Abstract P6-14-06: Prospective Testing of Three Different Gene-Signatures for Patient Selection for Dasatinib Therapy in Metastatic Breast Cancer

Abstract

Abstract Background: Several gene signature-based predictors of response to targeted drugs have been proposed in the literature but none has been prospectively tested as patient selection tools in the clinic. The goal of this trial is to assess the positive predictive value of 3 conceptually different multi-gene signatures as predictors of response to the multitargeted kinase inhibitor dasatinib. Methods: This clinical trial requires biopsy of a metastatic lesion for gene expression profiling and employs a parallel, multi-arm, two-step, phase II design. Three markers are assessed including a (i) cell-line derived dasatinib-sensitivity signature, (ii) a src-pathway activity signature and (iii) a dasatinib target index calculated as the weighted average expression of all known dasatinib targets. Only markerpositive patients are treated with dasatinib 100 mg po daily and each marker arm is considered as a separate study with early stopping rules for futility (minimum sample size 9, maximum sample size 40/marker arm). A predictor is considered worthy of further study if the clinical benefit rate (i.e. positive predictive value) is ≥25%. Results: Forty seven patients were accrued from July 2009 through June, 2010, 49 biopsies were performed (soft tissues n=31, liver n=8, bone n=3, lung n=1, adrenal gland n=1), 6 samples had poor cellularity and 3 failed array QC. There was no patient recall, hospitalization or emergency room visit due to biopsy procedure. The median time from biopsy to genomic prediction result was 5 days (range 3-7). Twenty three (57%) patients had positive result for at least 1 predictor (5 were positive for 2) and 20 are receiving therapy (3 withdraw or progressed before therapy began). Responses as of June 2010; Target index arm (n=9): 5 PD (progressive disease), 4 SD (3 stable disease at 8 weeks 1 SD at 16 weeks); SRC Pathway arm (n=5): 3 PD, 2 SD at 8 weeks; Cell line predictor arm (n=6): 2 PD, 1 SD at 8 weeks, 3 not yet reached response evaluation. None of the 3 predictive marker arms have met early stopping yet and accrual is ongoing. Conclusion: Gene-expression signature based patient selection for targeted therapy is feasible and FNA biopsies of metastatic lesions for genomic testing are safe. Updated efficacy results will be reported. Citation Information: Cancer Res 2010;70(24 Suppl):Abstract nr P6-14-06.