Integrated Genomic Profiling of Endocrine Therapy Response in Advanced Breast Cancer.

Abstract

Abstract PurposeIn hormone receptor positive breast cancer the response rates for endocrine treatment, i.e. tamoxifen (TAM) or aromatase inhibitors (AIs), are only 50 to 70% in the advanced disease setting. The overall aim of this retrospective study is to identify a molecular signature using integrated genomic profiling to improve prediction of endocrine treatment outcome in the advanced disease setting.ObjectivesA) To compare mRNA expression profiles of TAM- and AI-treated patients and to identify genes and pathways associated with treatment outcome.B) To discover miRNA and mRNA signatures predictive for AI response.Patients and MethodsFresh frozen Estrogen Receptor (ER)-positive primary breast cancer specimens from patients with advanced disease treated with first-line AIs (N=55) or TAM (N=109) were analyzed. Expression profiles of 670 miRNAs and 44K mRNAs were generated using multiplex qRT-PCR and microarrays. Profiles were related to clinical response and time to progression (TTP). Statistical and bio-informatic tools were applied to discover and combine markers into an integrated genomic predictive signature. The nearest centroid prediction method of BRB-ArrayTools (Version3.7.0) was used to assess the predictive value.ResultsThe quality controlled and informative expression profiles of 277 miRNAs and 14112 mRNAs in 50 AI-treated tumors and 10433 mRNAs in 101 TAM-treated tumors were included for further analysis in the discovery phase.Global testing of mRNAs linked to Biocarta pathways demonstrated the involvement of the interferon pathway in endocrine therapy response in both AI- and TAM-treated patients. Using BRB-ArrayTools survival analysis to find genes associated with TTP (P<0.05), we identified 1002 mRNAs in AI-treated and 662 mRNAs in TAM-treated tumors to be significantly related with TTP.The overlap of 40 mRNAs between AI- and TAM-treatment was defined as a mRNA signature for endocrine treatment outcome. In TAM-treated patients this classifier has a 69% accuracy (63% sensitivity, 74% specificity), an odds ratio for clinical benefit of 4.69 (95% CI 1.99-11.05, P<0.001) and a hazard ratio for TTP of 0.17 (95% CI 0.10-0.29, P<0.001). In AI-treated patients, this 40mRNA signature has a performance of 78% accuracy (84% sensitivity, 62% specificity) and significantly predicts clinical benefit (odds ratio = 8.27, 95% CI 2.00-34.3, P=0.004) and TTP (hazard ratio = 0.07, 95% CI 0.02-0.22, P<0.001).After statistical analysis a 16 miRNAs classifier for AI-treatment outcome was identified with a performance of 78% accuracy (89% sensitivity, 46% specificity). This classifier significantly predicts clinical benefit (Odds ratio = 7.07, 95% CI 1.57-31.9, P=0.011) and TTP (hazard ratio = 0.24, 95% CI 0.09-0.61, P=0.003).The genomic mRNA and miRNA signatures are currently integrated and validated in additional samples as well as “in silico” on tumors treated with neo-adjuvant AI (Miller et al, JCO 2009).ConclusionThis is the first study that combines miRNA and mRNA profiling in an attempt to define an integrated genomic signature for endocrine treatment outcome. Additional prospective multicenter studies are needed to confirm the predictive value of this signature. Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 3029.