Skip to content

Hepatocellular carcinoma — liver transplantation

TL;DR — Liver transplantation is uniquely curative for selected HCC because it removes both tumor and the cirrhotic field, but it allocates a scarce organ and therefore requires population-level utility as well as individual benefit. The Milan criteria—one tumor ≤5 cm or up to three tumors each ≤3 cm, without macrovascular invasion or metastasis—produced 85% four-year survival in the landmark 1996 cohort and remain the reference benchmark (Mazzaferro 1996, PMID 8594428). Expanded systems such as UCSF and Metroticket 2.0 show that acceptable biology extends beyond Milan when size/number are integrated with AFP (Yao 2007, PMID 17868066; Mazzaferro 2018, PMID 28989060). Successful downstaging is both treatment and a biological selection period; randomized and multicentre evidence supports transplant for selected responders (Mazzaferro 2020, PMID 32615109; Mehta 2021, PMID 34331914). Post-transplant recurrence risk is not zero and can be stratified with explant variables using RETREAT (Mehta 2018, PMID 29068145).

Why transplant is different

Resection and ablation remove visible tumor but leave cirrhosis and future primary-HCC risk. Transplant removes:

  • the known tumor;
  • occult intrahepatic lesions;
  • the fibrotic/cirrhotic field;
  • portal hypertension and many complications of liver failure.

It replaces these risks with operative mortality, lifelong immunosuppression, infection, renal/metabolic toxicity, graft failure, and recurrent HCC under immunosuppression.

Milan criteria

The original prospective cohort included cirrhotic patients with a single HCC ≤5 cm or up to three tumors each ≤3 cm. Four-year overall survival was 85% and recurrence-free survival 92% among those meeting criteria on pathology (Mazzaferro 1996, PMID 8594428).

Milan component Boundary
Number One tumor, or up to three
Size Single ≤5 cm; if 2–3 tumors, each ≤3 cm
Macrovascular invasion Absent
Extrahepatic spread Absent

Milan's enduring value is not that biology changes abruptly at 5.0 cm. It created an auditable rule that restored transplant outcomes after poor unselected experience and enabled fairer allocation.

Expanded morphology

UCSF criteria expand the allowable burden to a single tumor ≤6.5 cm or up to three tumors with largest ≤4.5 cm and total diameter ≤8 cm (Yao 2002, PMID 12200775). Imaging-based validation showed outcomes comparable to Milan in selected patients, supporting cautious expansion beyond the original boundary (Yao 2007, PMID 17868066).

Framework Variables Contribution Limitation
Milan Number and size Simple benchmark and allocation rule Binary morphology only
UCSF Larger number/size envelope Demonstrates safe expansion in selected cohorts Still morphologic
Up-to-seven Sum of largest diameter and tumor number Continuous burden concept Biology incompletely represented
AFP model AFP plus morphology Adds biomarker of tumor behavior AFP-negative tumors
Metroticket 2.0 AFP, tumor number, largest size Predicts HCC-specific death as continuous risk Calibration and policy thresholds

Metroticket 2.0: from boundary to probability

Metroticket 2.0 used competing-risk analysis in international transplant cohorts and combined AFP with tumor size and number. The model's c-statistic was approximately 0.78 and it outperformed Milan and UCSF for predicting HCC-specific survival (Mazzaferro 2018, PMID 28989060).

The conceptual advance is to estimate a survival probability rather than assert that every patient inside a boundary has good biology and every patient outside has bad biology. Allocation still requires a threshold, because a continuous model does not decide how much recurrence risk society should accept for one graft.

Bridging versus downstaging

Strategy Starting state Goal
Bridging Already within accepted transplant criteria Prevent progression/dropout during waiting
Downstaging Initially beyond accepted tumor burden Reduce viable disease into an acceptable state and observe biology
Salvage transplant Prior resection/ablation Rescue recurrence or liver failure while within criteria

Locoregional choices include TACE, TARE, ablation, and radiation according to tumor anatomy, liver function, and centre experience. Radiographic response must be interpreted with viable enhancement rather than diameter alone.

Downstaging evidence

The multicentre MERITS-LT consortium prospectively evaluated uniform UNOS downstaging criteria across broad US regions. It measured successful downstaging, dropout, transplant, and post-transplant outcomes, providing transportability beyond a single expert centre (Mehta 2021, PMID 34331914).

The XXL randomized phase 2b/3 trial enrolled patients with HCC beyond conventional criteria who achieved sustained response after downstaging and randomized them to transplantation versus non-transplant therapies. Transplantation improved tumor-event-free and overall survival in selected responders, supporting downstaging as more than radiographic cosmetics (Mazzaferro 2020, PMID 32615109).

Response carries biological information:

  • rapid progression despite locoregional therapy suggests aggressive disease;
  • durable AFP decline and radiographic response suggest more favorable biology;
  • waiting time exposes occult metastatic potential;
  • treatment intolerance exposes limited liver reserve.

Wait-list management

Risk during waiting Monitoring/response
Tumor growth beyond criteria Serial cross-sectional imaging; bridging/downstaging
New vascular invasion/metastasis Delist or reassess candidacy
Liver decompensation Update medical urgency and procedural feasibility
AFP rise Re-image and reassess biology
Locoregional toxicity Protect liver reserve; avoid futile repetition
Competing illness Reassess operative benefit and candidacy

Allocation priority differs by jurisdiction and changes over time. A transplant criterion, an exception score, and a listing policy are not interchangeable.

Deceased versus living donation

Living donation can shorten waiting time and reduce dropout, but transfers surgical risk to a healthy donor. The ethical comparison includes:

  • donor morbidity and rare mortality;
  • recipient recurrence probability;
  • alternative deceased-donor access;
  • coercion and voluntariness;
  • centre experience;
  • whether expanded recipient criteria expose a donor to risk for limited benefit.

Scarcity is shifted, not eliminated. Living donation should not bypass biological selection or independent donor protection.

Explant pathology

The explanted liver provides the most complete tumor assessment:

  • number and size of viable tumors;
  • necrosis after downstaging;
  • microvascular invasion;
  • differentiation;
  • satellite lesions;
  • mixed histology;
  • background liver disease.

Discrepancy between pre-transplant imaging and explant pathology is expected. Imaging governs allocation; pathology refines recurrence risk after the graft has been used.

RETREAT and recurrence risk

RETREAT combines AFP at transplant, microvascular invasion, and the sum of largest viable tumor diameter plus number on explant. In 3,276 UNOS recipients within Milan by imaging, three-year recurrence rose from 1.6% at score 0 to 29% at score ≥5; three-year survival fell from 91% to 58% (Mehta 2018, PMID 29068145).

RETREAT input Timing Meaning
AFP at transplant Preoperative Marker burden/biology
Microvascular invasion Explant Metastatic competence
Largest viable diameter + viable number Explant Residual burden after therapy

RETREAT can rationalize surveillance intensity and trial eligibility, but prospective evidence that score-guided surveillance improves outcomes is limited.

Post-transplant recurrence

Recurrence can involve graft, lung, bone, lymph nodes, adrenal glands, or other sites. Management depends on distribution, graft function, immunosuppression, resectability, and prior therapy.

Immune checkpoint inhibitors create a distinctive risk because activating immunity can precipitate allograft rejection. Evidence in transplant recipients is largely observational, and treatment decisions require transplant-oncology collaboration.

Immunosuppression strategy may influence cancer risk, but regimen changes must balance rejection, renal toxicity, and uncertain antitumor benefit.

Equity and utility

Transplant decisions operate at three levels:

  1. Benefit: how much survival and quality of life this recipient may gain.
  2. Utility: whether expected graft-years and recurrence risk are acceptable compared with alternative recipients.
  3. Justice: whether access is fair across geography, socioeconomic status, etiology, race/ethnicity, sex, and living-donor availability.

Morphology-only criteria are transparent but biologically crude. Continuous biology models can improve prediction but become harder to audit and may incorporate access-dependent variables such as response to expensive locoregional therapy.

Practical selection synthesis

Question Evidence source
Is HCC diagnosis secure? LI-RADS/pathology and multidisciplinary review
Is there macrovascular or extrahepatic disease? Complete staging imaging
What is viable burden? Size, number, enhancement after therapy
What does biology suggest? AFP trajectory, growth, response, waiting stability
Is transplant medically feasible? Liver, cardiovascular, frailty, infection evaluation
Is the candidate within policy? Jurisdiction-specific allocation rules
Is donor risk independently acceptable? Separate living-donor evaluation

Selection beyond diameter and number

Milan remains a reproducible minimum-benefit framework, not a biological law. AFP-adjusted size criteria and Metroticket-type models treat tumor burden and biology as continuous variables; external cohorts support acceptable survival beyond Milan when AFP and response remain favorable (Meischl 2021, PMID 32741316; Barreto 2022, PMID 35681757). The ethical constraint is opportunity cost: any expansion must compare post-transplant benefit with the benefit forgone by another candidate.

Selection step Quantified evidence Decision implication
Downstaging overall 25-study meta-analysis: 55.16% (95% CI 45.49–64.46) successfully downstaged; 31.52% transplanted by intention to treat; post-LT recurrence 16.01% Report denominators from first treatment, not only transplanted responders (Tan 2023, PMID 35181565)
UNOS-DS versus all-comers Prospective cohort: two-year downstaging 82% versus 66% Broader entry is feasible but less efficient and requires explicit upper bounds (Xu 2025, PMID 39808828)
Response during bridging Progression despite bridging predicted mortality, HR 1.80 Dynamic behavior supplies information beyond baseline morphology (Renner 2021, PMID 33839747)
Imaging under-staging National data show meaningful center-level variation and worse outcomes when explant exceeds imaging criteria Imaging quality and interval are allocation-quality variables (Mahmud 2020, PMID 32363720)
Living versus deceased donor 35 studies/7,822 patients: similar five-, six-, and ten-year OS and no significant recurrence difference Living donation can reduce wait-list loss but transfers risk to a healthy donor (Elkomos 2023, PMID 36564609)
Salvage versus primary transplant Meta-analyses show feasible salvage outcomes but are dominated by selected patients who remain transplantable Intention-to-treat failure after non-transplantable recurrence must be counted (Yadav 2018, PMID 30072683)

Recurrence prediction after transplant

RETREAT combines AFP at transplant, microvascular invasion, and explant viable-tumor burden. It was developed in 721 and validated in 341 recipients, then tested in national and prospective cohorts (Mehta 2017, PMID 27838698; Li 2025, PMID 40067686). This is a post-transplant score: it cannot guide initial organ allocation because two components are only known from the explant. Its most defensible use is matching surveillance intensity and trial eligibility to recurrence risk.

The choice between upfront transplant and resection also depends on recurrence geometry. In an international series of transplantable HCC resected first, 38.6% within Milan recurred and 28.3% of recurrences were already non-transplantable; cirrhosis, tumor >3 cm, and multiplicity predicted this failure mode (Zhang 2022, PMID 34797558). A later solitary-HCC model identified AFP >10 ng/mL as one risk component, but retrospective MRI-derived scores need external prospective validation (Zhang 2024, PMID 38298271).

Checkpoint inhibitors before transplant: an active safety boundary

An individual-patient meta-analysis of 91 pre-transplant ICI-exposed recipients found rejection in 26.4%, recurrence in 9.9%, and death in 9.9%; each additional week of washout reduced the rejection hazard (aHR 0.92, 95% CI 0.86–0.99) (Rezaee-Zavareh 2025, PMID 38996924). In an international cohort of 119 recipients, rejection occurred in 20.2%, usually early; washout <30 days had OR 21.3 (95% CI 5.93–103) and 30–50 days OR 9.48 (2.47–46.8) versus >50 days (Moeckli 2025, PMID 40042053). A broader 386-patient global cohort found rejection rates of 17.5% before-LT and 22.1% after-LT exposure, with more graft loss/dysfunction after post-LT treatment (Ma 2025, PMID 40898181).

These are selected observational data, not proof of a universally safe washout. Drug half-life, receptor occupancy, agent, number of doses, donor type, immunosuppression, and urgency all matter. Pre-transplant ICI should therefore remain a transplant-center decision with explicit rejection counseling and a documented washout strategy.

Controversies

  • Milan versus expanded criteria. Expansion can preserve good individual outcomes, but allocation fairness requires net transplant benefit and transparent competing-risk modeling (Mazzaferro 1996, PMID 8594428; Barreto 2022, PMID 35681757).
  • Downstaging as treatment versus selection test. Tumor shrinkage can be therapeutic, while sustained response also selects indolent biology; observational outcomes cannot separate these effects (Tan 2023, PMID 35181565; Xu 2025, PMID 39808828).
  • Living donation. Recipient outcomes are broadly comparable, but shorter waiting time may explain some benefit and donor morbidity cannot be erased by recipient statistics (Liang 2012, PMID 22685095; Zhu 2019, PMID 30503300).
  • Bridging. Systematic review found no randomized trials and substantial heterogeneity; response is informative, but the optimal modality for wait-list survival is unresolved (Kulik 2018, PMID 28859222).
  • ICI-enabled conversion. Tumor response may create transplant opportunity, but rejection can destroy a curative graft; current evidence supports neither a categorical ban nor routine use (Cesario 2023, PMID 37511937; Rezaee-Zavareh 2025, PMID 38996924).

Open questions

  • What continuous recurrence-risk threshold should replace or supplement Milan in deceased-donor allocation (Mazzaferro 2018, PMID 28989060)?
  • How far beyond current UNOS downstaging limits can response safely select candidates (Mehta 2021, PMID 34331914)?
  • Does transplant after immunotherapy require a minimum washout period, and which immune markers predict rejection?
  • Can RETREAT-guided surveillance improve salvageability or survival rather than only prediction (Mehta 2018, PMID 29068145)?
  • How should living-donor risk alter acceptable recipient HCC recurrence probability?

References

  1. Mazzaferro V, et al. Liver transplantation for the treatment of small hepatocellular carcinomas in patients with cirrhosis. N Engl J Med. 1996;334:693-699. PMID 8594428
  2. Yao FY, et al. Liver transplantation for hepatocellular carcinoma: validation of the UCSF-expanded criteria based on preoperative imaging. Am J Transplant. 2007;7:2587-2596. PMID 17868066
  3. Mazzaferro V, et al. Metroticket 2.0 model for analysis of competing risks of death after liver transplantation for hepatocellular carcinoma. Gastroenterology. 2018;154:128-139. PMID 28989060
  4. Mazzaferro V, et al. Liver transplantation in hepatocellular carcinoma after tumour downstaging (XXL): a randomised, controlled, phase 2b/3 trial. Lancet Oncol. 2020;21:947-956. PMID 32615109
  5. Mehta N, et al. Downstaging outcomes for hepatocellular carcinoma: results from the MERITS-LT consortium. Gastroenterology. 2021;161:1502-1512. PMID 34331914
  6. Mehta N, et al. Validation of the prognostic power of the RETREAT score for hepatocellular carcinoma recurrence using the UNOS database. Am J Transplant. 2018;18:1206-1213. PMID 29068145
  7. Meischl T, et al. AFP-adjusted-to-HCC-size criteria and survival after liver transplantation. United European Gastroenterol J. 2021;9:209-219. PMID 32741316
  8. Barreto SG, et al. Expansion from Milan to UCSF criteria and justification for Metroticket 2.0. Cancers. 2022;14. PMID 35681757
  9. Tan DJH, et al. UNOS down-staging criteria for liver transplantation: systematic review of 25 studies. Clin Gastroenterol Hepatol. 2023;21:1475-1484. PMID 35181565
  10. Xu E, et al. Downstaging before liver transplantation: national multicenter prospective cohort. Hepatology. 2025;82:612-625. PMID 39808828
  11. Li PJ, et al. Prospective multicenter validation of RETREAT after liver transplantation. Hepatology. 2025;82:1450-1460. PMID 40067686
  12. Mehta N, et al. Development and validation of the RETREAT score. JAMA Oncol. 2017;3:493-500. PMID 27838698
  13. Liang W, et al. Living- versus deceased-donor liver transplantation for HCC: a meta-analysis. Liver Transpl. 2012;18:1226-1236. PMID 22685095
  14. Elkomos BE, et al. Living- versus deceased-donor liver transplantation for HCC: a systematic review. Hepatol Int. 2023;17:18-37. PMID 36564609
  15. Yadav DK, et al. Salvage versus primary liver transplantation for HCC. Ann Transplant. 2018;23:524-545. PMID 30072683
  16. Zhu B, et al. Living or deceased donors in liver transplantation for HCC: a meta-analysis. HPB (Oxford). 2019;21:133-147. PMID 30503300
  17. Rezaee-Zavareh MS, et al. Pre-transplant immune-checkpoint inhibitor use and post-transplant outcomes: IPD meta-analysis. J Hepatol. 2025;82:107-119. PMID 38996924
  18. Moeckli B, et al. Determining safe checkpoint-inhibitor washout before liver transplantation. Hepatology. 2025;82:1122-1137. PMID 40042053
  19. Ma D, et al. Immune-checkpoint inhibitor use in liver transplantation for HCC: a global cohort. BMC Med. 2025;23:515. PMID 40898181
  20. Kulik L, et al. Therapies for patients with HCC awaiting liver transplantation: a systematic review. Hepatology. 2018;67:381-400. PMID 28859222
  21. Renner P, et al. HCC progression during bridging before liver transplantation. BJS Open. 2021;5. PMID 33839747
  22. Mahmud N, et al. Risk factors and center-level variation in HCC under-staging for transplantation. Liver Transpl. 2020;26:977-988. PMID 32363720
  23. Cesario S, et al. Liver transplantation for HCC from stage migration to immunotherapy. Life (Basel). 2023;13. PMID 37511937
  24. Zhang XF, et al. Non-transplantable recurrence after resection for transplantable HCC. J Gastrointest Surg. 2022;26:1021-1029. PMID 34797558
  25. Zhang C, et al. Prediction of non-transplantable recurrence after resection for solitary HCC. J Hepatocell Carcinoma. 2024;11:229-240. PMID 38298271
  26. Yao FY, et al. Liver transplantation for hepatocellular carcinoma: comparison of the proposed UCSF criteria with the Milan criteria and the Pittsburgh modified TNM criteria. Liver Transpl. 2002;8:765-774. PMID 12200775