January 22nd, 2010

Chantix and Cardiovascular Risk: Another Weak Safety Study

In 2008, Dr. John Spangler of the Wake Forest University School of Medicine wrote a letter to the editor of Current Medical Research and Opinion expressing concern about a Pfizer-funded, randomized, placebo-controlled trial of the smoking-cessation drug varenicline (Chantix). By 1 year, the varenicline group had experienced a higher rate of serious adverse events than the placebo group; many were cardiovascular (CV) events. The difference was not statistically significant, but Spangler considered it clinically significant and deserving of further study.
When I initially read Spangler’s letter and reviewed the varenicline trial data, I was equally concerned. CardioExchange editor-in-chief Dr. Harlan Krumholz and I approached Pfizer about performing a pooled, subject-level meta-analysis of all clinical trial data on varenicline, to fully evaluate its CV safety using methods we were then applying to our investigation of rofecoxib. Pfizer assured us that the meta-analysis was unnecessary because CV-safety concerns about varenicline were being addressed in a forthcoming randomized trial.
That Pfizer-funded trial has now been published in Circulation. The investigators randomized 714 smokers with stable CV disease to receive varenicline or placebo for 12 weeks. The rate of adjudicated CV events at 1 year was 7.1% in the varenicline group and 5.7% in the placebo group, again a statistically nonsignificant difference. The authors conclude that “varenicline treatment did not increase the risk of cardiovascular events . . . However, trial size and duration preclude a definitive conclusion about the safety of varenicline.”
I applaud the authors’ candor about the limitations of their trial, but I have difficulty understanding its design given the goal of investigating CV safety. Three main areas concern me:
The trial excluded smokers who had undergone a CV procedure or exhibited CV instability, including myocardial infarction or unstable angina, in the previous 2 months. Those are precisely the patients who have the greatest motivation to quit smoking and are at the highest risk for CV events.
2. Varenicline use was examined over only 12 weeks, and serious adverse events were followed out to 1 year, consistent with previous trials. However, given that quit rates are below 50% at 12 weeks, it’s reasonable to expect that many patients will use the drug for longer than that.
3. My biggest (but also most technical) concern has to do with the 1.4% CV-event difference. Only a trial involving nearly 10,000 subjects could find such a difference to be statistically significant (at 80% power). A similarly powered trial involving about 700 subjects could detect a statistically significant difference only if the CV-event difference were at least 6.2% between the groups. That would be essentially double the CV-event rate — and would exceed the CV risk from smoking!
So what do I make of another underpowered safety study? My overall instinct is that varenicline is safe, particularly when I weigh the anticipated CV benefits from smoking cessation against the potential CV risk from the drug. But I would prefer to have stronger evidence on which to base my decision.
Are you concerned about varenicline’s CV risk? Do the findings from the new trial make you less concerned or persuade you to change how you view the drug in clinical practice? For your patients at highest CV risk, are you more inclined to use other smoking-cessation drugs (which have lower efficacy rates) or varenicline as first-line therapy? Please share your thoughts with your fellow clinicians here on CardioExchange.

2 Responses to “Chantix and Cardiovascular Risk: Another Weak Safety Study”

  1. I’d like to see more detail about the randomisation methods (stratified, use of blocks etc?) as there seemed to be some sizeable differences in some of the baseline measures (sex, non-cigarette tobacco use, diabetes).
    I find it implausible that anyone could assert that this trial was ever going to make a meaningful contribution to the cardiovascular safety knowledge. It was designed and powered around an efficacy outcome and that is what it tells us about.
    Assessing the CV safety in an RCT would be very expensive, but the previously suggested individual patient data meta-analysis would seem a reasonable requirement from a regulator. Then everyone can balance the risks and benefits and make an informed decision.

  2. Efficacy studies and equipoise

    Martin, I agree with you that the trial seemed designed and powered to examine efficacy, not safety. This raises the issue of equipoise, since varenicline (Chantix) had been proven effective in several randomized trials (3 published in JAMA in 2006), although from my reading of the literature, all of the previous trials excluded patients with cardiovascular disease. Nevertheless, why compare to placebo instead of to nicotine-replacement therapy? It remains to be seen whether the data in their totality would raise a sufficiently red flag to attract attention from regulators, but I agree that a meta-analysis (either summary-level or preferably subject-level) is warranted. As I said previously, given the health benefits from quitting tobacco, I can’t imagine that the drug’s (possible) risk outweighs it. But I would prefer to have an explicit examination on which to base my decision to use the medication.

    Competing Interests I was compensated for my work from 2006-2007 as a consultant at the request of plaintiffs in litigation against Merck & Co. Inc. related to rofecoxib.