Abstract 1498: Methylation landscape of cancers associated with immunogenicity

• Published in: Cancer research (Chicago, Ill.) Ill.), 2019-07, Vol.79 (13_Supplement), p.1498-1498
• Main Authors: Park, Changhee, Jeong, Kyeonghun, Park, Joon-Hyeong, Bae, Jeong Mo, Kim, Kwangsoo, Ock, Chan-Young, Kim, Miso, Keam, Bhumsuk, Kim, Tae Min, Jeon, Yoon Kyung, Kim, Dong-Wan, Kang, Gyeong Hoon, Chung, Doo Hyun, Heo, Dae Seog
• Format: Article
• Language: English
• ISSN: 0008-5472; 1538-7445
• DOI: 10.1158/1538-7445.AM2019-1498

Introduction: Determining immunogenicity of tumors is important in predicting response to cancer immunotherapy. Tumor mutation burden, the degree of copy number variation expressed as chromosomal instability score and gene expression profiles (GEP) of tumors, such as cytolytic activity score, interferon-gamma signature and immune signature, are known biomarkers of immunogenicity. However, correlation between methylation burden and signature of tumor immunogenicity is unknown. Methods: We used The Cancer Genome Atlas (TCGA) pan-cancer database to investigate association of RNA sequencing data and methylation signature of tumors generated by HumanMethylation450K BeadChip. The estimated methylation levels in the CpG sites were expressed in β-scores. To determine immunogenicity in a sample, we used cytolytic activity (CytAct) score defined by the summation of RNA transcript levels of granzyme A (GZMA) and perforin 1 (PRF1). In each CpG site, β-scores of top 5% and bottom 5% were defined as “hypermethylated” and “hypomethylated”, respectively. Methylation burden of a sample was defined as number of CpG sites having hypermethylated or hypomethylated β-scores. Results: In TCGA, total of 7,914 pan-cancer samples had both RNA sequencing data and methylation array data. Methylation burdens of pan-cancer samples were negatively correlated with CytAct scores. (Spearman’s correlation rho value -0.37, p < 2.2e-16). This negative trend was consistently observed in most of cancer types and methylation subtypes of individual cancer including breast cancer. Using multivariate linear regression model, methylation burden predicted CytAct score independently along with mutation burden and chromosomal instability score. Hypermethylations in CpG sites of genes related to interferon gamma response(CXCL10, STAT1, IFI6and IFI27), lymphocyte infiltration(CD8A, CD3, CD79, LCK and CCL5) and tumor antigen recognition were associated with decreased CytAct scores whereas hypomethylation in CpG sites of genes related to TGF-β(CTNNB1, COL1A2, IGF2R, ITGB2, MMP17, SPARC and SMO) and fibroblast response(PLAUR, PLOD2, LOXL2 and MET) were associated with decreased CytAct scores in pan-cancer analysis. Conclusions: The methylation burden of tumor has negative correlation with the immunogenicity of tumors in general pan-cancer. Correlation of the specific methylation pattern with the response of immunotherapy is warranted in further clinical study. Citation Format: Changhee Park, Kyeonghun Jeong