Establishment of metastasis-related prognostic models for colon cancer and bioinformatics analysis of the benefits of aspirin chemoprophylaxis

Abstract

Background: Invasion and metastasis are the main factors affecting the prognosis of patients with colorectal cancer. Therefore, it is urgent to explore the key metastasis genes and potential regulatory mechanisms that affect the prognosis of colorectal cancer, and provide new prognostic markers or therapeutic targets for colorectal cancer.Methods: Firstly, we obtained differentially expressed metastasis-related genes in cancer and normal tissues. By constructing WGCNA from differentially expressed genes, the modules most associated with colon cancer were identified. Univariate and lasso-cox regression analyses were used to identify key prognostic genes for the most relevant modules and to establish a risk score model. Nomogram model was constructed to predict OS of colon cancer patients by risk score. GO and KEGG enrichment analysis were performed to study the pathways and functions of these module genes. Iregulon plugin in cytoscape was used to explore the transcription factors of module genes. Finally, aspirin interacting genes were obtained through the Comparative Toxicology Genomics Database, and the clinical significance of aspirin regulation on these genes was clarified.Results: 2062 metastasis-related genes were involved in the construction of WGCNA, blue and yellow modules were significantly correlated with colon cancer. Analysis of GO and KEGG pathways of these two modules showed that the function of differential genes is closely related to many important processes in tumor genesis and metastasis. We used the blue module gene to build a risk score model that predicted overall survival (OS). In addition, nomogram combining risk score with age, stage, and M-stage were developed to better predict 1-, 3-, and 5-year survival. Transcription factor analysis showed that E2F family were the key regulator of blue module, while NFκB1 and STAT3 were the key transcription factors of yellow module. Finally, we identified that patients of colon cancer whose NOX4, CXCL8, CXCL5, GDF15, and MMP13 genes were continuously upregulated may not be suitable for aspirin chemoprophylaxis.Conclusions: We developed and validated a predictive model of metastasis genes. Patients with colon cancer whose NOX4, CXCL8, CXCL5, GDF15, and MMP13 genes are upregulated may not be ideal candidates for aspirin chemoprophylaxis.