Figure 4.

Figure 4A. Clustering analysis of gene amplification in 161 primary breast tumors. The log2ratios of the 17 genes listed in Table 1 for the 161 tumors were used to perform clustering analysis, generating the dendrogram at the top of the figure and the heat map at the bottom. The tested genes from Table 1 are listed to the left of the heat map. The labels to the left of the middle portion of the figure are: node (red for lymph node positive and blue for negative), invasion (red for invasive breast cancer and blue for noninvasive, or DCIS), size (red for tumor size over 5 cm and blue for less than 5 cm), and mutation status for PIK3CA, TP53, and AKT1 (red for the presence of mutation and blue for absence of mutation). Figure 4B. Survival plots calculated by Kaplan-Meier analysis using expression data of 4 genes POLD3, CCND1, FGFR1, and FGFR2. The four publicly available gene expression datasets from the GEO database are GSE4922, GSE2034, GSE1456, and GSE3494. The red and blue curves represent good and poor survival patient groups, respectively. The two patient groups were determined by hierarchical clustering analysis based on gene expression values of these 4 genes as described in Results and Discussion section.