A Worldwide Usable Scoring System for Patients With Hepatocellular Cancer Waiting for Liver Transplantation

For two decades, the Milan Criteria (MC) have represented the cornerstone in the selection of patients with hepatocellular cancer (HCC) as candidates for liver transplantation (LT). Since then, several Western and Eastern centres have tried to overcome MC stringency with the aim to expand the number of potentially transplantable patients without increasing the risk of post-LT tumour recurrence. Recently, variables correlated with HCC biology have been introduced to capture its aggressiveness and suitability for LT. Among the most commonly proposed, we can cite the alpha-fetoprotein (AFP) and the radiological response after neo-adjuvant loco-regional treatments (LRT). The effort to combine HCC morphology and biology is now emerging as the more promising approach for further refining the selection process of HCC candidates. The recently proposed prognostic indexes Metroticket 2.0, Time-Radiological-response-Alpha-fetoprotein-INflammation (TRAIN), Model Of Recurrence After Liver transplant (MORAL) and French AFP-model represent the most promising results of such planned endeavours.

With the intent to be clinically useful, a prognostic system must achieve not only the most common performance metrics, such as discrimination and calibration, but also an adequate coverage of the population in which it would be applied. It can be argued that a prognostic system based only on criteria available at the moment of LT could not satisfy such a pre-requisite when the entire population of HCC patients waiting in the list for LT is taken into account, considering an intention-to-treat (ITT) point of view. In this sense, only the TRAIN system developed an ITT analysis, whereas the remaining were all based on pre-transplant information. The second quality that a clinically useful prognostic system should have is its immediacy. Thus, it should be based on commonly available information, not requiring additional analyses or exams if not justified by a substantial improvement in accuracy. Last but not least, a prognostic system should also provide for an individual prediction in addition to risk-stratification, with the intent to offer personalized prognostication for each different patient. To date, only the Metroticket 2.0 fulfils these two last requisites. A final important quality of the Metroticket 2.0 is that it applied a competing-risk analysis able to clean up the HCC-related cause of death from other non-tumour-related causes.

On this background, we developed the present study with multiple aims: 1) to refine available models which would be able to cover the fate of HCC candidates from an ITT point of view; 2) to develop such an approach on cohorts coming from both Eastern and Western countries; 3) to maintain simplicity of use; 4) to provide individual prognostication taking into account different causes of death, through a competing-risk model; 5) to provide an external validation on cohorts coming from both Eastern and Western countries. All these aims converge at providing a comprehensive and useful assessment suitable for both candidates selection and allocation priority.