PD07-12: Identification of Chromatin Signatures Predictive of Prostate Cancer Progression

Michael Rothberg*, Jacob Enders, Cheyenne Williams, Michael Daneshvar, Nitin Yerram, Sohyoung Kim, Saori Fujiwara, Lorenzo Rinaldi, Songjoon Baek, Luke O'Connor, Qizong Kim, Patrick Gomella, Maria Merino, Bradford Wood, Peter Choyke, Baris Turkbey, Gordon Hager, Peter Pinto, Bethesda, MD

Introduction: Alterations of androgen receptor (AR) binding across the human genome and subsequent reprogramming of active regulatory sites are associated with prostate tumorigenesis and cancer progression. It is unknown whether early AR alterations are predictive of future aggressiveness of prostate cancer. We sought to identify chromatin landscape signatures predictive for low-grade, high-grade, and progressed prostate cancer to characterize patients at risk for aggressive disease after initial biopsy.

Methods: Samples from 42 patients with biopsy-confirmed prostate cancer were included in this study. Patients underwent MRI fusion-guided biopsies to obtain tissue from both prostate tumors and paired adjacent normal tissue at the time of diagnosis. Thirteen patients had low grade (grade group (GG) 1) lesions, 21 patients had high-grade (GG = 4) lesions, and 8 patients had initial GG = 2 lesions which later progressed to GG = 3 at follow-up biopsy. Median follow-up was 42 months (IQR: 28-54). Transposase-Accessible Chromatin sequencing (ATAC-Seq) of each fresh biopsy sample obtained genome-wide chromatin accessibility as a readout of active regulatory sites. Ultimately, 98 high-quality ATAC-Seq data sets (52 tumor, 47 adjacent normal) were obtained and analyzed.

Results: ATAC-Seq revealed two groups of regulatory sites possessing contrasting activities between low-grade and high-grade/progressed cancer. 2,078 sites were predominantly active in low-grade cancer, and 2,655 sites were predominantly active in high-grade/progressed cancer. Analysis of AR motifs showed similar levels of AR motifs in both groups. However, a comparison of AR binding in prostate primary tumors revealed that the high-grade/progressed sites had higher AR binding than the low-grade sites indicating additional co-factors may facilitate AR binding in the high-grade/progressed cancer. Binding motifs for ETS Transcription Factor (ERG) and Grainyhead Like Transcription Factor 2 (GRHL2) were enriched exclusively in the high-grade/progressed sites (enrichment P= 1e-94, 1e-54 for ERG and GRHL2 respectively in high-grade/progressed, and P= 1e-2, 1e-15 for ERG and GRHL2 respectively in low-grade).

Conclusions: We identified regulatory sites commonly active among high-grade and initially low-grade but later progressed PCa. Among high grade/progressed cancer, enriched ERG and GRHL2 binding motifs are associated with elevated AR binding, indicating the roles of ERG and GRHL2 as cofactors modulating AR binding. This suggests that ERG and GRHL2 modulating AR reprogramming may be an early event in PCa predictive of future progression.

Source of Funding: The National Institute of Health (NIH) Medical Research Scholars Program, Foundation for the NIH, NIH Intramural Research Program