Impact of Methylation Alterations in Colon Cancer: Epidemiology and Prognosis

We conducted a preliminary study in 2010 using the innovative Illumina GoldenGate (methylation assay) which has enabled us to characterize the level of methylation of 807 potential markers on a series of 200 adenocarcinomas of the colon (C18) and rectosigmoid junction (C19) resected between 1998 and 2001. The results, validated by pyrosequencing, allowed us to establish a panel of 12 markers to assess the level of methylation. The aim of this project is to validate the prognostic value of this panel in a second cohort of patients with adenocarcinomas (stage I to IV) resected between 2002 and 2006 in public and private hospitals (n=685).

This study relies on an original collection of surgical specimens of colorectal cancers resected in public and private hospitals among patients resident in the département of Côte-d'Or and followed by the cancer registry of Burgundy. Samples obtained from each cancer and from adjacent normal mucosa are stored in liquid nitrogen in the Ferdinand Cabane Biological Resources Centre(certified S 96900). Annotations are collected by the cancer registry staff from multiple sources (pathologists, gastroenterologists, oncologists and radiotherapists).

Pyrosequencing will be used to quantify the level of methylation. The percentage of methylated allele for each marker on cancerous DNA and the ratio of methylated markers to the number of markers analyzed will be determined. The molecular status of MSI, BRAF, KRAS, and PI3K will be taken into consideration.

To study the prognosis, relative and relapse free-survival will be calculated. A multivariate relative survival analysis will be performed to determine independent prognostic factors taking into account the characteristics of patients and tumours and epigenetic and molecular alterations. Epigenetic alterations will be described according to age group, sex and subsite. The epigenetic phenotypes will be evaluated according to already known molecular status using logistic regression models.