Extracellular matrix-based gene signature for predicting prognosis in colon cancer and immune microenvironment.

BACKGROUND

The extracellular matrix (ECM) plays a vital role in progression, expansion, and prognosis of malignancies. In this study, we aimed to explore a novel ECM-based prognostic model for patients with colon cancer (CC).

METHODS

ECM-related genes were obtained from Molecular Signatures database. Differential expression analysis was performed using the CC dataset from The Cancer Genome Atlas (TCGA) database. Four ECM-related genes related to overall survival were identified using the Cox regression and LASSO analysis. Then an ECM-related signature was developed and verified in three independent CC cohorts (GSE33882, GSE39582 and GSE29621) from the Gene Expression Omnibus (GEO). A prognostic nomogram was developed incorporating the ECM-related gene signature with clinical risk factors. CIBERSORT was used to explore the immune cell infiltration level. Human Protein Atlas (HPA) database was utilized to validate the expression levels of identified prognostic ECM genes.

RESULTS

Four ECM-related genes (, , and ) were identified to develop an ECM-based gene signature and demarcated CC patients into the high- and low-risk groups. In training and validation datasets, patients in the low-risk group had better overall survival outcomes than those in the high-risk group (log-rank P<0.001). In addition, ECM-related signature was significantly associated with consensus molecular subtype 4 (CMS4) as well as other known clinical risk factors such as a higher Tumor, Nodal Involvement, Metastasis (TNM) stage. Moreover, the risk score derived from the ECM-based gene signature could be utilized as an independent prognostic factor for CC patients. A nomogram including the ECM-related gene signature, age and stage was developed to serve clinical practice. CIBERSORT analysis showed immune cell infiltration was different between high- and low-risk groups. The immunohistochemical results derived from HPA indicated differential expression of prognosis-related ECM genes in CC and normal tissues.

CONCLUSIONS

In the present study, a novel risk model based on ECM-signature could effectively reflect individual risk classification and provide potential therapeutic targets for CC patients. Moreover, the prognostic nomogram may help predict individualized survival.