IGF2 DMR0 hypomethylation has been suggested as a surrogate biomarker for LOI with the potential to predict inherent predisposition to cancer. DNA methylation studies are more robust and quantitative compared to RNA expression studies and therefore amenable to high throughput analysis or use as clinical biomarkers. In this study, we have evaluated IGF2 methylation levels at the DMR0 region in DNA from breast and colorectal cancer tissue biopsies, as well as in large cohorts of peripheral blood DNA samples from cross-sectional and prospective populations using a well validated quantitative pyrosequencing assay.
We have found that IGF2
DMR0 hypomethylation is more prevalent than LOI and 36% of our colorectal tumour tissue samples with hypomethylation were monoallelic for IGF2
expression. Interestingly, the tumours with LOI had lower methylation levels than those without LOI. Since IGF2
DMR0 is methylated on the paternally inherited allele (36
), it is unlikely that the loss of methylation at this region would have a direct role in reactivation of the silent maternal allele. This is supported by our finding that IGF2
DMR0 hypomethylation and LOI are not tightly linked. However, IGF2
DMR0 hypomethylation was found in 80% of colorectal and 33% of breast cancer tissues with most of the colorectal and breast cancer patients having significantly lower methylation in their tumours compared to adjacent non-tumour tissues. These results suggest that hypomethylation at IGF2
DMR0 in tumour tissue should not be viewed as a surrogate biomarker for LOI but as a frequently identifiable attribute of colorectal cancer.
Our second finding was that in our prospective studies using the EPIC population samples only 9.5% of samples had ≤35% methylation. Based on previous reports in which IGF2 DMR0 hypomethylation in peripheral blood was found to be associated with colorectal cancer and the perception that loss of methylation is likely to be an early event in tumorigenesis, we expected to find a higher incidence of colorectal cancer and possibly also breast cancer in the 10% percentile proportion of the population with hypomethylation. However, this was not the case and hypomethylation of IGF2 DMR0 was present with similar incidences in controls and persons who later went on to develop cancer. This observation indicates that constitutive hypomethylation of IGF2 is actually rare and not a predetermining factor for colorectal or breast cancer.
It has recently been shown that Dnmt3b-mediated methylation in cancer targets specific genes rather than being a stochastic random event (45
). It is possible that demethylation is also targeting specific loci during cancer and that IGF2
DMR0 is one such region. Demethylation in cancer may be passively mediated by failure to maintain methylation during replication or active via DNA repair mechanisms (46
). Age-related DNA methylation changes at the IGF2
locus have been previously reported (48
) and there are also indications that heritable factors may determine methylation levels at this locus (49
). Within the EPIC sample set we found a strong correlation with age and hypomethylation (Table ). In our study, IGF2
DMR0 methylation decreased after 60 years of age. Thus, if genetic factors initially predispose to hypomethylation, environmental or age-related factors probably augment hypomethylation. We have recently reported that the methylation profiles in congenital growth disorders with IGF2
LOI and Wilms tumour cases with IGF2
LOI are not the same (36
). This would suggest that even if similar epigenetic reprogramming mechanisms operate in the germ-line and in cancer tissues the IGF2
DMR sequences respond differently.
The results of our study in the EPIC samples contradict reports that show an increased frequency of LOI in peripheral blood samples of colorectal cancer patients (13
). The peripheral blood samples in these authors' studies came from individuals who were known to have LOI in colon tissue. The difference in our study is firstly that our prospective study utilized blood samples that did not come from the same patients as our tumour samples. Secondly, we looked at IGF2
DMR0 methylation levels as an indicator for cancer predisposition, whereas other studies examined LOI in mRNA. IGF2
expression is very low in peripheral blood samples. RNA analysis by nested PCR in peripheral blood could enrich for a subset of IGF2
expressing cells with LOI that is beyond the range of detection of quantitative pyrosequencing methylation analysis.
Overall, our results show that hypomethylation of IGF2
DMR0 is an acquired cancer-specific epigenetic event. Hypomethylation of IGF2
DMR0 is more frequent than LOI and may be a consequence of LOI as well as other cancer-specific epigenetic-mediated events such as chromosomal and chromatin rearrangements at the locus. It is plausible that the detection of lower levels of methylation in cancer tissue samples reflects a relative sampling effect in tumours with less-differentiated cell populations. Routine single cell analysis of methylation in situ
is not yet feasible, but if loss of methylation at the DMR0 is a feature of less differentiated cells, then pyrosequencing assays for methylation may prove a valuable alternate tool for quantifying less differentiated cells in heterogeneous tumour samples. Because IGF2
DMR0 hypomethylation is so highly prevalent in colon tissue from colorectal cancer patients and rarely present constitutively, it has a value as a diagnostic indicator of colorectal cancer and should be added to the modest list of discriminant methylation markers suitable for cancer screening (50
). Currently, there is no national UK screening programme aimed at reducing colorectal cancer-related deaths. The Danish (54
) and the Nottingham (55
) randomized controlled trials and the colorectal screening pilot study (56
) have shown the efficacy of screening with faecal-occult-blood test (FOBT) in reducing colorectal cancer-related mortality. The PPV of FOBT ranges from 9 to 17% with a sensitivity of ~52% (56
). Such low sensitivity and PPV result in a large number of unnecessary colonoscopies and highlight the need for a test with better predictive value. Our IGF2
DMR0 hypomethylation assay is robust and requires only 100 ng of DNA which makes it suitable as a screening test to complement FOBT. While the clinical application of these results with regard to diagnostic value is clear, detailed analyses of hypomethylation and IGF2
expression levels in subgroups of colorectal tumours, and normal tissue at different stages of disease are still required in order to evaluate the prognostic value of IGF2