I enjoy learning about new developments in psychiatry just as much as the next guy. In particular, developments that promise to make treatment more effective or “individualized.” So I was intrigued by the title of a recent paper in Molecular Psychiatry, which seemed to herald the rise of a new use for genetic testing. But not for a biological therapy. No, this new use for genotyping is to predict which type of psychotherapy is best for a patient.
It even has a snappy new name: “therapygenetics.” The term was minted by Thalia Eley and her colleagues at Kings College London, authors of the study. Basically, the study suggests that variation in a particular gene sequence might predict patients’ responses to a psychotherapeutic intervention. And according to a recent editorial (pdf here) in Trends in Cognitive Sciences, this might be the first step in a new era of “personalized psychotherapy.”
“Personalized psychotherapy?” A colleague of mine happened to see the editorial on my desk. After looking at that phrase for a moment, puzzled, he asked, “Isn’t psychotherapy personalized already?” Good question. After all, psychotherapy is the quintessential personalized medicine, isn’t it? Haven’t we been criticizing “biomarker” studies because they try to personalize treatment by measuring chemicals in the blood, scanning people’s brains, and doing genetic tests? Basically everything except talk to the person???
Not so fast. Before long, your psychotherapist might ask you for a cheek swab or blood sample on your first visit. And who knows—you just might thank him for it.
In this prospective, observational study, Eley’s group studied 359 children in the UK and Australia who enrolled in a cognitive behavior therapy clinic for treatment of an anxiety disorder. Before starting therapy, the children underwent genomic testing, specifically for the genotype of the 5-HTTLPR, the promoter region of the serotonin transporter gene. They found that children with SS genotype (i.e., two copies of the “S” allele) were more likely to respond to cognitive behavior therapy (CBT) than the other children. It wasn’t an absolute benefit, but the data looked pretty good, particularly after six months of follow-up:
The above results may be reminiscent of some earlier work. The 5-HTTLPR has become the “workhorse” of psychiatric pharmacogenetics, ever since the publication by Caspi in 2003 (pdf here) that the SS genotype predisposes people to depression if they also experience stressful life events. That result was challenged in 2009, however, by larger studies showing no effect. Then, this result was overturned again by an even larger analysis (pdf here) showing that, indeed, the S allele might in fact mediate the stress-depression relationship. Confused yet? I agree, it’s enough to make one’s head spin. Or, at least, to make us conclude that we don’t really know whether it makes people more susceptible to depression.
But maybe Eley’s finding can lead us in a slightly different direction. Maybe the 5-HTTLPR genotype may make one more responsive to treatment. Maybe the S allele makes a person not just more sensitive to stressful life events (and therefore more likely to become depressed or anxious), but also more able to overcome them through therapy?
It’s an intriguing suggestion, but easy to dismiss. After all, who’s to say that another group won’t overturn this result and lead us right back to square one? And does this mean that we should test all our patients before subjecting them to CBT? Who’s going to do that? (And pay for it?)
Personally, I don’t believe Eley’s paper should be casually tossed aside. First of all, anything that improves patient outcomes (yes, even pharmacogenetics) deserves study. And while it’s probably premature to use genetic tests to assign people to psychotherapy interventions, it is encouraging to see this work. Specifically, this is a first attempt to take an endophenotype and use it to enhance treatment response.
An endophenotype is a heritable feature—biochemical, anatomical, psychological—that is simpler than a “diagnosis” like a depression or anxiety disorder, but which can be readily observed and measured. (See excellent review here.) In this case, the S allele of 5-HTTLPR might bias one’s attention toward emotional stimuli, an endophenotype that has been found in other research. (Note: in a previous naturalistic study of bulimic patients, the S allele was correlated with greater novelty seeking and insecure attachment.) If such an endophenotype can be found in other subgroups of depressed and anxious patients, then it makes sense that we might be able to employ treatment strategies that exploit this psychological feature.
This is purely theoretical, of course, but the beauty is that this theory is entirely testable, and the Eley paper is the first attempt to do so in a non-pharmacological setting. Behavioral and psychological endophenotypes offer a perfect opportunity to test the efficacy of psychosocial approaches, which, by definition, target patients’ behavior and psychology. Biological phenotypes can also be tested (e.g., with different pharmacological interventions), but there are always several steps between a change in biology and a person’s subjective report of effect—this is the bane of psychiatry.
In other words, using psychobehavioral endophenotypes to enhance treatment offers face validity. It just makes sense to both the clinician and patient. Most patients would willingly submit to a genetic test (or, for that matter, a battery of psychological tests — maybe a “psychomarker” is on the horizon?) to match them with a psychotherapeutic treatment. However, using a CYP450 genotype, or brain scan, or quantitative EEG, to predict which drug is best for them, just seems, well, weird.
In conclusion, I’ll quote a passage from the Trends in Cognitive Sciences editorial: “Genetic variation can (and should) be incorporated into psychosocial treatment research…. Doing so promises to deliver a fuller, more nuanced understanding of psychopathology which, in turn, could enhance the ability to tailor treatments to individuals based on genetic profile, increase the effectiveness of psychosocial treatments, and ultimately alleviate substantial suffering associated with psychiatric illness.”
Hopefully, this will turn out to be true. It seems like the best way to harness the inevitable (and money-driven) push toward genotyping, but to use such data in a way that maximizes patient response, rather than simply make our treatment more automated and algorithmic (the psychiatric industry’s version of “personalized”) than it seems to be headed right now.