Intra-Tumoral Angiogenesis Is Associated with Inflammation, Immune Reaction and Metastatic Recurrence in Breast Cancer

Abstract

Angiogenesis is one of the hallmarks of cancer. We hypothesized that intra-tumoral angiogenesis correlates with inflammation and metastasis in breast cancer patients. To test this hypothesis, we generated an angiogenesis pathway score using gene set variation analysis and analyzed the tumor transcriptome of 3999 breast cancer patients from The Cancer Genome Atlas Breast Cancer (TCGA-BRCA), Molecular Taxonomy of Breast Cancer International Consortium (METABRIC), GSE20194, GSE25066, GSE32646, and GSE2034 cohorts. We found that the score correlated with expression of various angiogenesis-, vascular stability-, and sphingosine-1-phosphate (S1P)-related genes. Surprisingly, the angiogenesis score was not associated with breast cancer subtype, Nottingham pathological grade, clinical stage, response to neoadjuvant chemotherapy, or patient survival. However, a high score was associated with a low fraction of both favorable and unfavorable immune cell infiltrations except for dendritic cell and M2 macrophage, and with Leukocyte Fraction, Tumor Infiltrating Lymphocyte Regional Fraction and Lymphocyte Infiltration Signature scores. High-score tumors had significant enrichment for unfavorable inflammation-related gene sets (interleukin (IL)6, and tumor necrosis factor (TNF)α- and TGFβ-signaling), as well as metastasis-related gene sets (epithelial mesenchymal transition, and Hedgehog-, Notch-, and WNT-signaling). High score was significantly associated with metastatic recurrence particularly to brain and bone. In conclusion, using the angiogenesis pathway score, we found that intra-tumoral angiogenesis is associated with immune reaction, inflammation and metastasis-related pathways, and metastatic recurrence in breast cancer.

Keywords: angiogenesis; breast cancer; epithelial-mesenchymal transition; gene set; metastatic recurrence; sphingosine-1-phosphate.

Figure 1

Association between tumor angiogenesis score and expression of Vascular endothelial growth factor (VEGF)-, blood endothelial cell markers-, vascular stability-, hypoxia-, and Sphingosine-1-Phosphate (S1P)-related genes in The Cancer Genome Atlas (TCGA) and Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) cohorts. Red and blue boxes stand for high and low angiogenesis pathway score groups, respectively. Tukey type boxplots show median and inter-quartile level values. (A) Gene expression levels of VEGF-related genes; VEGFA, VEGFB, VEGFR1 (FLT1), VEGFR2 (KDR) and VEGRR3 (FLT4). (B) Gene expression levels of endothelial cell marker-related genes; CD31 (PECAM1) and von-Willebrand factor (VWF). (C) Gene expression levels of vascular stability-related genes; TIE1, TIE2, ANGPT1, AMGPT2, VE-cadherin, Claudin5 and JAM2. (D) Gene expression levels of hypoxia-related genes; HIF1A and HIF1B. (E) Gene expression levels of S1P-related genes; SphK1, SphK2, S1PR1, and SPNS2. One-way ANOVA test was used to calculate p values. ANGPT1/2; Angiopoietin 1/2, CD31 (gene name: PECAM1), Claudin 5 (gene name: CLDN5), Hypoxia-inducible factor 1-alpha (HIF1A), HIF1B (gene name: ARNT), SPHK1/2; Sphingosine kinase 1/2, SPNS2; Spinster homolog 2, S1PR1; Sphingosine-1-phosphate kinase receptor-1, TEKVE-cadherin; TIE1; Tyrosine kinase with immunoglobulin-like and EGF-like domains 1, Vascular endothelial cadherin (gene name: CDH5), Junction Adhesion Molecule 2 (JAM2), VWF; von Willebrand factor (VWF), VEGFR1 (gene name: FLT1), VEGFR2 (gene name: KDR), VEGFR3 (gene name: FLT4), VEGF; Vascular endothelial growth factor.

Figure 2

Association between clinical features and the tumor angiogenesis score in TCGA and METABRIC breast cancer cohorts. The median was used as a cut-off to divide patients into high and low score groups within each cohort. (A) Boxplots of the angiogenesis score by breast cancer subtype, Nottingham pathological grade, and American Joint Committee on Cancer (AJCC) cancer stage in the TCGA and METABRIC cohort, and AJCC N and M category in TCGA cohort. One-way ANOVA test was used to calculate p values. Tukey type boxplots show median and inter-quartile level values. (B) Pathological complete response (pCR) rate for neoadjuvant chemotherapy between low (blue) and high (red) angiogenesis pathway score in estrogen receptor-positive/human epidermal growth factor receptor 2-negative (ER+/HER2-) and triple negative breast cancer (TNBC) in the GSE20194 (n = 197), GSE25066 (n = 467), and GSE32646 (n = 81) cohorts. Two tailed fisher’s exact test was used to calculate p values.

Figure 3

Association between tumor angiogenesis score and survival of breast cancer patients. Progression-free survival (PFS), Disease-Free (DFS), Disease-Specific (DSS), and Overall Survival (OS) of angiogenesis score low (blue) and high (red) in Whole breast cancer cohort, and each subtypes; estrogen receptor-positive/human epidermal growth factor receptor 2-negative (ER+/HER2-), triple negative breast cancer (TNBC), and HER2-positive, in the TCGA cohorts. The median was used as cut-off to divide into high and low score groups within each cohort. Log rank test was used to compare between two groups with Kaplan–Meier survival curves and to calculate p values.

Figure 4

Comparison of tumor infiltrating immune cells between low (blue) and high (red) angiogenesis score tumors. Boxplots of the comparison with (A) favorable immune cells: CD8, CD4 memory, dendritic cell (DC), T helper type 1 cells (Th1), M1 macrophages, and B cell, and (B) unfavorable immune cells: T helper type 2 cells (Th2), regulatory T cell (Treg), and M2 macrophage by low and high angiogenesis scores in the TCGA and METABRIC cohorts. (C) Comparison of low and high angiogenesis scores in scores of leukocyte fraction, tumor infiltrating lymphocyte (TIL) regional fraction and lymphocyte infiltration of TCGA cohort. The median was used as cut-off to divide into high and low score groups within each cohort. One-way ANOVA test was used to calculate p values. Tukey type boxplots show median and inter-quartile level values.

Figure 5

Gene Set Enrichment Assay (GSEA) comparing low and high angiogenesis pathway score tumors with enrichment gene sets in high angiogenesis pathway score group in both the TCGA and METABRIC cohorts. Correlation plot of (A) immune response gene sets; interferon (IFN)-γ, and IL2-STAT5 signaling, (B) Inflammatory response gene sets; Inflammatory response, IL6-JAK-STAT3 signaling, TNF-α signaling via NFkB, TGF-β signaling and hypoxia, (C) metastasis-related gene sets; epithelial mesenchymal transition (EMT), HEDGEHOG signaling, NOTCH signaling and WNT-β catenin signaling with normalized enrichment score (NES) and false discovery rate (FDR). The median was used as cut-off to divide into high and low score groups within each cohort. (D) Correlation plots of the angiogenesis score and EMT pathway score of both cohorts. p value was analyzed with spearman r correlation.

Figure 6

Association of the angiogenesis pathway score with metastatic recurrence in breast cancer. The Kaplan–Meier survival plots of metastasis-free survival for metastasis to bone, brain, or lung based on the pre-metastasis primary tumor in GSE2034 cohort (n = 286). The median was used as cut-off to divide into high (red) and low (blue) score groups within each cohort. To calculate p values, log rank test is used for comparing between two groups with Kaplan–Meier survival curves.