Genomics research into cancer has already yielded much, but genetic code only tells us so much. With new advances in epigenetics and the mapping of the epigenome, new information is set to greatly improve our understanding of this most complex group of diseases. Shea takes a look at some of these developments while reviewing Cristian Tomasetti and Bert Vogelstein’s recent Science paper.

Epigenetics deals with the biochemical processes ‘above’ or ‘before’ the genes which regulate the expression of genes in the genome, usually in response to influences in the immediate environment. What is particularly intriguing about epigenetics is how much it blurs the traditional boundaries we have erected between our genes and our environments. The novel complications introduced by epigenetics suggest—if not require—commensurately novel ways of conceptualizing our relationships with ourselves and with our environments that transcend this conventional dichotomization.

The unique nexus of genetics and the environment presented by epigenetics is of considerable practical relevance for diseases such as cancer which have as yet defied understanding via existing approaches that dichotomize genetics versus the environment (or vice versa). In this context, one major purpose of this article is to provide a brief survey of how the conventional emphasis on genetics in cancer research is being extended and empowered by epigenetics to perhaps finally realize the muchanticipated promise of cancer genomics.


The unique perspective from epigenetics in cancer as compared to the more conventional dichotomy of genes versus the environment are particularly noticeable in the discussions of a recent paper by Cristian Tomasetti and Bert Vogelstein on the causes of cancer which has already generated a significant amount of controversy.

In this paper the main question the authors attempt to answer is why there is such a disparity in the incidence of cancer between different kinds of tissues—e.g., as the authors note, “the lifetime risk of being diagnosed with cancer is 6.9% for lung, 1.08% for thyroid, 0.6% for brain and the rest of the nervous system, 0.003% for pelvic bone and 0.00072% for laryngeal cartilage,” just as “cancer risk in tissues within the alimentary tract can differ by as much as a factor of 24 [esophagus (0.51%), large intestine (4.82%), small intestine (0.20%), and stomach (0.86%)].” These disparities in risk of cancer between tissue types has been recognized for more than a century, but have not yet been reducible to either hereditary or environmental factors, which until now have been the only ways to parse the causes of cancer.

Building on Vogelstein’s previous pioneering work in somatic mutation, or mutational changes in cells’ DNA that are not passed along via the germ line but which occur during a person’s life, Tomasetti and Vogelstein hypothesized that the relative incidences of cancer in different kinds of tissues could be caused by random mistakes when DNA is copied during cell division. In other words, the more times cells in a particular tissue type divide, the more opportunities for such copying errors to occur, the greater the risk of cancer.

However, to test this idea Tomasetti and Vogelstein needed a way to assess the rates of cell division of different kinds of tissues. Because only stem cells (versus differentiated cells) live long enough to initiate a tumor, Tomasetti and Vogelstein plotted the rates of stem cell divisions of the 31 tissue types for which the rates of stem cell divisions are known against the lifetime risk for cancer for each type of tissue on a log-log axis, predicting that “there should be a strong, quantitative correlation between the lifetime number of divisions among a particular class of cells within each organ (stem cells) and the lifetime risk of cancer arising in that organ.”

As shown in the figure below, there is a clearly noticeable relationship between these two very different measures. Tomasetti and Vogelstein report a strong positive correlation (0.80) between the lifetime risk of cancer and the number of stem cell divisions for a particular tissue type. From this correlation, the authors thereby conclude that around two-thirds of the variation in cancer risk between tissue types can be explained by the total number of stem cell divisions unique to that tissue.

To distinguish this stochastic cell division from external environmental and heredity causes, Tomasetti and Vogelstein construct an “extra risk score” (ERS) as a function of lifetime risk and the total number of cell divisions (log10 values). Utilizing machine learning methods and unsupervised classification, the 31 cancers clustered into two groups, high ERS (9) and low ERS (22): the higher the ERS (basically, the higher the risk of cancer relative to the number of stem cell divisions), the more likely are external environment factors to play a role. The authors found that the high ERS cancers were those with known links to specific environmental or hereditary risk factors, with the low ERS cancers being more likely to be caused by these stochastic errors during DNA replication.Shea Robinson diagram

These findings are particularly noteworthy for a couple of reasons. First, because before now the term “environmental” in cancer epidemiology has been used to denote anything not hereditary, such that these kinds of developmental processes had been “grouped with external environmental influences in an uninformative way.” Now these stochastic errors in DNA replication can be distinguished from external environmental factors. Second, because these non-hereditary genetic causes were found to contribute more to cancer risk than either hereditary or external environmental factors. This is important because, as reiterated by Tomasetti in a follow-up interview with Science, “if you go to the American Cancer Society website and you check what are the causes of cancer, you will find a list of either inherited or environmental things. We are saying two-thirds is neither of them.”


What are the implications of this identification of a third way by Tomasetti and Vogelstein, and how is it related to epigenetics? To the first point, as explained by one of the reviewers of the Tomasetti and Vogelstein paper, it is—or should be—common knowledge that even though the somatic mutations identified by Tomasetti and Vogelstein are legitimately genetic phenomena, they “are not in the germ line…are not transmitted from parents to offspring…don’t generate family risk correlations [and therefore] can’t be found by GWAS or other studies based on sequencing inherited genomes.” This reviewer also describes how it is—or should be— common knowledge that “environmental or life-history risk factors, like diet or reproductive history and so on,” can affect the risk of mutations identified by Tomasetti and Vogelstein, but that because this exposure “has to affect a cell in a given tissue and in a particular relevant gene being used by that tissue,” the net effect of these mutagens, and hence their predictability, is usually very small. In the end, for this reviewer the Tomasetti and Vogelstein paper uses new data but doesn’t show much that wasn’t already understood; perhaps the most salient point of this paper is how it demonstrates that “the love affair with inherited genotypes, enabled, encouraged, and funded by a variety of enthusiasms, opportunities, and vested interests, has distracted attention from working from what we knew.”

However, this point about the effects of age on the rate of somatic mutation is what opens the door for an epigenetic explanation of Tomasetti and Vogelstein’s results. Although Tomasetti and Vogelstein do not explicitly identify the epigenetic components of their findings as such, the copying errors which are such an important component of their model likely have epigenetic causes. This oversight is more than a little curious as Vogelstein has been a central figure in cancer epigenetics from its very beginning.

To explain how this might work, the reviewer from before goes on to identify a very clear environmental factor related to cancer risk not addressed by Tomasetti and Vogelstein in their model: “If mutations arising by chance during cell division ultimately lead to transforming genotypes in some cells, the longer one lives the more likely such changes are likely to arise in at least one such cell in the person. This is generally why most cancer rates rise with age in ways correlated with rates of cell division…That is environmental causation, even if indirect!” This oversight about the causal influence of age, “though it won’t change the empirical fact that neither inherited genotypes nor most environmental exposures do not have highly predictive effects,” suggests that Tomasetti and Vogelstein missed something important.

There are a number of recent papers published on the connections between DNA methylation and aging which have relevance for this proposed connection between somatic mutations and cancer. In particular, a 2013 paper by Steve Horvath describes his discovery of a highly accurate epigenetic clock based on DNA methylation age as a measure of the cumulative effect of an epigenetic maintenance system which predicts not the age of the cells but of the person the cells inhabit. The median error of this clock is 3.6 years, which means it can predict the age of half the donors to within 43 months for a broad selection of tissues. Horvath also analyzed 6,000 cancer samples of 20 cancer types, all of which showed significant age acceleration, except for “a significant negative relationship between age acceleration and the number of somatic mutations.” Subsequent studies have also found an advanced methylation aging rate in tumor tissue, and that DNA methylation-derived measures of accelerated ageing predict mortality independently of health status, lifestyle factors, and known genetic factors.

That epigenetics could be playing such a significant role in this longstanding puzzle about the disparity between the cancer risks of different tissue types, is intriguing. These results are preliminary at best, but quite suggestive of the profound role of epigenetics in cancer. Tomasetti and Vogelstein provided one important piece by identifying the role of stem cell divisions in risk of cancer. The next step is suggested by the connection between DNA methylation, somatic mutation, aging and cancer. The next step remains to be seen, but with the recent release of the first full mapping of the human epigenome, new developments are likely to come even more frequently.



Lima, S. C., Hernandez-Vargasl, H., & Hercegl, Z. (2010). Epigenetic signatures in cancer: Implications for the control of cancer. Current Opinion in Molecular Therapeutics, 12(3), 316-324;

Verma, M. (Ed.). (2015). Cancer Epigenetics: Risk Assessment, Diagnosis, Treatment, and Prognosis. Humana Press;

Ling, H., Vincent, K., Pichler, M., Fodde, R., Berindan-Neagoe, I., Slack, F. J., & Calin, G. A. (2015). Junk DNA and the long non-coding RNA twist in cancer genetics. Oncogene 34(8).

Vad-Nielsen, J., & Nielsen, A. L. (2015). Beyond the Histone Tale: HP1α Deregulation in Breast Cancer Epigenetics. Cancer biology & therapy, (Just accepted for publication).

Tomasetti, C. and B. Vogelstein (2015). Variation in cancer risk among tissues can be explained by the number of stem cell divisions. Science 347(217), 78-81.

Couzin-Frankel, J. (2015, January 13). Bad luck and cancer: A science reporter’s reflections on a controversial story. Retrieved February 23, 2015, from http:// bad-luck-and-cancer-science-reporter-sreflections-controversial-story

Weiss, K. (2015, January 5). Is cancer just bad luck? Part II. It’s a genetic, but usually unpredictable, disease. Retrieved February 20, 2015, from http://ecodevoevo.blogspot. com/2015/01/is-cancer-just-bad-luck-partii-its.html.

Feinberg, A. P., & Vogelstein, B. (1983). Hypomethylation distinguishes genes of some human cancers from their normal counterparts. Nature, 301(5895), 89-92.

Weiss, K. (2015, January 5). Is cancer just bad luck? Part II. It’s a genetic, but usually unpredictable, disease. Retrieved February 20, 2015, from http://ecodevoevo.blogspot. com/2015/01/is-cancer-just-bad-luck-partii-its.html.

Horvath, S. (2013). DNA methylation age of human tissues and cell types. Genome biology, 14(10), R115. Hannum, G., Guinney, J., Zhao, L., Zhang, L., Hughes, G., Sadda, S., … & Zhang, K. (2013). Genome-wide methylation profiles reveal quantitative views of human aging rates. Molecular cell, 49(2), 359-367.

Marioni, R. E., Shah, S., McRae, A. F., Chen, B. H., Colicino, E., Harris, S. E., … & Deary, I. J. (2015). DNA methylation age of blood predicts all-cause mortality in later life. Genome Biol, 16(1), 25.