The Physician Genome Partnership
Name: Stephen Peroutka
Position: Founder, The Physician Genome Partnership, 2007
Background: director, department of neuroscience, Genentech, 1991; founder, Spectra Biomedical (acquired by Glaxo Wellcome), 1993; chief medical officer, Collabra Pharma, 2000; vice president of clinical research, Deltagen; 2001; president and CEO, Synergia Pharma, 2003; franchise development leader for pain, Johnson & Johnson, 2005.
Education: AB, Cornell University,1975; MD and PhD, Johns Hopkins University School of Medicine, 1979 and 1980.
Despite advances in genomics spurring the development of new types of drugs and diagnostics, regulatory and industry officials have been dismayed by physicians’ reluctance to adopt genetic tests.
Some blame an outmoded reimbursement system and pine for the day when drug/diagnostic combination products will help ease doctors into the so-called era of personalized medicine. Others point to the need for continued physician education.
But according to Stephen Peroutka, founder of the Physician Genome Partnership, doctors haven’t warmed to genomic medicine mainly due to a lack of evidence confirming the clinical relevance of certain genetic targets. In the absence of evidence, he argues, doctors should come together and decide how to manage and use the overwhelming amount of data that is coming from this rapidly advancing science.
Peroutka spoke to Pharmacogenomics Reporter this week about a pilot project, called the Physician Genome Partnership, that brings together doctors from academia, industry, and private practice in an online forum to discuss advances in genomic science and, as a group of medical experts, begin to establish standardized genomic treatment guidelines for the most clinically relevant genetic variants.
How and why was this partnership started?
This is a pilot project that I am exploring to try to determine if there is enough interest to get this going. The problem that I see is that we call things genomic medicine but it really isn’t. The vast majority of work that is called genomic medicine is really genomic pathophysiology. It’s really trying to understand diseases, genes that cause diseases, genes that are associated with diseases. … The difference between genomic medicine and genomic research, genomic science, genomic pathophysiology, is huge. When you think about genomic medicine, the only examples that are very clear are BRCA testing for breast and ovarian cancer. That’s an example where you would truly be led to a different clinical outcome with a patient. It would guide you in prevention in that case. TPMT … is another example where you would change your treatment based on that knowledge. There are some examples, but my point is that they are so minimal compared to, say, a million SNPs. My background is academic as well as industry, and there is a definite interest in the world of physicians. Physicians in biotech research and private practice are the most interested. The least interested in this whole concept are academic physicians, because that’s what academics do, they already interpret the genome.
What are the main aims of the group?
Let me give you a specific example. APOE44 is a very bad genotype to have in terms of increased risk for heart disease, Alzheimer’s, if you have any kind of head trauma, stroke recovery, and most recently, even kids with sleep apnea. In the whole world’s literature I cannot find any clinical recommendation on if a patient walked in and told me their APOE44 what could I do as a doctor? Two percent of society has this genotype. And the evidence about its deleterious implications is overwhelming. It’s not an issue. It’s not a good thing to have. So, my interest is, okay, what are we going to do about it?
My idea was that, obviously, medicine prefers evidence-based medicine. You want to have a study showing that whatever you do, it works. But the reality is that we’re not going to have that in genomics for another 50 years. We’re just at the beginning of that. So, what do we do in the meantime?
That’s the concept behind the Physician Genome Partnership. It’s a Bayesian approach. So, if thousands of physicians would weigh in and say, ‘Here is the state of the knowledge. Here is what we know. There are clearly things we don’t know. But at this point in time, with things that we do know, what’s the best way forward?’
Right now, the analogy that I’d like to use is if you’re driving and you see a sign that says “falling rocks,” what do you do? Do you go faster? Do you go slower? Do you look? So what does that mean? All it does is really scare people, right? So, my point is in the absence of really knowing exactly what to do, at least we should begin to do it in a consensus-building way amongst multiple physicians. Not just one person’s view, but a consensus of experts … and try and come up with some recommendations on what people could do.
Jay Flatley, CEO of Illumina, recently gave a talk [at a conference] and showed us a million SNPs. But what are you going to do with all that information? … There’s got to be, in all that data, some useful information, but at this point in time, I just don’t see the clinical utility of the information. Not because it’s not there necessarily, but because we haven’t addressed it.
In your experience within industry, what did you find to be the strongest factor deterring physicians from genomic medicine?
I think the concept that genomic medicine will revolutionize the practice, everybody agrees with that without really knowing the details of how. The concept I think is well accepted. All you really need to do is take family history.
Somebody once told me, and this is probably still true, ‘What is best genetic test that you can buy today? It’s your family history.’ If you have parents who died in their 40s, you have bad genes. If you have grandparents in their 90s, then you have good genes. In medicine, concept of family history plays a key role. It’s completely accepted. It’s not even an issue. And you look to family history for different clues as to the problems patients might have, be it migraine, or heart disease.
The concept that genomics will have an impact on medicine is overwhelmingly accepted. What’s not understood by anybody that I see is how exactly we’re going to get from 6 million data bits to DNA sequence to bedside treatment decisions.
How many do you have right now? Who are these members? And how are you recruiting new members?
Currently, under this pilot project, we have 33 members. It’s the beta of the beta version. It’s basically physicians that I know and physicians that they know, and we’re trying to build a web of physicians that are interested in this general area.
The complexity of the existing data is so immense. In one case, if I look at the literature, and there [are] 263 association studies, different people use different markers, the nomenclature of the SNPs change, they don’t use the same ones between labs, the patient populations they study are not identical. So, for anybody to delve into this and truly understand what the significance is, it’s extraordinarily complicated.
You’ve noted that the partnership is based around the concepts utilized in the collaborative online encyclopedia Wikipedia, the social networking site MySpace, and the Cochran Library, an online international collaboration that conducts systematic reviews of healthcare interventions and promotes evidence-based treatments. However, at least with Wikipedia, there have been numerous reports and complaints regarding inaccuracies. There have been misuses of MySpace reported as well. How do you plan to manage such problems within the academic setting of the partnership?
The concept is that we would have an editor for each gene. That editor would have five to 10 contributors or co-editors, or whatever you would call it. They are the only people that would be allowed to change what’s written. However, on the website you’d be able to add comments. Anyone can add a comment. Then it’s up to the editor of that gene, whether that comment should be incorporated, or considered, or posted. That would be my basic concept.
Again, this is all beta version. So, it’s all a little early and we’re still evolving it. I’m talking to numerous people and we’re trying to figure out how this would work.
One thing I’m trying to do is rank order genetic variance in terms of not cost to society but on impact on society. So, if you take the frequency of a given allele, so you multiply that by a score of clinical relevance, ‘0’ being of little clinical relevance and ‘4’ meaning if you have this you have to deal with this your entire life. … For instance, BRCA 1, that’s important. That would have a clinical relevance score of like a ‘3’ or a ‘4.’ If you do that, the gene that comes out number one in terms of clinical relevance is APOE.
So the MySpace aspect would be this: Who would be interested in using the APOE44 genotype? It would be the academics who work on it, it would be the neurologists in terms of Alzheimer’s, and cardiologists in terms of the cardiovascular implications. But it would also seem to me, that as physicians become genotyped, and 2 percent of people have his, that means that out of 800,000 US physicians 16,000 are walking around in theory with APOE44. But as time goes on and they know their own genotype, I think they’re going to be driven to learn more about what’s in it for them. So, the community of physicians would know that they have certain variances, are going to want to interact with one another to try to come up with a best way forward.
This is the problem. In a way we know the ending. We know the outcome of these variants. What we don’t know is what we can do to alter the course. That’s a huge challenge in medicine. That goes against the grain of current medicine, which is evidence-based.