Retrospective Clinical Validation of HepatoPredict

Liver cancer remains a global health challenge with growing incidence worldwide. Hepatocellular carcinoma (HCC) is the most common liver cancer, accounting for ~90% of the cases and for 75% of the deaths. HCC is the third cancer with the highest mortality rate worldwide. Surgery is the treatment of choice for HCC, leading to the best outcomes of any available treatment in well selected candidates, with 5-year survival rates of 60-80%. Liver resection (LR) and liver transplantation (LT) represent the first treatment option for patients with early to intermediate HCC.

Eligibility for LR comprises assessment of multiple factors including liver function, portal hypertension, extent of hepatectomy and surgical invasiveness. Moreover, LR is currently the treatment of choice for non-cirrhotic patients, allowing for larger and more complex resections. Nonetheless, a significant percentage of the patients undergoing resection recur within 5 years and although with 1 year morbidity and mortality lower than LT, recurrence free survival is higher in patients submitted to LR. When possible, LT is the best curative option for patients with HCC. However, due to the small number of organs available for LT concerning the actual need for patients on waiting lists, the selection criteria for candidates for transplantation are very strict and rigorous. Current selection criteria are mainly based on clinical variables and limit transplantation to HCC patients with oncologic disease within specific parameters (such as number and diameter of tumors). Although these parameters appear to be related to disease severity and the biology of the tumor, they do not unequivocally predict the favorable prognosis of a patient with HCC who is a candidate for transplantation. Therefore, not only do these selection criteria fail in more than 30% of transplanted patients who eventually relapse, but they also exclude from transplantation many patients outside criteria with favorable prognosis and who could potentially be cured with LT. Thus, it is of extreme importance to develop predictive tools that can provide solid support in the selection of patients with HCC for transplantation. In addition, different scores such as RETREAT, Moral and R3-AFP have been proposed to prognosticate HCC recurrence post-liver transplantation as discrepancies between pre-LT tumor assessment and explant have been frequently reported. These post-surgical scores are helpful to standardize post-LT patient surveillance schemes and select candidates for adjuvant therapies. However, just as the pre-LT criteria, post-LT criteria also have several pitfalls and more accurate models to predict chance of recurrence are needed.

HepatoPredict is an innovative medical decision-making prognostic tool aimed at helping physicians and surgeons to select patients with HCC for effective hepatic surgery. This kit does not rely solely on clinical variables but also on tumor expression of a selected set of genes. This in combination with a proprietary algorithm, results in a more accurate prediction of patient outcome. While the current criteria used for selecting patients for LT, such as Milan and San Francisco criteria, only offer up to 70% success rate, HepatoPredict achieved up to 94% success.

This study aims to retrospectively validate HepatoPredict results in an independent cohort of patients with early to intermediate HCC in the context of:

1. Liver transplantation for HCC (1.1) Pre-surgery: selection of patients and (1.2) Post-surgery: assess risk of recurrence
2. Liver resection for HCC (2.1) Post-surgery: assess risk of recurrence. Archived tumour tissue from HCC patients who underwent surgery, either LT or LR, will be analyzed retrospectively using a gene expression signature. Specifically, it will use one slide H\&E-stained with the tumoral area clearly labeled by a pathologist (at least 30% of tumor cells in the marked area), and two unstained slides with 5 micrometers thickness of the same area.

RNA will be extracted from the two slides with 5 micrometers thickness and gene expression signature will be evaluated by RT-qPCR. Prognosis prediction, based on proprietary algorithm, will then be compared to the real outcome to calculate accuracy measures.