomised trials compared radioembolisation (tiny radioactive beads placed in the liver's blood supply) with sorafenib tablets in people with locally advanced liver cancer. Median survival was roughly 8 to 9 months with radioembolisation and about 10 months with sorafenib, and the differences were not statistically significant.SARAH (France; N=467 randomised; open-label): median OS 8.0 months with SIRT vs 9.9 with sorafenib, HR 1.15 (95% CI 0.94-1.41; p=0.18). SIRveNIB (Asia-Pacific; N=360; open-label): median OS 8.8 vs 10.0 months, HR 1.1 (95% CI 0.9-1.4; P=0.36). Of those assigned to SIRT, 22% (SARAH) and 28.6% (SIRveNIB) did not receive it.
  6. 2018 · Treatment · TargetedCELESTIAL: cabozantinib after prior systemic therapyCabozantinib was compared with placebo in 707 people with advanced liver cancer who had already received sorafenib and could have had up to two earlier drug treatments. Median survival was about 10 months with cabozantinib and 8 months with placebo. Severe (grade 3 or 4) side effects were about twice as common with cabozantinib.CELESTIAL (double-blind phase 3, 2:1, ITT N=707: 470 vs 237): cabozantinib 60 mg daily vs placebo after prior sorafenib (up to two previous regimens for advanced disease). Second planned interim analysis: median OS 10.2 vs 8.0 months, HR 0.76 (95% CI 0.63-0.92; P=0.005); median PFS 5.2 vs 1.9 months, HR 0.44 (95% CI 0.36-0.52); ORR 4% vs under 1% (P=0.009). Grade 3 or 4 adverse events: 68% vs 36%.
  7. 2018 · Treatment · TargetedREFLECT: lenvatinib non-inferior to sorafenib first lineLenvatinib was compared with sorafenib in 954 people with unresectable liver cancer who had had no drug treatment for advanced disease. Median survival was about 14 months with lenvatinib and 12 months with sorafenib. The trial was built to show lenvatinib was not worse, and it met that aim. The FDA approved it in August 2018.REFLECT (open-label phase 3 non-inferiority, ITT N=954, no prior treatment for advanced disease): lenvatinib (n=478) vs sorafenib (n=476). Median OS 13.6 vs 12.3 months, HR 0.92 (95% CI 0.79-1.06); the non-inferiority margin of 1.08 was met. Primary endpoint OS; the abstract reports no superiority test.
  8. 2018 · Treatment · ImmunotherapyPembrolizumab: KEYNOTE-224, -240 and -394 and the FDA labelPembrolizumab was tested in three trials after earlier treatment, mostly sorafenib. In a single-arm study, 17% of 104 people had a tumour response. One placebo-controlled trial did not reach its prespecified thresholds; another, in people from Asia, reported longer median survival (14.6 versus 13.0 months). The FDA label was later revised to hepatitis B-related disease.KEYNOTE-224 (single-arm phase 2, n=104): ORR 17% (95% CI 11-26). KEYNOTE-240 (double-blind phase 3, N=413): median OS 13.9 vs 10.6 months, HR 0.781 (95% CI 0.611-0.998; one-sided P=0.0238 vs threshold 0.0174); authors: OS and PFS did not reach specified significance. KEYNOTE-394 (Asia, N=453): median OS 14.6 vs 13.0 months, HR 0.79 (95% CI 0.63-0.99; P=0.0180 vs threshold 0.0193).
  9. 2019 · Treatment · TargetedREACH-2: ramucirumab in AFP-selected patients after sorafenibRamucirumab was compared with placebo in 292 people with advanced liver cancer who had received sorafenib and whose blood alpha-fetoprotein (AFP) was high (400 ng/mL or more). Median survival was 8.5 months with ramucirumab and 7.3 months with placebo. An earlier trial without the AFP requirement did not show a significant survival difference.REACH-2 (double-blind phase 3, 2:1, ITT N=292, AFP of 400 ng/mL or more, prior sorafenib): ramucirumab (n=197) vs placebo (n=95), median OS 8.5 vs 7.3 months, HR 0.710 (95% CI 0.531-0.949; P=0.0199); median PFS 2.8 vs 1.6 months, HR 0.452 (95% CI 0.339-0.603). REACH (N=565, no AFP threshold): median OS 9.2 vs 7.6 months, HR 0.87 (95% CI 0.72-1.05; P=0.14).

Precision & engineered

  1. 2020 · Treatment · ImmunotherapyIMbrave150: atezolizumab plus bevacizumab versus sorafenibIn 501 people with advanced liver cancer who had not had drug treatment, atezolizumab plus bevacizumab was compared with sorafenib. About two-thirds were alive at 12 months with the combination, against about half with sorafenib (67% versus 55%). Median time without progression was 6.8 versus 4.3 months. The FDA approved the combination in May 2020.IMbrave150 (open-label phase 3, 2:1, ITT N=501, no prior systemic therapy): atezolizumab plus bevacizumab (n=336) vs sorafenib (n=165). Primary analysis: OS HR 0.58 (95% CI 0.42-0.79; P<0.001), 12-month OS 67.2% vs 54.6%; median PFS 6.8 vs 4.3 months (HR 0.59; 95% CI 0.47-0.76). Post hoc update (median follow-up 15.6 months): median OS 19.2 vs 13.4 months, HR 0.66 (95% CI 0.52-0.85).
  2. 2021 · Reversal · ImmunotherapyCheckMate 459 and withdrawal of nivolumab's HCC indicationIn 743 people with advanced liver cancer who had not had drug treatment, nivolumab did not reach its prespecified significance level for overall survival against sorafenib (median 16.4 versus 14.7 months). The single-agent nivolumab indication after sorafenib, accelerated in 2017, appears on the FDA's withdrawn list with a date of 23 July 2021.CheckMate 459 (open-label phase 3, ITT N=743, first-line): nivolumab (n=371) vs sorafenib (n=372), median OS 16.4 vs 14.7 months, HR 0.85 (95% CI 0.72-1.02; P=0.075); protocol-defined significance level P=0.0419 not reached; median follow-up 15.2 and 13.4 months. FDA list: single-agent nivolumab for hepatocellular carcinoma after sorafenib, accelerated approval 22 September 2017, withdrawal 23 July 2021; no reason stated.
  3. 2022 · Treatment · ImmunotherapyHIMALAYA: tremelimumab plus durvalumab (STRIDE) versus sorafenibIn 1,171 people with liver cancer that could not be removed, three treatments were compared: one priming dose of tremelimumab plus regular durvalumab, durvalumab alone, and sorafenib. Median survival was about 16 months with the combination and about 14 months with sorafenib. About three in ten on the combination were alive at three years, against two in ten on sorafenib.HIMALAYA (open-label phase 3, N=1171, no prior systemic therapy): STRIDE (n=393), durvalumab (n=389) or sorafenib (n=389). Median OS 16.43 vs 16.56 vs 13.77 months; STRIDE vs sorafenib HR 0.78 (96.02% CI 0.65-0.93; P=0.0035); durvalumab non-inferior (HR 0.86; margin 1.08). Median PFS not significantly different. Updates: 4-year OS HR 0.78 (95% CI 0.67-0.92); 60-month OS 19.6% vs 9.4% (HR 0.76; 95% CI 0.65-0.89).
  4. 2025 · Treatment · ImmunotherapyTACE plus immunotherapy: EMERALD-1 and LEAP-012 (PFS)Two randomised trials added immunotherapy-based drug combinations to chemoembolisation for liver cancer that could not be removed by surgery. Median time without progression was about 15 versus 8 months in one trial and about 15 versus 10 months in the other. Neither had shown a significant overall survival difference at the analyses reported.EMERALD-1 (N=616; TACE + durvalumab + bevacizumab vs TACE + placebo): BICR median PFS 15.0 vs 8.2 months, HR 0.77 (95% CI 0.61-0.98; p=0.032); durvalumab alone HR 0.94 (95% CI 0.75-1.19). LEAP-012 (N=480; TACE + lenvatinib + pembrolizumab vs dual placebo): median PFS 14.6 vs 10.0 months, HR 0.66 (95% CI 0.51-0.84); interim OS HR 0.80 (95% CI 0.57-1.11; one-sided p=0.087).
  5. 2025 · Treatment · ImmunotherapyCheckMate 9DW: first-line nivolumab plus ipilimumabNivolumab plus ipilimumab was compared with lenvatinib or sorafenib in 668 people with liver cancer who had not had drug treatment. Median survival was about 24 months with the combination and about 21 months with the comparison drugs. The FDA approved the combination as first-line treatment in April 2025.CheckMate 9DW (open-label phase 3, N=668, no prior systemic therapy, Child-Pugh 5 or 6): nivolumab plus ipilimumab (n=335) vs lenvatinib or sorafenib (n=333). Preplanned interim analysis, median follow-up 35.2 months: median OS 23.7 vs 20.6 months, HR 0.79 (95% CI 0.65-0.96; P=0.018); OS 49% vs 39% at 24 months, 38% vs 24% at 36 months. Curves crossed early (HR 1.65 for the first 6 months).
  6. 2026 · Reversal · ImmunotherapyIMbrave050: early recurrence-free benefit not sustained on updateA trial gave atezolizumab plus bevacizumab for one year, or active monitoring, to 668 people at high risk of recurrence after liver cancer surgery or ablation. A 2023 interim analysis reported fewer recurrences or deaths with the combination, but a 2026 update found no sustained benefit, and its authors do not support adjuvant use.Open-label phase 3, N=668 (334 per arm) after resection or ablation at high risk of recurrence: atezolizumab plus bevacizumab for 17 cycles vs active surveillance. Interim analysis (median follow-up 17.4 months): IRF-assessed RFS HR 0.72 (adjusted 95% CI 0.53-0.98; p=0.012). Updated post hoc analysis: RFS HR 0.90 (95% CI 0.72-1.12); second interim OS HR 1.26 (95% CI 0.85-1.87), immature.

Then and now

  • 5-year relative survival · United States, SEER 8 registries

    • 2.63%

      1975 (observed value; first year in the table)

    All races, both sexes; SEER Cancer Stat Facts category 'liver and intrahepatic bile duct cancer', which includes intrahepatic bile duct cancers (intrahepatic cholangiocarcinoma) as well as hepatocellular carcinoma; the page does not give a hepatocellular-only series. SEER 8 five-year relative survival series (1975-2018), all stages together. The page states no age range, stage mix or histology mix for this series.

    Modeled-trend value for 1975: 1.73%. Relative survival excludes deaths from other causes. 'Observed' is the table's label for the yearly value, as opposed to the Joinpoint modeled-trend value in the adjacent column. The column is headed 'Year' and the chart note says only 'SEER 8 5-Year Relative Survival Percent from 1975-2018'; the page does not say the year is the year of diagnosis. SEER 8 covers a limited set of registries, and year-to-year values move (observed range 2.14% in 1981 to 27.34% in 2015). The page does not describe changes in stage mix, imaging, surveillance, causes of liver disease, treatment or the mix of hepatocellular and bile duct cancers over time, and attributes the change to none of them. Read beside the mortality values below. Compare only with other values from this same SEER 8 series, not with the SEER 21 values.

    Source

  • 5-year relative survival · United States, SEER 8 registries

    • 27.34%

      2015 (observed value; highest observed value in the table)

    All races, both sexes; SEER Cancer Stat Facts category 'liver and intrahepatic bile duct cancer', which includes intrahepatic bile duct cancers (intrahepatic cholangiocarcinoma) as well as hepatocellular carcinoma; the page does not give a hepatocellular-only series. SEER 8 five-year relative survival series (1975-2018), all stages together. The page states no age range, stage mix or histology mix for this series.

    Included because the series is not monotonic: the observed value was 22.38% in 2016, 25.15% in 2017 and 22.81% in 2018, and the 2015 value is a single-year observation (modeled-trend value for 2015: 26.71%). Relative survival excludes deaths from other causes. 'Observed' is the table's label for the yearly value, as opposed to the Joinpoint modeled-trend value in the adjacent column. The column is headed 'Year' and the chart note says only 'SEER 8 5-Year Relative Survival Percent from 1975-2018'; the page does not say the year is the year of diagnosis. SEER 8 covers a limited set of registries, and year-to-year values move (observed range 2.14% in 1981 to 27.34% in 2015). The page does not describe changes in stage mix, imaging, surveillance, causes of liver disease, treatment or the mix of hepatocellular and bile duct cancers over time, and attributes the change to none of them. Read beside the mortality values below. Compare only with other values from this same SEER 8 series, not with the SEER 21 values.

    Source

  • 5-year relative survival · United States, SEER 8 registries

    • 22.81%

      2018 (observed value; last observed year in the table)

    All races, both sexes; SEER Cancer Stat Facts category 'liver and intrahepatic bile duct cancer', which includes intrahepatic bile duct cancers (intrahepatic cholangiocarcinoma) as well as hepatocellular carcinoma; the page does not give a hepatocellular-only series. SEER 8 five-year relative survival series (1975-2018), all stages together. The page states no age range, stage mix or histology mix for this series.

    Modeled-trend value for 2018: 25.21%. The table shows no observed five-year value for 2019 onward (the modeled-trend entries for 2019-2023, 24.71% to 22.76%, are not used). Relative survival excludes deaths from other causes. 'Observed' is the table's label for the yearly value, as opposed to the Joinpoint modeled-trend value in the adjacent column. The column is headed 'Year' and the chart note says only 'SEER 8 5-Year Relative Survival Percent from 1975-2018'; the page does not say the year is the year of diagnosis. SEER 8 covers a limited set of registries, and year-to-year values move (observed range 2.14% in 1981 to 27.34% in 2015). The page does not describe changes in stage mix, imaging, surveillance, causes of liver disease, treatment or the mix of hepatocellular and bile duct cancers over time, and attributes the change to none of them. Read beside the mortality values below. Compare only with other values from this same SEER 8 series, not with the SEER 21 values.

    Source

  • 5-year relative survival · United States, SEER 21 (excluding IL)

    • 21.9%

      2016-2022

    All races, both sexes; SEER Cancer Stat Facts category 'liver and intrahepatic bile duct cancer', which includes intrahepatic bile duct cancers (intrahepatic cholangiocarcinoma) as well as hepatocellular carcinoma; the page does not give a hepatocellular-only series. All stages together; the headline figure on the SEER Cancer Stat Facts liver page (SEER 21, excluding IL). The page states no age range beside the figure.

    A different registry set and a multi-year period from the SEER 8 series above, so it is not a continuation of that series and must not be subtracted from it. The page labels the period '2016-2022' and does not say whether it is year of diagnosis. Relative survival excludes deaths from other causes. The page notes that population statistics cannot predict what will happen to an individual patient.

    Source

  • 5-year relative survival · United States, SEER 21 (excluding IL)

    • 37.4%

      2016-2022; localized stage at diagnosis (45% of cases)

    All races, both sexes; SEER Cancer Stat Facts category 'liver and intrahepatic bile duct cancer', which includes intrahepatic bile duct cancers (intrahepatic cholangiocarcinoma) as well as hepatocellular carcinoma; the page does not give a hepatocellular-only series. SEER Combined Summary Stage 'localized' (confined to the primary site).

    Stage-specific value from the same SEER 21 table as the 21.9% figure. The page shows the stage mix as 45% localized (its text gives 44.7%), 23% regional, 21% distant and 10% unknown stage (the table percentages are rounded and sum to 99%). This is SEER Combined Summary Stage, not BCLC or AJCC/TNM stage, and the page does not separate hepatocellular carcinoma from bile duct cancer. Not comparable with SEER 8 values.

    Source

  • 5-year relative survival · United States, SEER 21 (excluding IL)

    • 13.4%

      2016-2022; regional stage at diagnosis (23% of cases)

    All races, both sexes; SEER Cancer Stat Facts category 'liver and intrahepatic bile duct cancer', which includes intrahepatic bile duct cancers (intrahepatic cholangiocarcinoma) as well as hepatocellular carcinoma; the page does not give a hepatocellular-only series. SEER Combined Summary Stage 'regional' (spread to regional lymph nodes).

    Same SEER 21 table and caveats as the localized-stage value: SEER summary stage rather than BCLC or AJCC/TNM stage, hepatocellular and bile duct cancers together, not comparable with SEER 8 values.

    Source

  • 5-year relative survival · United States, SEER 21 (excluding IL)

    • 3.6%

      2016-2022; distant stage at diagnosis (21% of cases)

    All races, both sexes; SEER Cancer Stat Facts category 'liver and intrahepatic bile duct cancer', which includes intrahepatic bile duct cancers (intrahepatic cholangiocarcinoma) as well as hepatocellular carcinoma; the page does not give a hepatocellular-only series. SEER Combined Summary Stage 'distant' (cancer has metastasized).

    Same SEER 21 table and caveats as the localized-stage value; the page also gives 11.6% five-year relative survival for unknown stage (10% of cases). SEER summary stage rather than BCLC or AJCC/TNM stage; hepatocellular and bile duct cancers together; not comparable with SEER 8 values.

    Source

  • Age-standardised mortality · United States (U.S. Mortality data shown on the SEER page; not a SEER registry series)

    • 2.8per 100,000

      1975 (observed value; first year in the table)

    All races, both sexes; SEER Cancer Stat Facts category 'liver and intrahepatic bile duct cancer', which includes intrahepatic bile duct cancers (intrahepatic cholangiocarcinoma) as well as hepatocellular carcinoma; the page does not give a hepatocellular-only series. Age-adjusted death rate per 100,000 from U.S. Mortality as tabulated on the SEER Cancer Stat Facts page (the page does not name the standard population); deaths are not drawn from the SEER registries.

    Death-rate series 1975-2024, observed. Modeled-trend value for 1975: 2.77. Mortality depends on incidence as well as survival: the SEER 8 rate of new cases was 2.60 per 100,000 in 1975, 9.45 in 2015 (the highest observed) and 8.75 in 2023; the page text says new-case rates have fallen on average 0.7% a year over 2014-2023 and death rates have been stable over 2015-2024. The page does not attribute the changes to a cause, and it covers liver and intrahepatic bile duct cancers together. Rates are age-adjusted; do not compare directly with rates age-standardised to another (for example European) standard population. Shown beside the survival series, not instead of it.

    Source

  • Age-standardised mortality · United States (U.S. Mortality data shown on the SEER page; not a SEER registry series)

    • 6.74per 100,000

      2016 (observed value; highest value in the 1975-2024 table, equal to the 2018 value)

    All races, both sexes; SEER Cancer Stat Facts category 'liver and intrahepatic bile duct cancer', which includes intrahepatic bile duct cancers (intrahepatic cholangiocarcinoma) as well as hepatocellular carcinoma; the page does not give a hepatocellular-only series. Age-adjusted death rate per 100,000 from U.S. Mortality as tabulated on the SEER Cancer Stat Facts page (the page does not name the standard population); deaths are not drawn from the SEER registries.

    The observed rate was 6.74 in both 2016 and 2018 (6.73 in 2017); modeled-trend value for 2016: 6.74. Mortality depends on incidence as well as survival: the SEER 8 rate of new cases was 2.60 per 100,000 in 1975, 9.45 in 2015 (the highest observed) and 8.75 in 2023; the page text says new-case rates have fallen on average 0.7% a year over 2014-2023 and death rates have been stable over 2015-2024. The page does not attribute the changes to a cause, and it covers liver and intrahepatic bile duct cancers together. Rates are age-adjusted; do not compare directly with rates age-standardised to another (for example European) standard population. Shown beside the survival series, not instead of it.

    Source

  • Age-standardised mortality · United States (U.S. Mortality data shown on the SEER page; not a SEER registry series)

    • 6.64per 100,000

      2024 (observed value; last year in the table)

    All races, both sexes; SEER Cancer Stat Facts category 'liver and intrahepatic bile duct cancer', which includes intrahepatic bile duct cancers (intrahepatic cholangiocarcinoma) as well as hepatocellular carcinoma; the page does not give a hepatocellular-only series. Age-adjusted death rate per 100,000 from U.S. Mortality as tabulated on the SEER Cancer Stat Facts page (the page does not name the standard population); deaths are not drawn from the SEER registries.

    Modeled-trend value for 2024: 6.59. The page's headline death rate for 2020-2024 deaths is 6.6 per 100,000, and the 2026 estimate (American Cancer Society projection) is 30,980 deaths. Mortality depends on incidence as well as survival: the SEER 8 rate of new cases was 2.60 per 100,000 in 1975, 9.45 in 2015 (the highest observed) and 8.75 in 2023; the page text says new-case rates have fallen on average 0.7% a year over 2014-2023 and death rates have been stable over 2015-2024. The page does not attribute the changes to a cause, and it covers liver and intrahepatic bile duct cancers together. Rates are age-adjusted; do not compare directly with rates age-standardised to another (for example European) standard population. Shown beside the survival series, not instead of it.

    Source

Survival figures are affected by earlier detection (lead time) and overdiagnosis, and by changes in who is diagnosed. For screened cancers, read them alongside mortality.

Evidence chains

A lane follows one defined population and setting, where each new trial tests against what was then standard. Compare only within a lane, and read every comparison as illustrative.

Advanced HCC, first-line systemic therapy

Adults with unresectable or advanced hepatocellular carcinoma who had not received systemic treatment for advanced disease (SHARP, the Asia-Pacific sorafenib trial, REFLECT, IMbrave150, CheckMate 459, HIMALAYA, CheckMate 9DW) · Metastatic · endpoint: OS

Results are not comparable across trials.

Eras, regions, underlying liver disease, liver-function and vascular-invasion eligibility, and control arms differ. SHARP and the Asia-Pacific trial were placebo-controlled; REFLECT was a non-inferiority comparison of lenvatinib with sorafenib; IMbrave150, CheckMate 459 and HIMALAYA used sorafenib as control; CheckMate 9DW used lenvatinib or sorafenib at the investigator's choice. Some trials stopped or reported at interim analyses (SHARP, CheckMate 9DW) and follow-up differs. IMbrave150's plotted medians come from its post hoc updated analysis (data cut-off 31 August 2020); its primary analysis reported 12-month survival and a hazard ratio but no medians. CheckMate 459 did not reach its prespecified significance level. HIMALAYA's plotted pair is tremelimumab plus durvalumab versus sorafenib; its durvalumab-alone arm is not drawn. Compare only each trial's own two arms; the chart is not a pooled or network estimate.

Each pair shows one trial: its control arm and its experimental arm. Yesterday's experimental arm is often today's control. Populations, definitions and follow-up differ.

051015202530OS, median (months)2007200720182020202120222025SHARPasia-pacific-sorafenib-hccreflect-hccIMbrave150CheckMate 459HIMALAYACheckMate 9DW
  • experimental arm
  • control arm
  • different endpoint from the lane's
  • step line through experimental values
OS, median, across 7 trials in this lane. SHARP (2007): experimental 10.7 months vs control 7.9 months; asia-pacific-sorafenib-hcc (2007): experimental 6.5 months vs control 4.2 months; reflect-hcc (2018): experimental 13.6 months vs control 12.3 months; IMbrave150 (2020): experimental 19.2 months vs control 13.4 months; CheckMate 459 (2021): experimental 16.4 months vs control 14.7 months; HIMALAYA (2022): experimental 16.43 months vs control 13.77 months; CheckMate 9DW (2025): experimental 23.7 months vs control 20.6 months.
Data table
TrialYearEndpointControl (months)Experimental (months)
SHARP2007OS7.9 months10.7 months
asia-pacific-sorafenib-hcc2007OS4.2 months6.5 months
reflect-hcc2018OS12.3 months13.6 months
IMbrave1502020OS13.4 months19.2 months
CheckMate 4592021OS14.7 months16.4 months
HIMALAYA2022OS13.77 months16.43 months
CheckMate 9DW2025OS20.6 months23.7 months

Advanced HCC, after sorafenib

Adults with advanced hepatocellular carcinoma previously treated with sorafenib, in placebo-controlled trials with best supportive care · Metastatic · endpoint: OS

Results are not comparable across trials.

Entry rules differ: RESORCE required sorafenib tolerance and progression on sorafenib; CELESTIAL allowed up to two previous systemic regimens; REACH-2 required an alpha-fetoprotein of 400 ng/mL or more while REACH had no threshold; KEYNOTE-240 enrolled patients previously treated with sorafenib and KEYNOTE-394 enrolled patients from Asia, including some who had received oxaliplatin-based chemotherapy. KEYNOTE-240 did not reach its prespecified one-sided significance level for overall survival; KEYNOTE-394 did. REACH (unselected) did not show a significant overall survival difference and is plotted beside REACH-2 for that reason. CELESTIAL's figure is from its second planned interim analysis. Single-arm records (KEYNOTE-224, and the CheckMate 040 milestone) are not drawn. Eras differ; compare only each trial's own two arms.

Each pair shows one trial: its control arm and its experimental arm. Yesterday's experimental arm is often today's control. Populations, definitions and follow-up differ.

05101520OS, median (months)201620182018201820192019RESORCECELESTIALKEYNOTE-240KEYNOTE-394REACHREACH-2
  • experimental arm
  • control arm
  • different endpoint from the lane's
  • step line through experimental values
OS, median, across 6 trials in this lane. RESORCE (2016): experimental 10.6 months vs control 7.8 months; CELESTIAL (2018): experimental 10.2 months vs control 8 months; KEYNOTE-240 (2018): experimental 13.9 months vs control 10.6 months; KEYNOTE-394 (2018): experimental 14.6 months vs control 13 months; REACH (2019): experimental 9.2 months vs control 7.6 months; REACH-2 (2019): experimental 8.5 months vs control 7.3 months.
Data table
TrialYearEndpointControl (months)Experimental (months)
RESORCE2016OS7.8 months10.6 months
CELESTIAL2018OS8 months10.2 months
KEYNOTE-2402018OS10.6 months13.9 months
KEYNOTE-3942018OS13 months14.6 months
REACH2019OS7.6 months9.2 months
REACH-22019OS7.3 months8.5 months

Resected or ablated HCC, adjuvant therapy

Adults with hepatocellular carcinoma after curative-intent resection or local ablation (STORM: complete radiological response; IMbrave050: high risk of recurrence) · Adjuvant · endpoint: RFS

Results are not comparable across trials.

STORM was double-blind and placebo-controlled; IMbrave050 was open-label with an active-surveillance control, enrolled people defined as high risk, and assessed recurrence by independent review. IMbrave050 reported hazard ratios without arm medians (median recurrence-free survival was not evaluable at its 2023 interim analysis), so only STORM is drawn; IMbrave050's 2023 interim result and its 2026 post hoc update are described in the milestone with their own analysis labels. Eras and risk definitions differ; compare only each trial's own two arms.

Each pair shows one trial: its control arm and its experimental arm. Yesterday's experimental arm is often today's control. Populations, definitions and follow-up differ.

010203040RFS, median (months)2015STORM
  • experimental arm
  • control arm
  • different endpoint from the lane's
  • step line through experimental values
RFS, median, across 1 trial in this lane. STORM (2015): experimental 33.3 months vs control 33.7 months.
Data table
TrialYearEndpointControl (months)Experimental (months)
STORM2015RFS33.7 months33.3 months

Embolisation-amenable HCC, TACE with or without systemic therapy

Adults with unresectable, non-metastatic hepatocellular carcinoma amenable to transarterial chemoembolisation (EMERALD-1, LEAP-012) · Locally advanced · endpoint: PFS

Results are not comparable across trials.

Both trials compared TACE plus an immunotherapy-based combination with TACE plus placebo, by blinded independent central review, but tumour-response criteria differ (LEAP-012 used a study-modified RECIST 1.1 with LI-RADS criteria for new intrahepatic tumours), as do the partner drugs, regions and tumour-burden rules. EMERALD-1's plotted pair is TACE plus durvalumab plus bevacizumab versus TACE plus placebo; its durvalumab-only arm is not drawn. Overall survival was immature or not significant at the analyses reported, so this lane shows progression-free survival only. Compare only each trial's own two arms.

Each pair shows one trial: its control arm and its experimental arm. Yesterday's experimental arm is often today's control. Populations, definitions and follow-up differ.

05101520PFS, median (months)20252025EMERALD-1LEAP-012
  • experimental arm
  • control arm
  • different endpoint from the lane's
  • step line through experimental values
PFS, median, across 2 trials in this lane. EMERALD-1 (2025): experimental 15 months vs control 8.2 months; LEAP-012 (2025): experimental 14.6 months vs control 10 months.
Data table
TrialYearEndpointControl (months)Experimental (months)
EMERALD-12025PFS8.2 months15 months
LEAP-0122025PFS10 months14.6 months

Locally advanced HCC, radioembolisation versus sorafenib

Adults with locally advanced or inoperable hepatocellular carcinoma (SARAH in France; SIRveNIB in the Asia-Pacific region) · Locally advanced · endpoint: OS

Results are not comparable across trials.

Both were open-label trials comparing yttrium-90 radioembolisation with sorafenib, with sorafenib as the control arm. Regions, eligibility (SARAH also allowed people with recurrence after a previous cure or after unsuccessful TACE), sorafenib dosing and the share of people assigned to radioembolisation who did not receive it differ. The differences in overall survival were not statistically significant in either trial. Compare only each trial's own two arms.

Each pair shows one trial: its control arm and its experimental arm. Yesterday's experimental arm is often today's control. Populations, definitions and follow-up differ.

02468101214OS, median (months)20172017SARAHSIRveNIB
  • experimental arm
  • control arm
  • different endpoint from the lane's
  • step line through experimental values
OS, median, across 2 trials in this lane. SARAH (2017): experimental 8 months vs control 9.9 months; SIRveNIB (2017): experimental 8.8 months vs control 10 months.
Data table
TrialYearEndpointControl (months)Experimental (months)
SARAH2017OS9.9 months8 months
SIRveNIB2017OS10 months8.8 months

Unresectable HCC, chemoembolisation versus conservative care

Adults with unresectable hepatocellular carcinoma not suitable for curative treatment (a Barcelona trial with Child-Pugh A or B disease; a Hong Kong trial in Asian patients) · Locally advanced · endpoint: OS

Results are not comparable across trials.

Both trials reported survival probabilities (here at 12 months, the first landmark each paper gives), not medians. The control was conservative treatment in the Barcelona trial (which stopped early at a sequential inspection and also had an embolisation-only arm, not drawn) and symptomatic treatment in the Hong Kong trial. Both were small (112 and 80 randomised) and selected from a larger number assessed. Regions, chemotherapy regimens and eras of supportive care differ. Compare only each trial's own two arms.

Each pair shows one trial: its control arm and its experimental arm. Yesterday's experimental arm is often today's control. Populations, definitions and follow-up differ.

020406080100OS at 12 months (%)20022002lo-2002-tace-rctllovet-2002-tace-rct
  • experimental arm
  • control arm
  • different endpoint from the lane's
  • step line through experimental values
OS at 12 months, rate, across 2 trials in this lane. lo-2002-tace-rct (2002): experimental 57 % vs control 32 %; llovet-2002-tace-rct (2002): experimental 82 % vs control 63 %.
Data table
TrialYearEndpointControl (%)Experimental (%)
lo-2002-tace-rct2002OS32 %57 %
llovet-2002-tace-rct2002OS63 %82 %

Standard of care, then and now

How the usual sequence of treatment changed for one setting. Choose a point in time; each step links to the milestone behind it.

Advanced HCC: first-line systemic therapy

Adults with unresectable or advanced hepatocellular carcinoma who had not received systemic treatment for advanced disease · Metastatic

  1. Targeted1SorafenibSorafenib: SHARP and Asia-Pacific trials in advanced HCC (2007)
  2. Targeted2Lenvatinib (non-inferior to sorafenib)REFLECT: lenvatinib non-inferior to sorafenib first line (2018)
  3. Immunotherapy3Atezolizumab plus bevacizumabIMbrave150: atezolizumab plus bevacizumab versus sorafenib (2020)
  4. Immunotherapy4Tremelimumab plus durvalumabHIMALAYA: tremelimumab plus durvalumab (STRIDE) versus sorafenib (2022)
  5. Immunotherapy5Nivolumab plus ipilimumabCheckMate 9DW: first-line nivolumab plus ipilimumab (2025)
2025. Based only on the trials cited in each step. Two more immunotherapy combinations were tested in untreated disease: tremelimumab plus durvalumab (FDA 2022) and nivolumab plus ipilimumab (FDA April 2025).
Text version
  • 2008: Sorafenib
  • 2018: Sorafenib → Lenvatinib (non-inferior to sorafenib)
  • 2021: Sorafenib → Lenvatinib (non-inferior to sorafenib) → Atezolizumab plus bevacizumab
  • 2025: Sorafenib → Lenvatinib (non-inferior to sorafenib) → Atezolizumab plus bevacizumab → Tremelimumab plus durvalumab → Nivolumab plus ipilimumab

Advanced HCC: treatment after sorafenib

Adults with advanced hepatocellular carcinoma whose disease progressed on, or who could not tolerate, sorafenib · Metastatic

  1. Targeted1Regorafenib (after sorafenib tolerance)RESORCE: regorafenib after sorafenib progression (2016)
  2. Targeted2CabozantinibCELESTIAL: cabozantinib after prior systemic therapy (2018)
  3. Immunotherapy3Pembrolizumab (HCC secondary to HBV)Pembrolizumab: KEYNOTE-224, -240 and -394 and the FDA label (2018)
  4. Targeted4Ramucirumab (AFP 400 ng/mL or more)REACH-2: ramucirumab in AFP-selected patients after sorafenib (2019)
  5. Immunotherapy5Nivolumab plus ipilimumab (converted 2025)CheckMate 9DW: first-line nivolumab plus ipilimumab (2025)
2025. Based only on the trials cited in each step. The FDA converted pembrolizumab to regular approval (2024, revised indication) and nivolumab with ipilimumab after sorafenib (April 2025).
Text version
  • 2017: Regorafenib (after sorafenib tolerance) → Nivolumab (accelerated approval)
  • 2019: Regorafenib (after sorafenib tolerance) → Nivolumab (accelerated approval) → Cabozantinib → Pembrolizumab (accelerated approval) → Ramucirumab (AFP 400 ng/mL or more)
  • 2021: Regorafenib (after sorafenib tolerance) → Cabozantinib → Pembrolizumab (accelerated approval) → Ramucirumab (AFP 400 ng/mL or more)
  • 2025: Regorafenib (after sorafenib tolerance) → Cabozantinib → Pembrolizumab (HCC secondary to HBV) → Ramucirumab (AFP 400 ng/mL or more) → Nivolumab plus ipilimumab (converted 2025)

Reversals and dead ends

Approaches that were abandoned or withdrawn, and trials whose results were later questioned, are part of the story.

  • 2015

    STORM: adjuvant sorafenib did not improve recurrence-free survival

    After surgery or local ablation for liver cancer, 1,114 people were randomly assigned to sorafenib tablets or a placebo for up to four years to see whether it delayed recurrence. Median recurrence-free survival was about 33 months in both groups, and the authors concluded that sorafenib was not effective in this setting.

    Phase 3, double-blind, placebo-controlled trial, N=1114 with a complete radiological response after resection or ablation: sorafenib 400 mg twice daily vs placebo for up to 4 years. Final analysis (464 events): median RFS 33.3 vs 33.7 months (HR 0.940; 95% CI 0.780-1.134; one-sided p=0.26). Grade 3 or 4 hand-foot skin reaction: 28% vs <1%.

  • 2017

    SARAH and SIRveNIB: no survival gain over sorafenib

    Two randomised trials compared radioembolisation (tiny radioactive beads placed in the liver's blood supply) with sorafenib tablets in people with locally advanced liver cancer. Median survival was roughly 8 to 9 months with radioembolisation and about 10 months with sorafenib, and the differences were not statistically significant.

    SARAH (France; N=467 randomised; open-label): median OS 8.0 months with SIRT vs 9.9 with sorafenib, HR 1.15 (95% CI 0.94-1.41; p=0.18). SIRveNIB (Asia-Pacific; N=360; open-label): median OS 8.8 vs 10.0 months, HR 1.1 (95% CI 0.9-1.4; P=0.36). Of those assigned to SIRT, 22% (SARAH) and 28.6% (SIRveNIB) did not receive it.

  • 2021

    CheckMate 459 and withdrawal of nivolumab's HCC indication

    In 743 people with advanced liver cancer who had not had drug treatment, nivolumab did not reach its prespecified significance level for overall survival against sorafenib (median 16.4 versus 14.7 months). The single-agent nivolumab indication after sorafenib, accelerated in 2017, appears on the FDA's withdrawn list with a date of 23 July 2021.

    CheckMate 459 (open-label phase 3, ITT N=743, first-line): nivolumab (n=371) vs sorafenib (n=372), median OS 16.4 vs 14.7 months, HR 0.85 (95% CI 0.72-1.02; P=0.075); protocol-defined significance level P=0.0419 not reached; median follow-up 15.2 and 13.4 months. FDA list: single-agent nivolumab for hepatocellular carcinoma after sorafenib, accelerated approval 22 September 2017, withdrawal 23 July 2021; no reason stated.

  • 2026

    IMbrave050: early recurrence-free benefit not sustained on update

    A trial gave atezolizumab plus bevacizumab for one year, or active monitoring, to 668 people at high risk of recurrence after liver cancer surgery or ablation. A 2023 interim analysis reported fewer recurrences or deaths with the combination, but a 2026 update found no sustained benefit, and its authors do not support adjuvant use.

    Open-label phase 3, N=668 (334 per arm) after resection or ablation at high risk of recurrence: atezolizumab plus bevacizumab for 17 cycles vs active surveillance. Interim analysis (median follow-up 17.4 months): IRF-assessed RFS HR 0.72 (adjusted 95% CI 0.53-0.98; p=0.012). Updated post hoc analysis: RFS HR 0.90 (95% CI 0.72-1.12); second interim OS HR 1.26 (95% CI 0.85-1.87), immature.

Today and tomorrow

Standard of care today

Prevention
No FDA or EMA action specific to preventing hepatocellular carcinoma is described here. Hepatitis B vaccination (Taiwan programme) is covered in the Foundations timeline. AASLD Practice Guidance 2023 (Hepatology, PMID 37199193) and the EASL Clinical Practice Guidelines 2025 (J Hepatol, PMID 39690085) are cited by identifier only; their text was not read, so no recommendation is attributed to them. A 2025 'Critical Update' to the AASLD guidance exists (Hepatology, PMID 39992051); it was not read and is not cited here, and whether the other documents are the current editions was not checked.
Guideline: Singal AG, Llovet JM, Yarchoan M, Mehta N, Heimbach JK, Dawson LA, et al 2023Guideline: European Association for the Study of the Liver 2025
Screening
No FDA or EMA action on surveillance for hepatocellular carcinoma is described here. The randomised surveillance trials from China and the later cohort and meta-analytic evidence are in the milestone timeline. Surveillance recommendations are addressed by AASLD 2023 (Hepatology, PMID 37199193) and EASL 2025 (J Hepatol, PMID 39690085), cited by identifier only; their text was not read, so no recommendation is attributed to them.
Guideline: Singal AG, Llovet JM, Yarchoan M, Mehta N, Heimbach JK, Dawson LA, et al 2023Guideline: European Association for the Study of the Liver 2025Tzartzeva K, Obi J, Rich NE, Parikh ND, Marrero JA, Yopp A, et al 2018
Localised
Resection, ablation and liver transplantation. No drug approval is described here for these options, and FDA device clearances for ablation or embolisation were not examined. The randomised resection-versus-ablation trials and the transplantation criteria are in the milestone timeline. The BCLC strategy 2022 update (J Hepatol, PMID 34801630), AASLD 2023 (PMID 37199193), EASL 2025 (PMID 39690085) and ESMO 2025 (Ann Oncol, PMID 39986353) are cited by identifier only; their text was not read, so no allocation rule or recommendation is attributed to them. A 2025 'Critical Update' to the AASLD guidance exists (Hepatology, PMID 39992051); it was not read and is not cited here, and whether the other documents are the current editions was not checked.
Reig M, Forner A, Rimola J, Ferrer-Fàbrega J, Burrel M, Garcia-Criado Á, et al 2022Guideline: Singal AG, Llovet JM, Yarchoan M, Mehta N, Heimbach JK, Dawson LA, et al 2023Guideline: European Association for the Study of the Liver 2025Guideline: Vogel A, Chan SL, Dawson LA, Kelley RK, Llovet JM, Meyer T, et al 2025
Adjuvant

After resection or ablation. No FDA or EMA approval of adjuvant systemic therapy for hepatocellular carcinoma was found in the documents read:

  • the Imfinzi prescribing-information record read (openFDA, label version 34, effective 31 August 2026) lists hepatocellular carcinoma only as use with tremelimumab in unresectable disease. Label records for other agents, including Tecentriq, were not read, so the absence of an adjuvant approval is not established beyond that. STORM (adjuvant sorafenib) and IMbrave050 (adjuvant atezolizumab plus bevacizumab) are in the milestone timeline. Guideline documents (AASLD 2023, PMID 37199193
  • EASL 2025, PMID 39690085) are cited by identifier only
  • their text was not read
FDA 2026Bruix J, Takayama T, Mazzaferro V, Chau GY, Yang J, Kudo M, et al 2015Qin S, Chen M, Cheng AL, Kaseb AO, Kudo M, Lee HC, et al 2023Guideline: Singal AG, Llovet JM, Yarchoan M, Mehta N, Heimbach JK, Dawson LA, et al 2023Guideline: European Association for the Study of the Liver 2025
Locally advanced
Unresectable disease amenable to transarterial chemoembolisation (TACE). The Imfinzi prescribing-information record read (openFDA, label version 34, effective 31 August 2026) lists hepatocellular carcinoma only as use with tremelimumab in unresectable disease, with no TACE-combination indication; label records for other agents were not read. Phase 3 trials of systemic therapy added to TACE (EMERALD-1 and LEAP-012) are in the milestone timeline. The BCLC strategy 2022 update (PMID 34801630) and ESMO 2025 (PMID 39986353) are cited by identifier only; their text was not read, so no recommendation is attributed to them.
FDA 2026Sangro B, Kudo M, Erinjeri JP, Qin S, Ren Z, Chan SL, et al 2025Kudo M, Ren Z, Guo Y, Han G, Lin H, Zheng J, et al 2025Reig M, Forner A, Rimola J, Ferrer-Fàbrega J, Burrel M, Garcia-Criado Á, et al 2022Guideline: Vogel A, Chan SL, Dawson LA, Kelley RK, Llovet JM, Meyer T, et al 2025
Metastatic

Unresectable or metastatic hepatocellular carcinoma, first-line treatment. FDA approval letters read:

  • sorafenib, 16 November 2007 (NDA 21-923/S-007
  • unresectable disease)
  • lenvatinib, 15 August 2018 (NDA 206947/S-007)
  • atezolizumab with bevacizumab, 29 May 2020 (for patients with no prior systemic therapy)
  • durvalumab with tremelimumab-actl, 21 October 2022 (effective date in the corrected tremelimumab letter, signed 31 October 2022)
  • nivolumab with ipilimumab, 11 April 2025 (BLA 125554/S-129 and 125377/S-133). None of these letters describes the first-line indication as an accelerated approval (the 2025 nivolumab and ipilimumab letters mention accelerated approval only for the earlier previously treated indication), so the type 'regular' is the project's inference from that absence. EMA: the Imjudo product page shows marketing authorisation on 20 February 2023 for tremelimumab with durvalumab as first-line treatment of adults with advanced or unresectable hepatocellular carcinoma (opinion adopted 15 December 2022)
  • the page does not say whether it is conditional or under exceptional circumstances. EU decisions for the other first-line agents were not read. ASCO 2024 (J Clin Oncol, PMID 38502889), APASL 2024 (Hepatol Int, PMID 39570557) and ESMO 2025 (PMID 39986353) are cited by identifier only
  • their text was not read
U 2007U 2018U 2020U 2020U 2022U 2022U 2025U 2025European Medicines Agency 2023Guideline: Gordan JD, Kennedy EB, Abou-Alfa GK, Beal E, Finn RS, Gade TP, et al 2024Guideline: Lau G, Obi S, Zhou J, Tateishi R, Qin S, Zhao H, et al 2024Guideline: Vogel A, Chan SL, Dawson LA, Kelley RK, Llovet JM, Meyer T, et al 2025
Relapsed / refractory

Hepatocellular carcinoma after prior systemic therapy, FDA approval letters read. Regorafenib, 27 April 2017 (previously treated with sorafenib); cabozantinib, 14 January 2019 (previously treated with sorafenib); ramucirumab, 10 May 2019 (alpha fetoprotein of 400 ng/mL or more and treated with sorafenib). Pembrolizumab, 9 November 2018 under accelerated approval (previously treated with sorafenib); on 25 January 2024 the indication was converted to regular approval and revised to hepatocellular carcinoma secondary to hepatitis B in patients who have received prior systemic therapy other than a PD-1/PD-L1-containing regimen (the letter does not state the reason for the narrower wording). Nivolumab with ipilimumab, 10 March 2020 under accelerated approval (previously treated with sorafenib), converted to regular approval on 11 April 2025 (the letters state that approval fulfils the commitments made under 21 CFR 601.41). Single-agent nivolumab:

  • accelerated approval 22 September 2017, withdrawn 23 July 2021 per the FDA list of withdrawn cancer accelerated approvals (content current as of 5 October 2026), which gives no reason. EU documents for these indications were not read. Guideline documents (ASCO 2024, PMID 38502889
  • ESMO 2025, PMID 39986353) are cited by identifier only
  • their text was not read
U 2017U 2019U 2019U 2018U 2024U 2020U 2020U 2025U 2025U 2017FDA 2026Guideline: Gordan JD, Kennedy EB, Abou-Alfa GK, Beal E, Finn RS, Gade TP, et al 2024Guideline: Vogel A, Chan SL, Dawson LA, Kelley RK, Llovet JM, Meyer T, et al 2025

Trials to watch

  • ARTEMIDE-HCC01 NCT06921785

    In first-line advanced hepatocellular carcinoma (BCLC stage B not eligible for locoregional therapy, or stage C; Child-Pugh class A), does rilvegostomig plus bevacizumab, with or without tremelimumab, differ in outcome from atezolizumab plus bevacizumab? Registry record: randomised, open-label, sponsor-blinded phase 3 with three arms (rilvegostomig, bevacizumab and tremelimumab; rilvegostomig and bevacizumab; atezolizumab and bevacizumab); status recruiting; enrolment 1,220 (estimated); the single primary outcome listed is overall survival for the tremelimumab-containing arm versus atezolizumab plus bevacizumab; estimated primary completion March 2030; no results posted.

  • CheckMate 9DX NCT03383458

    After curative resection or ablation of hepatocellular carcinoma at high risk of recurrence, does adjuvant nivolumab differ from placebo in recurrence-free survival? Registry record: randomised, double-blind phase 3; status active, not recruiting; enrolment 545 (actual); the primary outcome listed is recurrence-free survival; primary completion listed as 27 May 2026 (actual) and study completion estimated May 2031; no results posted on the registry.

  • Adjuvant camrelizumab plus rivoceranib NCT04639180

    After curative resection or ablation of hepatocellular carcinoma at high risk of recurrence, does adjuvant camrelizumab plus rivoceranib (apatinib) differ from active surveillance in recurrence-free survival? Registry record (no trial acronym): randomised, open-label phase 3; status active, not recruiting; enrolment 687 (actual); the primary outcome listed is recurrence-free survival by blinded independent review committee; estimated primary completion July 2026 (a date that has passed without a registry update since 30 December 2025); no results posted.

  • HELIO-RT NCT07166406

    In hepatocellular carcinoma with macrovascular invasion, does adding liver stereotactic body radiotherapy to immunotherapy-based systemic treatment (atezolizumab with bevacizumab, tremelimumab with durvalumab, or nivolumab with ipilimumab) differ in overall survival from the systemic treatment alone? Registry record: randomised phase 3 (NRG Oncology); status recruiting; enrolment 226 (estimated); the primary outcome listed is overall survival; estimated primary completion March 2029; no results posted.

  • KYLIN-02 NCT06618664

    In first-line advanced hepatocellular carcinoma, does the anti-CTLA-4 antibody SHR-8068 added to adebrelimab and bevacizumab differ in outcome from sintilimab plus bevacizumab? Registry record: randomised, open-label phase 3; status recruiting; enrolment 590 (estimated); the primary outcomes listed are progression-free survival and overall survival; estimated primary completion October 2026; no results posted.

References

Show all 89 references

Loading references… Open the list as a page

Every figure was checked against its primary source to the extent possible. Items that could not be fully confirmed are marked “needs review”. Educational history, not medical advice.

Report an error

Links open the primary source. A flag marked ⚑ needs review means a value or claim has not yet been independently confirmed.

2004TreatmentDetection & prevention

notablePrevention

Lamivudine in advanced hepatitis B: fewer HCC, a borderline finding

In a placebo-controlled trial of 651 people with advanced hepatitis B, the antiviral lamivudine slowed disease progression. Liver cancer was less frequent with lamivudine, about 4 in 100 versus 7 in 100, but that difference was borderline and the virus often developed resistance.

Randomised placebo-controlled trial (n=651; 98% Asian, 85% male) of lamivudine 100 mg daily versus placebo (2:1) in hepatitis B with cirrhosis or advanced fibrosis; stopped early at a median treatment duration of 32.4 months. Disease progression (composite including HCC): 7.8% versus 17.7% (HR 0.45, P=0.001). HCC: 3.9% versus 7.4% (HR 0.49, P=0.047). YMDD resistance mutations arose in 49% of lamivudine-treated patients.

Standard before

Whether antiviral therapy prevents disease progression in chronic hepatitis B with advanced fibrosis or cirrhosis was unknown, per the trial's background statement.

Why it mattered

Randomised, placebo-controlled evidence that antiviral treatment of advanced hepatitis B delayed clinical progression, with a lower incidence of HCC as one component of the composite.

Caveats

  • HCC was a component of the composite primary end point, not a separate primary end point; its P value (0.047) is borderline and the trial stopped early.
  • Confidence intervals for the hazard ratios are not given in the abstract.
  • Genotypic resistance (YMDD) mutations developed in 49% of lamivudine-treated patients, and the Child-Pugh score was more likely to increase in patients with these mutations.
  • The study population was 98% Asian; the trial predates newer nucleos(t)ide analogues, whose HCC data are not covered in this entry.