UCSF Researcher Tracks Genes that Predict Response to Antidepressants

By Jeffrey Norris Buoyed by the power of modern genetics, researchers thought it would not be difficult to discover why some people respond to the most popular antidepressants, while others do not. After all, drugs of this type, called selective serotonin reuptake inhibitors (SSRIs), do just what that name implies. Serotonin is a neurotransmitter needed to relay signals at the gap - the synapse - between certain nerve cells, including nerve cells in pathways that help maintain mood. SSRIs slow down a protein that returns serotonin to storage within these cells, allowing more of the neurotransmitter molecules to remain in the gap between cells to trigger more signaling. Researchers expected to find a role in antidepressant response for individual variations in the gene that encodes this serotonin transporter protein, or in genes that affect how rapidly the drugs are metabolized, or in genes for one of several varieties of serotonin receptors - the proteins that receive and pass on the signal from serotonin. The understanding gained could be used to help get the best treatment to individuals confronting grinding mental misery and distress. But sometimes what appears to be a slam dunk is not. "We have looked at the usual suspects - the genes we all think are involved in antidepressant response - and basically dismissed all of those," says UCSF psychiatrist and geneticist Steven Hamilton, MD, PhD. "We have admitted to ourselves that we really don't know the genes that are involved in how these drugs work." Within a basement lab at the Langley Porter Psychiatric Institute, Hamilton, along with several graduate students and a postdoctoral fellow, has expanded the search. In an unbiased way, he is analyzing the entire genetic codes of thousands of individuals - an approach called a genome-wide association study. He's making comparisons with clinical outcomes, and trying to identify common variations in human genes that may help determine whether depressed individuals are likely to respond to drug treatment. Nature Versus Nurture In the mid-20th century heyday of psychoanalysis, there were few ways to explore biological contributors to mental illness. Even after it became possible to detect genetic differences among individuals, there were many instances of false alarms, even in studying mental illnesses in which clear breaks from normal behavior suggested the possibility of a biological origin. For instance, for many years, discoveries of genes thought to play a role in schizophrenia could not be verified in follow-up studies. But in the last several years, with rapid improvements in gene mapping and tracking, genes that raise schizophrenia risk have been identified and confirmed. And you're no longer likely to hear anyone mention "schizophrenogenic" mothers, even though families may indeed often be the key to better understanding mental illness.

Steven Hamilton

Biological explanations for mental illness are clearly ascendant, but has the pendulum swung too far in favor of nature over nurture? Could a failure to account for environmental differences be why early success has been elusive in studies of depression and genetics? Hamilton thinks not. "Even depression - which for many years people thought of as purely environmentally determined illness - seems to cluster in families in a way that suggests it is also due to genetic factors, not just environmental stressors," he says. Hamilton still believes that tackling the genetic contributors to antidepressant response is a more modest goal than trying to explain the biological basis for the illness's origin. "There are many etiologies of depression," he says. "Some are biological, some are not. Most are a combination. Clinically, we notice that regardless of etiology, a substantial proportion of people will respond to antidepressant medication. It may be that a common set of proteins and the genes that encode the proteins are related to response. We believe that this set is much smaller than the larger array of genes that may be involved in depression itself." Large Study Piggybacking on a Still Larger Study The initial buzz surrounding the first SSRI, Prozac (fluoxetine), included talk of some users not only being freed of depression, but also feeling "better than normal." But subsequent clinical research has indicated that a slim majority of depressed patients actually respond to SSRIs. About one-third of patients have complete remission of their symptoms, according to Hamilton. Many who do not respond to a first drug will gain at least some significant relief after subsequently trying a second or third antidepressant. Treatment with an SSRI, though, often lessens sexual functioning and has other side effects. Officials at the National Institute of Mental Health (NIMH) earlier sponsored a now-completed, $35 million, seven-year study, called STAR*D (Sequenced Treatment Alternatives to Relieve Depression), to evaluate options for patients who do not respond to a first round of SSRI treatment. The initial treatment given was the SSRI Celexa (citalopram). Subsequent treatments included SSRIs and other drugs, as well as cognitive therapy. The population of patients studied was genetically and geographically diverse. At each office visit, patients filled out questionnaires intended to gauge mood. Researchers surveyed clinicians about the patients, as well. As the variable outcomes of SSRI treatment became apparent, the NIMH also solicited research proposals for studies of factors affecting drug responses. Hamilton submitted the winning, $3 million proposal. Fortunately, the STAR*D researchers had collected DNA from about half of the 4,000 patients enrolled in that earlier study - DNA that is key to Hamilton's own research. Now, three years into the ambitious project, Hamilton and colleagues - statistical geneticist Susan Slager, PhD, from the Mayo Clinic in Rochester, MN, and psychiatrist Patrick McGrath, MD, from Columbia University in New York - are analyzing data obtained from the DNA of 950 patients, comparing responders with nonresponders. Already, they have identified dozens of previously unsuspected genes in which DNA variations may be associated with a treatment response. They are finalizing plans for how to validate initial results in a second batch of DNA from 950 additional, similarly treated STAR*D patients. The researchers also will try to identify genes that play a role in predicting side effects of treatment. In addition, Hamilton is beginning to use the study data to try to identify the genetic determinants of depression itself. Data Debate Today, a hot topic among researchers is how to analyze the massive amounts of data generated during whole genome association studies. The high cost of the studies intensifies these discussions. "We have kind of gotten ahead of ourselves in that we can generate the data, but there is open debate about how to analyze it," Hamilton says. "Much of what we are seeing is statistical noise, which we expected to see." When so many bits of DNA are analyzed, the results can be difficult to interpret. Even with a high threshold for identifying significant associations, there will inevitably be many false positives - genetic variants that appear associated with treatment outcome just by chance. In this new frontier of genome-wide research, there is no convention for determining a threshold value to use in identifying gene variants where an association with treatment outcome warrants additional study. Another layer of complexity arises when one considers investigating interactions between genes that may augment or reduce risk. Ultimately, all significant risk factors identified need to be confirmed by Hamilton or others. If Hamilton's group has indeed identified dozens of common variations in genes that increase the likelihood of a treatment response, a single individual could have several of them. It may be possible to develop a predictive DNA profile that clinicians could use to guide antidepressant treatment. "The genetic contribution of any single gene may be small," Hamilton says. "But that's not bad. If you have a whole bunch of those genes, it adds up." Image adapted from "don't look at me" by Sascha Assbach, and used under the terms of the photo's creative commons license. Related Links: Long-Anticipated Gene Search Technique Is Now Powerfully Real UCSF Today, June 5, 2007 Steven P. Hamilton, MD, PhD Graduate Program in Pharmaceutical Sciences and Pharmacogenomics (PSPG), UCSF Department of Biopharmaceutical Sciences Q&A on STAR*D Clinical Trial National Institute of Mental Health (NIMH), National Institutes of Health (NIH), November 2006 STAR*D Trial Results Abstract American Journal of Psychiatry 163:1905-1917, November 2006 ponse to Antidepressants