Clinical validation of breast cancer gene expression signatures using customized microarrays

A81 Current histopathological methods for classifying breast tumors are associated with several limitations, including significant inter-observer variability and their reliance on measuring single biomarkers. Recent studies have suggested that gene expression profiles may represent a useful alternative for cancer classification in the clinical setting. Here, we used a customized microarray platform to assess the robustness of these gene signatures in a prospectively defined breast cancer patient cohort. Using electrochemical in situ synthesis, we fabricated microarrays containing approximately 200 genes covering five previously defined breast cancer expression signatures (Estrogen receptor (ER+), HER2, 9Basaloid9 status, low histologic grade, and the Nottingham Prognostic Index (NPI)), and confirmed both the inter-user and day to day technical reproducibility of the custom array platform by classifying 19 breast tumors and 4 breast cell lines with previously known molecular profiles. Using the custom array, we profiled 58 invasive ductal carcinomas, and compared their classification to standard genome wide Affymetrix arrays and immunohistochemical (IHC) techniques. Of the 58 tumors, 52 (90%) were classified by the custom array into distinct ER+ (23 tumors), HER2 (15), and basaloid (14) groups. The concordance between the custom array and Affymetrix platforms was 85%, with 44/52 cases exhibiting similar classifications. We also correlated the custom array profiles to ER and HER2 IHC. The ER+ signature showed a good concordance of 84% with reported IHC status, which subsequently increased to 90% after IHC reassessment. The HER2 signature initially exhibited a lower concordance of 73% with reported IHC status, however this too increased to 87% after IHC reassessment. We are currently correlating the custom array profiles to other clinical parameters including HER2 FISH, NPI, and patient outcome. Our preliminary results suggest that custom arrays have the potential to function as an alternative diagnostic tool to standard IHC. This may be advantageous as custom microarrays possess I) relatively simple operating procedures, II) confer multiple types of information in a single assay, and III) may improve classification robustness by the simultaneous measurement of multiple biomarkers.