TCPR: Dr. Perlis, can you help us understand the role of pharmacogenetics in psychiatric practice?
Dr. Perlis: The basic idea behind pharmacogenetics is that, just as genes influence characteristics such as eye color, hair color, and risk for breast cancer, we think that genes contribute to variations in the way people respond to medications. What we are looking at is whether we can find genetic markers that predict how specific people will respond to specific medications. The genes that have so far received the most attention are those that influence drug metabolism. Some people are genetically poor metabolizers at P450 2D6, while others are ultrarapid metabolizers. Poor metabolizers might require a lower dose of a particular drug to achieve the same effect as a normal metabolizer, whereas ultra-rapid metabolizers might require a higher dose. And within the last few years, there has been a test marketed for clinical use in P450 testing – Roche’s Amplichip CYP450.
TCPR: Is this test useful in clinical psychiatry? In a recent editorial in the British Medical Journal, you were skeptical (Perlis RH, BMJ 2007;334:759).
Dr. Perlis: That’s true. With any new laboratory test, the first crucial question we need to ask is, will the information from this test really lead me to change the way I am managing this patient? The second question is, can I get this information any other way? It turns out that when you are worried about the level of a particular medication, it is almost always quicker, cheaper, and more precise to check a blood level rather than to assess the genetic P450 profile.
TCPR: And why is that?
Dr. Perlis: Well, you have to remember that genes are only one of many things that influence the way that we metabolize medications. Variations in 2D6 and 2C19 and other genes do affect blood levels of particular drugs, but so do things like diet, smoking status, and other medications. And these other factors can change over time, making it useful in some cases to get repeat blood levels. So, for example, if someone is having trouble tolerating fluoxetine at 10 mg, or hasn’t fully responded at 80 mg, it is easy enough to get a fluoxetine and norfluoxetine level. The lab will have a reference range to let you see if someone has an extremely high blood level or an extremely low blood level compared to the norms, and you can make dosing adjustments.
TCPR: So the bottom line on testing genes related to pharmacokinetics (that is, the P450 enzyme system) is that the evidence to support clinical utility is actually fairly scant. Now what about genes related to pharmacodynamics? Your group at Massachusetts General has recently published something about this. How close are we to ordering lab tests to predict response to antidepressants?
Dr. Perlis: As a way of introducing our recent findings, let me talk about some of the earlier genetic findings from the STAR-D study. The STAR-D cohort is the largest group of antidepressant-treated patients for whom we have genetic information. Most previous genetic studies looked at 100-150 patients; this study has nearly 2000 people who were treated with the same antidepressant, citalopram. In other words, it is one of the first studies with enough statistical power to find genetic effects on response to medication. And the initial report on this, which was done by colleagues at NIMH, examined 68 genes and found that a variation in the gene coding for the serotonin 2A receptor was correlated with a better response to citalopram (McMahon FJ et al., Am J Hum Genet 2006 May;78(5):804-14).
TCPR: And what was the strength of the prediction?
Dr. Perlis: If you had two copies of a particular version of the gene, you were about 18% more likely to respond than if you had no copies.
TCPR: What does the serotonin 2A receptor do?
Dr. Perlis: It is one of the key receptors affected by serotonin, and it is downregulated by citalopram, suggesting it has a role in the antidepressant response. So, if you are looking for an obvious suspect to predict response, the serotonin 2A receptor gene would be high on your list, and this was a very exciting finding. Another very well-studied gene for antidepressant response has been the serotonin transporter, which a number of studies have suggested is associated with differences in how patients respond to SSRIs or other antidepressant treatments.
TCPR: And what is the serotonin transporter?
Dr. Perlis: The serotonin transporter is a protein embedded in the presynaptic cell membrane that is responsible for the reuptake of serotonin from the synapse. It is blocked by SSRIs. People with a version of the gene that makes fewer transporters have a higher risk of depression. Studies suggest that if you have this high-risk version of the gene, your risk of depression goes up as the number of stressful life events you experience goes up. Whereas if you have the lower risk form, you don’t seem to be terribly susceptible to developing depression, despite an increase in the number of stressful life events (Caspi A, et al., Science 2003;301:386-9). It is a very elegant example of a gene-environment interaction. The high-risk gene also makes you less likely to respond to SSRIs.
TCPR: But in some ways this finding doesn’t make sense. If you have a gene that makes fewer transporters, wouldn’t it be easier for the brain to keep more serotonin in the synapse? And then, wouldn’t you logically be more likely to respond to SSRIs?
Dr. Perlis: Actually, in STAR-D there was initially no association reported between the transporter gene and treatment response, which was quite puzzling to many of us, because it is the most consistently replicated finding in psychiatric pharmacogenetics. Well, a finding this month in Archives may well solve this puzzle (Hu XZ et al., Arch Gen Psychiatry 2007 Jul;64(7):783-92). Their paper shows that what serotonin transporter variations seem to influence is tolerability, rather than treatment response per se. Greater levels of the transporter may therefore make patients less susceptible to serotonergic adverse effects. I should stress, though, that this and the serotonin 2A finding are intriguing but still account for only a small amount of the variation in treatment response. In other words, we’re not yet ready to take these tests into the clinic.
TCPR: Tell us a bit about your paper, also from the STAR-D database, looking at treatment-emergent suicidality (Perlis RH et al., Arch Gen Psychiatry 2007 Jun;64(6):689-97).
Dr. Perlis: We became interested in studying suicidality because it is clear that some people who are started on antidepressants get worse, whether or not it’s specific to antidepressant treatment or just part of the course of the illness. We wondered if it would be possible to predict who might be at greater risk for worsening after they started treatment and in particular who might develop treatment-emergent suicidal thinking or behavior.
TCPR: And the gene you focused on was neither the serotonin 2A receptor gene nor the transporter gene, but the “CREB1” gene. Why were you interested in CREB1?
Dr. Perlis: One of the ways antidepressants work is to turn on and off certain genes. Part of the pathway by which that is accomplished involves the protein CREB1. We had done some earlier studies suggesting that CREB1 is involved in anger regulation and response to aversive or uncomfortable stimuli. If you think back to the old analytic concept of suicide as ‘anger directed inward,’ CREB1 was a clear candidate for predicting this particular side effect. Our main finding was that among the men in STAR-D, but not the women, there was a very striking association between a particular variation in the CREB1 gene and treatment-emergent suicidality. Males carrying the high-risk form of the gene were two to three times more likely to develop treatment-emergent suicidal thinking during the first four weeks of treatment. Now, it is important to point out that only one of these patients made a suicide attempt, and so we are not predicting suicide here, just suicidal thinking. But looking for this variation of CREB1 may help us identify a group that may be more vulnerable to suicidal ideation.
TCPR: So what do you see as the role of pharmacogenetic testing in the future?
Dr. Perlis: I don’t think pharmacogenetics is going to give us the entire answer – the art of clinical practice will always be important. What I think pharmacogenetic testing can do, like other tests in medicine, is to help us stratify risk for patients, and to help us calibrate our clinical thinking. So, for example, it doesn’t mean that patients with this CREB1 risk variation shouldn’t get treated with SSRIs, because many of these patients, in spite of having this risk allele and having this side effect, went on to recover. But these might be the patients who need closer follow-up. I might see them back in the office sooner. I might treat their anxiety more aggressively. I might be sure that I have a longer conversation with family members about warning signs to look for. If we can stratify risk, and if we can get some sense of how people are likely to do on treatment, I think that would be a major advance.