This article has been updated to clarify that a limited number of decision support alerts for PGx testing fire for patients who already don't have test results in the system as part of the Right 10K study, based on their ethnicity and the drugs they're on.
NEW YORK (GenomeWeb) – As of the beginning of this year, all 10,000 participants in a Mayo Clinic study investigating the utility of pharmacogenomics have been genetically sequenced for 77 genes, a subset of this information has been placed into their electronic medical records, while the bulk of the data is fueling more research.
The aim of the Right 10K study is to assess whether having PGx information preemptively in their EMRs, before patients need it, improves their outcomes and saves healthcare dollars from avoidance of adverse events.
With the sequencing now done, study participants with variants in 13 genes associated with drug response have had this information incorporated into their EMRs so that it is accessible to doctors at the point of care via clinical decision support alerts, while information related to the other genes is accessible for research. Richard Weinshilboum, a professor of pharmacology at Mayo Clinic who co-directs the Center for Individualized Medicine's pharmacogenomics program, said his colleagues are now focused on gathering evidence to demonstrate the clinical utility of PGx testing in the population and discover new drug/gene interactions.
Before starting Right 10K, Mayo conducted a pilot involving 1,000 patients and showed that with just five gene/drug pairs, 99 percent had an actionable variant."Eventually, pharmacogenomics should be an integral part of every therapeutic encounter," Weinshilboum said. "We see that coming [and] we'd like to prepare at Mayo for doing that."
However, Mayo's goal from the outset was not to just do research for the sake of research, but to implement a program in which it could take the learnings from those investigations and use it to improve patient care. That's why when Mayo collaborated with the Baylor College of Medicine to sequence the pharmacogenes of 10,000 of its patients for the study, they made sure to do the testing in a CLIA-certified lab. This would ensure that once there was evidence demonstrating the validity of new drug/gene interactions, that could be quickly added within patients' EMRs.
Having devoted his career to studying the impact of genetic variants on drug response, Weinshilboum has faith that pharmacogenomics will be the aspect of clinical genomics to reach the masses first. "That's not an exaggeration by somebody who's just been at this so long that he doesn't have a perspective," he joked. But for broad consumer adoption, Weinshilboum acknowledged that the cost of sequencing has to fall even further, there needs to be more evidence establishing the impact of common and rare variants on drug response, and physicians, as well as patients, need to recognize the value of this information.
Mayo has had these goals in sight for more than a decade while it has steadily invested in the resources it would take to perform a study like Right 10K. In 2009, for example, it set up a biorepository on its Rochester, Minnesota-campus to start collecting and storing samples that would fuel the research. Then, in 2012, it opened its Center for Individualized Medicine (CIM) to house the experts leading biomarker research and physician education efforts. Mayo started building PGx decision support alerts into its EMR system in 2013.
For Right 10K participants, because they've already been tested and their data is already in the EMR, the decision support system makes specific clinical recommendations about changing the drug or dosing. Once tested, the PGx results and a report from informatics company OneOme, which includes information on their ability to respond to some 400 medications, are added into the EMR and a copy of the report is mailed to patients and uploaded to their online patient portal. For Mayo patients who are not part of the study and who do not already have PGx results in the EMR, a limited number of PGx decision support alerts recommend genetic testing based on the medications they're on and their race or ethnicity.
As researchers conducted the Right 10K study, there were challenges along the way. "We were somewhat naïve about the complexity of what we were getting into," Weinshilboum said. What proved surprisingly difficult, he noted as an example, was importing the sequencing data from Baylor to Mayo, since the two institutions have systems that store this data in different formats.
The compatibility issues now resolved and sequence data in hand, the focus at Mayo is to see how having this information pre-emptively impacts patients' long-term outcomes and healthcare costs. This information is necessary for the kind of broad uptake PGx proponents like Weinshilboum would like to see, because that requires buy-in for insurers.
Unfortunately, few PGx tests have satisfied payors' evidence requirements, which remains a topic of debate and controversy among stakeholders. For example, a commentary recently published in Clinical Pharmacology & Therapeutics by Stanford University's Teri Klein and colleagues took issue with insurers' demands for randomized-controlled trials, arguing that such studies are often unnecessary, impractical, and even unethical to perform.
"It will be virtually impossible on the scale that we're talking about to do a randomized trial," Weinshilboum said, adding that Mayo has been in discussions with several insurers to try to gain a better sense of the types of data they'd like to see. While many insurers still have non-coverage policies in place for PGx tests, according to Weinshilboum, several firms have expressed interest in offering this type of testing for certain beneficiary groups. "But they want evidence [and] they want to be involved in designing the way in which the evidence is generated," he noted.
The Right 10K project, by sequencing genes instead of genotyping for specific SNPs, has also amassed data to kickstart dozens of research projects to explore the functionality of novel genetic variants that can be clinically actionable in the future. Mayo put out a call for PGx research proposals among its physician faculty and now has 30 institutional review board-approved protocols.
"We've been stunned by the number of proposals that have come forward from our faculty saying that they want to test drug X using these 10,000 patients," he said. "It's the [physicians] on the ground who are actually prescribing the drugs, and who have the most creative ideas about which drugs are a problem and which they want to test."
Generating research ideas from the faculty is also an opportunity to educate doctors at Mayo about pharmacogenomics. Physicians who come up with these projects often become involved in the study, and through the experience become the go to person for pharmacogenomics in a particular practice area and help other colleagues come up to speed.
"Of course, [physicians] read textbooks and journal articles, but we always go back to our colleagues and ask how they treated a patient or dealt with an issue. We learn from each other," said Konstantinos Lazaridis, associate director for the Center for Individualized Medicine. "If someone is a champion, works on a project, learns from it, and sees its value, then this individual is the best person to disseminate this information to the rest of the practice. That is the model we use."
The first time pharmacogenetics came into play for a patient with high blood pressure, Tamim Rajjo, a family physician at Mayo and a preventive medicine specialist, was a bit incredulous. Rajjo had prescribed this patient two common first-line medications, a diuretic and an ACE inhibitor, but when her blood pressure remained uncontrolled, he added the beta blocker metoprolol (Toprol). Within a week, the patient complained of being very tired, which isn’t an uncommon reaction to metoprolol, and is usually managed by lowering the dose, which Rajjo did in this case.
Still, this patient who was an otherwise active person, continued to complain of tiredness. Eventually, Rajjo decided to look at her PGx test report and saw that she carried a CYP2D6 variant that made her a poor metabolizer of metoprolol. Based on that, he decided to stop metoprolol and give her a drug in a different medication class. "It was like a switch. It was like night and day for her," he recalled. "I had a hard time believing this could be the case."
Although Rajjo had attended lectures on pharmacogenomics in the past, this was his first experience implementing it in his practice. "I was kind of forced to incorporate it, but it just made sense," he said. "This medication was not the best option for this condition in this patient. Period. It was an eye opener for me."
Meanwhile, for participants within the Right 10K Study, their doctors are now receiving alerts if they have actionable variants in 13 pharmacogenes that have been incorporated into the decision support. While PGx information may not come into play for every patient, Rajjo certainly pays attention and takes the appropriate actions when the decision support alerts him to a potentially important interaction. And for patients who are on five or more medications, who are at a greater risk for drug-drug or drug-gene interactions, he actively goes into their medical record to look for these potential problems.
Rajjo wouldn't describe himself as an early adopter of technologies. "If an iPhone 10 comes out, I'll buy it in five years," he quipped. He described himself as a stranger to pharmacogenomics before he encountered the patient with high blood pressure, and admitted to feeling guilty that he wasn't more aware of it so he could have helped her faster.
"When you see the change in patients' lives, you get curious," he said, noting that he particularly appreciates the alerts he gets about potential PGx interactions for patients enrolled in Right 10K and other studies. Although many of Rajjo's colleagues have yet to order testing for their patients, the best way to engage and educate physicians in pharmacogenomics, he said, is "to connect the science with the practice."
Mayo is not the only large academic health system working on implementing preemptive pharmacogenomics, and educating physicians on the discipline. In recent years, similar programs have been started around the country at Moffitt Cancer Center, NorthShore University HealthSystem, University of Florida, and Vanderbilt University. However, as these programs have taken shape, the US Food and Drug Administration, once a vocal proponent of pharmacogenetics, has shifted to a more circumspect position.
Last year, the agency issued a safety alert cautioning healthcare providers against using PGx tests that haven't been approved by the FDA. Because the agency has had practice enforcement discretion over lab-developed tests (LDTs) for around two decades, most PGx tests on the market, including those offered through preemptive testing programs within healthcare systems like Mayo, are offered as LDTs that are performed in a CLIA lab without FDA approval.
The agency's safety alert, and its more recent warning letter to Inova Health System's genomics lab for offering PGx testing without regulatory approval, has been criticized by lawyers and pharmacogenomics experts alike. Klein from Stanford and Mary Relling from St. Jude Children's Research Hospital, both of whom are co-principal investigators of the Clinical Pharmacogenetics Implementation Consortium (CPIC), took particular issue with the FDA's suggestion that the only source of "established" PGx information are found within drug labels it has updated or approved.
Counter to the FDA's claim that "the relationship between CYP2C19 genotype and drug response to escitalopram and sertraline is not established," for example, they asserted in a post on the PharmGKB website that CPIC guidelines issued by PGx experts are also a valid source of information and that the group has concluded after a review of the literature that CYP2C19 phenotypes are actionable for these two drugs.
Mayo formed a task force, which reviewed drug labels the FDA has updated with PGx information, as well as recommendations from the CPIC and the Dutch Pharmacogenetics Working Group, to identify the specific drug/gene pairs to alert doctors about. The drug/gene relationships are incorporated into EMRs and have decision support alerts supported by "sufficient clinical evidence," said Liewei Wang, who co-directs Mayo's Center for Individualized Medicine pharmacogenomics program with Weinshilboum.
Even when there is evidence supporting the use of many pharmacogenomic indications, in Lazaridis' view, the main rate-limiting step remains physician education and workflows that allow the integration of this information into clinical practice. One way Mayo is trying to address this is by embedding the "Ask Mayo Expert" feature within the decision support system to provide online educational information to doctors about drug/gene interactions.
There is also more work to be done in terms of research, Lazaridis said. He noted that while studies ongoing at Mayo are aimed at improving physician education and demonstrating clinical utility, there is a need to improve understanding of the interplay of genetics and other factors on drug response. "There is a lot of additional information about the metabolism of medications that goes above and beyond pharmacogenomics," said Lazaridis, noting that two people with the same genetic variant may still have very different responses to a particular drug.
"It's not only the genetic information that is important for metabolizing medications. It's what you eat, drink, and what you're exposed to," he said. "This is another level of knowledge that we don't even have the capacity to fully appreciate, but which is a significant component of how we metabolize medications."