DNA methylation epigenotypes in breast cancer molecular subtypes

Abstract

Introduction: Identification of gene expression based breast cancer subtypes is considered as a critical means of prognostication. Genetic mutations along with epigenetic alterations contribute to gene expression changes occurring in breast cancer. So far, these epigenetic contributions to sporadic breast cancer subtypes have not been well characterized, and there is only a limited understanding of the epigenetic mechanisms affected in those particular breast cancer subtypes. The present study was undertaken to dissect the breast cancer methylome and deliver specific epigenotypes associated with particular breast cancer subtypes.

Methods: Using a microarray approach we analyzed DNA methylation in regulatory regions of 806 cancer related genes in 28 breast cancer paired samples. We subsequently performed substantial technical and biological validation by Pyrosequencing, investigating the top qualifying 19 CpG regions in independent cohorts encompassing 47 basal-like, 44 ERBB2+ overexpressing, 48 luminal A and 48 luminal B paired breast cancer/adjacent tissues. Using all-subset selection method, we identified the most subtype predictive methylation profiles in multivariable logistic regression analysis.

Results: The approach efficiently recognized 15 individual CpG loci differentially methylated in breast cancer tumor subtypes. We further identify novel subtype specific epigenotypes which clearly demonstrate the differences in the methylation profiles of basal-like and human epidermal growth factor 2 (HER2)-overexpressing tumors.

Conclusions: Our results provide evidence that well defined DNA methylation profiles enables breast cancer subtype prediction and support the utilization of this biomarker for prognostication and therapeutic stratification of patients with breast cancer.

Figure 1

Supervised hierarchic clustering of 28 paired breast cancer samples by using methylation microarray data from differentially methylated genes. Heat map and clustering of 76 significant (FRD-corrected, P < 0.001) differences between tumors (red) compared with adjacent tissues (green) shows that 30 of the differences (corresponding to 25 genes) are increases of methylation, whereas 47 (corresponding to 37 genes) are decreases of promoter methylation in tumors. Heat-map colors symbolize DNA methylation, as indicated in the color key. The full list of genes is presented in Table S2 in Additional file 2.

Figure 2

Supervised hierarchic clustering of 187 paired breast cancer samples by using methylation data from selected genes in the validation study. (a) Heat map and clustering of significant (FRD-corrected, P < 0.01) differences between tumors (red) and adjacent tissues (green). Heat-map colors symbolize DNA methylation, as indicated in the color key. (b) Principal component analysis (PCA) plots showing separation between tumor and adjacent tissue. (c) Box plots showing lower methylation density in adjacent tissue compared with tumors.

Figure 3

Supervised hierarchic clustering of the five candidate genes (Let-7a, NPY, RASFF1, FGF2, and HS3ST2). Differential methylation of patterns is shown by red (hypermethylated) versus green (nonmethylated) colors. As noted, three differentiated clusters (C1 to C3) are generated by this statistical analysis. Underneath, box plots showing differential methylation density among tumor subtypes.