Nature Medicine | Spoonful of Medicine

Doubts about usefulness of gene testing for antiplatelet drug prompt video rebuttal

A textbook example of pharmacogenomics is the testing of the CYP2C19 liver enzyme gene to predict how well people will metabolize the antiplatelet drug Plavix (clopidogrel). Many cardiologists have such confidence in the link that they preemptively genotype people before administering the drug. Last year, US regulators even added a ‘black box’ warning to clopidogrel’s prescribing information noting the the link between certain genotypes and reduced drug efficacy. But a new meta-analysis published in the Journal of the American Medical Association (JAMA) this week reports that such preemptive moves might be premature.

“The totality of evidence suggests that there’s no association between genotype and clinical outcomes,” says Michael Holmes, a clinical pharmacologist at University College London (UCL) who led the study.

Holmes and his colleagues cast a wide net in the literature and found 32 studies involving more than 42,000 participants that evaluated whether the CYP2C19  genotype affected the risk of cardiovascular complications for clopidogrel. The researchers found that small studies typically reported the strongest gene link to these adverse events — suggesting that the link between CYP2C19  and the antiplatelet drug might be based on a publication bias. When they restricted their analysis to studies reporting at least 200 such adverse events, they found no overall association between CYP2C19 genotype and health outcomes. In light of the new analysis, “physicians should use CYP2C19 or platelet reactivity testing rarely, if ever,” Steven Nissen, chief of cardiovascular medicine of the Cleveland Clinic in Ohio, wrote in an accompanying editorial.

Many experts took exception with this conclusion. One fundamental problem with the analysis, notes Dan Roden, assistant vice-chancellor for personalized medicine at the Vanderbilt University Medical Center in Nashville, Tennessee, is that the authors problematically included people with conditions such as stable coronary heart disease or atrial fibrillation, for whom clopidogrel hasn’t shown any clinical benefit. “If you throw a bunch of those patients in, then, well, jeez, you’re going to get that kind of result,” he says.

According to Roden, people undergoing stenting for the treatment of acute coronary artery disease stand to benefit the most from CYP2C19 genotyping before taking clopidogrel. Of the groups analyzed in the meta-analysis, those whose DNA predicted that they would be poor clopidogrel metabolizers suffered disproportionately more from stent thrombosis.

“With the goal of trying to include every possible study, I think [the authors] did a disservice by including studies where one wouldn’t even expect to see a benefit of pharmacogenetics for clopidogrel.” says Marc Sabatine, a cardiologist at the Brigham and Women’s Hospital in Boston, who published his own meta-analysis last year in JAMA demonstrating the benefit of genotyping among people receiving stents and antiplatelet therapy. And yet, at the same time, the authors chose to exclude any studies that only looked at rates of stent thrombosis but didn’t include other endpoints directly related to heart problems in their analysis, he notes.

“Unfortunately,” Sabatine says, “this study will set the field potentially back a bit.”

Indeed, the press coverage of the report has largely presented the authors’ conclusions without challenge. CNN’s morning show yesterday featured a Columbia University cardiologist explaining why genetic testing is no longer necessary. The Wall Street Journal’s report didn’t include a single critical comment.

But some leading cardiologists aren’t taking the report’s conclusions lying down. The new meta-analysis so irked Eric Topol, chief academic officer of Scripps Health in La Jolla, California, that he, together with many of his colleagues, have openly challenged the report. In a letter and video posted on Topol’s blog, hosted by, he, Roden and others called the report “remarkably misleading,” and based on “a dataset that doesn’t hang together.”

In an email to Nature Medicine, Topol says he is now working with Sabatine, Roden and others to prepare “a different, even stronger letter” to send to the editors of JAMA. “This flawed meta-analysis,” he wrote, “coupled with uninformed interpretation, has the potential to negate a prototypic pharmacogenomic effect — a real disservice to patients, the FDA and the field of genomic medicine.”

For their part, the authors of the new JAMA study stand by the methods of their analysis. “Clopidogrel is used in a number of clinical settings,” Aroon Hingorami, Holmes’ UCL colleague, wrote in an email. “I am sure there will be a range of questions and views on this matter. If these are raised in specific terms through correspondence with JAMA there will be an opportunity to address them accordingly.”

Image: Wikimedia Commons


  1. Report this comment

    Ryan Smith said:

    Not to mention the broad brush used in the original article suggesting that genotypes as a biomarker are not useful. One success, for example with warfarin dosing, is proof that genetic biomarkers have great promise. But as Dr. Topol et al. are getting at, its critically important to apply the markers in the correct context.

    1. Report this comment

      Ryan Smith said:

      To clarify, when referring to the broad brush, I didn’t mean the JAMA article, I was referring to the blog post on the Nature website yesterday.