Distinct molecular mechanisms underlying clinically relevant subtypes of breast cancer: gene expression analyses across three different platforms

Abstract

Background: Gene expression profiling has been used to define molecular phenotypes of complex diseases such as breast cancer. The luminal A and basal-like subtypes have been repeatedly identified and validated as the two main subtypes out of a total of five molecular subtypes of breast cancer. These two are associated with distinctly different gene expression patterns and more importantly, a significant difference in clinical outcome. To further validate and more thoroughly characterize these two subtypes at the molecular level in tumors at an early stage, we report a gene expression profiling study using three different DNA microarray platforms.

Results: Expression data from 20 tumor biopsies of early stage breast carcinomas were generated on three different DNA microarray platforms; Applied Biosystems Human Genome Survey Microarrays, Stanford cDNA Microarrays and Agilent's Whole Human Genome Oligo Microarrays, and the resulting gene expression patterns were analyzed. Both unsupervised and supervised analyses identified the different clinically relevant subtypes of breast tumours, and the results were consistent across all three platforms. Gene classification and biological pathway analyses of the genes differentially expressed between the two main subtypes revealed different molecular mechanisms descriptive of the two expression-based subtypes: Signature genes of the luminal A subtype were over-represented by genes involved in fatty acid metabolism and steroid hormone-mediated signaling pathways, in particular estrogen receptor signaling, while signature genes of the basal-like subtype were over-represented by genes involved in cell proliferation and differentiation, p21-mediated pathway, and G1-S checkpoint of cell cycle-signaling pathways. A minimal set of 54 genes that best discriminated the two subtypes was identified using the combined data sets generated from the three different array platforms. These predictor genes were further verified by TaqMan Gene Expression assays.

Conclusion: We have identified and validated the two main previously defined clinically relevant subtypes, luminal A and basal-like, in a small set of early stage breast carcinomas. Signature genes characterizing these two subtypes revealed that distinct molecular mechanisms might have been pre-programmed at an early stage in different subtypes of the disease. Our results provide further evidence that these breast tumor subtypes represent biologically distinct disease entities and may require different therapeutic strategies. Finally, validated by multiple gene expression platforms, including quantitative PCR, the set of 54 predictor genes identified in this study may define potential prognostic molecular markers for breast cancer.

Figure 1

Unsupervised approach to identify breast cancer subtypes. (A) Correlation of breast tumor samples with the previously identified five subtypes of breast tumors. 526 out of 552 previously identified "intrinsic" genes were cross-mapped to Applied Biosystems Human Genome Survey Microarray and used for centroid correlation analysis and hierarchical clustering. Correlations with the centroids of the five subtypes were calculated for each sample from this study (the two microarray replicates for each sample were averaged). Samples were assigned to a subtype with which it showed the highest correlation using a cutoff value of 0.2. (B) Unsupervised hierarchical clustering of 20 breast tumor tissues analyzed by AB arrays using the 526 mapped intrinsic genes (the two microarray replicates for each sample are shown). The level of expression of each gene in each sample, relative to the median level of expression of that gene across all the samples, is represented using a red-black-green color scale as shown in the key (green: below median; black: equal to median; red: above median). (Left panel): Scaled down representation of the entire cluster of the 526 intrinsic genes and 20 tissue samples. (Right panel): Experimental dendrogram displaying the clustering of the tumors into three distinct subgroups. Branches are color-coded according to the subtype with which the corresponding tumor sample showed the highest correlation. Tumors with low correlation (< 0.2) with a specific subtype are indicated by gray branches.

Figure 2

Prediction of tumor subtype by Prediction Analysis of Microarrays (PAM). 428 genes were selected at a threshold of 1.0 that separated the two subtypes with the lowest overall misclassification rate of 5% (data not shown). Predicted probabilities of subtype for each tumor sample analyzed on Applied Biosystems Human Genome Survey Microarrays (top panel), Stanford Human cDNA arrays (middle panel) and Agilent Whole Human Genome Microarrays (bottom panel) were computed using the predictive model built using these 428 predictor genes.

Figure 3

Validation of the luminal A and basal-like subtypes using three microarray platforms. (A) Unsupervised hierarchical clustering of 20 breast tumor tissues analyzed by Applied Biosystems (two replicates per sample), Stanford cDNA and Agilent arrays using the 510 mapped intrinsic genes. The data from each platform were first transformed independently and then combined for clustering: the level of expression of each gene in each sample (Applied Biosystems microarrays: normalized signal intensity; Stanford cDNA and Agilent microarrays: normalized log2 ratio of the sample vs. the reference (UHR)) was transformed into a log2 ratio relative to the median level of expression of that gene across all the samples within the data set of the given platform. The experimental dendrogram displays the clustering of the tumors into distinct subgroups. Branches are color-coded according to the subtype with which the corresponding tumor sample showed the highest correlation. Tumors with low correlation (< 0.2) with a specific subtype are indicated by gray branches. Luminal A subtype (dark blue) and basal-like subtype (red). (B) Venn Diagram of the most differentially expressed genes identified by all three different array platforms; 319 genes were identified as the common signature genes using ANOVA analysis and the following criteria: (1) > 2-fold change between the two subtypes, and (2) False discovery rate < 5%. (C) PCA analysis of luminal A and basal-like samples using a minimum set of 54 genes identified by PAM analysis. Data sets generated from three array platforms (48 arrays total, two replicates per sample in the Applied Biosystems data set) on 6 luminal A and 6 basal-like tumor samples using the 319 common signature genes were used as training set for the PAM analysis.

Figure 4

Two-dimensional cluster diagram of the 1210 signature genes characterizing the luminal A and basal-like subtypes. ANOVA analysis was performed on 6 luminal A samples and 6 basal-like samples, coupled with Benjamini and Hochberg False Discovery Rate multiple testing corrections. The two subtypes of breast tumors (two replicates per sample) were clustered into distinct clusters with reversed gene expression patterns: highly expressed in luminal A (bottom, 613 probes representing 603 genes) and highly expressed in basal (top, 631 probes representing 607 genes). The color scheme of the heat map was described in legend to Figure 1B and is shown in the key. Branches in the dendrogram are color-coded according to the subtypes: blue, luminal A; red, basal-like.

Figure 5

Validation of potential prognostic markers by Taqman^® assay-based real-time PCR. 85 marker genes including the minimal set of 54 genes identified to best distinguish the luminal A and the basal-like subtypes were validated by TaqMan^® Gene Expression assays. Profile correlation analysis showed the expression profile across the 20 tumor samples determined by the three microarray platforms and by TaqMan^® Gene Expression assays are highly correlated with median correlation coefficient R > 0.9 and a rate of good correlation (R > 0.8) of 81–88%.

Figure 6

Hierarchical clustering analysis of expression data from the TaqMan^® assays and the three microarray platforms across the 85 marker genes. The same tumor sample analyzed by the three microarray platforms and the TaqMan^® Gene Expression assays were clustered together except for one sample (MicMa185). For the Applied Biosystems microarrays, the mean expression values of the two microarray replicates were presented. The transformed z-scores were represented using a red-black-green color scale as shown in the key (green: below mean; black: equal to mean; red: above mean).