LAUSR

Primary prostate cancer (PCa) is frequently multifocal (genetically independent(GI) clones) and is further characterized by extensive intratumoral heterogeneity (genetically related clones). In metastatic castrate resistant PCa (mCRPC) monoclonal and polyclonal seeding of metastatic sites by related clones has been reported. However, the limited sequencing depth and/or sampling of sites is insufficient to rule out alternative evolutionary routes to drug resistance and metastatic dissemination. Here, we performed deep multi-site sequencing of 22 mCRPC patients to investigate the clonal architecture of mCRPC. Methods We obtained and profiled 81 metastatic biopsies from 22 patients using RNA and whole-exome sequencing (average depth of 600x). For each tumor we reconstructed its clonal phylogeny based on an integrated analysis of mutations (SNV) and copy-number variation (CNV) based on an explicit evolutionary model. Cancer Cell Fractions (CCF) were computed by adjusting variant allele frequencies (VAFs) for tumor purity and CNV multiplicity, followed by joint identification of clones through clustering of variants with similar CCFs and phylogenetic reconstruction using PhylogicNDT, which implements a nonparametric Drichlet process model. To assess overall genetic similarity between metastatic sites (and clones) we quantified the proportions of shared somatic mutations (present in all sites) and private mutations (mutations observed in only one site). Results We noted a high degree of genetic heterogeneity across metastatic sites, with a median of 35% (IQR:[21% - 51%]) shared somatic mutations, and 41% (IQR:[31% - 62%]) site-private mutations. No anatomical site was enriched for private mutations (anova p-value=0.747). Next, we characterized the clonal architecture across metastatic sites. We found a median of 10 clones per patient (IQR:[7.25-14.25]) and a median of 5.5 clones(IQR:[4-8]) per site with an average of 77 new synonymous and non-synonymous mutations per site. The number of subclones did not differ between metastatic sites as grouped by anatomical locations (anova P-value = 0.432). Strikingly, for 5 patients (23%) we found evidence for GI clones or extremely early branching clones metastasizing to different sites. In 2 of the 5 cases genetically unrelated clones metastasized to different sites, shared no common mutations and had distinct CNV profiles. For the remaining 3 cases metastatic sites only a median of 3% of mutations were shared across all sites, but displayed similar CNV profiles, indicating that chromosomal instability drove mutagenesis and preceded the acquisition of driver mutations. Conclusion In line with prior results our data confirms the high heterogeneity of mCRPC. We find metastatic sites to be multi-clonal with evidence of cross-site seeding. The limited number of truncal mutations, patterns of early-branching, suggest the need for multiple biopsies to fully characterize a patient's mCRPC. Our results also demonstrate, for the first time, that mCRPC can metastasize from more than one GI clone in the primary tissue. Citation Format: Noshad Hosseini, Zachery Reichert, Rohit Mehra, Arul Chinnaiyan, Marcin Cieslik. Multi-site sequencing of lethal prostate cancer reveals metastatic dissemination by independent and early-branching clones [abstract]. In: Proceedings of the AACR Special Conference: Advances in Prostate Cancer Research; 2023 Mar 15-18; Denver, Colorado. Philadelphia (PA): AACR; Cancer Res 2023;83(11 Suppl):Abstract nr A040.