193 Background: Genomic heterogeneity has been observed in a number of tumor types including prostate cancer. However, how subclonal tumor diversity changes during metastasis and progression to lethality remains unexplored. Large scale genomic analyses have reported the most prevalent somatic aberrations associated with the dominant clone of the tumor without permitting an analysis of subclonal complexity or how this complexity impinges on metastatic potential or resistance to treatment. Methods: To understand and track the evolution of lethal prostate cancer from initial therapy to end stage metastases, we performed longitudinal and multiregional sampling of tumors from 7 patients with lethal prostate cancer. We performed whole-genome sequencing, RNA sequencing, and SNP profiling. Computational approaches were used to reconstruct the genetic relationships and evolution of the tumors. These evolutionary tree reconstructions allowed us to observe the dynamics of chromoplexy and mutational processes along specific branches of tumor evolution. To refine the genetic landscape and spatial connections between subclones, we employed deep, targeted re-sequencing of variant loci in the original sequenced samples, in additional FFPE sites sampled from the organ confined tumors, and from blood. Results: We show that while all primary and metastatic prostate tumors share a single ancestral clone, metastases arise from subclones present at minor frequencies in the primary tumor. We reveal that individual metastases comprise mixtures of subclones indicative of intra-metastatic heterogeneity. We provide evidence for cross-metastatic site seeding and dynamic remolding of subclonal mixtures in response to therapy suggesting a distinct metastatic hierarchy. Ultra-deep sequencing of end-stage blood reveals the presence of diverse subclones with metastatic potential derived from various stages in the evolution of the tumor. Conclusions: Our results demonstrate unexpected complexity in the origins of both primary and metastatic prostate cancer, with distinct implications for treatment of advanced disease.
Dynamic CTC phenotypes in metastatic prostate cancer models visualized using magnetic ranking cytometry