DNA methylation profiling: a promising tool and a long road ahead for clinical applications

Biology is navigating one of its most exciting times. Elucidating the genetic basis of heredity opened a new era, and directed significant resources towards understanding the link between mutations and phenotypes. Almost in parallel, multiple lines of evidence revealed the existence of heritable factors that are not encoded in the nucleotide sequence and make important contributions to health and disease. These factors became the focus of a rapidly expanding field, epigenetics, which studies gene expression and phenotypic changes mediated by mechanisms that do not change the sequence of the DNA, and include DNA methylation, histone post-translational modifications, and small interfering RNA-mediated silencing pathways.

Non-genetic influences explain why only 5–10% of all cancers are currently attributed to mutations, while 90–95% are linked to environmental factors and lifestyle, which act by epigenetic mechanisms (1). Epigenetic changes are becoming clinically relevant in multiple ways that, until recently, seemed unimaginable. Ushijima and Sasako (2) reported that in gastric cancers, three genes, CDKN2A (p16), CDH1 (E-cadherin), and hMLH1, are more frequently inactivated by aberrant methylation than by genetic mechanisms such as mutations or deletions. Tanemura et al. (3) revealed for the first time an association between CpG methylation and cutaneous melanoma progression, and Nobeyama et al. (4) found that tissue factor pathway inhibitor-2 (TFPI-2), a protein that suppresses malignant melanoma invasiveness, was methylated in 29% of metastatic lesions but in none of the primary tumours examined.

As more details are being unveiled about epigenetic modifications in malignant tumours, the possibility to exploit them for developing biomarkers, indicative of very early modifications during malignancy, emerges as an attractive option. In the current issue of the IJCP, Luo et al. (5) conduct a meta-analysis to examine the feasibility of using DNA methylation biomarkers for detecting colorectal cancer or adenomas. The authors included 19 studies published between 2004 and 2009 that were conducted in 10 countries or regions and enrolled 2356 patients. Thirteen studies examined methylation of a single gene, and six studies examined multiple genes. Hypermethylated genes occur frequently in colorectal carcinoma, and the possibility to detect them in the stool promises to provide biomarkers with diagnostic, therapeutic, and prophylactic applications. With an overall 62% sensitivity and 89% specificity for detecting colorectal cancer, Luo et al. note that DNA methylation biomarkers have a similar or higher diagnostic accuracy than faecal occult blood testing or the carcinembryogenic antigen. While this high accuracy is promising, and as the authors acknowledge, hypermethylated gene panels are not yet accurate enough to confirm or exclude malignancy or adenoma when used in isolation. Their strength, at present, is when they are used in combination with other tests, and the discovery and inclusion of additional biomarkers that would improve sensitivity and specificity represent an important research goal.

For several malignant tumours, epigenetic modifications were shown to precede histological changes, providing one of the earliest markers (6–9). Promoter hypermethylation was described in breast cancer, not only in the tumour itself but sometimes extending as far as 4 cm into the surrounding histologically normal tissue and also in the contralateral, unaffected breast (10). In addition to providing a non-invasive and early-stage biomarker, DNA methylation profiling has an additional, unique advantage: unlike mutations, epigenetic changes are reversible, and the possibility to therapeutically manipulate them promises a rational approach for prophylactic interventions based on molecular modifications that precede histological changes.

Studies at the interface between nutrition and epigenetics provide a powerful exemplification of the possibility to exploit the reversible nature of epigenetic marks. Exposure of pregnant mice to the synthetic estrogen bisphenol A (10 mg/kg of body weight/day) caused stable hypomethylation at several CpG dinucleotides in the foetal epigenome and changed the offspring fur colour towards yellow. This colour change was reversible by supplementing the maternal diet with methyl donors or with the phytoestrogen genistein (11). In addition, it was recently shown, for the first time, that in utero genistein supplementation changes gene expression by epigenetic mechanisms and decreases susceptibility to obesity once the mice reach adulthood (12).

A significant learning point is the necessity to integrate genetic and epigenetic contributions into our understanding of health and disease. We got used to the idea that mutations shape everything, which is far from the truth. The extent of epigenetic influences is illustrated by Fraga et al. (13) who found, in a comparison on monozygotic twins, increasingly divergent DNA methylation and histone acetylation patterns with age, which made the almost identical gene expression profiles of three-year-old twin pairs become very dissimilar by the time they reached age fifty. Thus, even in genetically identical individuals, the phenotype is differentially sculptured, in time, by epigenetic marks. Epigenetics promises to help design and implement preventive strategies and guide treatment, and it harbours a wealth of resources for every medical field. Taking a step back from the trees, to admire the forest in its splendor and complexity, we should appreciate the additional layer that epigenetics, alongside genetics, brings into shaping gene expression and the phenotype, and the numerous tools that already started to enter the clinical arena. Of the four epigenetic drugs recently approved by the FDA, which appear to be less toxic than conventional chemotherapy, two compounds, 5-azacitidine and decitabine, approved in May 2004 and 2006 respectively, demethylate CpG sequences in the DNA and correct abnormal methylation and gene expression patterns associated with disease (14,15). Epigenetics changed some of our long-held beliefs, and once again confirms, as Vera Rubin so eloquently remarked, that [s]cience progresses best when observations force us to alter our preconceptions.

Disclosures

None.