An anoikis-related signature predicts prognosis and immunotherapy response in gastrointestinal cancers.

BACKGROUND

Gastrointestinal (GI) cancers have high incidence rates and mortality rates. Anoikis is a special type of cell apoptosis, and anoikis resistance has been reported to be associated with tumor malignancy. We aimed to explore the roles of anoikis-related genes (ARGs) in the GI cancer prognosis.

METHODS

We extracted RNA sequencing and clinical data from The Cancer Genome Atlas and Gene Expression Omnibu databases for patients with esophageal cancer, gastric cancer, colon cancer and rectal cancer and identified ARGs from GeneCards and Harmonizome. Anoikis-related patterns were identified via unsupervised clustering analysis. We constructed a prognostic signature (Anoscore) based on prognostic ARGs through univariate, LASSO, and multivariate Cox regression analyses. The model was validated and evaluated using Kaplan-Meier analysis, receiver operating characteristic curves, univariate Cox regression analysis, multivariate Cox regression analysis, column charts, and calibration curves. We also performed a single-cell sequencing analysis of candidate genes via TISCH2. A correlation analysis between the Anoscore, the tumor microenvironment and drug sensitivity was conducted in GI cancers. The expression and function of some candidate genes were validated .

RESULTS

In terms of prognostic ARGs, two anoikis-related patterns, ARG clusters A and B, were identified. ARG cluster B had a worse prognosis than did ARG cluster A. Subsequently, the Anoscore was developed as an independent prognostic factor. It demonstrated the robust predictive capability for the prognosis of patients with GI cancers. Notably, patients with high Anoscores exhibited poor outcomes. In addition, we established a nomogram (Ano-nomogram) based on the Anoscore and clinicopathological factors of patients to predict the 3-year and 5-year survival probabilities. Moreover, patients with high Anoscores had higher levels of immune cell infiltration and higher immune checkpoint expression. The drug sensitivity analysis revealed that patients with high or low Anoscores were sensitive to different chemotherapies and targeted drugs. S100A11 and TLR3, representative candidate genes, exhibited different expression patterns and biological functions.

CONCLUSION

This study highlighted the significant potential of the Anoscore in predicting prognosis and guiding the selection of personalized therapeutic regimens for patients with GI cancers.