Abstract 3514: DNA methylation markers for risk of metastasis in a cohort of men with localized prostate cancer

Abstract

Abstract Background: Most localized prostate cancers (PCa) are indolent and will not require treatment, but a small proportion can progress to metastasis. Distinguishing aggressive from indolent tumors can better inform treatment decisions. There has been interest in cancer genomics to predict PCa aggressiveness but with no consensus on methylation's role. We examined whether DNA methylation status of selected candidate genes can predict risk of PCa metastasis in the absence of curative treatment. Methods: A retrospective cohort of men diagnosed with PCa at Kaiser Permanente Southern California 1997-2007 who met the following inclusion criteria were identified: (1) diagnosed at stage I and II and (2) did not receive PCa treatment for at least 6 months after diagnosis. Men were followed for metastasis and censored at the initiation of prostatectomy or radiation, non-PCa related death, membership termination, or end of 2016. Potential metastasis was identified by an algorithm, then manually reviewed and confirmed. For each metastasis case, up to 4 controls were selected using density sampling, matched on age, race (black vs. non-black) and Gleason grade. For each case and control, FFPE blocks of diagnostic prostate biopsy cores were retrieved. The study pathologists reviewed the H/E slides and circled cancerous areas, which were macro-dissected for DNA extraction. Methylation status was obtained using Illumina's Infinium Methylation EPIC BeadChip. The candidate genes (N=118) were selected from the literature in the functional categories of cell cycle control; tumor suppressors; cell signaling; cell adhesion; angiogenesis; immune function; and stem cell markers. In total, 4,518 CpG markers from the Infinium assay covered regions of the candidate genes. Three statistical methods were used to detect the genetic associations, including univariate, principal components (PC)-based, and cluster-based analysis. To adjust for multiple comparisons within a gene and across genes, a double false discovery rate (DFDR) procedure was applied. The DFDR-adjusted p-value for each marker was calculated at 0.05 alpha level. Results: In total 163 cases and 311 matched controls were included in the analyses after excluding those with insufficient DNA or that failed the assay. For each analysis, univariate, PC-based, and cluster-based, the following number of candidate genes were found significant after DFDR procedure: 5 genes (PLAU, RASSF1, SLC5A6, KDR, CCL2), 8 genes (BCL2, KLF4, CTNND2, NRP2, RAP1GAP, LPL, EFEMP1, HSPA1A), and 7 genes (ETV4, ANGPTL4, BRMS1, LPL, BCL2, NANOG, ADM), respectively. The overall agreement of gene-level testing significances was at 84% between PC-based and cluster-based analysis. Conclusions: Among those showing consistent statistical significance, two genes, BCL2 and LPL, were identified in two analyses. Further study may provide insight to predict risk of prostate cancer metastasis. Citation Format: Kim Cannavale, Jeff Slezak, Yu-Hsiang Shu, Gary W. Chien, XuFeng Chen, Feng Shi, Kim Siegmund, Stephen Van Den Eeden, Jiaoti Huang, Chun R. Chao. DNA methylation markers for risk of metastasis in a cohort of men with localized prostate cancer [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 3514.