Lorcaserin -- Rejected by FDA for Being Too Safe?

Andrew J. Vickers, PhD, DPhil


February 08, 2011

Late last year, the FDA rejected an application for lorcaserin, a weight-loss drug, on the grounds that its benefits were outweighed by safety concerns. Although part of the decision was related to carcinogenicity studies in animals, a key issue involved side-effect data from the randomized trials. Drugs with a similar mechanism of action to lorcaserin were known to raise the risk for heart valve disease and, as a result, the trials were specifically designed to determine whether lorcaserin was associated with this toxicity. The investigators prespecified that for lorcaserin to be considered safe, the trial had to exclude the possibility that it raised the risk for cardiac valvular disease by 50% or more. To put this in statistical terms, the upper bound of the 95% confidence interval for relative risk had to be less than 1.5.

When the results came in, lorcaserin was associated with a nonsignificant 7% increase in the relative risk for valvular disease, with a 95% confidence interval of 0.74-1.55. As such, the trial missed its target and the drug's safety could not be confirmed.

One reason for this result was that the rate of cardiac valvular disease was much lower than expected and, as it turns out, the fewer events you have, the less sure you can be about relative risk. The lorcaserin data included about 2500 patients in each group, with about 2% experiencing valvular disease. Imagine the situation in which a drug is almost totally safe, with only 2 patients on the drug and 2 on placebo diagnosed with heart problems. The 95% confidence interval for absolute risk difference would be about ± 0.15%; that is, in the worst-case scenario, the drug would cause just 1 case of heart valve disease for every 650 patients treated. However, the 95% confidence interval for relative risk is very wide: 0.14-7 -- that is, a 7-fold reduction or increase in risk. In other words, an exceptionally safe drug would miss the target for drug safety.

Now imagine another scenario, in which the incidence of valvular disease was 10 times higher than observed, but lorcaserin led to a 20% increase in relative risk. The difference between groups would be statistically significant, with an estimate of 1 additional case of heart valve disease for every 25 patients prescribed lorcaserin. Although these results essentially would be a disaster, the trial would nonetheless meet the safety criterion with plenty of room to spare, with a 95% confidence interval for relative risk of 1.08-1.33.

The point isn't that the FDA made a bad decision -- weight loss on lorcaserin was less than spectacular -- but that it is irrational to make decisions based on relative risk. In a previous post. I pointed out that just as you shouldn't necessarily buy something just because it is 30% off in price, you shouldn't necessarily take a drug that promises 30% off your risk. The lorcaserin example shows that the opposite is true: You shouldn't necessarily avoid a drug that increases your risk in relative terms. For example, I enjoy the occasional glass of beer, and I can't see myself turning down a freshly poured Brooklyn Pilsner, even if I was told that it would double my risk for, say, a lightning strike. What matters in life is what you get and what you have to pay for it; how much more or less in relative terms is really neither here nor there.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: