The Food and Drug Administration (FDA) assesses the efficacy and safety of medical treatments and, since starting in March 2009, the current Commissioner has had a focus on safety. This may partly be a reaction to apparent "misses" by the FDA—the infamous Vioxx affair and the noise about rosiglitazone. Fear of other misses has led the FDA to generally mandate a large clinical trial to evaluate safety of any drug that might impact the CV system; they (and we) do not want to be "blindsided" again. Some rave and say, "It's about time!"

However, what has ensued is a nearly constant flow of "safety warnings" from the FDA. One group of researchers documented that, in 2009, the FDA added 135 new boxed warnings, contraindications, or warnings; updated 43; and downgraded only three.¹ But, I’m beginning to wonder, has the pendulum swung too far? Is the FDA now sometimes crying wolf? Is the threshold for placing a safety warning too low? The level of evidence seems to be much lower than that for claiming efficacy—but is that the right balance?

One example we’re all familiar with: in January 2009, the FDA called out proton pump inhibitors and clopidogrel in an "initial communication," followed by a full warning in November 2009 stating that "co-administration of clopidogrel and omeprazole be avoided because omeprazole reduces the effectiveness of clopidogrel." However, the randomized COGENT trial found no difference in cardiovascular outcomes, and in the nearly 3 years since the warning’s publication, the FDA has not updated their guidance and only published a letter to the editor of the New England Journal of Medicine saying that there were too few major CV events to prove no adverse effect.

What I find difficult here is that they say the evidence showing no harm is not sufficient (despite 3,500 patients in a randomized trial) yet the evidence supporting an adverse effect is weaker; it is based on platelet inhibition data from 50 patients and clinical data only from observational studies—studies which have since been refuted as showing differences in event rates only due to confounding. Shouldn’t the randomized data supplant the confounded observational data?

I wonder: Should the FDA include a grading of the level of evidence for safety warnings? I think this might actually be a good first step.

A related story is the CYP2C19 polymorphism that leads to reduced metabolism of clopidogrel: the FDA issued a warning initially, and a "black box warning" in March 2010, based on platelet function and pharmacokinetic data in 40 patients, and additional observational analyses that had also been published. However, when this issue was looked at in the genetics substudy of the randomized CURE trial—you guessed it—the prior findings didn’t appear to hold up. Clopidogrel was equally beneficial in reducing events in those with or without the genetic polymorphism.² This information, from more than 5,000 patients in a randomized trial, has not been added to the label for clopidogrel—why not?

Regarding statins, the FDA has gotten some of their warnings right, but one that gives me great worry is the warning on the decline in cognitive function associated with statins from February 2012. They appropriately added the new risk of development of diabetes, and even went out of their way to also note that the associated benefits appeared to outweigh the new diabetes risk. They also appropriately added the increased risk of rhabdomyolysis in the first year of using simvastatin 80 mg.

But, on top of all that, they also added "cognitive impairment (e.g., memory loss, forgetfulness, amnesia, memory impairment, confusion)" to the list of adverse effects. The FDA said they reviewed reports of their adverse events database (i.e., MedWatch forms), observational studies, and randomized trials. However, there are two large randomized trials (in 5,800 and more than 20,000 patients treated for 3.5 and 5.3 years, respectively) that demonstrated no change in cognitive function on formal testing. Yet the FDA still put out this warning. I had many patients and friends ask if they should stop taking their statins because of this insufficiently evidence-based warning—and I fear that some people did after reading this in the newspaper, which may have increased their risk of MI, stroke, or death.

Another area that puzzles me is the FDA's non-approval of the lower (110 mg) dose of dabigatran. The FDA said they saw better efficacy with the 150 mg dose, so they only approved the higher dose. But then the bleeding reports started coming in from real world—if we only had the lower dose available could the bleeding have been reduced? Ironically, in this situation, the safer dose was not approved by FDA!

So, should I rant about all these examples, or rave about all the excellent safety issues they have brought forth? I guess both. But I hope the FDA will be able to adjust its stance and update warnings as new information comes forth, even if that means they have to retract a warning. Hopefully, a matching of the warnings to the level of evidence will provide the best possible safety information for our patients and ourselves.

References

1. Moore TJ, et al. Arch Intern Med. 2012;172:78-80.
2. Pare G, et al. N Engl J Med. 2010;363:1704-14.

Christopher P. Cannon, MD, is a professor of medicine at Harvard Medical School in Boston, Massachusetts. He is also the Editor-in-Chief of CardioSource Science and Quality.

Keywords: Polymorphism, Genetic, Platelet Aggregation Inhibitors, Confusion, United States Food and Drug Administration, Memory Disorders, Proton Pump Inhibitors, Diabetes Mellitus, Rhabdomyolysis

< Back to Listings