Analysis for drug metabolism-related prognostic subtypes and gene signature in liver cancer

Abstract

Liver cancer is highly heterogeneous and has a poor prognosis. We aimed to identify a drug metabolism-related prognostic subtype and a gene signature as references for prognosis and therapy options for patients with liver cancer. Patient information was collected from online databases. Drug metabolism-related genes were obtained from previous studies and were used to screen differentially expressed prognostic genes. The patients were divided into different clusters and differences in clinical features, immunity, pathways and therapy responses between the clusters were analyzed. LASSO analysis was performed to identify the optimal prognostic genes and establish a risk score model. Finally, the risk score distribution in different subtypes was investigated. A total of 54 prognostic genes were identified to categorize the patients into cluster 1 and cluster 2. Cluster 1 showed worse survival than cluster 2, and cluster 1 also showed high levels of malignancy. Furthermore, cluster 1 exhibited a higher TIDE (tumor immune dysfunction and exclusion) score and lower IC50 response to paclitaxel, gemcitabine and camptothecin, indicating that cluster 1 individuals may derive more benefit from immunotherapy but less benefit from chemotherapy. The risk score, based on the six optimal prognostic genes, demonstrated an adequate prognostic capability. The high-risk group showed worse survival; meanwhile, cluster 1 contained the majority of high-risk samples. Our results should be useful for prognosis and specific therapy for patients with liver cancer. Patients with the features of cluster 1 and a high risk score will tend to exhibit worse survival. Furthermore, immunotherapy may be more suitable for cluster 1-type patients while chemotherapy may be more suitable for cluster 2 patients.