Pan-cancer methylation analysis reveals an inverse correlation of tumor immunogenicity with methylation aberrancy.
Changhee ParkKyeonghun JeongJoon-Hyeong ParkSohee JungJeong Mo BaeKwangsoo KimChan-Young OckMiso KimBhumsuk KeamTae Min KimYoon Kyung JeonSe-Hoon LeeJu-Seog LeeDong-Wan KimGyeong Hoon KangDoo Hyun ChungDae Seog HeoPublished in: Cancer immunology, immunotherapy : CII (2020)
Tumor immunogenicity is driven by various genomic and transcriptomic factors but the association with the overall status of methylation aberrancy is not well established. We analyzed The Cancer Genome Atlas pan-cancer database to investigate whether the overall methylation aberrancy links to the immune evasion of tumor. We created the definitions of hypermethylation burden, hypomethylation burden and methylation burden to establish the values that represent the degree of methylation aberrancy from human methylation 450 K array data. Both hypermethylation burden and hypomethylation burden significantly correlated with global methylation level as well as methylation subtypes defined in previous literatures. Then we evaluated whether methylation burden correlates with tumor immunogenicity and found that methylation burden showed a significant negative correlation with cytolytic activity score, which represent cytotoxic T cell activity, in pan-cancer (Spearman rho = - 0.37, p < 0.001) and 30 of 33 individual cancer types. Furthermore, this correlation was independent of mutation burden and chromosomal instability in multivariate regression analysis. We validated the findings in the external cohorts and outcomes of patients who were treated with immune checkpoint inhibitors, which showed that high methylation burden group had significantly poor progression-free survival (Hazard ratio 1.74, p = 0.038). Overall, the degree of methylation aberrancy negatively correlated with tumor immunogenicity. These findings emphasize the importance of methylation aberrancy for tumors to evade immune surveillance and warrant further development of methylation biomarker.
Keyphrases
- genome wide
- dna methylation
- papillary thyroid
- risk factors
- public health
- metabolic syndrome
- type diabetes
- childhood cancer
- high resolution
- squamous cell carcinoma
- machine learning
- young adults
- mass spectrometry
- electronic health record
- single cell
- deep learning
- newly diagnosed
- big data
- lymph node metastasis
- high density
- protein kinase