Multi-omic profiling to clarify mechanisms of therapy-induced prostate cancer lineage plasticity.

Abstract

e17077 Background: Prostate adenocarcinomas are the most common form of the disease. Targeting the androgen receptor (AR) is the principal treatment strategy. However, resistance is common. Lineage plasticity (LP)—or change in cell state—is increasingly recognized as a mechanism of resistance to AR-targeting therapies. LP is a continuum, including amphicrine, double negative prostate cancer (DNPC), and neuroendocrine prostate cancer (NEPC). Despite the increasing incidence of LP due to more widespread use of potent AR inhibitors, mechanisms by which LP emerges after therapy have been understudied due to the paucity of matched patient tumor biopsies. To address this deficit, we performed multi-omic and spatial profiling on matched treatment-naïve prostate biopsies and post-treatment metastatic biopsies. Methods: We examined matched biopsies from 16 patients including seven whose progression tumors had evidence of LP. We performed DNA-sequencing to identify mutations and copy number alterations. We performed RNA-sequencing to transcriptionally subtype all tumors. We also performed spatial transcriptomic profiling with 10X Visium in a subset of matched samples with evidence of LP to infer copy number alterations and to understand the interactions between specific tumor populations and the tumor microenvironment (TME). Results: RNA-sequencing identified pathways linked to proliferation and inflammatory signaling to be highly activated in baseline tumors that eventually underwent LP and clarified distinct LP trajectories. DNA-sequencing revealed specific genomic patterns in TP53 and RB1 at baseline and progression linked to distinct LP trajectories vs. maintenance of an AR-driven phenotype. We confirmed that the genomic patterns seen in the distinct LP trajectory samples from our cohort were also present in similar therapy-induced LP samples from independent cohorts. Unbiased clustering of spatial transcriptomics data revealed seven tumor clusters. Inferred copy number analysis implicated either clonal selection or divergent clonal evolution in specific tumors that underwent LP. Conclusions: Multi-omic analysis clarifies baseline molecular features linked to specific LP trajectories and mechanisms that may contribute to the emergence of those trajectories. These results provide clues about why certain patient tumors may be at risk of undergoing specific LP trajectories with treatment. TME analysis to understand the influence of specific immune populations on cell state heterogeneity is ongoing.