It was a review but with a lot of primary data in it as well. In particular, by the late 1990s, we thought that this hypermethylation might be a molecular marker of cancer you could use to look at the DNA—from tumor, or urine, or sputum, or whatever—with diagnostic purposes in mind. This became apparent not only from a basic understanding of how tumor-suppressor genes work and how genes get silenced, but through the constant position of this hypermethylation in the promoter region of these genes, and the ability of investigators to develop very sensitive techniques to detect it. So the idea was that these specific hypermethylated promoter regions could be used as a molecular marker of tumor suppressors, and by putting together tumor panels of four or so hypermethylated genes, we might be able to detect lung cancer, colon cancer, breast cancer, and leukemias. We showed we could get up to 80% coverage of the genome with such panels. In other words, one or more of the markers are likely to be methylated in at least 80% of tumors. Now people are getting closer to 100%. So the idea is that this panel of markers might be utilized in screening for early detection of cancer.

You would take your PCR procedure, such as the one that we reported in a 1996 PNAS paper. You would isolate DNA, say, from the sputum. That procedure is extremely sensitive. If there is one hypermethylated tumor cell in there—one in 1,000 or 2,000 or maybe even 10,000—we’d be able to detect it and that would be the marker. So you look at the hypermethylation of these promoter regions of specific genes. And you look in a specific repertoire of genes for any given tumor type.

There are lots of small observations that have built up over the last several years indicating the tremendous promise of this. It’s going to take bigger studies, though, exploring hypermethylation in large numbers of patients in given situations—such as screening sputum for lung cancer, or screening urine for prostate cancer. All these things are under active exploration from lots of groups. It’s going to be the data from these kinds of studies that will tell us how far things will go.

I think the challenge and excitement of this research is that it spans everything from basic science all the way to clinical possibilities. The basic science dovetails into the exploration of how chromatin, the proteins that interact with DNA, dictates the gene expression pattern that emerges from our genome. This has also been a very exciting, explosive area over the past few years; how the DNA methylation associated with gene silencing is intimately linked to understanding how chromatin dictates gene expression status. So why I am mentioning this? The biggest challenges in the field at the moment are understanding two things: how does DNA methylation participate in gene silencing (and that includes what molecules target the DNA methylation to these promoter regions), and why is the cancer cell doing this? At the molecular level, why is this taking place? What’s happening during tumor progression? Understanding these two processes constitutes the biggest fundamental challenge in the field. That’s what my lab works on a lot.

They’re pretty obvious. For one thing, unlike mutations, which are permanent changes in genes that lead to improper protein readout and function, this is a reversible change. Certainly, experimentally, you can make this gene expression come back on. You can strip away the DNA methylation, and change chromatin from the inactive to the active pattern around these promoters. There’s an entire enterprise growing up around this idea. As with all things related to cancer, however, it’s not so simple. What drugs can specifically do this to a sufficient extent to get these genes back on and actually have a good therapeutic effect or actually work to prevent cancer is an open question. This hypermethylation is a very early change in cancer cells, though, so we’re optimistic that this might be utilized.

Sometimes it does. Definitely in some genes, it’s given us exciting leads to understand the progression of cancer cells. Here’s an example: about 15% of all colon cancers have a phenotype in which you have multiple regions of mutation in and outside of the genes in the genome of those tumors—it’s called microsatellite instability. Researchers first discovered, in families who inherit this form of tumor, that they inherit it because they have mutations in the enzymes that repair mutations that arise during DNA replication. If these repair enzymes are defective, then they build up other mutations. The leading way this happens in the non-familial form of the cancer, in the 15% of these sporadic cancers, is hypermethylation in a promoter of a gene called MLH1, which is a mismatch repair gene. So 15% of the colon cancers from patients have this microsatellite instability, and about 70% of that 15%—the majority—have hypermethylation of MLH1. So here’s an epigenetic event that would set up a genetic series of changes that people think are fundamental to why that type of colon cancer gets started. That’s one example. VHL is another. The hypermethylation and mutations in VHL are very specific, mostly for renal cancer. They clearly play a role. Since families inherit mutations in VHL, we know that methylation and mutations are equally causative in the progression of the cancer. The functional significance of many of these genes is now becoming apparent.

Now that people are devising ways to screen cancer genomes for these hypermethylated genes, we then have to say "OK, if we find them, are they important to function?" One way we can do that is to knock that gene out in a mouse model. We have done that for one gene we found in this kind of screen, and showed that it was important to embryonic development. When one copy of the gene is knocked out in mice, the mice go on to develop tumors later in life as the other copy of the gene becomes hypermethylated.

I think I’ll say what we are likely to know in five years. We will certainly be a lot farther along in understanding the promise of this marker technology. Some people see this promise as very high right now, but we have to test it. In five years we should either really be using these markers and knowing which direction to go with them, or we will have obtained data for whatever reasons that tell us the promise is not as high as we thought. So we will have a much better understanding of the clinical promise of markers. By the same token, we will have drugs that can block DNA methylation, such as 5-azacytidine, which will probably be approved by the FDA for at least one condition, a pre-leukemic syndrome known as myelodysplastic syndrome, or MDS. Patients with MDS have anemia and deficiency of other blood elements, and they have a high incidence of developing leukemia. They don’t have proper function of the bone marrow, and we have known for some time now that they respond to this drug with long-term improvement and normalization of marrow function. The FDA probably will approve 5-azacytidine for use in this syndrome. The challenge will be to find out whether it works through reactivation of genes. If so, it will further open up the question of how efficacious reversal of gene silencing may be for cancer in general. I think in the next few years we will get farther along in understanding that possibility.

That’s what I’m excited about. The research is real and the clinical possibilities are real. And we’re making progress. Over the next five years we will see how far it is likely to go.

Stephen Baylin, M.D.
Johns Hopkins University
Baltimore, MD, USA