Microdissecting the Hypoxia Landscape in Colon Cancer Reveals Three Distinct Subtypes and Their Potential Mechanism to Facilitate the Development of Cancer.

BACKGROUND

Hypoxia contributes to tumor progression and confers drug resistance. We attempted to microdissect the hypoxia landscape in colon cancer (CC) and explore its correlation with immunotherapy response.

MATERIALS AND METHODS

The hypoxia landscape in CC patients was microdissected through unsupervised clustering. The "xCell" algorithms were applied to decipher the tumor immune infiltration characteristics. A hypoxia-related index signature was developed via the LASSO (least absolute shrinkage and selection operator) Cox regression in The Cancer Genome Atlas (TCGA)-colon adenocarcinoma (COAD) cohort and validated in an independent dataset from the Gene Expression Omnibus (GEO) database. The tumor immune dysfunction and exclusion (TIDE) algorithm was utilized to evaluate the correlation between the hypoxia-related index (HRI) signature and immunotherapy response. Quantitative reverse transcription polymerase chain reaction (qRT-PCR) and western blotting were performed to verify the mRNA expression levels of five key genes. The Cell Counting Kit-8 (CCK-8) assay and flow cytometry were performed to examine the cell viability and cell apoptosis.

RESULTS

Patients were classified into hypoxia-high, hypoxia-median, and hypoxia-low clusters in TCGA-COAD and verified in the GSE 17538 dataset. Compared with the hypoxia-low cluster, the hypoxia-high cluster consistently presented an unfavorable prognosis, higher immune scores, and stromal scores and elevated infiltration levels of several critical immune and stromal cells. Otherwise, we also found 600 hypoxia-related differentially expressed genes (HRDEGs) between the hypoxia-high cluster and the hypoxia-low cluster. Based on the 600 HRDEGs, we constructed the HRI signature which consists of 11 genes and shows a good prognostic value in both TCGA-COAD and GSE 17538 (AUC of 6-year survival prediction >0.75). Patients with low HRI scores were consistently predicted to be more responsive to immunotherapy. Of the 11 HRI signature genes, RGS16, SNAI1, CDR2L, FRMD5, and FSTL3 were differently expressed between tumors and adjacent tissues. Low expression of SNAI1, CDR2L, FRMD5, and FSTL3 could induce cell viability and promote tumor cell apoptosis.

CONCLUSION

In our study, we discovered three hypoxia clusters which correlate with the clinical outcome and the tumor immune microenvironment in CC. Based on the hypoxia cluster and HRDEGs, we constructed a reliable HRI signature that could accurately predict the prognosis and immunotherapeutic responsiveness in CC patients and discovered four key genes that could affect tumor cell viability and apoptosis.