Identification of an immune-related gene pair signature in breast cancer

BACKGROUND: Although breast cancer outcome has improved significantly with the recent use of molecularly targeted agents, reliable prognostic signatures are still unavailable because of tumor heterogeneity. Immune processes play an important role in tumor progression. Therefore, the aim of this study was to construct a prognostic signature based on immune-related genes (IRGs). METHODS: Clinical information and gene expression of 3,496 patients were extracted from eight public data sets. A total of 2,498 IRGs associated to 17 immune processes were downloaded from the ImmPort database. RNA sequencing (RNAseq) datasets [The Cancer Genome Atlas (TCGA) and GSE96058] were used as the training set (n=2,736) and all microarray datasets were used as validation set (n=760). IRGs related to prognosis were screened out from the training set and used to construct gene pairs. The Cox regression model was used, based on the immune-related gene pairs (IRGPs). The risk score of each patient was calculated and patients were stratified into high- and low-risk groups according to the optimal threshold of the risk score. Immune cell infiltration was evaluated between both groups. RESULTS: Among the 129 prognostic-related immune genes, 8,256 IRGPs were constructed. After screening, 89 IRGPs, including 86 unique IRGs, were used in the prognostic prediction model. Patients in the high-risk group exhibited a significantly poorer overall survival (OS) both in the training set [hazard ratio (HR): 5.9, 95% confidence interval (CI): 4.61–7.54] and validation set (HR: 1.52, 95% CI: 1.16–1.98) compared to the low-risk group. In addition, patients in the high-risk group showed a significantly lower infiltration of CD8(+) T cells than patients in the low-risk group. CONCLUSIONS: An independent IRGP signature was constructed. Through pairwise comparison of a set of genes, the OS of patients could be predicted. This method avoids the impact of the batch effect caused by different sequencing platforms and has a promising application prospect.