January 2009
Feature Story

William Check, PhD

A major element in the innovative painting technique developed by American abstract expressionist artist Jackson Pollock was his use of synthetic enamel paints, which were cheaper than traditional artist’s oils. Pollock described his use of these new materials as “a natural growth out of a need.” People who advocate using pharmacogenomic testing to guide warfarin dosing see the genetic test in a similar way, as a natural response to an urgent need—the need to reduce the risk of major bleeding with an anticoagulant that has a relatively narrow therapeutic index. Despite the seemingly inevitable logic of this idea, adoption of pharmacogenomic testing for warfarin sensitivity in laboratory and clinical practice has been slow, even in the face of the FDA’s promotion of this safety innovation, such as by its revised labeling of warfarin.

One reason for the lagging adoption of pharmacogenomic, or PGx, testing for warfarin dosing is the lack of robust evidence that such testing will improve health outcomes. While studies of the impact of genetic information on accuracy of warfarin dosing have provided generally positive results, most have been retrospective, and almost all prospective trials have been observational cohort studies. “A few prospective studies have shown that genotype-guided dosing is associated with fewer dosing changes, faster time to target INR, and decreased out-of-range INRs,” says Karen Weck, MD, associate professor of pathology and laboratory medicine and director of molecular genetics at the University of North Carolina Medical Center. “Yet overall results are somewhat inconclusive.”

University and community clinicians appear to be reluctant to adopt this new approach until they see cogent proof that it improves warfarin dosing over clinical judgment. The largest randomized controlled trial to date, which compared an algorithm incorporating PGx to usual practice in an anticoagulation clinic, did not achieve its primary endpoint—reduction in out-of-range INRs (Anderson JL, et al. Circulation. 2007;116:2563–2570). “What that tells me,” says Alan Wu, PhD, “is that their practice is cutting edge. They already do a very good job at warfarin dose prediction. There is not that level of expertise in the community. That is where I think pharmacogenomics testing will be most useful.” Dr. Wu is professor of laboratory medicine at the University of California, San Francisco, and section chief of clinical chemistry and toxicology at San Francisco General Hospital.

Several professional groups have found current evidence insufficient to recommend routine use of this screen. A guideline from the American College of Chest Physicians says, “At the present time, for patients beginning [vitamin K antagonist] therapy, without evidence from randomized trials, we suggest against the use of pharmacogenetic-based initial dosing to individualize warfarin dosing (Grade 2C)” (Hirsh J, et al. Chest. 2008;71S–109S). Similarly, the California Technology Assessment Forum said “the use of genetic testing to guide initial warfarin dosing does not meet Technology Assessment Criteria 3 through 5 for safety, effectiveness and improvement in health outcomes” (www.ctaf.org/content/general/detail/814). An American College of Medical Genetics Working Group reached a more ambivalent conclusion: “[T]here is insufficient evidence, at this time, to recommend for or against routine [genetic] testing in warfarin-naive patients.” The ACMG statement also said, “Prospective clinical trials are needed that provide direct evidence of the benefits, disadvantages, and costs associated with this testing in the setting of initial warfarin dosing” (Flockhart DA, et al. Genet Med. 2008;10:139–150). A recent independent literature review concluded: “Recent randomized clinical trials validate the ability of genotype-guided dosing to estimate warfarin maintenance doses. However, to date, we do not have supporting evidence to suggest that genotype-guided therapy will improve anticoagulation control and prevent or reduce the risk of hemorrhagic or thromboembolic complications” (Limdi NA, DL Veenstra. Pharmacotherapy. 2008;28:1084–1097).

Several studies are planned or ongoing, including a large multicenter randomized controlled trial sponsored by the National Institutes of Health and National Heart, Lung and Blood Institute that is scheduled to get underway this year and an observational trial sponsored by Medco and the Mayo Clinic.

Another reason for the lack of uptake of PGx testing for warfarin is that it is not obvious how to translate genetic data into drug doses rapidly in clinical practice. “Reluctance in the coagulation community to incorporate genetic information is due particularly, I think, to a lack of demonstrated clinical utility and to the fact that there are not clear guidelines for dosage recommendations based on genotyping,” Dr. Weck says. “A question remains how genetic information will be utilized clinically.”

At a session on “Controversial Topics in Coagulation” at CAP ’08, Gwen McMillin, PhD, DABCC, presented the basics of PGx testing for warfarin. Summarizing her main messages in an interview with CAP TODAY, Dr. McMillin, assistant professor of clinical pathology at the University of Utah School of Medicine and medical director of toxicology at ARUP Laboratories, says, “It has been demonstrated many times that genetics plays a role in appropriate warfarin dosing.” Two main genes have been shown to affect warfarin action: cytochrome P4502C9 (CYP2C9), which governs warfarin metabolism; and vitamin K epoxide reductase complex 1 (VKORC1), which determines sensitivity to warfarin. Mutations in the CYP2C9 gene termed *2 and *3 reduce warfarin metabolism, while the 1639 mutation in VKORC1 alters sensitivity. Both are associated with lower warfarin dose requirements.

Despite the existence of complicated algorithms, Dr. McMillin says, “How to actually apply genetic information remains controversial. We need validated tools to help physicians in busy clinical practices, and we need large prospective randomized trials with well-designed dosing algorithms that are easy to use, fast, and safe.” Referring to the cautious tone of current guidelines, she adds, “No one is endorsing it for routine care right now.”

Many well-validated commercial platforms are available for PGx testing for warfarin. Four are approved by the FDA: AutoGenomics, Paragon Dx, Osmetech, and Nanosphere. “The methods vary a lot in terms of time required to generate results,” Dr. McMillin says. “By far the fastest are those based on real-time PCR.” Kits from Paragon Dx and Idaho Technology fall into this category. Arrays, such as the Luminex, AutoGenomics, Nanogen, and Osmetech products, take longer. Dr. McMillin has evaluated five of these platforms—AutoGenomics, Osmetech, Idaho Technology, Third Wave (which has since left this market), and Luminex. “All were 100 percent accurate and precise for detecting the three common variants,” she says.

Her laboratory offers PGx testing for warfarin using the Idaho Technology platform. The laboratory report suggests whether warfarin dose or elimination kinetics, or both, are expected to be affected based on the PGx results, but does not give a specific starting dose. The test was introduced two years ago and uptake has been slow, Dr. McMillin says. “We do get regular orders, but it is not a high-volume test.” Dr. McMillin’s explanation for this is that PGx for warfarin is not mandated, it is hard to interpret and apply, and the turnaround time may not be fast enough to affect patient care, a particularly challenging issue for a reference laboratory.

At the 2008 meeting of the Association for Molecular Pathology, Charles S. Eby, MD, discussed PGx testing for warfarin. Dr. Eby, associate professor in the Division of Laboratory and Genomic Medicine in the Department of Pathology and Immunology at Washington University School of Medicine, has worked from the outset with a group that has pioneered the study of the pharmacogenetics of warfarin and that developed the so-called Gage algorithm, which is now available online (warfarindosing.org) (Gage BF, et al. Clin Pharmacol Ther. 2008;84:326–331).

“There are a lot of controversial angles to this topic,” Dr. Eby says. He summarizes two opposing viewpoints: “From FDA’s perspective, drug safety is a strong motivation to provide genetic information to health care providers and more encouragement to use molecular testing prior to or at the time of initiating warfarin. From the clinical academic medical world there is a strong drive to determine whether pharmacogenomics-based dosing will affect important clinical outcomes and to better assess its cost-benefit ratio.”

There is very little data now, Dr. Eby says, to evaluate the clinical benefit of pharmacogenomics testing at the time of starting warfarin. He is on the steering committee of the NIH/NHLBI-sponsored randomized controlled trial, which will try to remedy this. Though the details are not yet finalized, Dr. Eby warns, “The design of the study will not make everyone happy.”

He agrees that current evidence and reasoning about the value of PGx testing for warfarin are “inferential.”

“That implies that it will lead to greater safety, an evidence gap that has so far not been filled,” he says. Advocates of routine PGx testing for warfarin assume that improving accurate dose prediction to 50 percent or higher will reduce the number of bleeding and clotting complications. “That’s an expensive assumption,” Dr. Eby points out. Charges run about $250 to test for the three pertinent genetic variants, he says. About 800,000 patients start warfarin annually (a hard number to verify), for a total annual cost of about $200 million. People who complain about the cost of doing a randomized controlled trial need to take this cost into account.

Dr. Eby notes that this cost is only for warfarin-naive patients. “I wouldn’t be surprised if some clinicians started ordering these tests for patients already on warfarin,” he says. “We have no indication to support pharmacogenomics in those patients. So it’s ­really important to nail this down as a test case for personalized medicine.” As one of the first applications, warfarin should be done “by the book,” he says, “especially since some people argue that we can’t afford to do randomized trials on every pharmacogenomics test.”

Dr. Eby and his colleagues have evaluated four platforms for warfarin PGx—AutoGenomics, Third Wave, Luminex (King CR, et al. Am J Clin Pathol. 2008;129: 876–883), and Osmetech. “We have confirmed excellent agreement between the four commercial platforms and our primary genotyping method, pyrosequencing,” he says. “I am confident that commercial methods are currently providing accurate genotyping information.” The CAP is offering proficiency testing for PGx-based warfarin dosing. “To date, accuracy seems to be very good for both FDA-approved and in-house methods,” Dr. Eby says.

His concern is not that laboratories will have difficulty providing accurate genotyping data, but how that information will be provided to clinicians and how labs will help interpret the data. “As it stands right now,” Dr. Eby says, “giving clinicians a better tool without appropriate education is like giving a bad driver a more expensive car without driver education—it doesn’t make them a better driver. Giving clinicians pharmacogenomic information without giving them fairly specific guidelines for how to use that information—I’m not sure it will translate into better care.”

Dr. Eby’s laboratory does warfarin PGx only for clinical research. He uses the Gage algorithm to convert genotypes into dose recommendations. A separate molecular diagnostics laboratory in the hospital does testing for clinicians, providing a general recommendation but not specific doses. “The volume [in that laboratory] has been low,” Dr. Eby says. “Major organizations have not recommended genotyping for initial warfarin dosing and I think that carries considerable influence in our academic hospital.”

Dr. Weck points out two subtle but important aspects of PGx algorithms for warfarin dosing. First, they are mostly based on studies of Caucasians, with a few exceptions (Limdi NA, et al. Pharmacogenomics. 2008;9:1445–1458; Wu AH, et al. Pharmacogenomics. 2008;9:169–178). “Those two algorithms performed somewhat better in our patient population,” Dr. Weck says, perhaps because African-Americans make up 30 percent of UNC patients. Second, “All algorithms so far have looked at prediction of final stable warfarin dose. It is unclear at this time how well those algorithms will do at predicting optimal initial dose. More work needs to be done in this area.”

Dr. Weck and her colleagues at the University of North Carolina Medical Center have started a single-center study of initial warfarin dosing guided by an algorithm incorporating PGx information compared with the same clinical algorithm without genotyping. “We are asking specifically what is the effect of adding genotyping to an algorithm rather than comparing PGx to current practice,” she says. They are enrolling all inpatients started on warfarin as well as patients seen at the Doppler clinic for workup of deep vein thrombosis. The goal is to recruit 100 patients in each arm and to follow them for three to six months.

Outcomes measures include the number of therapeutic-range INRs, time needed to get to a stable INR, and number of INRs greater than four, a surrogate for adverse bleeding events. They are asking, too, whether genotype-guided warfarin dosing will be cost-effective. Parameters for cost are the number of visits to the coagulation clinic, number of hospitalizations for anticoagulation, number of procedures for over or undercoagulation, and the cost of laboratory testing. Dr. Weck is using the Idaho Technology kit, which she validated in-house. One reason for this choice is its rapid turnaround time.

“It has never been demonstrated that a rapid turnaround time is required,” Dr. Weck says, “but our experience has been that clinicians are not willing to wait to start a patient on this important therapy. In principle, you would want to know about genetic variants that will affect a patient’s response prior to putting them on therapy, so a fast turnaround time could be a crucial factor.”

Some of the clinicians at UNC have been enthusiastic about PGx and others are more reluctant. “But all of them have agreed that there needs to be evidence, so they are enthusiastic for us to do this trial,” Dr. Weck says. “That’s why we decided to do a clinical trial rather than simply bringing the test online. Clinicians want to find out more about the test’s clinical utility.”

She describes as “great” the collaboration in this trial among the laboratory, geneticists, coagulation specialists, internal medicine physicians, and cardiologists—and particularly with the clinical pharmacy. “We do genotyping and pharmacy runs the algorithm and makes dosing recommendations. With pharmacogenomics in general, collaboration with pharmacy is key,” Dr. Weck says. “There is more acceptance by clinicians for pharmacists to make dosage and drug recommendations than the laboratory.”

Howard L. McLeod, PharmD, is director of UNC’s Institute for Pharmacogenomics and Individualized Therapy, which is coordinating the trial, and it is he who has been “crucial in forging these collaborations,” Dr. Weck says.

“We took a close look at what has gone right and wrong with personalized therapy,” Dr. McLeod says, “particularly in the use of genetic markers. We found a number of bottlenecks. It’s not only that there was not data from large clinical trials. In addition, some people who normally get involved in the process were missing.” For instance, health economists and people experienced at changing behavior at the policy and practice level hadn’t participated. “This institute draws across the entire university to pull those people together,” Dr. McLeod says.

About warfarin PGx specifically, Dr. McLeod says, “We found that the endpoints most exciting to academic physicians would not necessarily move the field.” Severe cerebral bleeds are one egregious example. “From community physicians’ viewpoint they are costly but rare, so they are not a big mover,” he explains. “The health system is much more interested in clinical efficiency. We need to manage warfarin with fewer INRs and 6 PM urgent calls, better dosing from the start. That is something they will adopt right away.”

Dr. McLeod says very few physicians understand whether CYP2C9*3 is good or bad. “That information needs to be put into a clinical decisionmaking algorithm using pharmacists as intermediaries. We spent a fair amount of time looking at what will happen when this is a routine test,” he says. They found that the group most equipped to apply information and asked to do it in normal practice is clinical pharmacists. “So we decided even in the context of the clinical trial to include them. They already individualize drug dosages based on drug levels or drug interactions or organ function. It is not a big deal for them to add genetics.”

Dr. Wu is also an advocate of clinical pharmacy. At a luncheon seminar at CAP ’08 he discussed setting up a clinical PGx service. “My principal message,” he says, “was that pharmacogenomics requires a collaborative effort among many different disciplines, including, and perhaps most critical, clinical pharmacy.” From his laboratory, PGx results are sent to the pharmacists for interpretation. “This cannot be done by the lab alone,” Dr. Wu says. “We are not involved with clinical decisionmaking as clinical pharmacists are.”

One feature of Dr. Wu’s practice at San Francisco General makes warfarin PGx for initial dosing particularly important. “I practice in a county hospital with a large indigent population,” he notes. “Their objective is where are they going to sleep tonight, not coming in regularly for INR adjustments. In this scenario you have one chance of getting the dose right or patients are likely to get lost to followup and end up having trouble.”

Dr. Wu says his experience might be a model for other people thinking about starting a PGx service for warfarin and other drugs. “I had the focus and vision to capitalize on the situation,” he says, “but my expertise was not preexisting. It was acquired in the last three to four years since I arrived here. It hasn’t been simple. I’ve required a lot of support from my department to make this happen, which is very difficult to get today. People say they want to do it but are not willing to provide resources.”

In addition, there was strong interest from clinicians. “They came to us. We had to respond to them,” he says. “Most people will not encounter that. In some places you will run into heavy resistance instead.” Dr. Wu attended a session at the 2008 AACC meeting in which four speakers presented pros and cons for warfarin PGx. “Pro” arguments came from two PhD laboratory directors. “They said that the evidence is there to move forward with this technology,” Dr. Wu recalls. Two clinical practitioners gave the contra side. “They felt the technology is not ready,” he says, “and that anticoagulation is not something they thought they needed help with.

“That was a very interesting dynamic,” Dr. Wu reflects. “It is a microcosm of what really happens at many places. Clinicians believe they have a sufficient knowledge base. That may be true in specialized centers, but the majority of warfarin is not prescribed in major medical centers. It is prescribed by community doctors, who do need help. But if they don’t see the experts using the method, they will be reluctant to try it.”

People who came to the CAP ’08 seminar enjoyed the lectures and gave the speakers high marks, Dr. Wu says. But when asked if the information was immediately applicable to their current practice, many attendees said no. “That tells me that many people are interested in the topic but are not actually in a position to do testing right now,” he says. “Perhaps they lack knowledge or resources or demand from clinicians.” Dr. Wu concludes that it is early to start thinking about a pharmacogenomics lab in every hospital. “I don’t think that will ever happen,” he says. “Right now even the majority of major medical centers are not doing this testing.” Among the 12 sites in the NIH/NHLBI trial, most are not offering warfarin PGx in their clinical practice. “This testing is still very much in its infancy,” he says.

Further evidence of this comes from the CAP PGx Survey, which was launched in 2007. Enrollment jumped from 17 in 2007 to 90 participants in 2008, suggesting spread of the technology. However, the number of laboratories returning Survey results in 2008 was only 25 or 30. “People are ordering Surveys, maybe doing development, but not actually returning results,” Dr. Wu says. “So there is interest, yes, but actual practice is still scant.”

Dr. Wu’s lab will be one of the 12 sites in the NIH/NHLBI ­trial, in which PGx will be used to establish the initial dose for high-risk patients those surviving myocardial infarction or stroke and some of those with atrial fibrillation. He agrees that, while there are a lot of data showing that warfarin pharmacogenomics can help better predict a stable anticoagulant level, it hasn’t been fully shown that it improves clinical outcomes. “These are very difficult trials to do,” Dr. Wu says. “A major reason is that the incidence of bleeding and other clinical side effects is not that great, so we have to have a couple of thousand patients to demonstrate statistical significance. It costs millions of dollars and it is a lab test, not a drug, so there is no company that will sponsor that kind of study.” Fortunately, NIH/NHLBI stepped in to fill the gap.

Stephen E. Kimmel, MD, MSCE, of the Center for Clinical Epidemiology and Biostatistics and associate professor of medicine and epidemiology at the University of Pennsylvania School of Medicine, will head the NIH/NHLBI trial, in which an algorithm incorporating genetic information will be compared with a clinical algorithm to set initial doses. Which specific algorithms will be used has not yet been finalized, Dr. Kimmel tells CAP TODAY. Endpoints also are still under discussion. Two other aspects of the trial still being decided are the selection of a PGx assay and inclusion of a cross-comparison of available assays on a subset of patient samples, says Maryellen de Mars, PhD, director of clinical biomarkers at the Critical Path Institute, or C-Path. C Path is coordinating the diagnostics aspects of the study for the NHLBI, the results of which will help support the FDA’s critical path effort to help standardize the evaluation of diagnostics.

More rigorous evidence is necessary for the routine adoption of warfarin PGx than for typical laboratory tests because “information from this test is going to be used to alter dosing of a drug with a narrow therapeutic index,” Dr. Kimmel says. “It is not like C-reactive protein (whose utility has been demonstrated, by the way). It is not a lab test for diagnostic purposes, but one that will alter the dose of a potentially dangerous drug.” His own view is this: “We know pretty well at least two genes associated with variable response to warfarin, but we don’t know if using information from those genes to help dosing will have any clinical benefit to our patients. We need a proof-of-concept study to show that this is the case.” At least among a segment of the anticoagulation community, he adds, “there is equipoise about whether this will be a useful method. Many practitioners want good, solid evidence that what they practice will benefit their patients. We have been fooled enough in clinical medicine by hypotheses of what is going to work. It is important to test this innovation in clinical trials.”

In addition to warfarin, Dr. Wu is offering three other PGx tests as clinical services: UGT1A1 to prevent hematologic complications from irinotecan; HLA-B to avoid severe hypersensitivity reactions from abacavir; and CYP2D6 to determine which breast cancer patients will benefit from tamoxifen as opposed to an alternative therapy, such as an aromatase inhibitor. Evidence for PGx for irinotecan and abacavir is clear, Dr. Wu says, while the tamoxifen application is still controversial. The FDA mandated that these drugs and others be labeled to indicate that PGx testing is available and might help. “In my opinion, FDA has been the driver for introduction of clinical pharmacogenomics in the U.S.,” Dr. Wu says.

Explaining the FDA’s position on improving the safety of warfarin through PGx specifically, Lawrence Lesko, PhD, director of the FDA’s Office of Clinical Pharmacology, cites several important facts about warfarin: It has a narrow therapeutic index; final stable doses cover a 20-fold range; and warfarin is one of the drugs that has the most adverse events reported, both in the literature and via FDA’s Medwatch system. “Age, gender, and body weight explain only about five percent of the variation in dose requirement but are nevertheless used routinely to dose warfarin,” Dr. Lesko says. “Still, all physicians use those criteria to help dose warfarin without the rigorous validation applied to genetic testing. And they accept and use INR to monitor it even though we know it doesn’t work very well, so INR is a clinically acceptable endpoint.”

Aggravating these problems, three of four patients treated with warfarin in the U.S. are not seen in high-quality anticoagulation clinics, Dr. Lesko says. “They are seen in community primary care settings. So they do not have the high-quality oversight that patients get in specialty clinics.” Dr. Lesko calls this situation “a very ineffective way of dosing. That overall is the problem,” he says. “It takes weeks or months to get patients stabilized on this drug. In that time there is a well-documented incidence of adverse events.”

Enter PGx. “Over the last five to seven years it has been demonstrated in many studies that variants in two genes account for approximately 35 percent of the variation in warfarin dose ­requirement,” Dr. Lesko says. “As a result, FDA updated the warfarin package insert to include genetic factors” (www. fda. gov/ cder/ drug/ info page/ warfarin/ default.htm). “The evidence we used to support the labeling revision came from numerous studies demonstrating improved INR control—increased time in range, reduction in the number of tests necessary to reach a stable dose, reduction in dosing adjustments, and other parameters,” Dr. Lesko says. “And many prospective studies demonstrated a nice correlation between predicted dose and observed dose” (Wadelius M, et al. Blood. 2008;Jun 23 [Epub ahead of print]).

Dr. Lesko concedes that most of these studies have been observational and some retrospective. He points out that they were “conducted in a natural setting.” In his view, that makes them easier to generalize to community practice. “Many people feel that, when normal bias is controlled, prospective observational studies reflect more accurately how drugs are actually used,” he says. He calls randomized controlled trials, on the other hand, “contrived” based on the inclusion/exclusion criteria and the selection of the comparative arm. In addition, Dr. Lesko says, in randomized controlled trials “often the standard-of-care arm that genomics is compared to is not realistic in terms of typical practice settings. So they are more difficult to generalize to the population than observational trials.”

He recognizes that interpreting the data is challenging. “Physicians are not trained in genetics. We do have a translation gap. I believe in time somebody will fill that gap,” he says. Perhaps the FDA? “We are evaluating the possibility of putting genotype-specific dosing information in the label based on meta-analysis of many studies,” Dr. Lesko says. “That would provide physicians with clear-cut information to make actionable decisions.”

Gregory Tsongalis, PhD, of Dartmouth Hitchcock Medical Center, calls the FDA’s labeling changes “very mild.”

“Pharmacogenomics was not even recommended, much less required,” he says. “They just said there was a pharmacogenomics test out there.” Making PGx testing mandatory “would change everything,” Dr. Tsongalis says. “Whether they have sufficient data to do that is another question.”

Demand for warfarin PGx testing at Dartmouth has been “very, very slow,” says Dr. Tsongalis, director of molecular pathology, co-director of the pharmacogenomics program, and associate professor of pathology at the medical school. “I don’t think anybody could argue with the fact that warfarin is probably one of the most difficult therapeutics to dose properly. I think the problem with creating a market or implementing a pharmacogenomics test for it is that physicians have a lot of experience in using this drug. And there are a number of lab tests they already use to monitor patients that they are relatively comfortable with. And the drug itself is inexpensive. So all the arguments we’ve used to promote pharmacogenomics we can’t make for warfarin, other than some severe adverse events on occasion. So I’m not sure warfarin was the best drug to have started PGx testing with.” Oncology drugs would have been a better class of therapeutics for this purpose, in Dr. Tsongalis’ opinion. “They are just the opposite of warfarin,” he explains. “They are quite expensive, there is no other type of monitoring for evaluating patient response, and it’s a disease where any help in managing the patient is welcome.”

In this time of incomplete information, Dr. Eby says, “It is difficult to tell everybody to be patient and wait and see how the randomized trial comes out. Results of that trial will probably not be available for two-and-a-half to three years. In the interim perhaps there will be some event we can’t predict now that will rapidly accelerate the use of pharmacogenomics-based warfarin dosing—a medicolegal case or a high-profile individual having an adverse effect. Or,” he speculates, “it could be FDA making package insert changes that are so compelling they become standard of care.”

Or some event could occur that would cause warfarin PGx to be superseded, just as pop art superseded abstract expressionism. One possibility: the introduction of oral anticoagulants that do not require monitoring—for example, rivaroxaban or dabigatrin, two direct factor Xa inhibitors that are now in phase three clinical trials. If one of these drugs is approved in the next couple of years, the whole issue of warfarin PGx could become moot.