Abstract LB-153: Pan-cancer patterns of DNA methylation dysregulation and copy number alteration burden

Abstract

Abstract Alterations both in the genome and epigenome are required for cancer development. In this study, we developed and investigated a summary measure of global DNA methylation dysregulation, the Methylation Dysregulation Index (MDI) in data from fourteen cancers (n = 5,603: bladder, breast, renal clear cell, renal papillary cell, esophagus, thyroid and uterine corpus endometrial carcinomas; colonic, lung, pancreatic, and prostate adenocarcinomas; head and neck and lung squamous cellular carcinomas; and hepatocarcinomas). We then compared the burden of DNA methylation dysregulation with somatic genetic alteration cumulative burden summary measures. We accessed the Level 1 Infinium HumanMethylation450K BeadChip and clinical data from 5603 tumors and 699 adjacent normal tissues in The Cancer Genome Atlas (TCGA) database. Information of fraction of the genome with copy number alterations (FGA) was retrieved from cBioportal. The raw DNA methylation data was preprocessed using RnBeads. Beta-values were background corrected using methylumi-noob and normalized using the functional normalization to increase comparability and reduce the noise of potential batch effects. Probes marked as Non-CpG, allosomes, cross-reactive probes or containing SNPs, and those with low quality were filtered before the analyses. We calculated the difference between the beta value per probe of tumor DNA methylation and the median beta value for the same probe for the adjacent normal tissue DNA methylation samples for each tumor type. The average absolute difference across all the interrogated probes corresponds to the MDI value. We examined the relation of MDI across tumors types as well as with age, gender, tumor stage, and the estimated percentage of tumor cells in the sample. We observed that MDI was significantly associated with tumor stage in head and neck squamous cell (P-trend = 4.24E-03), kidney clear cell (P-trend = 1.35E-08) and kidney papillary (P-trend = 4.86E-11) cancers, after adjusting for age, gender and estimated tumor purity. Our findings were consistent when the MDI calculation was restricted to probes stratified by genomic context (CpG islands, N-shores, N-shelves, open sea, S- shores, and S-shelves). Tumor stage in prostate adenocarcinoma and thyroid carcinoma were borderline significant (P-trend = 1.04E-02, and P-trend = 1.20E-02 respectively), however, the genomic context findings for these cancer types were inconsistent. The FGA was positively correlated to MDI for all tumor types (ρ = 0.54, P<2.2E-16). When analyzing by tumor type, MDI and FGA spearman correlation ranged between 0.21 and 0.80 and was significant for all the tumors except uterine cancer. This work contributes to the understanding of the impact of epigenetic alterations and their relation with genetic alterations across human cancers. DNA methylation dysregulation seems particularly important for kidney (clear cell and papillary) and head and neck squamous cellular cancers, and it might be related independently to cancer progression after adjusting for age and gender. Citation Format: Lucas A. Salas, Kevin C. Johnson, Brock C. Christensen. Pan-cancer patterns of DNA methylation dysregulation and copy number alteration burden. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr LB-153.