March 2007: Volume 33, Number 3
Ready or Not, Here Comes Pharmacogenomics
Will FDA’s Expected Change to the Warfarin Label Make Testing Mainstream?
By Deborah Levenson
In recent years, personalized medicine has been the subject of much hype in the media, but not part of routine healthcare. Now the emerging field of pharmacogenomics stands poised to make a difference in the lives of an unprecedented number of patients. Changes being considered by the Food and Drug Administration (FDA) to the label on warfarin—one of the most widely prescribed oral anticoagulants—could call for testing to identify certain genetic variations that affect how the drug is metabolized, its effectiveness, and safety. Although warfarin is widely prescribed for patients with disorders like venous thrombosis, atrial fibrillation, pulmonary embolism, and acute myocardial infarction, clinicians are keenly aware that it puts patients at risk of severe bleeding and even fatal events. Pharmacogenomic studies have led to a better understanding of why certain patients respond so differently to warfarin, and the expectation is that clinicians will be able to use that advice to better prescribe safe doses. A warfarin label that discusses testing could alter medical practice for anticoagulation therapy and put labs at the forefront of tailoring drug treatment protocols to patients’ genotypes.
While traditional genetic tests simply diagnose a disease or identify carriers of a particular gene involved in a condition, testing for genotypes that inform warfarin dosing is a much more complex undertaking, said Gregory Tsongalis, PhD, Director, Molecular Pathology at Dartmouth-Hitchcock Medical Center in Lebanon, N.H. “This is different from traditional lab testing in that there are different interpretations, responses to information, and potential for significant consequences in patient management,” he explained.
Laboratorian experts in pharmacogenomics and FDA officials agree: the forthcoming change to the warfarin label positions labs to take a leading role in educating clinicians about personalized medicine and to translate laboratory results into clinical practice. Laboratorians should prepare—even if they don’t plan to do pharmacogenomic testing in their own labs.
FDA Contemplates the Change
The FDA’s decision about the warfarin label could come “sometime in the second quarter” and as early as April, according to Lawrence J. Lesko, PhD, Director of the FDA’s Office of Clinical Pharmacology and Bioinformatics. He added, however, that the date is not firm. Lesko wouldn’t reveal where the label might mention pharmacogenomic testing or what it would say, noting that any update to a drug label is a regulatory action related to a supplement that a company submits in confidence.
FDA is still discussing the evidence related to the change and the quality of the data with its own medical staff, noted Lesko, who added that the next steps include getting clinical pharmacologists concerned with dosing and physicians interested in efficacy and safety to agree on certain issues. While both camps agree “in principle” that pharmacogenomics is important, he added, “The devil is in the details.”
While Lesko would not reveal details about the label’s language and its placement, he did say both considerations will have a major impact on clinical practice. Stronger language may imply liability if clinicians don’t genotype every patient before prescribing the drug, while weaker language may not change practice much, he said. Regarding placement, Lesko noted that FDA’s options range from a black box warning indicating that patients with certain genotypes are at higher risk of toxicity to a recommendation in the dosing and administration section of the label to lower doses for these patients.
Some laboratorians experienced in personalized medicine speculated that FDA is delaying the label change as it contends with pressure from drug companies concerned that genotyping warfarin patients will diminish profits. Others said FDA might be giving unusual consideration to how calling for genotyping could affect medical practice, or is busy keeping the many companies selling generic versions of warfarin “in the loop.”
Lesko refuted all of these conjectures. The idea that drug companies want to slow adoption of personalized medicine isn’t necessarily accurate, he said, noting that the problem for companies is a lack of breakthroughs that make identifying subpopulations that can benefit from genetic testing feasible. Neither is FDA holding off because of concerns about medical practice, because that’s the responsibility of state practice boards, not his agency. Rather, FDA is charged with weighing benefits and risk of drugs and labeling them appropriately. Widespread off-label prescribing “is an indication that we can’t control physician practice,” Lesko explained. Further, FDA doesn’t need to discuss changes to the warfarin label with companies selling generic versions because alterations are linked to the new drug application held by Bristol-Meyers (New York, N.Y.), Lesko added.
Genes, Tests, and Technology
The tests at issue identify genetic variations of the enzymes cytochrome P450 2C9 (CYP2C9) and vitamin K epoxide reductase complex subunit 1(VKORC1). While CYP2C9 metabolizes warfarin and is involved in how much of the activated drug circulates in the blood, VKORC1 is the drug’s target and determines the effectiveness of the circulating dose.
More than half of the U.S. population has VKORC1 genotypes that affect safe and effective warfarin dose, while between 3% and 20% of individuals, depending on ethnicity, have CYP2C9 alleles that do. Patients with CYP2C9 variants and impaired warfarin clearance need lower doses, and those with VKORC1 variations require dose adjustments to optimize their response to the drug. Variant CYP2C9 and VKORC1 alleles combined with certain patient characteristics, other drugs the patient is taking, and clinical status may predict more than 50% of dosing variability, according to a 2005 presentation on pharmacogenomics and warfarin available on the ARUP Laboratories (Salt Lake City, Utah) Web site (see box, below).
For clinical labs, examples of technologies that are most cost-effective and useful for identifying genetic variations that affect warfarin dosing are based on real-time polymerase chain reaction (PCR) and signal amplification, according to Gwen McMillin, PhD, Medical Director of ARUP Laboratories’ Toxicology Lab. Real-time PCR is the fastest—giving results in less than an hour—because it detects variations as it amplifies. The LightCycler instrument marketed by Roche Applied Science (Indianapolis, Ind.) uses this technology. Signal amplification requires a separate instrument—such as a spectrophometer or luminometer—to identify variations and takes a little longer. However, results usually are available within a normal shift. The Invader chemistry marketed by Third Wave (Madison, Wis.) is an example of a technology that uses signal amplification to detect genetic variants. Other technologies, such as arrays, can detect more genetic variants, but are much more expensive and unnecessary to identify the three most common alleles now used to predict how the body processes warfarin, according to McMillin.
Many labs that perform pharmacogenomic testing currently use their own homebrew assays to detect CYP2C9 and VKORC1 variations, as few assays are commercially available. Assay reagents are available through Third Wave and Tm Bioscience (Toronto, Ontario), but they do not yet have in vitro diagnostic status, McMillin added.
But several companies have submitted 510(k) applications for warfarin tests to the FDA for clearance, noted Alan Wu, Director of Clinical Chemistry and Toxicology at University of California at San Francisco and director of a pharmacogenetics lab that will soon open there. Today, tests that identify genetic variations that affect warfarin dosing are relatively expensive, as they require special expertise and are time consuming, but Wu and others expect a steep drop in prices once more tests come to market.
However, a patient’s genotype isn’t the only piece of information that clinicians will need to prescribe safe doses of warfarin. At least three algorithms can help clinicians determine warfarin dose, and laboratorians should become familiar with these. Elizabeth Sconce, PhD, of University of Newcastle in the U.K., and Brian Gage, MD, at Washington University in St. Louis, Mo. have both published dosing algorithms (See box). Roland Valdes, Jr., PhD, Professor and Senior Vice Chairman at University of Louisville’s Department of Pathology and Laboratory Medicine and colleague Mark Linder, PhD, who run a pharmacogenomic lab, also have a dosing algorithm that hasn’t yet been published.
Laboratorians will also need to look at reimbursement for the pharmacogenomic tests. According to laboratorians already offering tests that identify CYP2C9 and VKORC1 variations, payment is good, although the tests lack specific CPT codes. However, every step involved in performing the tests has an existing CPT code that allows labs to receive adequate reimbursement for the entire test. While labs already doing the test have figured out how to get paid for it, some lab directors said that lack of a single code could present a barrier to labs that are inexperienced with the methods.
Most major medical centers already have the molecular diagnostic capability to test for genetic variants that affect processing of warfarin, Wu noted. However, while more hospital labs have added this capability, it isn’t mainstream yet in part because pharmacogenomic testing requires a large investment in equipment for DNA amplifications and extractions. In order to prepare for the expected label change, labs will need to decide if they can find the money for doing the test in house or if they need to send samples to a reference lab.
Have the Right Expertise
Whether labs perform their own tests for CYP2C9 and VKORC1 variants or send them out, they still need to consider several issues. First and foremost is the expertise needed to make genomic information useful to clinicians. “Just because there’s a box [instrument] that makes it easy to do, doesn’t mean that you’re really qualified to do it,” Tsongalis said. “With pharmacogenomics, you aren’t looking at a disease state, but rather a patient’s response to therapy.” Doing so requires examining drug-drug interactions, and for warfarin specifically, physical traits like gender, weight, and body mass, as well as environmental factors.
While researchers at Tsongalis’s institution had intended to start a study early in 2006 to examine the cost-effectiveness of pharmacogenomics, the team has yet to start. “Data from these tests has been more complex than we first thought,” commented Tsgongalis. This complexity is why results of Dartmouth-Hitchcock’s tests get attention from a multi-disciplinary team that includes Tsongalis, a molecular pathologist, a clinical chemist/toxicologist, and a clinical pharmacologist.
In addition, patients on warfarin therapy are difficult to treat and to study because they often have diabetes and underlying cardiovascular disease or take medications that complicate dosing algorithms, said Thomas Moyer, PhD, Chair of External Affairs at Mayo Clinic’s Department of Laboratory Medicine and Pathology in Rochester, Minn. and primary investigator of a study that will examine genetic testing in primary care involving warfarin therapy. “Our experience shows that at least 50% of patients who take warfarin are on at least one medication that complicates the dosing algorithm. Singling out the impact of warfarin is difficult,” he commented.
A Learning Curve for Physicians
But do the anticipated changes on the warfarin label signal the beginning of the pharmacogenomic era in medicine? Laboratorians said that this move by the FDA could prompt a major change in clinical practice. However, before the tests become routine, many clinicians will first need to understand the information themselves, agreed those interviewed. “Physicians don’t even know what genotyping is, never mind ask for it,” Wu quipped.
Simply giving physicians information isn’t enough, explained Valdes. Laboratorians should show them how to use it. “What physicians need to be shown is evidence that pharmacogenetic information related to warfarin is valuable. And they need information packaged for immediate use, so they don’t have to figure out what a certain genotype means relative to a concentration or dosing scheme.”
In addition to giving physicians this “actionable” information, they also need information about cost savings associated with genotyping patients who will take warfarin. “Cost savings is based on safety, frankly,” Valdes emphasized.
Those savings could indeed be huge. Integrating genetic testing into routine warfarin therapy could reduce healthcare spending by approximately $1.1 billion annually, according to a November 2006 report issued by the American Enterprise Institute-Brookings Institute Joint Center for Regulatory Studies. That figure is based on avoidance of an estimated 85,000 serious bleeding events and 17,000 strokes each year, the report said.
Understanding the value of the test for the population a particular lab or institution serves is also crucial, Lesko noted. Laboratorians should know and communicate to physicians the rate of specific variants among the ethnic groups they serve. While research on warfarin and genetic variations has thus far focused on Caucasians, alleles identified as having an effect on warfarin dosing are present in other ethnic groups, with the same effects.
To ensure that information about genetic variations is properly integrated into medical decisions, physicians should understand that test results are not a replacement for monitoring the patient’s International Normalized Ratio, Lesko suggested. He recommends that laboratorians explain the lab’s measures of sensitivity and specificity so that physicians understand the test’s limitations and predictive power. Lesko also emphasized that all labs have ongoing responsibility to ensure quality under CLIA.
Taking a Few Extra Steps
Wu and other laboratorians worry that some labs may be ill prepared for the expected recommendation. “The FDA committee jumped the gun a bit when it recommended something that’s not yet widely available, except in reference labs. Right now there is no FDA-cleared test for both CYP2C9 and VKORC1. It’s still possible to do the test, but you must take extra steps to show you can do it,” he said.
Right now labs must also devise their own proficiency testing (PT), although a new pharmacogenomic testing survey is expected this spring from the College of American Pathologists (Northbrook, Ill.). Until then, directors can get sample cell lines or share samples with other labs. At an absolute minimum, labs should blind some of their samples and run them internally to determine that results are the same, Tsongalis advised.
Other forthcoming tools will aid laboratorians as they navigate changes brought on by more widespread testing for variants that affect warfarin dosing. The American College of Medical Genetics (Bethesda, Md.) plans to issue recommendations on validity and utility of tests used to detect variations involved in warfarin dosing, while the National Academy of Clinical Biochemistry (Washington, D.C.), under the leadership of Valdes, is working on guidelines for lab analysis and application of pharmacogenetics to clinical practice.
Warfarin and Pharmacy Benefit Managers
The impending label change has caught the attention of pharmacy benefit managers as well. A study being jointly conducted by Mayo Clinic’s Center for Individualized Medicine and the pharmacy benefit management company Medco (Franklin Lakes, N.J.) will examine the impact of giving primary care physicians information about pharmacogenomic testing for treating patients with new warfarin prescriptions.
The year-long, community-based Mayo/Medco study will enroll 1,065 patients and assess whether physicians adjust warfarin dose based on patients’ genotypes, according to Moyer. Thus far, research on warfarin and genetic testing has focused on patients under the care of specialists in academic institutions, but this study “will look at real-time, standard patient care in a primary care setting.” The information on personalized medicine and warfarin in community settings across the nation will be an important contribution to peer-reviewed literature, he added.
Medco hopes the study will lead to “precision in drug dose and drug management” that will both increase patient safety and help the company better manage benefits for its clients, which include healthcare plans and large companies, said Russell Teagarden, Medco’s Vice President for Clinical Practice and Therapeutics.
Other studies will yield data on using genetic tests to aid safe and effective dosing, according to published reports. Projects at Harvard Medical School and Marshfield Clinic Research Foundation in Wisconsin are also examining warfarin patients. Pharmacare Management Services, Inc., a subsidiary of CVS Corporation, is collaborating on research with a subsidiary of the pharmacogenetic testing firm Clinical Data Inc. to incorporate testing into clinical programs. The project will focus on warfarin and the antipsychotic drug clozapine.
If the data from such studies is positive, clinicians, laboratorians, and regulators can all look forward to a time when knowing a patient’s cytochrome P450 genotype means less worry about adverse events. “If we can improve physician management of warfarin patients through genomics, we will improve patient outcomes,” Moyer said.