One of the questions I get asked most often, by people outside of the drug industry, is whether generic medications really are the same as the original branded ones. My answer has always been the same: that yes, they are. And that's still my answer, but I'll have to modify it a bit, because we're seeing an exception right now. Update: more exceptions are showing up in the comments section.

Unfortunately, "right now" turns out, in this case, to mean "over the last five years". The problem here is bupropion (brand name Wellbutrin), the well-known antidepressant. A generic version of it came on the market in 2006, and it went through the usual FDA review. For generic drugs, the big question is bioequivalence: do they deliver the same ingredient in the same way as the originally approved drug and formulation? The agency requires generic drug applications to show proof of this for their own version.

For bupropion/Wellbutrin, the case is complicated by the two approved doses, 150mg and 300mg. The higher dose is associated with a risk of seizures, which made the FDA grant a waiver for its testing - they extrapolated from the 150mg data instead. And right about here is where the red flags began to go up. The agency began to receive reports, almost immediately, of trouble with the 300mg generic dose. In many cases, these problems (lack of efficacy and/or increased side effects) resolved when patients switched back to the original branded formulation. That link also shows the pharamacokinetic data comparing the two 150mg dosages (branded and generic), which turned out to have some differences, mostly in the time it took to reach the maximum concentration (the generic came on a bit faster).

At the time, though, as that link shows, the FDA decided that because of the complicated clinical course of depression (and antidepressant therapy) that they couldn't blame the reported problems on a difference between the two 300mg products. A large number of patients were taking each one, and the number of problems reported could have been explained by the usual variations:

The FDA considers the generic form of bupropion XL 300 mg (Teva Pharmaceuticals) bioequivalent and therapeutically equivalent to (interchangeable with) Wellbutrin XL 300 mg. Although there are small differences in the pharmacokinetic profiles of these two formulations, they are not outside the established boundaries for equivalence nor are they different from other bupropion products known to be effective. The recurrent nature of (major depression) offers a scientifically reasonable explanation for the reports of lack of efficacy following a switch to a generic product. The adverse effects (e.g., headache, GI disorder, fatigue and anxiety) reported following a switch were relatively few in number and typical of adverse drug events reported in drug and placebo groups in most clinical trials. . .

But they seem to have changed their minds about this. It appears that reports continued to come in, and were associated most frequently with the generic version marketed by Teva (and produced by Impax Pharmaceuticals). That FDA page I've quoted above is not dated, but appears to come from late 2007 or so. As it turns out, the agency was at that time asking Teva to conduct that missing bioequivalence study with their 300mg product. See Q12 on this page:

FDA continued to review postmarketing reports throughout 2007. In November 2007, taking into consideration reports of lack of efficacy, FDA requested that Impax/Teva conduct a bioequivalence study directly comparing Budeprion XL 300 mg to Wellbutrin XL 300 mg. The study protocol stipulated the enrollment of patients who reported problems after switching from Wellbutrin XL 300 mg to Budeprion XL 300 mg. Impax/Teva began the study, but terminated it in late 2011, reporting that despite efforts to enroll patients, Impax/Teva was unable to recruit a significant number of affected patients.

The agency apparently was continuing to receive reports of problems, because they ended up deciding to run their own study, which is an uncommon move. This got underway before Teva officially gave up on their study, which gives one the impression that the FDA did not expect anything useful from them by that point:

In 2010, because of the public health interest in obtaining bioequivalence data, FDA decided to sponsor a bioequivalence study comparing Budeprion XL 300 mg to Wellbutrin XL 300 mg. The FDA-sponsored study enrolled 24 healthy adult volunteers and examined the rate and extent of absorption of the two drug products under fasting conditions. In that study, the results of which became available in August 2012, Budeprion XL 300 mg failed to demonstrate bioequivalence to Wellbutrin XL 300 mg.

That FDA-sponsored study is what led to the recent decision to pull the Imapax/Teva 300mg product from the market. Their 150mg dosage is still approved, and doesn't seem to have been associated with any increased reports of trouble (despite the small-but-real PK differences noted above). And it's also worth noting that there are four other generic 300mg bupropion/Wellbutrin products out there, which do not seem to have caused problems.

How big a difference are we talking about here? There are several measurements that are used for measuring blood levels of a drug. You have Cmax, the maximum concentration that is seen at a given dosage, and there's also Tmax, the time at which that maximum concentration occurs. And if you plot blood levels versus time, you also get AUC (area under the curve), which is a measure of the total exposure that a given dose provides. There are a lot of ways these measurements can play out: a very quickly absorbed drug will have an early Tmax and a large Cmax, for example, but that concentration might come back down quickly, too, which could lead to a lower AUC than a formulation of the same drug (at the same nominal dose) that came on more slowly and spread out over a longer time period. To add to the fun, some drugs have efficacy that's more driven by how high their Cmax values can get, while others are more driven by how large the AUCs are. And in the case of bupropion/Wellbutrin, there's an additional complication: some of the drug's efficacy is due to a metabolite, a further compound produced in the liver after dosing, and such metabolites have their own PK profiles, too.

So in this case, it turns out that the AUC just missed on the low side. The FDA wants the statistical 90% confidence interval to fall between 80 and 125% compared to the original drug, and in this case the 90% CI was 77-96%. The Cmax was definitely lower, too - 90% CI was 65-87% of the branded product. And while the agency doesn't provide numbers for the metabolite, they also state that it missed meeting the standards as well. There are drugs, it should be said, that would still be effective at these levels, but Wellbutrin clearly isn't one of them.

My own take is that the FDA was willing to consider the adverse reports as just the usual noisy clinical situation with an antidepressant until the other generics were approved, at which point it became clear that the problems were clustering around the Impax/Teva product. Here's how the FDA addresses the "Why didn't we find out about this earlier?" question:

Q17. In retrospect, were FDA’s decisions regarding the approval and ongoing monitoring of Budeprion XL 300 mg appropriate?

A17. A less cautious approach in studying the bioequivalence of Budeprion XL 300 mg could have brought the data to light earlier. The FDA-sponsored study was completed only weeks ago, which is a very short time for data from a clinical experiment to be announced to the public.

Bupropion is associated with a risk for seizures, which was the basis of the Agency's cautious approach with regard to the early Budeprion XL bioequivalence studies, in which data were extrapolated from Budeprion XL 150 mg in patients to the projected consequences of exposure to Budeprion 300 mg. In retrospect, it is clear that this extrapolation did not provide the right conclusion regarding bioequivalence of Budeprion XL 300 mg. FDA also has much more knowledge today of the seizure-associated risk of bupropion-containing drugs. The trial design of the sponsor-initiated study of 2007 could have been successful, had it been replaced by the trial design employed in the recent FDA-sponsored study.

Of course, the trial design in the sponsor-initiated study of 2007 was that requested by the FDA. But Teva, for their part, does not appear to have been a ball of fire in getting that study recruited and completed, either. It's quite possible, though, that they couldn't round up enough patients who'd had trouble with the generic switch and were also willing to go back and experience that again in the cause of science. Overall, I think that the FDA is more on the hook here for letting things go on as long as they did, but there's plenty of blame to go around.

Still, I find this post at Forbes to be full of unnecessary hyperventilation. You wouldn't know, from reading it, that the FDA initially waived the requirement for 300mg testing in this case because of the risk of seizures. There's a line in there about how the agency is making patients their guinea pigs by not testing at the higher dose, but you could have scored the same debating points after a 300mg study that harmed its patients, which is what it looked at the time would happen. You also wouldn't know that the other generic 300mg formulations don't seem to have been associated with increased adverse-event reports, either.

And that post makes much of the way that these bioequivalence tests are left up the manufacturers. That they are: but if you want to change that, you're going to have to (1) fund the FDA at a much higher level, and (2) wait longer for generic switches to occur. The generic manufacturers will run these tests at the absolute first possible moment, since they want to get onto the market. The FDA will run them when they get around to it; they don't have the same incentives at all. Their incentives, in fact, oscillate between "Don't approve - there might be trouble" and "Definitely approve - we might be missing out on benefit". The winds of fortune blow the line between those two around all the time.

In this case, I think the FDA should have exercised its court-of-last-resort function earlier and more forcefully. But that's easy for me to say, sitting where I am. I don't have to see the mass of noisy adverse event reports coming in over the transom day after day. If the agency acted immediately and forcefully on every one, we'd have no drugs on the market at all. There's a middle ground, but boy, is it hard to find.