When Kaplan Meier analysis was performed using relapse as an endpoint, patients with breast tumors having high expression of these genes showed a significantly poorer outcome. To further assess the relevance and applicability of this 14 3 3 signature, we selected ER positive tumors from three other independent breast tumor microarray datasets with clinical information selleck products available. For all these three datasets no information on hormonal treatment was reported. Figure 2a shows the heat maps and den drograms for expression of the 14 3 3 signature genes from these three studies. The red group ing in the dendrogram represents breast tumors with high expression of the signature genes. The Wang et al. dataset includes data from 209 patients, and used the same Inhibitors,Modulators,Libraries Hu133A Affymetrix microarray platform used by Frasor et al.

All 29 genes from the 14 3 3 sig nature were retrieved and used for data mining. Unsu pervised clustering analysis identified the red group Panel I as a poor prognosis group driven by high expression of the Inhibitors,Modulators,Libraries signature genes. In a similar fashion, we analyzed the ER positive Inhibitors,Modulators,Libraries breast Inhibitors,Modulators,Libraries tumors included in the vant Veer data set. Given the different microarray platform used, a reduced number of genes were retrieved, 17 out of the 29 genes in the 14 3 3 gene sig nature. The signature genes not retrieved by our analysis were not present on those arrays. However, the subset of patients characterized by high expression of the 14 3 3 signature showed a significantly earlier relapse. We also examined the dataset of Sorlie et al. which used cDNA Stanford arrays con taining 8,102 genes.

Expression Inhibitors,Modulators,Libraries data for 19 genes of the gene signature were recovered and used for the analysis. The findings confirmed once again that overexpression of the 14 3 3 signature was significantly associated with a poorer disease free survival. Breast cancer subtypes and the 14 3 3 gene signature We next examined the distribution of the five major breast cancer molecular subtypes in the set of patients that showed high expression of the 14 3 3 gene signa ture and a poor clinical outcome in the different clinical studies by using a centroid mediated clustering algo rithm. All datasets showed enrichment for luminal B subtypes in tumors with elevated expression of the 14 3 3 signature genes, ranging from 53 to 58% of all tumors. In addition, 7 to 21% of total ER positive breast cancers showing high expression of the 14 3 3 gene signature were represented by the basal breast cancer subtype. For comparison we also classified tumors characterized by low expression of the 14 3 3 gene signature, and found that luminal A fda approved was the most abundantly represented molecular subtype in the differ ent datasets.