3613
Background: Colorectal cancer is one of the most common malignancies with a high mortality rate. Patients with stage I and stage II colorectal cancer have limited options for treatment. Hypoxia affects the activation and regulation of colorectal cancer cells and participates in its invasion and migration. However, there is lack of an accurate and non-invasive method for assessing tumor hypoxia. The aim of this study was developing and validating a hypoxia gene signature for predicting the outcome in stage I/II colorectal cancer patients. At the same time , we hypothesized that analysis of database of CIT microarray dataset could identify important biomarkers for stage I/II colorectal cancer patients. Methods: A total of 309 colorectal cancer patients of early stage with complete clinical information were enrolled for construction generation of hypoxia-related gene signature (HRGS) based on the CIT microarray dataset. 1877 colorectal cancer patients with complete prognostic information in 5 independent datasets were divided into a training cohort and two validation cohort (TCGA and meta-validation). Prognostic analysis was assessed in these cohort to evaluate the predictive value of HRGS. Results: A model of prognostic HRGS containing 14 hypoxia-related genes was developed. In training cohort and two validation cohorts, patients in hypoxia high-risk group satisfied by our HRGS had significant poor disease free survival compared with those in the in the low risk group (HR=4.35, 95% CI=2.30-8.23, P<0.001 in training cohort, HR=2.14, 95% CI=1.09-4.21, P=0.024 in TCGA cohort, HR=1.91, 95% CI=1.08-3.39, P=0.024 in meta-validation cohort). When compared with Oncotype DX, HRGS achieved an improved survival correlation in the training cohort (mean C-index, 0.80 vs 0.65, P<0.05) and the validation cohort (mean C-index, 0.70 vs 0.61 in the TCGA cohort, mean C-index, 0.68 vs 0.73 in the meta-validation cohort). Analysis of the data found that patients with low survival rates have significant relationships with genes regulated by the cell cycle pathway, such as mTROC1, E2F, G2-M, mitotic, oxidative phosphorylation, MYC, PI3K-AKT-mTOR (P<0.005). Conclusions: HRGS was a satisfactory prognostic prediction model for early stage colorectal patients. Hypoxia-related genes that regulate the cell cycle pathway were associated with prognosis in patients with stage I and stage II colorectal cancer. Further researches are needed to assess the clinical effectiveness of the system and the treatment options for biological targets.