When I wrote about the placebo effect a couple of months ago, scientists didn't have any real understanding of why placebo works for some people but not others. Some patients can think themselves out of pain (the best-known placebo effect), but others cannot. Some patients with Parkinson's disease can take a sugar pill and, through the power of belief and hope, see their symptoms improve—but not all. Some patients, having experienced the respiration-depressing effects of morphine, will find their breathing becoming shallower even when they're injected with an inert solution, not morphine; others experience no such placebo effect. The difference, it turns out, may come down to the levels of particular neurotransmitters that carry messages through the brain, and those levels may reflect genetic differences.
In the August issue of the Journal of Clinical Psychopharmacology, scientists led by Andrew Leuchter of UCLA will report that in patients with major depressive disorder, variants of two genes affect whether someone will respond to a placebo. (The genes have no effect on whether someone will develop major depression in the first place.)
The genes that matter are those that make enzymes called catechol-O-methyltransferase (COMT) and monoamine oxidase A (MAO-A). That makes sense if you believe that one way the placebo effect works is by goosing the brain's natural reward pathways. Those pathways run on two neurotransmitters, dopamine and norepinephrine. "Most research on how placebos work now focuses on the brain's reward system and on dopamine signals," Leuchter told me by e-mail. "Our work suggests that norepinephrine should be examined as well. Dopamine and norepinephrine actually work hand-in-hand to manage reward information. One way to think of it is that dopamine helps an individual expect a reward, and norepinephrine helps you sustain attention on the possible reward and figure out how it can be achieved. We theorize that a person has to have the optimal level of norepinephrine in order to sustain the placebo response." COMT breaks down dopamine; MAO-A breaks down norepinephrine. The scientists therefore guessed that levels or forms of these enzymes would affect brain levels of the reward chemicals and thus whether that brain is more or less likely to respond to a placebo.
That's basically what they found. In the study, they examined the link between people's genotypes for the COMT and MAO-A enzymes and whether their depression was alleviated by a sugar pill. People with the genetic variants that give them a particularly voracious form of MAO-A—a form that busts up norepinephrine like a jackhammer does a sidewalk—had much less response to a placebo than people whose genotype gave them a form of MAO-A that had a more hands-off approach to norepinephrine: their score on a commonly used index of depression fell only a point or so (from an average of 21 points before they took the sugar pill) vs. nine or 10 points in people with the lazy form of MAO-A and therefore lots of norepinephrine. That suggests you need lots of norepinephrine for hope and belief to translate into something physiological.
The contrast was similar for COMT. People with the genetic variant that gave them COMT with low levels of activity, and therefore high levels of dopamine, had a lower placebo response. This is the opposite of what you might expect—namely, that if you have lots of dopamine, the brain's reward chemical, you'll respond to a placebo. But if the placebo response occurs when the expectation of improvement gives you a dopamine rush, and if you already have lots of dopamine, maybe the brain doesn't notice that extra dose. Only if the brain is low on dopamine—as it is in people with a high level of COMT activity—will the extra pulse of dopamine be noticeable. It's the difference between a downpour in a desert or an ocean.
"If a person has less COMT to break down dopamine, there are higher background dopamine levels all the time," explains Leuchter. "When a possible reward comes along, this person would be relatively insensitive to recognizing a spike in dopamine levels. We theorize, therefore, that this person would be 'resistant' to having a placebo response."
The predictive power of the genotypes isn't perfect. Of people with the form of the MAO-A gene that supposedly keeps them from responding to a placebo, 29 percent do anyway. And as Leuchter cautions, genotype isn't the only thing that affects whether someone will respond to a placebo. Still, he says, "the data suggest that individual differences in response to placebo are significantly influenced by individual genotypes."
Interestingly, this study linking genotypes to placebo response was funded in part by Eli Lilly and Pfizer. (Leuchter and several of his coauthors also receive speaking fees from the drug companies and serve as paid consultants.) The placebo effect drives drug companies crazy. It is so powerful--in some studies, accounting for as much relief from symptoms as the experimental new compound that's being tested--that any new drug a company wants the Food and Drug Administration to approve has to be really, really effective to beat the placebo (which is what the FDA looks for). This is true for antidepressants, as studies such as this and this have shown (with a smart commentary here). The placebo response plagues drugs for other illnesses too, such as lupus, in which the placebo response is significant enough that a promising new drug seems to make a difference in no more than 14 (and possibly only eight) patients out of 100.
It could be very advantageous to a drug company to weed placebo responders out of a clinical trial. Then it would be left only with people who can't, or can't as easily, think themselves out of their depression or other illness, allowing the drug to compete against, essentially, nothing. "It is difficult to determine whether a new medication is working when patients also are getting better with placebo," says Leuchter. "If these companies can understand who is getting better with placebo and why, it will make it easier for them to identify 'true' medication responders and determine whether a new medication really is effective."