Researchers at the University of Chicago's Center for Personalized Therapeutics have begun a 1,200-patient feasibility study intended to assess whether physicians with access to patient genotypes actually use that information in the routine healthcare setting.
The study, called the 1200 Patients Project, has so far consented just over 600 subjects and has begun genotyping around 400. These patients are under the routine care of 13 physicians that the investigators have "hand-selected" from the university's department of medicine, explained Mark Ratain, director of the Center for Personalized Therapeutics.
Ratain, who outlined the project at the American Association of Cancer Research conference earlier this month, noted that one of the project's primary aims is to determine whether "preemptive genotyping" of patients can overcome one of the main criticisms of genomically guided treatment — a long wait for test results.
Ratain cited as an example the case of the blood thinner warfarin, where there are well-established genetic variants associated with response to the drug — not to mention the US Food and Drug Administration's blessing in terms of using these variants to guide dosing — but pharmacogenomic testing has not been widely adopted because "when you want to start warfarin, you usually want to do it today. You don't want to wait for test results."
The U of Chicago investigators hypothesized that if patients are genotyped for key pharmacogenomic variants up front, and this information is made available to physicians via a web-based portal, then doctors would be more willing to use the data to inform treatment.
"By genotyping in advance, we have the information when we need it," he said. "We're doing this before it's necessary so the results are available yesterday."
Ratain stressed that the study is not looking at whether PGx data improves patient outcomes, but whether access to such an infrastructure — which the researchers have dubbed the Genomic Prescribing System — impacts treatment decisions in routine healthcare.
"This is a Phase I study, a feasibility study," he said. "The first question is, 'Can we even get physicians to go to the portal to look up information on their patients?'"
Ratain said the project began two years ago. The first step was to identify participating clinicians, "who are also our research subjects because we have to study how [they] will interact with the system."
He said that for the current phase of the study, the investigators sought out physicians who were "motivated" and in some cases already using pharmacogenomics in practice. He acknowledged that it will be necessary to eventually extend the project to a broader population of physicians, but noted that the researchers haven't yet made plans to do so. "We're still trying to figure out if what we're telling patients is valuable," he said. They will collect data through January 2016.
The project began enrolling patients a little more than a year ago and has so far consented 619 participants. "Patients provide blood at the time of a routine blood draw, so it's not even an extra stick," he said.
The Genomic Prescribing System will officially "go live" in June, at which time participating physicians will have access to patient genotype data. Enrollment and genotyping will continue as the project progresses, Ratain said.
Red Light, Green Light
The 1200 Patients Project is seeking participants over the age of 18 and under the age of 65 who are receiving ongoing outpatient care from one of the study's participating physicians. They must be taking at least one regularly used prescription medicine, but not more than six, "because we believe the risk of polypharmacy will overwhelm the risk of genotype-drug pairing," Ratain said.
Participants must have a life expectancy of at least three years. Patients with metastatic cancer or those who have had liver or kidney transplants are excluded.
Once patients are genotyped, this information is fed into the Genomic Prescribing System, which correlates genetic variants with their known impact on drug response and safety.
Physicians who sign into the system are presented with the patient's name, sex, date of birth, and medications. Response and safety information on drug-gene pairs is represented by traffic lights: "Green is good, yellow is caution, and red is stop," Ratain said. Physicians can also retrieve this information by searching for specific genes or drugs.
As an example, Ratain noted that a yellow light might appear for the cholesterol drug simvastatin for a patient in the CC genotype group, which represents 2 percent of the population with high risk of statin myopathy. "We're telling the physician that this patient is in the highest risk group for myopathy," but leaving it up to the doctor's discretion whether to prescribe the drug, he said.
In practice, Ratain said it might make sense for physicians to use the system to make a decision in the case of a suspected safety issue. "If this was a yellow light and the patient was having an adverse event, the physician would probably take the patient off the drug," he said. Another use case would be prior to prescribing a drug, in which case a physician could query the system about potential risks and benefits of that specific therapy for an individual patient.
The system also informs physicians about the level of evidence available for the pharmacogenomic data in the system. "Level one is a large replicated study, level two is a smaller study, and level three may be a much smaller study," he said. Physicians can also click through the initial results to get additional details about the associations — all the way through to the published literature on the genes.
While the primary entries in the Genomic Prescribing System don't include specific details on variants and their role in drug response, "it's not hidden from them either because all they have to do is click through to the research studies to deduce what their patient is taking," he said.
"We're not telling them the gene, we're not telling them the allele. We're giving them a consult. Because if you ask a physician what they want, that's what they want. They don't really care what the gene is or what the variant is. They want to know what to do for their patient."
The Genomic Prescribing System is currently separate from the medical center's electronic medical record system because it is a research study, but also due to some "technical issues" related to the hospital's EMR. For the time being, "we have research assistants maintaining the drug listing in this database from the EMR," Ratain said. "We are not providing all clinical information here, just the basic information. The only thing that can change are the drugs."
Sticking with Genotyping
Ratain acknowledged the growing hype around next-generation sequencing as a diagnostic tool, but noted that the U of Chicago team intends to stick with genotyping for the foreseeable future.
This decision is based on a number of factors, he said. "One, it gets you information that is very hard to analyze. Two, it gets you information you don't want to know" — namely information about disease risk.
"I'm not saying that nobody wants to know about disease risk, but we didn't want to know about disease risk and we didn't think our patients want to know about disease risk. And we didn't want to have to deal with all the IRB issues of disease risk" for this project, he said.
He added that there may be longer-term issues that might arise from sequencing for a study like this, since the Genetic Information Nondiscrimination Act of 2008 prohibits genetic discrimination in employment and healthcare, but does not prevent discrimination for life insurance, disability insurance, or long-term care insurance.
"If we enrolled patients in this program and they wanted to then apply for disability insurance, and they were asked, 'Did you ever have genetic testing of any kind?' the answer would have to be yes, and if they were asked to provide that information, they would have to provide it."
Ratain noted that whole-genome sequencing of patients could actually pose a "barrier to pharmacogenomics" since it provides much more information than is necessary.
"If I'm managing a diabetic, I don't need the whole metabolome. I just need a glucose," he said. "It's the same thing here. If one wants to implement pharmacogenomics, I don't need the whole genome. I just want the pharmacogenome, and the pharmacogenome is a defined set of variants."
Ratain said the U of Chicago team considers the pharmacogenome to be around 1,000 "potentially useful variants" that are available for clinical implementation.
He declined to provide details on the variants that are currently included in its genotyping assay, but Peter O'Donnell, assistant director of the Center for Personalized Therapeutics and principal investigator for the project, told PGx Reporter via e-mail that the panel it includes "hundreds of variants" at this time.
The group is partnering with Oregon Health and Sciences University for the genotyping aspect of the project. OHSU's CLIA-certified lab is performing all the genotyping using a custom panel on the Sequenom MassArray system.
Ratain said that the cost of running the panel is in the "low hundreds" of dollars, which provides "everything we want for less than the cost of the Mayo Clinic's HLA test, for example."
Have topics you'd like to see covered in Pharmacogenomics Reporter? Contact the editor at btoner [at] genomeweb [.] com.