Aberrant DNA methylation patterns in microsatellite stable human colorectal cancers define a new marker panel for the CpG island methylator phenotype.
Claudia GebhardRoger Mulet-LazaroDagmar GlatzLucia Schwarzfischer-PfeilschifterPeter SchirmacherJochen GaedckeWilko WeichertEdith ReuschelWolfgang DietmaierMichael RehliPublished in: International journal of cancer (2021)
A distinct group of colorectal carcinomas (CRCs) referred to as the "CpG island methylator phenotype" (CIMP) shows an extremely high incidence of de novo DNA methylation and may share common pathological, clinical or molecular features. However, there is limited consensus about which CpG islands (CGIs) define a CIMP, particularly in microsatellite stable (MSS) carcinomas. To study this phenotype in a systematic manner, we analyzed genome-wide CGI DNA methylation profiles of 19 MSS CRC using methyl-CpG immunoprecipitation (MCIp) and hybridization on 244K CGI oligonucleotide microarrays, determined KRAS and BRAF mutation status and compared disease-related DNA methylation changes to chromosomal instability as detected by microarray-based comparative genomic hybridization. Results were validated using mass spectrometry analysis of bisulfite-converted DNA at a subset of 76 individual CGIs in 120 CRC and 43 matched normal tissue samples. Both genome-wide profiling and CpG methylation fine mapping segregated a group of CRC showing pronounced and frequent de novo DNA methylation of a distinct group of CGIs that only partially overlapped with previously established classifiers. The CIMP group defined in our study revealed significant association with colon localization, either KRAS or BRAF mutation, and mostly minor chromosomal losses but no association with known histopathological features. Our data provide a basis for defining novel marker panels that may enable a more reliable classification of CIMP in all CRCs, independently of the MS status.
Keyphrases
- dna methylation
- genome wide
- copy number
- gene expression
- mass spectrometry
- single molecule
- wild type
- machine learning
- endothelial cells
- air pollution
- high grade
- high resolution
- deep learning
- multiple sclerosis
- single cell
- risk factors
- ms ms
- electronic health record
- young adults
- big data
- metastatic colorectal cancer
- circulating tumor
- circulating tumor cells