Skip to main content
Premium Trial:

Request an Annual Quote

Coriell, OSU Launch Personalized Medicine Study that Combines Genetic Risk Information with EMRs

Premium

By Uduak Grace Thomas

This week, the Coriell Institute for Medical Research and Ohio State University Medical Center embarked on a project that aims to understand how physicians can use patients' genomic information included in electronic medical records to personalize clinical care.

As part of the study, Coriell and OSU plan to incorporate genetic risk information into the EMRs of 1,800 patients who have been diagnosed with congestive heart failure or hypertension.

The aim of the effort is to figure out whether genomic risk data, in combination with genetic counseling, leads to changes in patient behavior. In addition, the study will observe whether and how doctors act on the information in EMRs to make clinical care decisions. Approximately 30 to 35 OSU cardiologists and primary care physicians are participating in that aspect of the study.

The project is expected to reveal whether genome-informed medicine is useful in practice and how likely doctors are to use the information when it is available to them.

The OSU project is part of the much broader Coriell Personalized Medicine Collaborative launched in December 2007, which aims to enroll 10,000 patients to determine whether knowledge about their genomic risk for diseases causes changes in behavior.

CPMC participants submit saliva samples, which are analyzed using the Affymetrix 6.0 GeneChip and the DMET Plus drug metabolism platform for SNPs that are associated with common complex diseases, such as age-related macular degeneration, colon cancer, coronary artery disease, melanoma, and type 1 and type 2 diabetes. The Coriell project also tracks genetic markers for response to certain medications, including the blood-thinner warfarin.

Participants' family, medical, and lifestyle history information are collected via a web portal and linked with the information on relevant SNPs.

Scott Megill, Coriell's chief information officer, explained to BioInform that the project built its entire information technology infrastructure in house in order to maintain participants' confidentiality. In addition, he said, there aren't many off-the-shelf packages created for personalized medicine studies.

"When [participants] log into the portal ... their authentication information and anything that’s personally identifying to the participants is kept in a separate database [from] the actual genetic information and risk reports," Megill explained. "We do that for confidentiality and security purposes [so that] if we ever had a breach of our databases, there is really no way to marry those together without being logged in as that user to our portal."

Megill said that analyzing each participant's sample and creating the associated risk reports generates more than one gigabyte of information per patient. Currently, the collaborative has data on between 5,000 and 6,000 patients, not including the participants in the OSU study.

Coriell uses an IBM storage technology called XIV, which is expected to eliminate the need to invest resources in storage infrastructure as the number of participants increases.

Megill explained that typical deep network storage would store a 100-megabyte file on two "very fast spinning hard drives [within] a bank of existing disks." IBM's XIV system, however, spreads the file across all the disks and operates them in concert to store it "with much slower spinning disks."

He noted that the partnership with OSU affords Coriell an opportunity to "pilot test the integration of risk reports and personalized medicine into actual [patient] medical records" stored in OSU's electronic health record system.

"What we felt was important ... was to be able to present the genetically informed risk reports at the time in which a doctor is actually seeing a patient," he said. "In order to do that, we didn’t want the doctors to have to log into our portal to see that information because it's another username and password they would have to remember [and] we would have to do an association with the doctor and the patient to make sure the information was available in the doctor's log-in."

Educating Physicians and Patients

In addition to incorporating genomic information into the clinical care mix, "what's unique about the Ohio State study is that the physicians will themselves be studied," Michael Christman, Coriell's president and CEO, told BioInform. "The actions they take using the genetic information will be tracked ... this is kind of the first look at something like that."

The doctors will also receive genetic risk information about themselves and the researchers hope that having this data, in addition to information about each associated disease that will be provided to all the participants, will help them better use the information for their patients.

"We [will] present educational content associated with each condition that we report on. [For example,] before we tell you your likelihood of getting age-related macular degeneration ... we will tell you how common the disease is, what can be done about it," Christman said. "The physicians will be receiving this and we will be tracking how effectively they adopt it."

He explained that "in the adoption of personalized medicine probably the rate-limiting step overall is the education of physicians and other providers like nurse practitioners or physicians' assistants" who aren't familiar with genomic information.

He cited a survey that pharmacy benefits manager Medco conducted in 2009, which found that while 98 percent of 10,000 doctors who responded thought that personal genetic information will be important for treating patients, only about 10 percent felt confident ordering genetic tests and making decisions based on them.

"By their own description, [physicians] don’t know about personalized medicine" Christman said, but as these approaches become more mainstream, "they are going to have to come up to speed."

The study will also explore the role of genetic counseling on patient behavior.

Participants will be asked to complete surveys regarding the understanding of their risk, knowledge of genetics, their actions upon learning of their risk, and with whom they shared their results. Some patients will receive counseling via phone or e-mail, while others will be required to attend counseling in person. The investigators will compare the two groups to gain insights into the role that genetic counselors play as educators in personalized medicine.

Although the patients in the study know they have heart disease, Christman explained that they will receive results for other complex diseases that they may be at risk for as well as information about genes that impact how they respond to different prescription drugs for these conditions. Based on what they learn, patients can opt to receive counseling.

Obstacles to EMRs

Several recent partnerships in the clinical space highlight the importance of including genetic information in EMRs to personalized treatments as well as to help stratify patient populations and select participants for clinical trials.

For example, Aurora Health Care is collaborating with Oracle to create methods of mining patient samples stored in its biorepository to help researchers find samples of particular interest for biomarker discovery or clinical trials. Aurora's Open-Source Robotic Biorepository & Informatics Technology, or ORBIT, holds consenting patients’ DNA and is linked to patients' de-identified electronic medical records.

Furthermore, the International Serious Adverse Events Consortium said that it will work with the HMO Research Network to use HMO electronic medical records to identify genetic markers linked to three drug-related adverse events (BI 02/04/2011).

In spite of its promise for patient care, merging genetic information with EMRs is not without some drawbacks.

One challenge, Megill noted, is that currently there isn't a standard way to deliver risk reports and that the context and education that needs to go along with it are lacking.

"It's not a fortune-telling kind of exercise. You can't just tell somebody that at the age of 62, you are going to get coronary artery disease," he said, adding that an individual may be judged to have an elevated risk for the disease, but these measurements are "relative" to a particular baseline.

Moreover, Megill said, there are important IT challenges associated with moving data into disparate systems and delivering it in a standard way.

"When we are dealing with OSU, that’s one medical record system but different hospitals have different EMR systems," he said.

To get around this challenge, Megill said some groups are considering cloud-based models, and, as such, "we are dealing less and less with conversations around people with big implementations of software at their site and now trying to partner up with companies like Microsoft and IBM that are pushing for cloud-based or Internet-based EMR systems."

Even though genotyping arrays only measure a small fraction of the genome, Megill pointed out that "there is no way that all that of that raw information can be included in a patient record in a typical EMR." He added that the quantity of data for each patient will increase further when whole-genome sequencing becomes affordable enough for routine use.

One solution, Megill said, is to create what he called "third-party data escrows" that would store raw genomic information for future use, "but that can continue to grow as we move closer and closer to full-genome sequencing."

Christman added while that concerns about data privacy have been somewhat assuaged with the passage of the Genetic Information Nondiscrimination Act, which he said addresses some of the real risks associated with the adoption of personal genome information, it may be "unrealistic" to imagine a world where genetic information is completely private.

"As much as one wants to be as careful as possible with security issues, I think we have to face, probably, as a society, that genetic information will get out there at some point," he said.

He likened the model to online shopping, which many people were concerned about in the early days of the Internet. "It basically works and it is a system that does get abused rarely and people who abuse it break the law and go to jail. I think genetic privacy has to be addressed at that level too."


Have topics you'd like to see covered in BioInform? Contact the editor at uthomas [at] genomeweb [.] com.