Hierarchical clustering of primary estrogen receptor-positive breast cancers based on genome-wide DNA copy number aberrations. Unsupervised hierarchical clustering of 71 primary estrogen receptor-positive breast cancers, diagnosed in younger women (age ≤ 45 years) or older women (age ≥ 70 years), based on genome-wide DNA copy number aberrations. As previously reported for BAC-based array comparative genomic hybridization analyses of human breast cancer samples , columns represent individual tumor samples and rows represent individual genome probes (BAC clones), ordered by chromosome and genome position with 1pter at the top and 22qter at the bottom. Chromosome p-arms and q-arms are shown as different shades of the same color (blue = odd numbered chromosomes, yellow = even numbered chromosomes). As indicated in the color scale at the bottom, genome copy number losses are indicated in red (-0.5) and copy number gains are indicated in green (0.5). Yellow dots represent high-level genomic amplifications. Colored and grey-toned upper bars identify the age cohort, progesterone receptor (PR) status, nodal status and grade status of the estrogen receptor-positive samples in each column. The dendrogram shows unsupervised classification of the 71 samples into two primary clusters and four secondary clusters, with no significant cluster bias according to age, PR status, nodal status or grade status (P > 0.3, Fisher exact test).