March 16, 2012 — Editor's note: Cancer genomics is at the frontier of genomic medicine. Since the first complete cancer genome — acute myeloid leukemia (AML) — was sequenced in 2008, researchers have sequenced the genomes of many human cancers, including breast cancer, glioblastoma multiforme, lung cancer, melanoma, prostate cancer, and a broad spectrum of pediatric cancers. A current avenue of research is the way a cancer genome changes over the course of therapeutic treatment.
In an interview with Medscape Medical News, Elaine R. Mardis, PhD, professor of genetics and molecular microbiology, and codirector of The Genome Institute at Washington University in St. Louis, Missouri, discusses highlights, including her own presentation on cancer genome evolution, of the recent Future of Genomic Medicine (FGM) V. The meeting held March 1 and 2 in La Jolla, California, and was sponsored by the Scripps Translational Science Institute.
Medscape: How was the FGM V meeting?
Dr. Mardis: Great! I'm always glad to go to that meeting because it's so progressive in terms of the way people are thinking about medicine and using new technology toward better diagnosis. Really, a new philosophy about medicine in general is what I love hearing more about. It's not just cancer-specific, which is what I was there to talk about, but has very broad-reaching implications for the way patients are diagnosed and treated.
Medscape: What is the FGM philosophy?
Dr. Mardis: The emphasis is on what I call an "n of 1." Each patient is looked at as an individual, rather than as something that you can compartmentalize into a "what you know" generality. That was emphasized this year by the book that Eric Topol, the organizer of the conference, recently published: The Creative Destruction of Medicine. With new technologies, physicians can now start looking at patients as an n of 1, rather than trying to fit them into a description that fits preconceived notions of how to diagnose patients. It expands the data or information that a physician has access to, in terms of getting the diagnosis right the first time.
I have been at the last 3 meetings, and I think the change has been pretty remarkable. When I first spoke there in 2010, we had already done sequencing on the whole cancer genome, so that wasn't new. But we were just starting to expand into clinical trial samples. We could glean so much more information from the samples than ever before, because of the clinical data. The changes over that 3-year period have been remarkable, in that the technology is becoming more mature and the bioinformatics approaches to analyzing the data — across disease types — are becoming much more mature. Overall, I think that's what has led to the sea change in terms of how we make medicine more personalized.
Medscape: What did you discuss in your presentation on the cancer genome evolution?
Dr. Mardis: I described how we're taking the digital nature of next-generation sequencing and expanding that to understand and characterize the heterogeneity of the disease, or how the disease changes over the course of therapeutic treatment.
For example, in acute myeloid leukemia, we compared the initial presentation that a patient has and the relapse of that disease. We showed that different patients present with more or less heterogeneous disease. But once they go through chemotherapy, the relapsed population of cells in their tumor is genetically related to the initial presentation of cells in their tumor. This really addresses a major question that's been around for a while: What is the source of the relapse?
We showed that we can actually detect a pattern of mutations posttreatment — especially with DNA-damaging chemotherapy, which is commonly used in multiple types of tumors — that shows a signature of DNA damage that is really ascribable to the chemotherapy itself.
Medscape: Is the DNA damage so specific that with one type of chemotherapy you might find a certain type of genome in the recurrent cancer cells, and with another type of chemo a different type of cells might make it through?
Dr. Mardis: In this case, the standard of care is actually one type of chemotherapy. So you can look at the difference between patients just before and after they were treated, but in general, everybody gets the same type of DNA damage from chemo because it's the standard of care.
Typically this DNA-damaging reagent, cytarabine, is used in different proportions with daunorubicin, which is not DNA-damaging.
The other type of evolution that I talked about was in breast cancer patients with estrogen-receptor-positive disease. The clinical trial we were sequencing treated every patient with 1 of 3 randomized aromatase inhibitors. Because this is estrogen-receptor-positive disease, aromatase inhibitors are used to remove estrogens and their 4 major stimuli to the cancer cells from the system. But we know that not all women respond to aromatase inhibitors, so our goal with this study was to understand the mechanisms by which that could be determined in advance, rather than after the fact.
We use a cell-proliferation assay that's done in pathology, called Ki67, which is a protein that indicates the level of proliferation in the cells. Patients who show a decrease in that Ki67 index are patients who are responding — their tumor cells are dying. What we see is a clear pattern; if we sequence before and after the aromatase inhibitor using our whole-genome method, it's clear that the heterogeneity of that tumor is decreasing over time. You see clear evidence of subpopulations of cells that are no longer present in the posttreatment biopsy. The contrast to that is patients who are resistant, where you basically see no difference between pre- and posttreatment levels.
We're now working to provide real-time information to oncologists who are treating cancer patients. We sequence a patient's whole genome, tumor, and normal tissue. We also do exome capture, and then we sequence RNA, an RNA-seq sample from the tumor only (RNA-seq sequences complementary DNA to obtain information about the RNA content of a sample). If we combine those data, we can answer fundamental questions about whether there are druggable mutations, what proportion of the cells contains those druggable mutations, and then what the RNA signature is telling us.
The 2 patients that I profiled were diagnosed successfully through RNA-seq, and have responded to the drug indication that we got from the RNA-seq data. However, at the genome level, there is no clear evidence of why they were overexpressing the genes that were ultimately druggable. This is our new area of focus. We feel that RNA-seq data are going to provide a really valuable comparator in this type of analysis for patients.
Medscape: How much will funding affect genomic medicine in the next 5 years?
Dr. Mardis: Aside from the cost of sequencing a genome or an exome, that's a general concern overall. People are struggling in the extreme with how to compensate for funding that's being lost. It will obviously become part of the equation. It's a strange tension, because there are all these new sequencing instruments that have just been announced that are going to be a reality during this calendar year. We have to make hard decisions like we've never had to before about what to buy. All these manufacturers are working very hard to get the next latest, greatest thing out to us. We can now produce a lot of data in a shorter period of time, but I don't think we're going to be analyzing it any faster. In a tight funding environment, you're now faced with the problem of having difficulty finding enough samples, for example. Sequencing is that much cheaper, but then will your machines sit idle because you don't have enough samples to sequence?
Dr. Mardis reports that her research is supported by government funding, and that she uses Illumina sequencers in her work and occasionally gives a talk for Illumina, for which she receives a small honorarium.