Q: What was the major factor that enabled the Marshfield Clinic to become a pioneer in the field of personalized medicine years before many others jumped in?

A: The primary factor for Marshfield Clinic’s leadership is a highly sophisticated electronic medical record.  Marshfield Clinic started developing parts of their electronic medical record in the late 1970s with widespread usage taking place in the late 1980s. They made a concentrated effort to back-load information on patients with clinical data from the 1960s so that it was available for patient care.  The electronic medical record is used by all of Marshfield Clinic’s 41 sites and three affiliated hospitals.  The Marshfield Clinic sees about 400,000 unique patients annually, who collectively log about 1.8 million visits per year, and all of this activity, both out-patient and in-patient, is captured electronically. Currently, the Clinic has clinical data on over 2 million of our patients dating back to the early 1960s.  The breadth and depth of this database permits us to get a more comprehensive picture of disease and disease progression in the population.

Q: What is the Personalized Medicine Research Project?

A: The concept of the Personalized Medicine Research Project was a fairly simple one – to collect as many of the people in the epidemiological study area as we could, who would allow us to have access to their electronic medical record and would then give us samples of DNA, serum, and plasma so that we could perform studies in pharmacogenetics, genetic epidemiology, and population genetics.  We received funding from the federal government to determine and address the ethical and security issues related to this type of data collection.  In addition, these funds supported the formation of a scientific advisory group to help us look at various aspects of the collection to make sure that we were not generating any bias in the way that we actually collected the samples or the patients and that the resource would have the most utility for hypothesis generation.  Then we received funding from the state of Wisconsin to allow us to actually recruit patients and collect the samples.  Today, we're still collecting and have accumulated well over 19,000 samples.

Q: You’ve had tremendous success in getting patients to participate in this project.  Tell us about your recruiting process.

A: Before starting the patient recruitment and sample collection, we convened focus groups from different communities within our epidemiological study areas, and then talked to them about the whole process – what we were planning to do, why we were doing it – to learn what they thought the ethical or social issues might be, what their concerns were likely to be as far as participating in a study like this, and how they thought we could best get patients to participate.

That process was extremely helpful.  We also formed an advisory group of folks from different communities who understood what we were trying to do and who could help us to get the information out to individuals within the community and keep us sensitive to the community’s needs. 

Our focus groups told us that word of mouth was probably going to be the best way to collect patients. So once we started the project, we made a large number of visits to VFWs, Lions Clubs, and other local social meetings where we presented the project, let people understand what it was about, and importantly, what it was not about, what the safeguards were, and how privacy issues had been addressed.

Fortunately the state of Wisconsin has laws against the use of genetic information to discriminate in employment or insurance, which was another big help for us. Also, we were fortunate to obtain a Certificate of Confidentiality from the NIH to further safeguard the project. At our peak of collection, we were recruiting over 100 patients a day. 

Q: Can you describe some of the research being done under the Personalized Medicine Research Project?

A: We have validated, with a much larger sample size than had been done before, the interrelationship between the dose of warfarin and polymorphisms in cytochrome P450 2C9, and also the variation that occurs when you overlay the polymorphisms in VKORC1 on the P450 2C9 genotypes.  As a spin-off from the Personalized Medicine Research Project, we have performed the first prospective study to see if you can genotype patients in real time to adjust their warfarin dose. In that study we also showed, that if you have a mutant allele, you're far better off with genotypic dosing of warfarin than without it. 
 
Russ Wilke has recently published on CYP3A5 variants and their effect on statin induced myopathic complications; Cathy McCarty is looking at genes that are involved in macular degeneration and at pharmacogenetic aspects of topical beta-blockers used to treat glaucoma and intraocular hypertension;  Khemissa Bejaoui is searching for genes that are involved in multiple sclerosis and ALS; Phil Giampietro is examining genetic variants involved in vertebral malformations; Nader Ghebranious is looking at genetic variants linked to Alzheimer's disease and at candidate genes in hypertensive heart disease; and Seth Dobrin is collaborating with colleagues in Ireland and at the Stanley Neuropathology Consortium to discover genes involved in neuropsychiatric disorders.

Q: Will others be able to access the vast databases that the Marshfield Clinic has built?

A: There is no reason that we can't help the research efforts of many investigators in many areas, and our approach from the very beginning has been to make this a national resource. We realized from the outset that this is an enormous resource that we could never effectively use alone. We envision and are developing a library of phenotypes (the physical, observable manifestations of disease) for the research community, where investigators could visit and say, "I'm interested in finding cases with a particular phenotype, controls without that phenotype, and obtaining the genetic information necessary for association studies while still maintaining patient privacy.” Therefore, one of our current priorities is to genotype the entire Personalized Medicine cohort with a whole genomic scan. Obtaining this information would allow us, and many others, to ask questions and develop hypotheses on disease-gene relationships without re-collecting the data each time someone conceives of a study. It would also save everyone a lot of time and individual study efforts could be directed toward fine mapping. Our expectation is that those who use our databases will have a period of time to protect or publish their data and then the newly derived biomarker data will ultimately return to the database for everyone to use.

Obviously to enable studies with rare alleles, we need to continue to expand our cohort. Another useful attribute of our cohort is the relative homogeneity of the population. While homogeneity is good from a discovery perspective, we are very interested in collaborating with others who have similar resources in other ethnic groups to examine the universality of discoveries in our population.

Q: How do you envision personalized medicine evolving over the next several years? 

A: I think that the early work in pharmacogenetics is going to be in the area of drug safety. There need to be a couple of home runs that get incorporated into common medical practice, and it's one of the reasons that I've been so interested in pushing the warfarin project forward.  Warfarin is something that most physicians have used in their clinical life and understand the difficulty in prescribing the drug. If you can give physicians a genotype-based solution, they will immediately associate it with the power of personalized medicine.  

Then, I think we'll start seeing pharmacogenetics move into the areas of efficacy. This area will also be financially important to the healthcare system but is going take longer to realize than applications to safety, which are strongly driven by a public wanting to avoid unnecessary, harmful complications.  As both of these areas move forward, we're going to get much better at predicting outcomes.