Open Access Highly Accessed Open Badges Research article

A genome-wide DNA methylation study in colorectal carcinoma

Muhammad G Kibriya1*, Maruf Raza2, Farzana Jasmine1, Shantanu Roy1, Rachelle Paul-Brutus1, Ronald Rahaman1, Charlotte Dodsworth1, Muhammad Rakibuz-Zaman3, Mohammed Kamal2 and Habibul Ahsan1456

Author Affiliations

1 Department of Health Studies, The University of Chicago, Chicago, IL 60637, USA

2 Bangabandhu Sheikh Mujib Medical University (BSMMU), Dhaka 1000, Bangladesh

3 Columbia University and University of Chicago Research Office in Bangladesh, Dhaka, Bangladesh

4 Department of Human Genetics, The University of Chicago, Chicago, IL 60637, USA

5 Department of Medicine, The University of Chicago, Chicago, IL 60637, USA

6 Comprehensive Cancer Center, The University of Chicago, Chicago, IL 60637, USA

For all author emails, please log on.

BMC Medical Genomics 2011, 4:50  doi:10.1186/1755-8794-4-50

Published: 23 June 2011



We performed a genome-wide scan of 27,578 CpG loci covering 14,475 genes to identify differentially methylated loci (DML) in colorectal carcinoma (CRC).


We used Illumina's Infinium methylation assay in paired DNA samples extracted from 24 fresh frozen CRC tissues and their corresponding normal colon tissues from 24 consecutive diagnosed patients at a tertiary medical center.


We found a total of 627 DML in CRC covering 513 genes, of which 535 are novel DML covering 465 genes. We also validated the Illumina Infinium methylation data for top-ranking genes by non-bisulfite conversion q-PCR-based methyl profiler assay in a subset of the same samples. We also carried out integration of genome-wide copy number and expression microarray along with methylation profiling to see the functional effect of methylation. Gene Set Enrichment Analysis (GSEA) showed that among the major "gene sets" that are hypermethylated in CRC are the sets: "inhibition of adenylate cyclase activity by G-protein signaling", "Rac guanyl-nucleotide exchange factor activity", "regulation of retinoic acid receptor signaling pathway" and "estrogen receptor activity". Two-level nested cross validation showed that DML-based predictive models may offer reasonable sensitivity (around 89%), specificity (around 95%), positive predictive value (around 95%) and negative predictive value (around 89%), suggesting that these markers may have potential clinical application.


Our genome-wide methylation study in CRC clearly supports most of the previous findings; additionally we found a large number of novel DML in CRC tissue. If confirmed in future studies, these findings may lead to identification of genomic markers for potential clinical application.