Construction and validation of a prognostic marker and risk model for HCC ultrasound therapy combined with WGCNA identification

Background: Hepatocellular carcinoma (HCC) is a malignant tumor with a highly aggressive and metastatic nature. Ultrasound remains a routine monitoring tool for screening, treatment and post-treatment recheck of HCC. Therefore, it is of great significance to explore the role of ultrasound therapy and related genes in prognosis prediction and clinical diagnosis and treatment of HCC. Methods: Gene co-expression networks were developed utilizing the R package WGCNA as per the expression profiles and clinical features of TCGA HCC samples, key modules were identified by the correlation coefficients between clinical features and modules, and hub genes of modules were determined as per the GS and MM values. Ultrasound treatment differential expression genes were identified using R package limma, and univariate Cox analysis was conducted on the intersection genes of ultrasound differential expression genes and hub genes of key HCC modules to screen the signatures linked with HCC prognosis and construct a risk model. The median risk score was used as the threshold point to classify tumor samples into high- and low-risk groups, and the R package IOBR was used to assess the proportion of immune cells in high- and low-risk groups, R package maftools to assess the genomic mutation differences in high- and low-risk groups, R package GSVA’s ssgsea algorithm to assess the HALLMARK pathway enrichment analysis, and R package pRRophetic to analyze drug sensitivity in patients with HCC. Results: WGCNA analysis based on the expression profiles and clinical data of the TCGA LIHC cohort identified three key modules with two major clinical features associated with HCC. The intersection of ultrasound-related differential genes and module hub genes was selected for univariate Cox analysis to identify prognostic factors significantly associated with HCC, and a risk score model consisting of six signatures was finally developed to analyze the prognosis of individuals with HCC. The risk model showed strength in the training set, overall set, and external validation set. The percentage of immune cell infiltration, genomic mutations, pathway enrichment scores, and chemotherapy drug resistance were significantly different between high- and low-risk groups according to the risk scores. Expression of model genes correlated with tumor immune microenvironment and clinical tumor characteristics while generally differentially expressed in pan-cancer tumor and healthy samples. In the immunotherapy dataset, patients in the high-risk group had a worse prognosis with immunotherapy, indicating that subjects in the low-risk group are more responsive to immunotherapy. Conclusion: The 6-gene signature constructed by ultrasound treatment of HCC combined with WGCNA analysis can be used for prognosis prediction of HCC patients and may become a marker for immune response.