Name: Stephen Chanock
Position: Co-director of the NCI Cancer Genetic Markers of Susceptibility project; senior investigator at the NCI; director of the NCI core genotyping facility, 2001-the present
Background: Principal investigator at the NCI, 1991-2001
Education: MD, Harvard Medical School, 1981
The US National Cancer Institute this week released data from a genome-wide association study. The data, believed to be the first GWAS to be released to the public, come from the NCI’s Cancer Genetic Markers of Susceptibility study on prostate cancer.
The NCI said it hopes the data will help researchers identify genetic markers of disease risk and give a boost to drug development — including targeted therapies.
The release of the data comes three weeks after the National Human Genome Research Institute said it would pay the Genetics and Public Policy Center $2 million to find out what ordinary Americans think of such population studies [see PGx Reporter 10/4/2006].
The NCI data are available through the Cancer Biomedical Informatics Grid here.
Begun in February, CGEMS is the largest study of genetic risk factors and is focused on the two most common forms of cancer in the United States — breast and prostate cancer.
In the US this year there will be an estimated 234,460 new cases of prostate cancer, which is the third-leading cause of cancer-related death in men, according to the NCI.
The prostate study population comprises more than 2,200 men — with and without prostate cancer — and constitutes more than 680 million individual genotypes, the NCI said in a statement.
Pharmacogenomics Reporter spoke to Stephen Chanock, co-director of the CGEMS project, to discuss the potential impact of this data on pharmacogenomics research.
How do you see the CGEMS prostate susceptibility data release helping people involved in developing pharmacogenomic treatments and diagnostics?
Well, one of the major purposes of CGEMS was to efficiently and effectively conduct a large, whole-genome-wide scan using SNPs to identify susceptibility regions for prostate cancer — to take this so-called agnostic approach with a whole-genome scan.
In our minds, the idea is to accelerate research by making it available at the earliest possible time, because we know a dataset of this volume and complexity can’t be analyzed by any one group at any one time, for all of its richness.
To come to your specific point, I think the key issue is to identify new regions that had heretofore not been identified by the candidate gene approach, but instead to look across for unexpected regions, as well as unknown regions, where we would hope to identify susceptibility loci that could then be the basis for further interventional or preventive or therapeutic approaches.
So really, this is the first step in discovery of putting together the puzzle of the complex set of genetic contributing factors for prostate cancer.
We’ve done this in a large population-based study — the Prostate, Lung, Colon, Ovarian Screening Trial, run by the NCI — and the other thing that distinguishes our approach quite differently from some of the other publicly described and soon-to-be available studies … is that we have tied the replication scheme to the actual genome scan. So, we did 1,150 men with prostate cancer and 1,150 without, roughly, and we have as many as 7,000 cases and 7,000 controls that we are going to sequentially work our way through, so that at the end, we will have ferreted out the false positives. [That] is really the challenge of this field for looking for low-penetrance, high-frequency alleles.
So, it’s a discovery tool that I think the leadership at the NCI felt very strongly that it was in the best interest to make this available in the soonest possible time. So, by posting this, the initial scan results are now available, and then as subsequent studies are done, executive summaries of those scans showing the top 25,000 SNPs will be carried through a series of additional studies. Then we’ll be able to say these are the 10 or 50 or 100 or 200 [SNPs] — we just don’t know — that look to be really robust, and meet certain criteria for confirmation.
Whether it’s commercial or academic or institutional groups, they’ll be able to say, ‘Alright, we’re going to go try and figure out what this means, and perhaps use this as a first step toward a targeted therapy or a set of clinical diagnostics.’
I see a possibility for a pharmacogenomic approach with targeted therapies. Is there any reason that you might pick up on some drug-response-related loci as well?
That’s a terrific question. The study that we’re looking at comes out of a screening trial, so it will be some time before that information will have matured — that enough people will have been followed long enough that we could make observations about certain therapies in prostate cancer.
We’re very interested in pursuing this, but I think in the initial phase, the study has not gone long enough to reach all those endpoints. If you took a study that was developed in the early ‘80s or the late ‘70s, and you had a 15-year follow-up, you could say, ‘What loci are going to be important for survival?’ Then you’d look and segregate those who had radiation therapy versus those that didn’t have radiation therapy; or those who received androgen ablation, and ask the questions related to specific therapeutic approaches.
I think the real value of CGEMS right now — if you’re talking about your 2006 readership — is really on the discovery side of identifying new pathways and novel changes that can take place in prostate cancer. This is driving the discovery side of the process.
Recently, the US National Institutes of Health released a request for information related to genome-wide association studies. Are there specific privacy issues that might affect access to data?
I’ve been part of that here at NIH. … There’s a public dialogue going on right now about the GWAS. So, the key issue is having the dialogue with many different sectors of the American public. That information will all be configured in this final set of NIH guidelines that will be developed sometime in 2007. And that addresses issues of confidentiality, databases, data access — those kinds of very important questions.
I think GWAS has certainly raised those issues, and all possible steps are being taken to ensure the confidentiality and privacy of the individual, as well as the sanctity of the [institutional research board], under which samples have been collected, and that’s a key issue as well.
There are going to be different degrees of data access. So, if there’s a pharmaceutical company that’s interested in exploring something, they should be able to have access to some of these large public events. But it will be in a restricted way, in that you can only look at so many variables at a given time — to protect the confidentiality of the individual, you have to be sure that someone doesn’t take enough individual fields or outcomes and start crossing them, saying, ‘Oh, we can figure out who the gentleman is in Iowa who is a certain age, certain profession, a non-smoker, and six or seven additional things,’ and you can target and know exactly who that person is.
So, to protect against that, there will be access issues that will [get you] a certain amount of genomic information. I think in pharmacogenomics, it should in no way inhibit that in the first generation of saying, ‘I want to look at the dataset with respect to those who have or have not received adriamycin for breast cancer,’ or those who either have or have not received Gleevec for chronic myelogenous leukemia.
As these scans come along, I think that those individual questions will be very easy to address. What will be harder is if you want to ask eight or 10 questions simultaneously, because then you really start to move in the realm of well, can you use enough of that information and cross it, and say, ‘Ah, now we’re really able to identify individuals.’
I would say a key question is going to be the development of suitable and practical forms of access to the data that will be guided by the local IRBs and the informed consents that individuals have signed. And that’s going to vary from study to study. There is not a blanket IRB approach or form that’s out there.
That’s why we have IRBs — because they’re local. People in California think differently than in Alabama and in Boston in subtle, but really important ways.