An intention-to-treat competing-risk model for candidates with hepatocellular cancer waiting for liver transplantation

Background: Since the introduction of the Milan Criteria (MC), all systems, which describe post-transplant prognosis of patients with hepatocellular cancer (HCC), are exclusively based on characteristics available at surgery, and neglect the intention-to-treat principles. This study, based on a large international HCC patient cohort, aimed to develop comprehensive intention-to-treat models through a competing-risk analysis. We used data available at first referral to predict the risk of delisting and HCC-related death after liver transplantation (LT). Methods: Twelve centres in the United States, Europe and Asia created a Derivation Set (n=2,318) and an external Validation Set (n=773) of HCC patients listed for LT between January 2000-March 2017. The study was registered at http://www.ClinicalTrials.gov (ID:NCT03595345). Findings: In the Derivation Set, the competing-risk analysis identified three independent covariates predicting delisting (Model#1): age (sub-hazard ratio, SHR=1·049; p=0·001), MELD (SHR=1·033; p=0·002) and living donation availability (SHR=0·422; p-value=0·001), all evaluated at first referral. The risk of post-transplant HCC-related death (Model#2) was predicted by the combination of Metroticket2.0 (SHR=1·724; p=0·001) and MELD (SHR=0·970; p=0·045), both evaluated at first referral. In the external validation, Model#1 was compared with MELD, deMELD, MELDeq, and HCC-MELD, and Model#2 with the AFP French Model and the TRAIN score. Both exhibited the highest diagnostic performances (c-statistic=63·3% and 67·7%, respectively). The identified upper limit of post-transplant HCC-related death was 13%, corresponding to the different combinations of alpha-fetoprotein (AFP) and morphological indicators: AFP≤20 ng/mL and up-to-twelve as sum of diameter/number of lesions; AFP=21-200 and up-to-ten; AFP=201-500 and up-to-seven; AFP=501-1,000 and up-to-five. Interpretation: This study presents a scoring system based on a large international cohort of HCC patients awaiting LT. A freely accessible web calculator has been created to estimate the individual risks of delisting and HCC-related death after LT. Furthermore, a risk recalculation after neo-adjuvant treatments can improve patient selection and indications. Funding: None in relation to the here presented work. Declaration of Interest: The authors have no conflicts of interest to declare in relation to the present study. Ethical Approval: The Institutional Ethical and Scientific Review board of the coordinating centre approved the study, registered at http://www.ClinicalTrials.gov (ID: NCT03595345).