Abstract 2100: The impact of race on the relationship between poly-CpG DNA methylation and mRNA expression in cancer with application to epigenome-wide association studies

Abstract

Abstract DNA methylation is a known regulator of mRNA expression. Recent studies in cancer have shown that significant proportions of the variability in mRNA expression can be explained by variation in DNA methylation. However, none of these studies have investigated the influence of race on this relationship. Here, we develop poly-CpG site models of gene expression that facilitate the quantification of race-specific effects on the regulatory connection between methylation and transcriptional gene expression. Paired Illumina 450k methylation and RNA-Seq gene expression data from The Cancer Genome Atlas (TCGA) was used to develop poly-CpG site elastic net penalized Poisson regression models of gene expression that account for race-by-CpG interactions for each of the eight TCGA tumor types with African American (AA) representation (>40 tumors each). Race-specific effects were summarized as the percentage of CpGs in a gene model with retained interaction terms. The method was applied to methylation data from an AA radical prostatectomy prostate cancer cohort (n=261) to generate predicted gene expression estimates that were used to perform a mechanistic epigenome-wide association study (EWAS) of Gleason grade (≤3+4 and ≥4+3 tumors). High percentages of variability in mRNA expression can be explained by poly-CpG site models, incorporating race specific effects, with tumor type ranges of 62-90% overall, 64-87% for AAs, and 64-92% for European Americans (EAs). Race played a significant role in the relationship between methylation and mRNA expression, with 86-97% of gene models containing at least one CpG site with a race-specific effect. Further, the median percentage of CpGs in a gene model with race-specific effects ranging from 11-18%. For these race-specific effects, the vast majority (>99%) had the same direction of association between the CpG site and gene expression in both AAs and EAs but with differing magnitudes. Application of this approach identified unique methylation alterations (97%) associated with Gleason grade compared with single site and comb-p region-based EWAS. In summary, this poly-CpG site modelling approach revealed a significant impact of race in the regulatory relationship between methylation and gene expression. The high percentage of genes -and CpG sites within those genes- demonstrating race-specific effects was a general phenomenon observed across all eight tumor types investigated. In addition to revealing this previously underappreciated contribution of race, the application of the predicted gene expression output from these poly-CpG site models to perform a gene-based EWAS of Gleason grade demonstrated that it provides a) complementary information to commonly used EWAS methods and b) direct expression-based mechanistic interpretation of findings from an AA methylation-based EWAS of prostate cancer. Citation Format: Ian M. Loveless, Yalei Chen, Sudha Sadasivan, Indrani Datta, Nilesh Gupta, Pamela Paris, Jia Li, Benjamin A. Rybicki, Albert M. Levin. The impact of race on the relationship between poly-CpG DNA methylation and mRNA expression in cancer with application to epigenome-wide association studies [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2100.