On May 10, an FDA advisory committee will be trying to balance the effectiveness vs. the risks of Lorcaserin, Arena Pharmaceutical's (ARNA) weight loss drug. A lot of people think that Lorcaserin is only 3% effective, compared, for example, to Qnexa, the candidate from Vivus Pharmaceuticals (VVUS), which is 10% effective. The implication is that Lorcaserin is barely better than a placebo, in other words, barely better than nothing. But in my recent article, I quoted Lorcaserin as losing 8% of body weight. This difference is crucial to the FDA's upcoming decision. So how effective is Lorcaserin and why is efficacy so confusing?
In the usual placebo study, the drug is given to one group, and a sugar pill is given to the other. We can then discuss placebo-adjusted results (the effects in the drug group after subtracting the placebo effects) to rule out the effect of simply being in a study, which will not be the case in the real world. This model is based on thinking of the placebo as a mostly inert substance.
The 3% number is Lorcaserin's placebo-adjusted result. However, Lorcaserin's major phase III studies had an unusual design. The FDA wanted to rule out the effects of lifestyle interventions, so they included 12 sessions with a nutritionist in both groups. This causes two problems in talking about the placebo-adjusted results.
- Placebos that work too well
Arena decided to stretch out their 12 sessions with a nutritionist into once a month over the course of the trial year, and neglected to rule out simultaneous enrollment in programs like Weight Watchers or Jenny Craig. This is, I think, the fault of the company.
Subjects effectively signed contracts to be monitored for a year. The usual problem of losing focus and going back to normal living without the diet is mitigated. The subject forms a relationship with the nutritionist and wants to show a good result. The interval of one month between visits may be short enough in many cases to maintain motivation.
The end result was that the company stumbled into a "placebo arm" that was significantly effective. Those people lost 3% of body weight over the course of a year.
Vivus was smarter about it and crammed their 12 sessions into the first 3 months so that their subjects got nine months without supervision, and they also specifically ruled out "[P]articipation in a formal weight loss program (including: Weight Watchers and related dietary/lifestyle intervention programs; prepared food programs; prescribed or over the counter weight loss medications;…or any supervised fast or very low calorie diet)." Under those conditions, the "placebo" subjects lost only 1.5% of body weight, probably closer to what was intended.
Overall, Lorcaserin's subjects lost 6% of body weight, but as a placebo-adjusted result, it is only 3%.
A placebo-adjusted result estimates the overall real-world effectiveness only if the placebo is truly inert, having no effect, or if the effect of the drug and the placebo are independent with no overlap. But that's not realistic here.
A thought experiment: Another study includes not just 12 sessions with a nutritionist, but also personal trainers for three hours a week, inspirational DVDs once a week, and takes $5,000 from each subject, agreeing to give it back only if weight loss is achieved (which has been shown to work, by the way). Imagine what would happen to the placebo-adjusted results for Lorcaserin and Qnexa. They would go down, wouldn't they? Would that mean the drugs suddenly became less effective?
OK, so the FDA's idea was to compare to the current standard of care. Let's examine that.
- Lifestyle comparisons that are idealized and unrealistic
The lifestyle intervention selected is not in any way what the meds would be competing with in the real world, people dieting and exercising on their own, which we know generally doesn't work well at all. However, for those dieting and exercising, adding a pill every day is something that they can do.
Hourly rates for a private nutritionist consultation typically costs at least $150/hour. For 12 sessions, $1800. In addition to the costs, the idea of inviting a stranger into a private area of one's life is unpalatable for many people, and there is the time and effort involved in finding one, contracting, and meeting. That would be versus taking a pill for perhaps $2/day, $730 a year, without all the other requirements.
You probably know plenty of people who have dieted and exercised to try to lose weight. Do you know anyone who contracted their own nutritionist for multiple consultations?
Decades ago, a major study by the National Institute of Mental Health compared the best possible psychosocial interventions against Ritalin for kids with Attention Deficit Disorder (ADD). They found that the psychosocial interventions were as effective as the pills. The combination was not much better (like about 3%, sound familiar?). The conclusion: The psychosocial interventions were idealized and unrealistic for most people. The average adult with a kid with ADD isn't going to hire a family therapist, an individual therapist for the kid, someone to follow the kid around while he is in school, etc. A pill is easier, cheaper, and more practical. Because of that decision, millions of kids can now do well at school and not have every adult in their life yelling at them.
It doesn't make sense to say that we don't want these obesity pills on the market because it's theoretically possible for people to hire their own nutritionists and come close to the same weight loss. That is not going to happen. Without the pill, people keep going with attempts at diet and exercise on their own. With the pill, there is help available, which will increase people's motivation for diet and exercise.
I think that this unrealistic comparison design was part of the FDA's earlier bias against obesity meds because of the perception that obesity is merely a lifestyle choice. There is reason to think that is now changed.
- What is the best estimate of Lorcaserin's effectiveness?
Best for comparing across drugs is the percent of body weight lost. In an intent-to-treat analysis (ITT), which is everyone that entered the trial, including those that dropped out after one day, the Lorcaserin groups lost 6% of body weight. In a more complete analysis, which is everyone that finished the one-year trial, the Lorcaserin groups lost 8% of body weight. There is a reasonable argument people dropping out may have done so because they were not losing weight. Therefore, 8% might be a bit too high, so the true number seems somewhere between 6 and 8%.
How do we know what Lorcaserin would do without a nutritionist? Unfortunately, we don't to a high degree of scientific certainty. But we can make some informed judgments. In 2009, a 12-week study of Lorcaserin in 469 obese subjects showed about 3.5% of weight-loss . The charts of the weight loss from the year-long studies show that almost all of the weight loss for Lorcaserin is achieved by week 24. Between week 12 and 24 there is a doubling of the loss. Thus, applying that doubling to the 2009 study, it is reasonable to surmise that Lorcaserin subjects would lose approximately 6% of body weight by week 24. This agrees with the other findings.
In conclusion, placebo control groups in a case like this should be used only for proving that the intended intervention has the intended effect, over and above the placebo. That question should be asked in a binary yes/no question, "was the intended effect demonstrated or not." Lorcaserin did demonstrate an effect, although just barely, by the criteria set up by the FDA beforehand, and Qnexa clearly did. The next question should be, "What is the best estimate for the real-world effectiveness?"
- Investment Considerations
No one outside the FDA knows for sure how the FDA will look at this, certainly not I. But let's assume they are intelligent, well-meaning people. A good clinician, like a good FDA reviewer, should not always use the standard protocols. That is why we don't only enter data into a computer read by clerks when we go to our doctor's offices. A good clinician uses his or her judgment to decide when a particular patient may be an exception.
In the same way, FDA reviewers should not blindly stick to placebo-adjusted results in a case like this. Is it the fault of the drug or the public that the company accidentally stumbled onto a too-effective placebo, or that the design demanded in an earlier time was biased? If not, then why punish them? If the FDA uses the 6-8% estimate of Lorcaserin's effect, the risk vs. efficacy decision for Lorcaserin would be no worse than 50/50 for approval.
VVUS is obviously a safer investment than ARNA. Qnexa is well positioned to pass the FDA review. VVUS just got approved for a promising erectile dysfunction drug, and even if the FDA approves both Lorcaserin and Qnexa, large numbers of general practice docs may be prescribing Qnexa for those who don't respond well enough to Lorcaserin.
Whether you use intent-to-treat or completer analysis, Lorcaserin provides about 75% of the weight loss that Qnexa does. Qnexa is very likely to pass the FDA's upcoming review, but a number of intelligent people seem to think that Lorcaserin is a long-shot, if that.
ARNA is a more speculative stock, but specs have a place. ARNA has one fifth VVUS's market cap, and it should quadruple if approved. Even if it has only a 30% chance of approval, it is a bet with positive expected value, for money that you can afford to lose. Plus, think of the fun of being in it and watching it climb as the lines form.