Pharmacogenomics of psychotropic drugs.

Receptionist: "Hallo, Dr X's office, can I help you?" Patient: "I heard that your clinic performs individualized therapy on a genetic basis, is it true?" Receptionist: "Yes, it is true, would you like to make a reservation?" Patient: "Yes thanks". The patient goes to the clinic where a laboratory test is performed. One week later, in the doctor's office, after a detailed interview, the doctor is looking at the results of the genetic analysis. Doctor: "Well, your condition needs a specific drug treatment, I suggest you use the drug Y, one tablet twice a day, because your genetic analysis tells us that you are 80% likey to have a favorable outcome". Patient: "But, doctor, I read that in cases like mine, the chances of recovery with drug Y are about 70% regardless of laboratory analysis. Why did you tell me to perform the genetic analysis and to wait one week?" The above conversation is an example of the application of pharmacogenetic studies in clinical practice. Pharmacogenetic studies claim that they will change our lives with the promise of individualized therapies; however, initial enthusiasm should be moderated in the light of certain considerations. As the patient in the hypothetical clinical case says, what is the point of having a prediction from a costly analysis if its clinical impact is so small? We should consider firstly the probabilistic nature of the genetic information. We are used to reading laboratory data offering well defined information, for example, if a specific bacteria is sensitive or not to a certain antibiotic. Conversely, genetic information is much less deterministic; we should never forget that genetic data provides us with information regarding the initial input received from patients, and that a number of modulating events follow. The phenomenon 'response to drug'is therefore a probability of a certain individual having a certain reaction, and this probability is influenced by a number of factors, such as the environment, and drug-drug interactions amongst others. In particular, the environmental influence should not be overlooked; two cells with an identical genome develop differently depending on the nutritional environment, two monozygotic twins are not phenotypically identical. Therefore the information derived from a pharmacogenetic analysis is able to tell us that a specific drug has a certain probability of being effective in a given individual, not certainty. The clinical utility of this information depends on how much this data will change the a priori clinical knowledge, or, in statistical terms, how much is the variance explained; in clinical practice only explained variances of 40-60% have substantial clinical impact. In other fields of medicine, such as oncology treatment, polygenic kits are already marketed and used for clinically relevant prediction; however, in the case of oncology, the phenotype is simpler and environmental factors influence treatment much less. Unfortunately, in the case of antidepressants to date, gene variants influencing response explain less than 10% of the total variance, and they have not been unequivocally replicated [1]. Another complication arises from the fact that gene variants do not just influence response to drugs. The most notable example is the 5-HTTLPR*s variant. This variant is the most common factor influencing response to antidepressants [2]; however, carriers of this variant display a range of effects which go much further than the reduced antidepressant (selective serotonin reuptake inhibitors) response. These carriers have a different brain morphology and reactivity (e.g., in the hippocampus and amygdala), a reduced ability to cope with stressors and a more rigid and less flexible serotonin turnover leading to a higher risk of developing mood disorders which are in turn resistant to certain treatments. Conversely, 5-HTTLPR*s carriers are more adapted and successful in life in the absence of stressors [3]. This example underlines the complexity of the drug response modulation. Results are similar regarding response to other psychotropic drugs such as lithium or antipsychotics. Notwithstanding these limitations, recently there have been some initial attempts to apply pharmacogenetic findings; some centers are routinely applying the information derived from gene variants to prescribe individualized treatments; however, this is still an investigational procedure. Are we therefore close to a change in our everyday practice? Most probably not yet. First, pharmacogenetic prediction will not be useful in conditions where the drug has a very high expected efficacy. What is the point of increasing the rate of response from 70 to 80%? The main issue for the choice of the drug will for focus on other factors, such as side effects or toxicity or interactions. Recently, laboratory kits investigating the kinetics of drugs have been marketed; however, widespread use of them is still lacking for a variety of reasons, ranging from incomplete coverage of all modulating variants, to cost, time-lapse before prescription and knowledge of the procedure. Furthermore, for each individual, a careful analysis of cost-benefit ratio of a genetic test should be performed. Recently, we and others have investigated the potential economic benefit in the case of antidepressants [4]. Results suggest that a widespread use is only advisable when sound and replicated data is obtained regarding the effect of the gene variants, the cost of the test is below US $100 and the time of the analysis is between 1 and 2 days. Another issue is linked to ethical problems; what is the risk-benefit ratio of genetic knowledge for people? What are the problems concerning DNA banking? How could we conduct research with human beings respecting confidentiality? Those issues are now being dealt with for the first time [5,6]. Overall, we should consider that, whatever the difficulties may be, the development of such analyses should not be stopped, owing to their potential large use in clinical practice. A recent strategy points to the use of plasma biomarkers; this strategy avoids all of the complications related to DNA analysis, including the recent demonstration of a large number of new mutations in affected subjects. Conclusion Biological data will be useful for predicting psychotropic drug response only when identified factors are able to explain a large part of the variance of drug response and/or side effects, when it is technically feasible in a sufficient number of centers, and when common ethical and legal issues are developed. Until then, the preliminary data we have will only be of interest in a research context.