Abstract P008: Native Cell-Free DNA sequence analysis reliably infers epigenetically diverse subtypes of prostate cancer

Abstract

Abstract Background: With the adoption of potent AR signaling inhibitors, there has been a prominent rise in the frequency of treatment-resistant tumor phenotypes. Highly aggressive transdifferentiated tumors may emerge as AR low or AR null prostate cancers (PCa) with or without neuroendocrine signatures. An accurate determination of these epigenetically diverse phenotypic subtypes has direct clinical implications, including therapy choice. Current approaches for disease subtype determination often require tissue biopsies, limiting due to the high risk of comorbidity. For a temporally evolving cancer, a minimally invasive molecular assay is urgently needed to determine disease subtypes accurately to guide clinical decisions. Methods & Results: Cell-free DNA assay emerging as a nextgen tool for cancer early detection and real-time tumor evolution monitoring tool. Cell-free DNA primarily consists of DNA protected by nucleosomes and DNA binding elements and is shed into the bloodstream by cells that undergo apoptosis. We hypothesized that a nucleosome protection map and transcriptional regulation could be reliably determined from circulating tumor DNA (ctDNA) analysis. This analysis can accurately classify advanced prostate cancer into epigenetically divergent subclasses such as adenocarcinoma (ARPC) and neuroendocrine (NEPC). For developing robust phenotype-specific signatures, we performed whole-genome sequencing on plasma cfDNA from 30 PDX lines to achieve moderately deep coverage (>10X) of “pure” human ctDNA sequence data. We applied an analysis framework “Griffin” to determine transcription factor binding sites and transcription start site nucleosome protection signatures, predictive of specific transcription factor activity. We also generated H3K27ac, H3K27me1, and H3K27me3 histone profiles from resected PDX tumors and observed that ctDNA fragmentation and coverage patterns are consistent with histone modifications. Guided by differential nucleosome profiles, we derived a novel ctDNA fragment size variability signature that accurately predicted expression deregulation for phenotype-defining genes. These signatures helped us build an unsupervised statistical model that accurately classified PCa subtypes. We applied our model to 2 clinical cohorts and attained 96% accuracy for classifying patients with well-defined subtypes of advanced PCa, using low coverage sequencing. Conclusion: We have developed a ctDNA analysis method that can accurately classify epigenetically diverse subtypes of advanced prostate cancer, spanning AR active and neuroendocrine subtypes. This scalable and cost-effective approach is also suitable for longitudinal monitoring and prediction of treatment-emergent prostate cancer resulting from therapeutic resistance. Citation Format: Navonil De Sarkar. Native Cell-Free DNA sequence analysis reliably infers epigenetically diverse subtypes of prostate cancer. [abstract]. In: Proceedings of the AACR Special Conference: Precision Prevention, Early Detection, and Interception of Cancer; 2022 Nov 17-19; Austin, TX. Philadelphia (PA): AACR; Can Prev Res 2023;16(1 Suppl): Abstract nr P008.