I started as a medical student. I wanted to do some work related to cancer research. Later, in 1993, I started as a fellow in Steve Baylin’s lab, and we were looking at methylation changes. This was a time when many tumor-suppressor genes were being cloned and identified. We began to investigate whether these methylation changes could be involved in classic tumor-suppressor genes. We started looking at Von Hippel-Lindau genes in renal cancer and found that the gene was silenced.

Up to that point we looked at methylation changes in tumor suppressor genes in different cancers. We were really interested at that time in looking at precancerous lesions, at the early stages of cancer, and looking at cancer progression of normal tissue into fully malignant cells. We’d done all the work previously in full-blown cancers and we simply wanted to study earlier lesions. The problem was that we really needed an assay to look at possible methylation in small lesions with very little DNA and sometimes in DNA from a mixture of cells. It was out of necessity that we came up with this technique. We needed something sensitive and specific to methylation changes, so we had to go out and invent it.

I guess it was believing that something that simple would actually work.

The previous assays for doing this were based on restriction enzymes, and not on the bisulfite method. We had been trying to use restriction-based methods and we couldn’t get them to be reproducible. And we were then working with other bisulfite genomic sequencing, which had been in use for a couple of years. That technique treats DNA with bisulfite and that causes sequence changes you can determine by sequencing analysis. The methylation-specific PCR was a way of using those sequence changes to directly determine methylation changes. We would put primers that would recognize sequence differences caused by methylation patterns, and they would be in whatever the gene of interest was.

I don’t know if it was easy. It was simple.

To some degree, but not that it would be used as much as it is currently. I think we thought it would be a new way to look at things. It certainly made things a lot easier. The fact that we could quickly look at methylation changes in different genes and different tumor samples in these early lesions was really encouraging. But we were not thinking about a lot of the possible applications at the time.

The main one is the potential to use this in molecular detection of cancer, in early detection. That wasn’t something we were initially thinking about.

We were just getting comfortable with the technique, and we realized how sensitive it was for picking up a very small amount of methylation in samples. That’s really what you need for a molecular detection approach to cancer—you need very sensitive and specific detection. And that idea evolved from looking at early lesions to looking at body fluids—in blood and in sputum—for evidence of cancer. It wasn’t a complete shift from what we were doing, but rather a recognition that this would be another potential application.

Mainly we have to get a population at risk for cancer and then we have to find out whether the test can predict the cancer early enough to intervene. We are doing that in some different cancers. But the limiting factor is really getting large enough patient populations and then following up on them.

Initially we were able to detect one in 1,000 cells. But subsequently it’s been improved to one in 50,000 cells or even one in 100,000. We’ve done some recent studies where we’ve picked up a single DNA copy.

We were able to determine that when we got a positive signal that was originally from a single allele of DNA, a single copy.

It’s a very good journal. It’s widely read in a broad scientific area. I guess we thought it was the best match of exposure and novelty. If you look at the higher-profile journals, Nature and Science and so forth, usually you’re dealing with some biological question, and this was purely a technique paper. We were establishing a new technique. So PNAS was the first journal we sent it to.

I think we will continue to identify new genes and new pathways involved in cancer that are silenced by methylation changes. That’s already happening, and I think it will continue. I think the screening approach will be tested in large populations and really vigorously examined to see whether it will be useful for early detection of cancer. So diagnostic uses will be further tested and, I hope, validated. We’ll see.

I guess the most unexpected thing was how prevalent some of these changes in methylation turned out to be. When we started looking at gene silencing in cancer, the understanding at the time was that genetic changes were the most important changes in cancer, and they are clearly still very important. When we first started studying genes for methylation changes, we thought we’d find a few genes that might show this alteration and maybe just a few cancers. Now that list has just exploded. The surprise has been how prevalent the silencing of genes in cancer really is. That’s probably why some of these papers are so highly cited. People continue to expand some of the initial findings and continue to find this mechanism in many different tumor systems and many different genes.

I don’t even know the number—maybe 50 to 100. In terms of the type of cancers, I think every type of cancer has been shown to have methylation changes in different genes.