Background: The advent of microarray technology has enabled robust, high throughput analysis of breast cancer (BC) transcriptomes. Indeed, molecular classification of BC has been revolutionized by the advent of Gene Expression Profiling (GEP). FFPE tumor samples have presented a technical challenge for GEP studies due to degradation of extracted RNA. Newer technologies have overcome this challenge, and have lead to generation of quality GEP data and thus, new insights using archived tissues. Of particular interest to our group has been application of these techniques to the study of triple negative breast cancer (TNBC). TNBC is a BC sub-type characterized by a lack of erbB2 gene amplification and estrogen and progesterone receptor expression. This clinically-defined BC sub-type carries a poor prognosis, is insensitive to hormonal or HER-2 targeted therapies, and displays different incidences among ethnic groups. A better understanding of the genetic and molecular mechanisms underlying TNBC is critical to improving clinic outcomes and developing tailorized therapies.

Study Objective: We demonstrate utility of FFPE BC samples in obtaining consistent, reproducible GEP data, and apply this technology to validate the ability to identify intrinsic BC subtypes in unselected specimens, as well to identify differentially expressed genes in TNBC.

Methods: RNA isolation and labelled cDNA preparation were performed from freshly cut FFPE sections. Samples were hybridized to a breast cancer focused gene expression array (Breast Cancer DSA Research Tool, Almac Diagnostics Inc). DSA chip quality was assessed on parameters selected automatically from GCOS report files per chip using MATLAB script based web application developed by Almac. Data pre-processing used the Resolver Error Model. All parameters including Raw Q, Background, Scaling Factor and all controls met quality criteria set by Affymetrix and Almac Dx SOPs. Hybridization results were assessed with Principal Component Analysis and Cluster Analysis in a Rosetta Resolver Gene Expression Data Analysis System to identify potential outliers, contamination, or intra-tumor heterogeneity. In total, 47 FFPE breast cancer samples covering a range of hormonal receptor status and sub-types were profiled, as were 28 TNBC FFPE tumor samples.

Results: Cluster analysis demonstrated that the Almac Breast Cancer DSA was able to clearly separate the 47 tumor samples of the mixed subtype group into the previously described intrinsic subgroups. Moreover, the DSA array contains 167 probesets which correspond to the 40 of the PAM-50 gene set used as a subtype predictor (Parker et al 2009). Analysis of the TNBC samples using the 167 probeset (based on mean intensity for probes representing each of the 40 genes) showed 100% consistency with published results demonstrating a basal-like gene expression signature. Summary: We have shown that our study methodology used can reliably measure gene expression in FFPE BC samples, and that the Breast Cancer DSA can be used to evaluate intrinsic subtypes of BC specimens. Analysis of the TNBC cases showed complete concordance between the PAM-50 gene set and the corresponding genes in the DSA. These study results are being validated in a larger data set.

Citation Information: Cancer Res 2010;70(24 Suppl):Abstract nr P4-08-12.