A Novel Artificial Neural Network Prognostic Model Based on a Cancer-Associated Fibroblast Activation Score System in Hepatocellular Carcinoma

Abstract

Introduction: Hepatocellular carcinoma (HCC) ranks fourth as the most common cause of cancer-related death. It is vital to identify the mechanism of progression and predict the prognosis for patients with HCC. Previous studies have found that cancer-associated fibroblasts (CAFs) promote tumor proliferation and immune exclusion. However, the information about CAF-related genes is still elusive. Methods: The data were obtained from The Cancer Genome Atlas, International Cancer Genome Consortium, and Gene Expression Omnibus databases. On the basis of single-cell transcriptome and ligand–receptor interaction analysis, CAF-related genes were selected. By performing Cox regression and random forest, we filtered 12 CAF-related prognostic genes for the construction of the ANN model based on the CAF activation score (CAS). Then, functional, immune, mutational, and clinical analyses were performed. Results: We constructed a novel ANN prognostic model based on 12 CAF-related prognostic genes. Cancer-related pathways were enriched, and higher activated cell crosstalk was identified in high-CAS samples. High immune activity was observed in high-CAS samples. We detected three differentially mutated genes (NBEA, RYR2, and FRAS1) between high- and low-CAS samples. In clinical analyses, we constructed a nomogram to predict the prognosis of patients with HCC. 5-Fluorouracil had higher sensitivity in high-CAS samples than in low-CAS samples. Moreover, some small-molecule drugs and the immune response were predicted. Conclusion: We constructed a novel ANN model based on CAF-related genes. We revealed information about the ANN model through functional, mutational, immune, and clinical analyses.