Predictive biomarkers in circulating tumor cells (CTC) from patients with castration-resistant prostate cancer (CRPC) through genomic analysis.

Abstract

179 Background: Although designed to retain activity against known androgen receptor (AR) mutants, it is predicted that new MDV3100-resistant mutations may predict for sensitivity to the drug in clinic. To predict for tumor sensitivity to treatment with novel AR targeted therapies, we explored the frequency of mutation detection and copy number alteration in CTC isolated from patients with CRPC enrolled on a trial with MDV3100. Methods: We used fluorescence-activated cell sorting (FACS) methodology to enrich EpCAM+, CD45−, DAPI− cells. For mutation detection and genomic copy number alteration in CTC in low number of cancer cells found in clinical samples, we optimized next-gen deep sequencing by Illumina. Results: In patients with progressive CRPC, >10 or >50 EpCAM+ events (EPE) were isolated by FACS in 88% or 58% of patients, in whom 32% and 10% had unfavorable (>5 cells/7.5 ml) CTC counts using CellSearch. EPE, expressing prostate-specific mRNAs, provide sufficient high quality DNA for genomic sequencing and copy number analysis. Adequate coverage was obtained from as few 50 EPE, with a recovery rate of 89% from FACS sorted samples. The detection threshold of a mutation was established at 1:4 alleles. To further expand genomic profiling in CTC, we optimized Nimblegen exome mutation detection by deep sequencing on HiSeq PE101. Our initial analysis established the polymorphism frequency detection thresholds in heterogeneous cell populations, and confirmed the sequencing coverage. Somatic missense mutations in AR, APC and TP53 found in CTC but not in paired WBC were confirmed by Sanger sequencing. In parallel, copy number alterations in CTC are studied. Conclusions: Somatic mutations detected in CTC isolated from patients with CRPC can serve as predictive markers of tumor sensitivity to targeted therapies. We established standard operating procedures for specimen processing, and confirmed the sequencing coverage and polymorphism detection thresholds in heterogeneous cell population. Currently we are proceeding to clinical samples to study the associations between specific molecular alterations in CTC as predictive markers of sensitivity and clinical outcomes.