The construction and analysis of a prognostic assessment model based on P53-related multi-genes in breast carcinoma

Breast cancer ranks second in female tumor mortality, with an estimation of 2 million new cases diagnosed each year worldwide.

In our current study, we screened 13 genes highly distributed on the P53 phenotype which were significantly expressed and had a strong correlation with survival in the Cancer Genome Atlas breast cancer dataset. Least absolute shrinkage and selection operator Cox regression was conducted to construct the risk assessment model. Based on bioinformatics and statistical methods, we confirmed the credibility and validity of the model by training set and testing set.

The result of comparing the other two previous hypoxia models was also satisfying. We also verified the model on one of the Gene Expression Omnibus datasets-GSE20685. Using clinical data from patients in the Cancer Genome Atlas, we acknowledged the risk score as an independent influence on breast cancer survival prognosis, and strong relevance was suggested between risk signature and age, lymphatic metastasis, tumor size and clinical stage by performing univariate and multivariate analysis. Immunology analysis demonstrated that the macrophages subset was positively associated with a risk score and other immune cell types had a negative effect with the risk score increases. The risk score was also emerging as a valuable prognostic factor for the prediction of chemotherapy drug curative effect because Gemcitabine, vinorelbine, paclitaxel and cisplatin known as a generic drug for breast cancer had more pleasing sensitivity in high-scored patients than low-scored patients.

The P53-related risk assessment model is promising to be a potential predictor for the prognosis of patients with breast cancer and a powerful guide for the selection of therapeutic strategies.