University of Pittsburgh scientists have developed a novel genome prediction model to select patients with hepatocellular carcinoma (HCC) who are most suitable for liver transplant.
Description
HCC is one of the most lethal human cancers with a survival rate at five years of less than 20%. While transplantation is an effective treatment for many people, for others HCC can recur in the newly transplanted liver leading to subsequent death. Identifying patients at most risk of recurrence could improve decision making for clinicians when planning transplantation surgery and aftercare. From historical data, this genome prediction model used transcriptome and whole exome sequencing analyses with machine learning processes to predict HCC recurrence in patients. Optimizing and validating this model has the potential to offer a new method of selecting HCC patients for liver transplant.
Applications
• Hepatocellular carcinoma (HCC)
• Post-transplant survival
Advantages
Liver transplantation can effectively treat HCC; however, HCC recurrence can occur in up to 20% of cases. There is a global shortage of livers suitable for transplant and with a growing demand for liver transplant it is vital that suitable organs are transplanted to those patients most likely to gain long-term benefit. Several selection criteria based on HCC lesion size and number of tumor nodules have been developed, but no reliable way to predict HCC recurrence currently exists.
This novel prediction model based on transcriptome and exome analyses has an accuracy of over 80% in correctly predicting recurrence of HCC post liver transplant surgery. It has the potential to offer clinicians a reliable and robust tool to identify patients who are most likely to benefit from liver transplant and could be used to assess patients who may not meet selection criteria using other approaches.
Invention Readiness
Analyses of transcriptome expression at RNA-level and pathway mutation at DNA-level from a training cohort and based on historical data were used to predict a probability of recurrence. Using various machine learning models, a prediction model was developed, and a testing cohort based on a different group of patients was used to validate the model.
Further development of the prediction model in combination with a traditional selection tool, the Milan criteria, improved prediction accuracy to over 80%. Further optimization and validation of this prediction tool could aid clinical decision making in selecting patients with HCC most likely to benefit from liver transplant surgery.
IP Status
https://patents.google.com/patent/WO2023034955A1
