Description of each patient cohort

1) SOTIRIOU et al (http://www.pnas.org/cgi/content/full/1732912100/DC1) - This cohort contained 99 node-negative and node-positive breast cancer patients. All of the tumor samples were invasive ductal carcinomas. 46 individuals were node negative and 53 were node positive. 16 patients with tumor grade 1, 38 patients with tumor grade 2 and 45 patients with tumor grade 3. 65 patients were ER positive and 34 ER negative patients. Two patients received PMF Chemotherapy; 30 patients received CMF; and two received Adrymycin, CMF chemotherapy.

2) BILD et al (GSE3143) - This cohort contained 157 patients, 110 with ER level 1 and 47 with ER level 0.

3) WANG et al (GSE2034) - This cohort contained 286 lymph-node-negative patients of which, 146 were of stage T1, 132 of stage T2, and 8 of stage T3/4. This cohort contained 209 ER positive and 77 ER negative patients. There were 165 PR positive, 111 PR negative and 10 with unknown PR status. 148 patients were with poor grade, 42 with moderate grade, 7 with good grade and 89 with unknown grade. There were 139 pre-menopausal and 147 post-menopausal patients.

4) VAN DE VIJVER et al (http://www.rii.com/publications/2002/nejm.html) - There were 295 consecutive patients with primary breast carcinomas, 151 with lymph node negative disease, and 144 with lymph node positive disease.

5) MILLER et al (GSE3494) – This cohort contained 236 patients, 62 patients with Grade I, 121 with Grade II, 51 with Grade III and 2 patients with unknown grade information. 201patients were ER positive and 31 patients were ER negative. There were 179 PR positive patients and 57 PR negative patients. The cohort contained 78 lymph node positive patients and 149 lymph node negative patients.

6) LOI et al (GSE6532) – This cohort contained 137 untreated patients and 277 tamoxifen treated patients. Gene expression profiles of 327 patients were screened on GPL96 Affymetrix Gene Chip Human Genome U133 Array Set HG-U133A platform and 87 patient expression profiles were generated on GPL570 Affymetrix GeneChip Human Genome U133 plus 2.0 Array. The cohort contained 250 lymph node negative patients and 143 lymph node positive patients. There were 82 patients with Grade I, 182 patients with Grade II, and 76 patients with Grade III. There were 349 ER positive patients and 45 ER negative patients.

7) IVSHINA et al (GSE4922) - This cohort contained patient and tumor samples of the Uppsala and Singapore sets. The Uppsala set was composed of 249 patients. The Singapore set contained 40 patients. There were 211 ER positive patients and 34 ER negative patients. This cohort contained 81 lymph node positive patients and 159 lymph node negative patients.

8) SORLIE et al (GSE4335) - This cohort contained 122 tissue samples of which, 77 carcinomas and 7 nonmalignant tissues were previously published. There were 83 ER positive patients and 32 ER negative patients. There were 34 Lymph node negative patients and 46 lymph node positive patients. The cohort contains 11 patients with Grade I, 49 patients with Grade II, and 53 patients with Grade III.