By: Dr. Geoffrey Modest
-A new article just came out of the Program in Placebo Studies in Boston that link the “placebo effect” to specific genes in the newly-defined “placebome” (see here), also noted in a more popular forum.
Background, promoting the concept that placebo effects are legitimate biological responses to environmental cues:
–studies on placebos have dated back decades. Old studies have found red placebos and those in capsule form to be more effective. Another found that blue placebos are more effective for sleep, across cultures (though, per NPR, not so only for males but not females in Italy, where blue perhaps evokes images of their national soccer league and may increase arousal).
–other old studies have found that in some patients pain could be suppressed by placebo, equivalent to up to 8mg of morphine, and blocked by the opioid receptor antagonist naloxone
In this setting, this article explores what is known about the genetic component of the placebo response, noting that behavioral/personality assessments have had limited success in predicting placebo response.
–A very large limitation in assessing the placebo response is the lack of NTC (no-treatment control) trials. Just as we compare active drugs with placebo to assess the added effect of the active drug, we need to compare placebo to NTC to see if placebo has additional effects over doing nothing (thereby accounting for “regression-to-the-mean” as well as other changes in the natural history). As a real-world example I use in my lipid talk, there was a study of hyperlipidemic patients with elevated ALT levels put on a statin, compared to hyperlipidemic patients with normal ALT put on a statin, but they also had a 3rd group with elevated ALT given no statin: results were that there was an increase in ALT in the first group on statin, as compared to those in the 2nd group with baseline normal ALT, which suggested that statins in the setting of an elevated ALT are associated with higher ALT levels. But in fact there was no difference between the 1st and 3rd groups, suggesting that there is enough variability in ALT values in those with an elevated ALT that the statin actually had no additional effect on ALT levels…).
–There are several studies suggesting that there is a “placebome“, areas of the genome which may predispose individual patients to a placebo response:
–much of the research has assessed areas of the genome that affect the opioid (pain suppression) and dopamine (reward) systems.
–pain studies have found that those patients with more profound placebo response have higher opioid and dopamine receptor activation; patients with higher pain sensation with placebo had decreased signaling in both systems
–several studies have looked at genetic polymorphisms of the catechol-O-methyltranserase gene (COMT is responsible for dopamine metabolism), finding in different studies that homozygotes for a less-active allele (a common polymorphism in the population, varies with ethnicity/race but is in 20-25% in the Causasian population) have higher levels of dopamine in the prefrontal cortex; patients with this polymorphism and irritable bowel syndrome had the greatest placebo response; a pain study found increasing placebo-induced pain suppression in those with increasing numbers of this allele; and, conversely, in those with high levels of the “normal” allele, there were more nocebo effects (adverse effects from the placebo). Similar results were found with variants of the monoamine oxidase A polymorphisms (MAO-A metabolizes monoamines, including dopamine and serotonin), where people with a particular polymorphism had decreased MAO-A activity and higher basal dopamine levels; and, in patients with clinical depression who had this polymorphism and were given placebo vs one of 3 selective serotonin reuptake inhibitors (SSRIs), they had a higher placebo response. A study of genetic variations of dopamine beta-hydroxylase (converts dopamine to norepinephrine) has found that patients with alcohol dependence who had this polymorphism did better with placebo and worse on naltrexone. Other studies have looked at variations in the mu opioid receptor, finding variability in placebo-induced pain perception. There are also some data on placebo response variability with endocannabinoid and seratonergic signaling genetic variants.
I don’t mean to overplay this, but I do think this article raises a few interesting points:
–there is utility in having a no-treatment control in studies, to help control for clinical variability in symptoms over time, and making it easier to tease out a placebo effect.
–identifying those with genetic variants associated with less placebo response could, at some point in the future, lead to needing smaller studies to determine medication efficacy
–identifying those with more profound placebo response could lead to more effective non-medical (or placebo) interventions
–it is important to be aware that in those instances with known high placebo-response (eg depression, irritable bowel), it is much harder to assess drug efficacy, with results biased to null even if the medication were really effective in those patients genetically predisposed to minimal placebo response (more likely to be an issue in smaller trials). Awareness of placebo-sensitive genotypes might be particularly useful in these cases.
–there may be important gene-drug-placebo effects. There is the potential that in some patients with specific genetic pro-placebo polymorphisms and given an active drug, they may then get both the drug effect as well as the placebo effect, and these 2 effects could interact with each other in different ways: eg, these placebo effects could significantly augment or compete with the drug effect (especially if the drug and placebo focus on the same target in the body). The point here is that even the active drug’s effects may be very different in those predisposed to a placebo effect than those not, and that the placebo and drug effects are not necessarily simply additive. The above example of alcohol dependence/naltrexone may indicate an adverse placebo-drug interaction.