CD8+ T Lymphocyte Coexpression Genes Correlate with Immune Microenvironment and Overall Survival in Breast Cancer

Abstract

Purpose: To identify CD8+ T lymphocyte-related coexpressed genes that increase CD8+ T lymphocyte proportions in breast cancer and to elucidate the underlying mechanisms among relevant genes in the tumor microenvironment.

Method: We obtained breast cancer expression matrix data and patient phenotype following information from TCGA-BRCA FPKM. Tumor purity, immune score, stromal score, and estimate score were calculated using the estimate package in R. The CD8⁺ T lymphocyte proportions in each breast carcinoma sample were estimated using the CIBERSORT algorithm. The samples with p < 0.05 were considered to be significant and were taken into the weighted gene coexpression network analysis. Based on the CD8⁺ T lymphocyte proportion and tumor purity, we generated CD8⁺ T lymphocyte coexpression networks and selected the most CD8⁺ T lymphocyte-related module as our interested coexpression modules. We constructed a CD8+ T cell model based on the least absolute shrinkage and selection operator method (LASSO) regression model and robust model and evaluate the prediction ability in different subgroups.

Results: A breast carcinoma CD8+ T lymphocyte proportion coexpression yellow module was determined. The coexpression genes in the yellow module were determined to increase the CD8+ T lymphocyte proportion levels in breast cancer patients. The yellow module was significantly enriched in the antigen presentation process, cellular response to interferon-gamma, and leukocyte proliferation. Subsequently, we generated CD8+ T cell-related genes lasso regression risk model and robust model, and eight genes were taken into the risk model. The risk score showed significant prognostic ability in various subgroups. Expression levels of proteins, encoded by CD74, were lower in the breast carcinoma samples than in normal tissue, suggesting expression differences at both the mRNA and the protein levels.

Conclusion: These eight CD8+ T lymphocyte proportion coexpression genes increase CD8⁺ T lymphocyte in breast cancer by an antigen presentation process. The mechanism might suggest new pathways to improve outcomes in patients who do not benefit from immune therapy.

Figure 1

The flow chart of the experimental sequence.

Figure 2

(a) We evaluated 860 breast cancer samples accurately using CD8+ T lymphocytes. (b) Hierarchical clustering tree was built using the dynamic hybrid cutting method. (c) Twenty-two coexpression models were identified. The yellow module had the strongest correlation with CD8+ T lymphocyte proportion in the TCGA-BRCA cohort (Cor = -0.41; p = 1e − 28). (d) The yellow module showed a significant correlation to CD8+ T Cell (Cor = 0.78, p = 9.7e − 259), (e) tumor purity (Cor = 0.86, p = 1.7e − 83), (f) immune score (Cor = 0.98, p = 1.2e − 197), and (g) stomal score (Cor = 0.28, p = 1.9e − 06).

Figure 3

(a) The 28 CD8+ T lymphocyte proportions positively coexpressing mRNA were most significantly enriched in the antigen processing and presentation and response to interferon-gamma, which suggested these biological processes might promote CD8+ T lymphocyte infiltration in the breast cancer microenvironment. (b) The CD8+ T lymphocyte negatively coexpressing module was most significantly enriched in the extracellular matrix organization.

Figure 4

The patients in low expression groups for GZMA (TCGA :  p < 0.001), CD74 (TCGA : p < 0.001), IL2RG (TCGA : p=0.009), CD3E (TCGA : p < 0.001), CCL5 (TCGA : p < 0.001), CD3D (TCGA : p < 0.001), CORO1A (TCGA : p < 0.001), HLA-DMA (TCGA : p=0.003), SELPLG (TCGA : p=0.002), HCST (TCGA : p < 0.001), HLA-DPB (TCGA : p=0.001), GZMK (TCGA : p=0.001), CD48 (TCGA : p < 0.001), PAMB9 (TCGA : p=0.005), CD2 (TCGA : p=0.003), CD27 (TCGA : p=0.003), IRF1 (TCGA : p=0.003), CD8A (TCGA : p=0.005), GBP4 (TCGA : p=0.048), TNFRSF1B (TCGA : p=0.011), GMFG (TCGA : p=0.006), CST7 (TCGA : p=0.001), GZMB (TCGA : p=0.049), PSMB10 (TCGA : p=0.002), and HLA-E (TCGA : p=0.046) showed survival risk against high expression groups.

Figure 5

(a-b) Establishment of a risk signature using the lasso regression curve and verification. (c) The samples in high-risk score level samples showed worse clinical survival outcomes for breast cancer patients (TCGA : p < 0.001; HR = 1.83). (d) The risk score was evaluated in various subgroups, including gender, stage, metastasis status, Ki-67, and EGFR. The results were the of the same significance in these subgroups.

Figure 6

(a) The CD8+ T cell proportion level in different gene expression patterns. (b) The gene expression level in different survival statuses. (c) The gene expression level in different tumor purity. (d) The gene expression level in different immune scores.

Figure 7

The correlation between CD8+ T cell coexpression genes and clinical stages (a), CD8+ T lymphocytes (b), and M2 macrophages (c).

Figure 8

Antigen processing and presentation, the chemokine signaling pathway, B cell receptor signaling pathway, and the T cell receptor signaling pathway were related to the high expression group in CD74, GIMAP4, HCST, and HLA-DMA.

Figure 9

We compared the various expression levels of these genes between normal and tumor tissues. HPA010592 was the antibody of CD74, which showed higher intensity in the tumor tissue against normal tissue.

Figure 10

GIMAP4 can be used as an independent prognostic factor after immunotherapy.