Skip to content

Hepatocellular carcinoma — statistics

Last curated: 2026-08-30

Figures are kept with source year, population, and method. Estimates from unlike populations are shown side by side and are not averaged.

Global burden and substrate

Statistic Estimate Source year/population Method/source
All-cancer incident cases ~20 million 2022; 185 countries IARC GLOBOCAN modeled estimates (Bray 2024, PMID 38572751)
All-cancer deaths 9.7 million 2022; 185 countries IARC GLOBOCAN modeled estimates (Bray 2024, PMID 38572751)
Liver-cancer share of cancer deaths 7.8% 2022; global IARC GLOBOCAN modeled estimates (Bray 2024, PMID 38572751)
Global chronic liver disease prevalence ~1.5 billion people Contemporary global review Published data plus CDC/GBD synthesis (Moon 2020, PMID 31401364)
Chronic liver disease/cirrhosis incidence 20.7 per 100,000 Global Review of epidemiologic sources (Moon 2020, PMID 31401364)
Change in CLD/cirrhosis incidence +13% since 2000 Global Review synthesis (Moon 2020, PMID 31401364)

Etiologic risk and prevention

Statistic Estimate Source year/population Method/source
NAFLD prevalence ~25% of world population Evidence summarized through 2020 Epidemiologic review (Huang 2021, PMID 33349658)
HCC incidence in NASH cirrhosis 0.5%–2.6% per year International published cohorts Review range (Huang 2021, PMID 33349658)
HCC incidence in non-cirrhotic NAFLD ~0.1–1.3 per 1,000 patient-years International published cohorts Review range (Huang 2021, PMID 33349658)
Non-cirrhotic NAFLD HCC relative risk adjusted HR 2.69 (95% CI 1.33–5.44) South Korean national cohort; published 2025 Claims/ICD cohort refined by fatty liver index (Lee 2025, PMID 40345323)
HCC incidence in alcohol-associated cirrhosis 0.9%–5.6% per year International cohorts Review range (Huang 2023, PMID 36258033)
Alcohol share of global HCC deaths ~20% 2019; global Modeled burden synthesis (Huang 2023, PMID 36258033)
Alcohol share of global cirrhosis deaths ~25% 2019; global Modeled burden synthesis (Huang 2023, PMID 36258033)
Alcohol per-capita consumption 5.5 L (2005); 6.4 L (2016); 7.6 L projected (2030) Global Modeled trend/projection (Huang 2023, PMID 36258033)
HCC occurrence after HCV SVR 1.54 per 100 person-years 23 studies; 29,395 treated patients Random-effects meta-analysis (Lv 2024, PMID 38965190)
HCC occurrence after HCV nonresponse 7.80 per 100 person-years Same meta-analysis Random-effects meta-analysis (Lv 2024, PMID 38965190)
HCC occurrence after SVR with cirrhosis 2.47 per 100 person-years Treated HCV cohorts Meta-analysis subgroup (Lv 2024, PMID 38965190)
HCC occurrence after SVR without cirrhosis 0.85 per 100 person-years Treated HCV cohorts Meta-analysis subgroup (Lv 2024, PMID 38965190)
Taiwan HBV vaccination evaluation 1,958 HCC cases aged 6–29; 5,524,435 vaccinees in post-program comparison 1983–2004 Taiwan Population registries and vaccination records (Chang 2009, PMID 19759364)

Surveillance performance and outcomes

Statistic Estimate Source year/population Method/source
Shanghai surveillance trial population 18,816 HBV-infected adults Shanghai; published 2004 Randomized surveillance trial (Zhang 2004, PMID 15042359)
Trial adherence 58% Shanghai HBV trial Trial follow-up (Zhang 2004, PMID 15042359)
HCC mortality effect Rate ratio 0.63 (95% CI 0.41–0.98) Shanghai HBV trial Randomized comparison (Zhang 2004, PMID 15042359)
Early detection with surveillance RR 1.86 59 studies; 145,396 cirrhosis patients Observational meta-analysis (Singal 2022, PMID 35139400)
Curative treatment receipt RR 1.83 Same meta-analysis Observational meta-analysis (Singal 2022, PMID 35139400)
Survival after lead-time adjustment HR 0.67 Same meta-analysis Observational meta-analysis (Singal 2022, PMID 35139400)
Ultrasound sensitivity, any-stage HCC 84% Cirrhosis surveillance studies Meta-analysis (Tzartzeva 2018, PMID 29425931)
Ultrasound sensitivity, early-stage HCC 47% Cirrhosis surveillance studies Meta-analysis (Tzartzeva 2018, PMID 29425931)
Ultrasound + AFP sensitivity, early-stage 63% Cirrhosis surveillance studies Meta-analysis (Tzartzeva 2018, PMID 29425931)
Modeled deaths averted ~13 per 1,000 over 5 years Simulated cirrhosis cohort Decision model (Taylor 2017, PMID 28605060)
Modeled false-positive experiences ~150 per 1,000 over 5 years Same simulated cohort Decision model (Taylor 2017, PMID 28605060)
Physical-harm model input 15%–20% Compensated cirrhosis surveillance Literature-informed Markov model (Parikh 2020, PMID 32530829)
Surveillance-intervention meta-analysis 9 studies; 4,550 patients Program studies through 2020 Systematic review/meta-analysis (Ramai 2023, PMID 34999648)
PREMIUM planned enrollment 4,700 Recruiting US veterans with cirrhosis; ultrasound+AFP vs abbreviated MRI+AFP Mortality-powered randomized trial; no outcomes as of 2026-08-30 (NCT05486572; Ioannou 2026, PMID 41624484)

Diagnostic accuracy

Statistic Estimate Source year/population Method/source
CT/MRI LR-5 sensitivity 67% (95% CI 62%–72%) 14 studies; 2,056 patients; 2,589 observations Random-effects meta-analysis (Lee 2020, PMID 32145134)
CT/MRI LR-5 specificity 92% (95% CI 88%–95%) Same dataset Random-effects meta-analysis (Lee 2020, PMID 32145134)
CEUS LR-5 sensitivity 0.71 (95% CI 0.69–0.72) 8 studies; 4,215 focal lesions Meta-analysis (Li 2021, PMID 33307595)
CEUS LR-5 specificity 0.88 (95% CI 0.85–0.91) Same dataset Meta-analysis (Li 2021, PMID 33307595)
CEUS LR-M sensitivity for other malignancy 0.85 (95% CI 0.81–0.88) Same dataset Meta-analysis (Li 2021, PMID 33307595)
CEUS LR-M specificity for other malignancy 0.86 (95% CI 0.85–0.87) Same dataset Meta-analysis (Li 2021, PMID 33307595)
Needle-biopsy seeding 0.62%; range 0%–7.77% 37 studies; 13,959 HCC biopsies Systematic review (Nie 2025, PMID 39692333)
Chinese GALAD validation sample 602 HCC; 923 controls Five Chinese centres Multicentre case-control/longitudinal study (Huang 2022, PMID 34679250)
Early BCLC share in GALAD HCC cases 34.1% BCLC 0–A Same study Case characterization (Huang 2022, PMID 34679250)

Curative and transplant outcomes

Statistic Estimate Source year/population Method/source
Milan four-year overall survival 85% Selected cirrhotic HCC transplant cohort Prospective cohort (Mazzaferro 1996, PMID 8594428)
Milan four-year recurrence-free survival 92% Same cohort Prospective cohort (Mazzaferro 1996, PMID 8594428)
Metroticket 2.0 discrimination c-statistic ~0.78 International transplant cohorts Competing-risk development/validation (Mazzaferro 2018, PMID 28989060)
RETREAT three-year recurrence, score 0 1.6% 3,276 UNOS transplant recipients External validation (Mehta 2018, PMID 29068145)
RETREAT three-year recurrence, score ≥5 29% Same cohort External validation (Mehta 2018, PMID 29068145)
RETREAT three-year survival, score 0 91% Same cohort External validation (Mehta 2018, PMID 29068145)
RETREAT three-year survival, score ≥5 58% Same cohort External validation (Mehta 2018, PMID 29068145)
RFA vs MWA complete ablation 96.7% vs 96.9% 5 RCTs; 413 RFA and 431 MWA RCT meta-analysis (Yu 2021, PMID 33870454)
RFA vs resection evidence base 3 RCTs + 20 retrospective; 15,482 patients Small HCC studies through 2014 Meta-analysis (Feng 2015, PMID 24889505)

Locoregional therapy

Statistic Estimate Source year/population Method/source
TARE vs TACE overall-survival difference −0.55 months (95% CI −1.95 to 3.05) 17 studies; 2,465 patients Aggregate meta-analysis (Brown 2023, PMID 35943116)
Mean time to progression TARE 17.5 vs TACE 9.8 months 3 studies with TTP Comparative meta-analysis (Brown 2023, PMID 35943116)
Mean TTP difference 4.8 months (95% CI 1.3–8.3) Same studies Comparative meta-analysis (Brown 2023, PMID 35943116)

Systemic phase 3 anchors

Trial/statistic Estimate Population Method/source
SHARP median OS 10.7 vs 7.9 months Advanced HCC; sorafenib vs placebo Phase 3 RCT (Llovet 2008, PMID 18650514)
SHARP OS hazard ratio 0.69 Same trial Phase 3 RCT (Llovet 2008, PMID 18650514)
REFLECT median OS 13.6 vs 12.3 months Unresectable HCC; lenvatinib vs sorafenib Phase 3 non-inferiority RCT (Kudo 2018, PMID 29433850)
REFLECT OS hazard ratio 0.92 Same trial Phase 3 RCT (Kudo 2018, PMID 29433850)
RESORCE median OS 10.6 vs 7.8 months Post-sorafenib; regorafenib vs placebo Phase 3 RCT (Bruix 2017, PMID 27932229)
RESORCE OS hazard ratio 0.63 Same trial Phase 3 RCT (Bruix 2017, PMID 27932229)
CheckMate 9DW median OS 23.7 vs 20.6 months 668 randomized patients; nivolumab-ipilimumab vs lenvatinib/sorafenib Global phase 3 interim analysis (Yau 2025, PMID 40349714)
CheckMate 9DW OS hazard ratio 0.79 (95% CI 0.65–0.96) Median follow-up 35.2 months Global phase 3 interim analysis (Yau 2025, PMID 40349714)
CheckMate 9DW early/late OS HRs First 6 months 1.65 (95% CI 1.12–2.43); thereafter 0.61 (0.48–0.77) Non-proportional survival effect Prespecified global phase 3 analysis (Yau 2025, PMID 40349714)
IMbrave050 updated RFS HR 0.90 (95% CI 0.72–1.12) 668 high-risk patients after resection/ablation; atezolizumab-bevacizumab vs surveillance Initial benefit not sustained (Yopp 2026, PMID 41580093)
IMbrave050 updated OS HR 1.26 (95% CI 0.85–1.87) Second interim analysis; OS immature Benefit–risk did not support adjuvant use (Yopp 2026, PMID 41580093)
LEVIATHAN post-atezolizumab-bevacizumab OS 11.9 vs 7.4 months; HR 0.67; p=0.018 Lenvatinib vs sorafenib; 230-patient observational comparison Propensity adjustment does not replace randomization (Lombardi 2025, PMID 41321927)

Disparities and access

Statistic Estimate Population/method Source
Black vs White survival pooled HR 1.08 (95% CI 1.05–1.12) 35 US studies; 563,097 HCC patients Meta-analysis (Rich 2022, PMID 33387668)
Black vs White early-stage detection OR 0.66 (95% CI 0.54–0.78) 20 studies reporting stage Meta-analysis (Rich 2022, PMID 33387668)
Curative treatment receipt 18.9% (2,617/13,874) SEER-Medicare, age ≥65, 2001–2015 Retrospective cohort (Wagle 2022, PMID 34796703)
Black vs White curative treatment OR 0.76 (95% CI 0.64–0.91) Same cohort Adjusted logistic model (Wagle 2022, PMID 34796703)
Low-SES Black vs White curative treatment OR 0.64 (95% CI 0.49–0.84) High-poverty neighborhoods Stratified adjusted model (Wagle 2022, PMID 34796703)
SEER disparity cohort 45,789 HCC cases Histologically confirmed, 2000–2015 Population registry (Flores 2021, PMID 33737301)
Incidence extremes in SEER analysis 12.1 low-SES API vs 3.2 high-SES White Rates per 100,000 Age-adjusted SEER analysis (Flores 2021, PMID 33737301)

Patient experience

Statistic Estimate Population/method Source
Qualitative review 11 studies; 5 thematic domains Systematic qualitative synthesis Norman 2022, PMID 35974658
International journey study 50 patients + 12 clinicians Semi-structured interviews Wörns 2024, PMID 39052152
HCC interview study 25 patients; median age 63; 68% men Qualitative concept elicitation Patel 2022, PMID 34115280
Terminal HCC longitudinal study 14 patients; 45 interviews Monthly interviews up to 6 months Hansen 2015, PMID 25122134
Caregiver longitudinal study 13 caregivers; 39 interviews Monthly interviews up to 6 months Hansen 2017, PMID 28820518

2026-08-30 epidemiology and prevention additions

Statistic Estimate Population/method Source
Global HBV-attributable fraction of HCC 52%; ~345,434 cases 857 publications; 81 countries/territories; applied to 2022 burden Meta-analysis (Cao 2026, PMID 42135055)
Global HCV-attributable fraction of HCC 21%; ~134,418 cases Same analysis Meta-analysis (Cao 2026, PMID 42135055)
HBV-attributable fraction, Eastern Asia 70% Regional pooled prevalence among HCC Meta-analysis (Cao 2026, PMID 42135055)
HCV-attributable fraction, Northern Africa 77% Regional pooled prevalence among HCC Meta-analysis (Cao 2026, PMID 42135055)
Anti-HDV prevalence among HBsAg-positive people 4.5% (95% CI 3.6%–5.7%) 282 studies; 120,293 HBsAg-positive people Meta-analysis (Stockdale 2020, PMID 32335166)
HDV population-attributable fraction for HCC among HBsAg-positive people 20% (95% CI 8%–33%) Preliminary random-effects estimate Meta-analysis (Stockdale 2020, PMID 32335166)
Aflatoxin-attributable HCC fraction 17% (14%–19%) overall; 21% HBV-positive; 8.8% HBV-negative 17 studies; 1,680 cases and 3,052 controls Meta-analysis (Liu 2012, PMID 22405700)
Aflatoxin plus HBV interaction OR 73.0 (95% CI 36.0–148.3) Sensitivity analysis excluding one heterogeneous study Meta-analysis (Liu 2012, PMID 22405700)
Coffee, per additional two cups/day RR 0.65 (95% CI 0.59–0.72) 18 cohorts plus 8 case-control studies Dose-response meta-analysis; very-low GRADE certainty (Kennedy 2017, PMID 28490552)
Diabetes in chronic HBV HR 1.26 (95% CI 1.20–1.32), fixed effect; 1.36 (1.23–1.49), random effects 40 observational studies; DM subset meta-analysis Campbell 2021, PMID 33305479
Smoking in MASLD RR 1.30 (95% CI 1.08–1.57) 31 studies; 1.02 million people across modifiable-factor analysis Meta-analysis (Xiao 2024, PMID 39047929)

2026-08-30 surveillance, diagnosis, and biomarkers

Statistic Estimate Population/method Source
Six- versus twelve-month surveillance, early-stage detection RR 1.17 (95% CI 1.08–1.26) 13-study interval/modality meta-analysis Yang 2023, PMID 36921104
Six- versus twelve-month surveillance, five-year survival RR 1.39 (95% CI 1.07–1.82) Same meta-analysis Yang 2023, PMID 36921104
Annual HBV screening, corrected mortality HR 0.74 (95% CI 0.60–0.91) prevalent round; 0.52 (0.40–0.68) incident rounds 14,426 HBsAg-positive adults; prospective community cohort Zeng 2023, PMID 37667043
Severe visualization limitation in NAFLD cirrhosis Ultrasound 35% vs abbreviated MRI 19% 54-person prospective paired study Huang 2022, PMID 35229334
Obesity and severe ultrasound limitation OR 5.1 (95% CI 1.1–23.1) Same study Huang 2022, PMID 35229334
LR-5 sensitivity, MRI versus CT 61% (95% CI 43%–76%) vs 48% (31%–65%) 7 comparative studies; 1,145 observations Meta-analysis (Kim 2022, PMID 35849177)
LR-5 specificity, MRI versus CT 93% (95% CI 88%–96%) vs 96% (92%–98%) Same analysis Meta-analysis (Kim 2022, PMID 35849177)
CEUS LR-5 sensitivity/specificity 69% (95% CI 64%–73%) / 92% (83%–96%) 11 studies; 5,535 observations Meta-analysis (Shin 2020, PMID 32722894)
Longitudinal GALAD sensitivity at 90% specificity 66.7%; AFP 40.5% 397 cirrhosis patients; 42 incident HCC Prospective-specimen blinded evaluation (Singal 2022, PMID 34618932)
HES V2.0 incremental TPR over GALAD +7.2% overall at 10% false-positive rate 2,331 patients; 125 HCC Phase-3 biomarker study (El-Serag 2025, PMID 38899967)
External HES V2.0/GALAD AUROC 0.79 for both HEDS cohort; 1,485 patients, 119 HCC PRoBE validation (El-Serag 2026, PMID 41043723)
cfDNA diagnostic meta-analysis Sensitivity 0.54; specificity 0.90; AUC 0.82 38 studies; HCC versus healthy-control dominated Zhang 2021, PMID 33470842

2026-08-30 treatment, transplant, and safety additions

Statistic Estimate Population/method Source
Downstaging success 55.16% (95% CI 45.49%–64.46%) 25 studies; 3,997 patients Meta-analysis (Tan 2023, PMID 35181565)
Transplant after downstaging, intention to treat 31.52% (95% CI 24.03%–40.11%) Same analysis Tan 2023, PMID 35181565
Recurrence after transplant among downstaged recipients 16.01% (95% CI 11.80%–21.37%) Same analysis Tan 2023, PMID 35181565
Adjuvant TACE, broad stage I/II RFS HR 0.88 (95% CI 0.62–1.24), p=0.468 Phase 3 RCT; 332 patients Ma 2025, PMID 39808820
Adjuvant TACE, selected HBV/high-risk RFS HR 0.68 Phase 3 RCT; 280 patients Wang 2018, PMID 29420221
SBRT versus RFA for recurrent small HCC Local-PFS HR 0.45 (95% CI 0.24–0.87); two-year local PFS 92.7% vs 75.8% Randomized trial; 166 patients Xi 2025, PMID 39693584
HIMALAYA five-year OS 19.6% STRIDE vs 9.4% sorafenib; HR 0.76 (95% CI 0.65–0.89) Phase 3 five-year update Rimassa 2025, PMID 40222621
KEYNOTE-240 OS 13.9 vs 10.6 months; HR 0.781 (95% CI 0.611–0.998); p=0.0238 Phase 3; prespecified alpha 0.0174 Finn 2020, PMID 31790344
Variceal bleeding on atezolizumab-bevacizumab 7.0%; six-month 6.3%; twelve-month 7.4% 640 pre-treatment-endoscoped patients Retrospective cohort (Park 2025, PMID 39871662)
Acute renal failure after TACE 8.6% (12/140) Prospective cohort Huo 2004, PMID 15189271
ICI-exposed transplant rejection 26.4% (24/91) Individual-patient meta-analysis Rezaee-Zavareh 2025, PMID 38996924
Rejection with pre-LT ICI washout <30 days OR 21.3 (95% CI 5.93–103) versus >50 days International retrospective cohort; 119 recipients Moeckli 2025, PMID 40042053

2026-08-30 disparities and patient priorities

Statistic Estimate Population/method Source
Treatment delay >3 months 14.3% (1,205/8,450) SEER-Medicare treated within 12 months Wagle 2023, PMID 35933076
Black race and treatment delay OR 1.96 (95% CI 1.21–3.15) Adjusted model Wagle 2023, PMID 35933076
High-poverty neighborhood and treatment delay OR 1.55 (95% CI 1.25–1.92) Adjusted model Wagle 2023, PMID 35933076
Treatment delay and mortality HR 1.15 (95% CI 1.05–1.25) Five-month landmark analysis Wagle 2023, PMID 35933076
Early BCLC 0/A presentation 42.0% Four-system US cohort; 2,263 patients Rich 2024, PMID 39666898
Black versus White early presentation OR 0.63 (95% CI 0.45–0.89) Fully adjusted model Rich 2024, PMID 39666898
Preference importance: surveillance sensitivity 51.3% (95% CI 49.0%–53.4%) Discrete-choice study; 179 completed surveys Woolen 2022, PMID 33618022
Simulated abbreviated-MRI preference 29.0%; ultrasound alone 3.4% Same study Woolen 2022, PMID 33618022
Daily function versus survival tradeoff Ten months maintained function valued at least as much as ten months OS 200 US patients with unresectable HCC Discrete-choice study (Li 2023, PMID 36900262)

Known conflicts and caveats

  • Global burden sources often count all liver cancer, not HCC histology alone.
  • Annual incidence ranges mix etiologies, fibrosis definitions, and surveillance intensity.
  • Post-HCV-cure estimates depend strongly on cirrhosis ascertainment and prior HCC.
  • Surveillance survival estimates are observational and remain vulnerable to lead-time, length-time, and healthy-user bias.
  • Diagnostic meta-analyses often use lesion-level rather than patient-level denominators.
  • Transplant outcomes are highly selected and should not be applied to all HCC.
  • Systemic RCTs predominantly enrolled Child-Pugh A, performance-preserved patients.
  • Cross-trial median survival comparisons are not treatment rankings.
  • Race/ethnicity disparity estimates reflect social and structural pathways, not biological racial effects.