DNA Methylation Levels Measured Globally And At LINE-1 DNA Repetitive Elements In Pre-Diagnosis Peripheral Blood-Derived DNA And Risk Of Colorectal Cancer
Dallas ENGLISH, University of Melbourne, Australia
KARAHALIOS A. 1
, JOO E. 3
, WONG M. 3
, WALTERS R. 4
, ROSTY C. 5
, CLENDENNING M. 3
, MAKALIC E. 1
, SCHMIDT D. 1
, JUNG C. 6
, DUGUE P. 2
, WILLIAMSON E. 7
, HOPPER J. 1
, MILNE R. 1,2
, SOUTHEY M. 3
, GILES G. 1,2
, BUCHANAN D. 1,3
1 School of Population and Global Health, University of Melbourne
2 Cancer Epidemiology Centre, Cancer Council Victoria
3 Department of Pathology, University of Melbourne
4 QIMR Berghofer Medical Research Institute
5 University of Queensland
6 VLSCI, University of Melbourne
7 London School of Hygiene and Tropical Medicine
Purpose: The aim of this study was to associations between colorectal cancer (CRC), overall and by molecular subtype, and DNA methylation measured at LINE-1 DNA repetitive elements and globally from the mean CpG methylation levels of the Illumina HumanMethylation450k array.
Methods: A nested case-control study was conducted within the Melbourne Collaborative Cohort Study. Cases were diagnosed with CRC during follow-up (until 2010). Controls were selected using risk set sampling with age as the time variable, and matched by sex, year of baseline attendance, country of birth and baseline sample type for DNA (Guthrie card, lymphocytes, buffy coat). Tumours were characterised for microsatellite instability (MSI) and CpG island methylator phenotype (CIMP). Bisulphite converted DNA from baseline blood samples was assayed for LINE-1 methylation using pyrosequencing and at more than 485,000 CpG sites across the genome using the Illumina HumanMethylation450k Beadchip array. Global methylation was estimated by calculating the mean beta values across all CpG sites. We estimated odds ratios (ORs) and their 95% confidence intervals (CI) per standard deviation increase using conditional logistic regression with adjustment for potential confounding variables.
Results: DNA methylation values were obtained for 824 CRC cases and 824 matched controls (50.6% female; mean age at blood draw=59.8 ± 7.6 years). No association with CRC risk overall was observed for mean global methylation levels or for LINE-1 (OR=0.93; CI=0.76-1.14; p=0.49 and OR=0.86; CI=0.67-1.12; p=0.23, respectively). However, when stratified by CIMP status, increasing levels of LINE-1 methylation were associated with a 2-fold increased risk of CIMP positive CRC (OR=2.34; CI=1.02-5.35; p=0.04). A weaker, but consistent, association was observed for CIMP positive CRC in regard to mean global methylation (OR=1.67; CI=0.89-3.12; p=0.11).
Conclusions: Higher levels of LINE-1 methylation measured in peripheral blood-derived DNA were associated with CIMP-positive CRC.
Funding source: NHMRC (1027505)