Circulating tumor DNA (ctDNA) to detect neuroendocrine prostate cancer genomic and DNA methylation changes.

Abstract

8 Background: Loss of androgen receptor signaling dependence occurs in approximately 20% of treatment resistant prostate cancers, and this may manifest as transformation to castration resistant neuroendocrine prostate cancer (CRPC-NE). The diagnosis of CRPC-NE relies on a metastatic tumor biopsy, which is invasive for patients. Here we study ctDNA as an approach to identify CRPC-NE-associated genomic and epigenomic changes. We also use this as a tool to understand tumor heterogeneity and clonal dynamics that occurs during CRPC-NE progression. Methods: 64 patients with metastatic prostate cancer (10 hormone naive, 35 castration resistant adenocarcinoma (CRPC-Adeno), 17 CRPC-NE) were prospectively enrolled for matched metastatic biopsy and blood collection. Twenty-four had multiple time-points, for total of 69 plasma and 98 metastatic tissues. Whole exome sequencing (WES) and whole genome bisulfite sequencing(of subset) of ctDNA,germline DNA, and metastatic biopsies were performed. Results: Overall there was high concordance of alterations between ctDNA and patient-matched metastasis, though this was highest in CRPC-NE. There was also less heterogeneity across patients with CRPC-NE. Alteration frequencies were consistent with published tissue-based studies with AR alterations enriched in CRPC-Adeno and TP53and RB1loss enriched in CRPC-NE ctDNA. The prognostic significance of these alterations differed based on histologic subtype. Allele-specific copy number analysis and serial sampling allowed for the detection and tracking of clonal and subclonal tumor cell populations. cfDNA methylation was reflective of methylation patterns in biopsies and detected CRPC-NE-associated epigenetic changes.A targeted set combining genomic ( TP53, RB1, CYLD, AR) and epigenomic(20 differential DNA methylation sites) was capable of identifying patients with CRPC-NE. Conclusions: WES and whole genome methylation of ctDNA is feasible, concordant with biopsy tissues, and identifies the spectrum and frequency of CRPC-NE genomic and epigenomic changes. A targeted panel extended and validated in larger cohorts could potentially improve the detection of CRPC-NE using non-invasive ctDNA profiling.