Identification of the Molecular Subtype and Prognostic Characteristics of Breast Cancer Based on Tumor-Infiltrating Regulatory T Cells

Abstract

Background: T regulatory cells (Tregs) are essential for preserving immune tolerance. They are present in large numbers in many tumors, hindering potentially beneficial antitumor responses. However, their predictive significance for breast cancer (BC) remains ambiguous. This study aimed to explore genes associated with Tregs and develop a prognostic signature associated with Tregs. Methods: The gene expression and clinical data on BC were obtained from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases. The integration of CIBERSORT and weighted correlation network analysis (WGCNA) algorithms was utilized to identify modules associated with Tregs. The consensus cluster algorithm was utilized to create molecular subtypes determined by genes associated with Tregs. Then, a prognostic signature associated with Tregs was constructed and its relationship to tumor immunity and the prognosis was evaluated. Results: The blue module genes exhibited the most significant correlation with Tregs, and 1080 genes related to Tregs were acquired. A total of 93 genes from the TCGA dataset were found to have a significant impact on patient prognosis. Samples from BC were categorized into two clusters by consensus cluster analysis. The overall survival, immune checkpoint genes, molecular subtype, and biological behaviors varied significantly between these two subtypes. A 10-gene signature developed from differentially expressed genes between two subtypes demonstrated consistent prognostic accuracy in both TCGA and GEO datasets. It functioned as a standalone prognostic marker for individuals with BC. In addition, patients with low risk are more inclined to exhibit increased immune cell infiltration, TME score, and tumor mutation burden (TMB). Meanwhile, Individuals classified within the low-risk group showed better responses to immunotherapies compared to their counterparts in the high-risk group. Conclusions: The prognostic model derived from Tregs-related genes could aid in assessing the prognosis, guiding personalized treatment, and potentially enhancing the clinical outcomes for patients with BC.

Keywords: breast cancer; immunotherapy; prognostic model; regulatory T cells; tumor microenvironment.

Figure 1

Discovery of genes linked to Tregs through WGCNA. (a) Sample clustering analysis to detect outliers. (b) Analysis of the network topology for various soft-thresholding powers. (c) The cluster dendrogram with the gene modules and module merging. (d) The correlations between gene modules and immune cells.

Figure 2

Cluster analysis for Tregs-related genes. (a) Consensus clustering analysis revealed the presence of two molecular clusters associated with Tregs. (b) The PCA analysis of distinct Tregs-related molecular clusters. (c) The comparison of survival rates among the two clusters. (d) The variances in the biological behaviors of two clusters by GSVA analysis. (e) Percentage of patients in two clusters with various molecular subtypes of BC. (f) The variations in the expression of genes related to immune checkpoints among the two clusters.

Figure 3

Development and validation of the Tregs-related prognostic model. (a) Volcano plot of the differentially expressed genes (DEGs) between two clusters. (b) Coefficients of the LASSO analysis. (c) Multivariate Cox regression analysis of 10 prognostic DEGs. (d) Difference in the risk scores among the two clusters. (e–h) Differences in overall survival of BC patients in training cohort (e), GSE20685 cohort (f), GSE21653 cohort (g), and GSE22219 cohort (h). (i–l) Time-dependent ROC curves analysis in training cohort (i), GSE20685 cohort (j), GSE21653 cohort (k), and GSE22219 cohort (l).

Figure 4

Development and validation of a novel nomogram in the TCGA cohort. (a) The correlation between risk score and clinic–pathologic feature. (b) Univariate and multivariate (c) cox analyses. (d) A prognostic nomogram predicting 1-, 3-, and 5-year overall survival of BC. (e) The calibration curve of the nomogram predicts the 1-, 3-, and 5-year survival rate. (f) ROC analysis showing the AUCs of nomogram in predicting the 1-, 3-, and 5-year overall survival.

Figure 5

Immune infiltration differences between high- and low-risk groups. (a) Correlation between the risk score and immune infiltrating cells by the CIBERSORT algorithm. (b) Differences between immune infiltrating cells and immune function (c) between the two risk groups by the ssGSEA algorithm. (d) Variations in the TME score among the two risk groups.

Figure 6

Differences in immunotherapy responses between the groups. (a) Comparison of the TIDE score in the two risk groups. (b–e) Differences in IPS among the two risk groups. (f) Differences in the expression of antigen presentation and (g) immune checkpoint genes. (h) Differences in TMB among the two risk groups.

Figure 7

Chemotherapeutic sensitivity and functional enrichment analysis. (a–e) The box plots of the estimated IC50 for four drugs between the high- and low-risk groups. (f) Volcano plot of the DEGs. (g) Dot plot of GO terms, including BP, CC, and MF analysis. (h) Dot plot of the KEGG enrichment analyses.