Skip to main content
Figure 1 | Breast Cancer Research

Figure 1

From: DNA methylation profiling in the Carolina Breast Cancer Study defines cancer subclasses differing in clinicopathologic characteristics and survival

Figure 1

Consensus clustering of methylation β values in breast tumors using the GoldenGate Cancer Panel I array. DNA methylation profiles in 517 breast tumors and 9 normal breast tissues are shown. Columns represent tissue samples; rows represent CpG (cytosine preceding a guanosine) loci. Beta (β) value, indicating the fraction of DNA methylated, varies from 0 (blue, unmethylated) to 1 (red, highly methylated), with intermediate values shown in yellow. (A) Unsupervised clustering of the 167 most variable CpG sites having standard deviation of methylation β values of more than 0.2 from among the 935 CpG sites evaluated after filtering (see Materials and methods). The four tumor clusters are numbered 1 (n =123), 2 (n =108), 3 (n =99), and 4 (n =187). Primary tumor characteristics are indicated at the top of the heatmap as intrinsic subtype (defined by immunohistochemistry, or IHC): luminal A (red), luminal B (green), basal-like (black), HER2+/ER- (blue), unclassified (gray), missing values (white); p53 mutation status: mutant (red), wild-type (black); and hormone receptor (HR) status: HR+ (red), HR- (black). (B) Methylation in the same 167 CpG sites in 9 normal breast tissues, with probes ordered as in the consensus clustered heatmap. (C) Relationship between methylation cluster and intrinsic tumor subtype, shown according to intrinsic subtype (top panel) or according to methylation cluster (bottom panel). (D) Box-and-whisker plot showing differences (P <0.0001) in methylation (β) of the four consensus clusters, with numbers of tumors within each cluster shown along the top of the boxplot. Luminal-enriched tumor cluster 3 exhibits distinctly higher methylation than other clusters. ER, estrogen receptor; HER2, human epidermal growth factor receptor 2.

Back to article page