Pharmacogenomic analysis of HER2 amplified breast cancer treated with preoperative trastuzumab and paclitaxel, 5-fluorouracil, epirubicin, cyclophosphamide (T/FEC) chemotherapy
545 Background: We performed gene expression analysis to identify molecular predictors of resistance to preoperative concomitant trastuzumab and T/FEC chemotherapy. Methods: Pretreatment fine needle aspirations from 21 patients with HER2 amplified, stages II-III breast cancer were available for transcriptional profiling with Affymetrix U133 A chips. Results: Overall pathologic complete response (pCR) rate was 71%. Age, nuclear grade, tumor size, nodal status or quantitative estrogen and HER2 receptor mRNA expression showed no association with response in univariate and multivariate logistic regression. We tested the accuracy of a 30-gene pCR predictor that was previously developed from patients who received T/FEC only preoperative chemotherapy. This predictor was accurate in validation in T/FEC treated patients (n=51) but showed low sensitivity in patients who also received trastuzumab (sensitivity 53% versus 92%). We could not identify any differentially expressed genes between pCR (n=15) and residual disease (RD, n=6) at a false discovery rate (FDR) <90% in the HER2 amplified trastuzumab-treated cases. Hierarchical clustering using the “Perou intrinsic gene set” also failed to separate pCR from RD. Gene Set Enrichment Analysis with 22 genes from trastuzumab-resistant cell lines showed a modest association with RD (FDR=9%). Conclusions: Clinical variables and pharmacogenomic predictors that predict pCR in the absence of trastuzumab are no longer accurate when trastuzumab is included in the treatment. Trastuzumab-resistance associated genes identified in vitro are also associated with resistance in vivo. Support: Ellence Foundation. No significant financial relationships to disclose.