Construction and validation of a hypoxia-related gene signature to predict the prognosis of invasive ductal breast carcinoma

Abstract

Abstract Background: Invasive ductal carcinoma (IDC), the most common type of breast cancer, was known for its aggressive nature. Recent research revealeda lack of oxygen, known as hypoxia, wascrucial in forming breast cancer. This research aimed to create a robustsignature with hypoxia-related genes to predict the prognosis of IDC patients. The function of hypoxia genes was further studied through cell line experiments. Materials and Methods: In the bioinformatic part, transcriptome and clinical information of breast IDC were obtained from The Cancer Genome Atlas. Hypoxia-related genes were downloaded from the Genecards Platform. Differentially expressed hypoxia-related genes (DEHRGs) were identified. Training and test sets were defined with a 1:1 ratio. Prognostic-related DEHRGs were identified through Cox regression; then the signature was established and validated. The nomogram was created by incorporating the signature and clinicopathological characteristics. The predictive value of the nomogram was evaluated by C-index and receiver operating characteristic cure. Immune microenvironment and mutation burden were also examined. In the experiment part, the function of the two most significant hypoxia-related genes were further explored by cell-line experiments. Results: In the bioinformatic part, 37 up-regulated and 55 down-regulated DEHRGs were screened out. A prognostic signature was constructed containing nine hypoxia genes (ALOX15B, CA9, CD24, CHEK1, FOXM1, HOTAIR, KCNJ11, NEDD9, PSME2) in the training set. Low-risk patients exhibited a much more favorable prognosis tthan higher-risk ones(P<0.001). The signature was double-validated in the test set and GSE131769 (P=0.006 and P=0.001). The nomogram showed excellent predictive value with 1-year OS AUC: 0.788, 3-year OS AUC: 0.783, and 5-year OS AUC: 0.817. Patients in the high-risk group had a higher tumor mutation burden when compared to the low-risk group. In the experiment part, the down-regulation of PSME2 inhibited cell growth ability and clone formation capability of breast cancercells, while the down-regulation of KCNJ11 did not have any functions. Conclusion: Based on 9 DEHRGs, a reliable signature was established through the bioinformatic method. It could accurately predict the prognosis of breast IDC patients. Cell line experiment indicated that PSME2 played a protective role. Summarily, we provided a new insight to predict the prognosis of breast IDC by hypoxia-related genes.