Multigene Predictors Developed On Breast Cancer Cell Lines To Predict Patient Chemotherapy Response: A Validation Study On The Nsabp B-27 Trial

1029 Background: We developed multigene predictors (MGPs) of chemotherapy sensitivity based on breast cancer cell lines exposed to combination chemotherapy regimens. Performance was assessed by the prediction of pathological complete response (pCR) in National Surgical Adjuvant Breast and Bowel Project (NSABP) Protocol B-27 clinical trial breast cancer patients. Methods: B-27 was a phase III trial of women with operable primary breast cancer assigned to receive 4 cycles of preoperative doxorubicin and cyclophosphamide (AC) (Grp 1) or 4 cycles of AC followed by 4 cycles of docetaxel (either preop [Grp 2] or post-op [Grp 3]). MGPs associated with AC and ACT sensitivity were separately developed based on the in vitro drug response and publicly available microarray genomic profiles of 40 breast cancer cell lines. Validations of the two genomic predictors were conducted blindly by NSABP statisticians to assess accuracy in predicting breast cancer patient pCR to preoperative therapy. Results: 219 patients treated with AC and 102 with ACT were included in this study. For the AC therapy, the MGP score was significantly associated with pCR from univariate analysis (p=0.0002) and multivariate analysis adjusted (p=0.015), and MGP remained the most important factor in predicting chemotherapy response compared to other clinical and biologic variables; prediction accuracy was 79%, sensitivity was 72%, and specificity was 80%; the area under the ROC curve was 0.75 [95% confidence interval: 0.64, 0.86]. For ACT therapy, the prediction accuracy, sensitivity, and specificity were 69.6%, 60%, and 73%, respectively; the area under the ROC curve was 0.64 [0.51, 0.77]. The association between MGP and pCR was statistically significant from univariate analysis, but less significant from multivariate analysis adjusted for covariates. Conclusions: Cell line-derived multigene predictors show promise in the prediction of chemotherapy response in breast cancer patients treated with neoadjuvant chemotherapy.The B-27 study was funded by NCI PHS grants U10-CA−37377, U10-CA−69974, U10-CA−12027, U10-CA−69651, and U24-CA−114732, and with additional support from sanofi-aventis.