DNA, Drugs and Depression
Genetic discoveries point the way to personalized treatment for depression, but researchers are still on the first leg of the journey

by Brenda Patoine

March 21, 2008

Imagine the day that you go to your doctor, whether it be for depression or diabetes, and your treatment is determined by a swab of cells taken from inside your cheek. A DNA analysis of those cells would tell your doctor which medications would be most likely to work for you, which would not—and which might have serious side effects.

This futuristic scenario—the goal of “pharmacogenomics,” or the science of using genetic information to predict drug response—is not so far-fetched. A series of recent discoveries has nudged researchers closer to that goal, including a new report identifying a genetic variation that predicts response to many antidepressants. The hope is that such information can be used to guide medication choices, which is especially important in depression treatment, where choosing the right drug can be a drawn-out exercise in trial and error.

“Among the three or four areas of medicine where we’re trying to develop pharmacogenomics, depression is right up there,” says Tom Insel, director of the National Institute of Mental Health.

He is quick to add that there are huge hurdles to overcome before “personalized medicine” is realized for depression. Still, the medical world is beginning to face the reality of DNA-driven drug selection, starting with new advice on a blood thinner taken by about 2 million people in the United States annually.

In August, the Food and Drug Administration approved new labeling for warfarin, used to prevent strokes and heart attacks, that reflects how individuals’ genetic variations influence the drug’s effectiveness and risk profile. One-third of people on warfarin metabolize it differently than the other two-thirds do, in part because they carry one or two specific gene variants—called single nucleotide polymorphisms, or SNPs—in their genetic code.

While the warfarin label change shows that pharmacogenomics has clinical applications, the science is far less clear in the case of depression. In the past two years, published reports have identified more than a dozen SNPs that affect risk and response to antidepressant therapy, including one associated with an increased risk of treatment-related suicidal thinking.

A report published in Neuron in January by Manfred Uhr and colleagues at Munich’s Max Planck Institute of Psychiatry adds a significant new wrinkle. The researchers focused on a protein called P-glycoprotein, which regulates the entry of certain antidepressants into the brain via the blood-brain barrier, a tight layer of cells and tissue that separates the brain from the rest of the body.

In experiments with genetically altered mice that don’t produce P-glycoprotein, Uhr’s team identified which antidepressants rely on the protein to gain entry to the brain. They then showed that certain variations in the gene encoding P-glycoprotein predicted treatment course and clinical outcomes in people treated with these antidepressants.

Insel calls the work fascinating because it suggests that, at least for antidepressants that use P-glycoprotein, the gene variations “look like they may have some relevance for clinical response to the drugs.”

New layer of complexity

Carmine Pariante, a neuroscientist at the Institute of Psychiatry, King’s College London who studies P-glycoprotein, is exuberant about the findings. “This has brought to everyone’s attention the relevance of P-glycoprotein from a clinical point of view," he says. "If the antidepressant doesn’t get into the brain in the first place, then of course the clinical efficacy will not be as good as you would hope.”

Up until now, most pharmacogenomics research in depression has focused on one of two areas: enzymes in the liver that help metabolize antidepressant drugs, and neurotransmitter systems in the brain, including receptors and transporters for serotonin.

“This adds another layer of complexity,” says Pariante. “For the first time, [Uhr’s research] identifies a biological target with a completely different action that is relevant to antidepressant response.”

Pariante is even more excited by the implications the work has for future drug discovery. “Pharmacogenomics research is not only about understanding how we can predict response to the antidepressants we currently have. It is also about understanding how we can develop better antidepressants.” Knowing that P-glycoprotein influences antidepressant response by denying entry to the brain by certain drugs, it makes sense to focus drug-development efforts on compounds that do not have a strong interaction with P-glycoprotein.

Recent work from Pariante’s lab indicates that P-glycoprotein also controls brain levels of glucocorticoids, a family of powerful stress hormones that are thought to be involved in the development of depression. This opens the possibility, he says, that targeting P-glycoprotein directly could reduce depressive symptoms.

A long road

The research by Uhr’s group underscores how complicated it will be to personalize depression treatment on the basis of genetics. “We need to be thinking about a whole panel of genes that might be relevant to antidepressant response,” says Insel, more akin to a “bio-signature” of multiple SNPs than a marker for a single variant. Such a panel could then be compared to the signature of individual patients to predict how they might respond.

Even then, Insel adds, personalized medicine will require more than genetic information. In depression, clinicians also will need to take into account factors such as family history, neurocognitive function and brain structure-function relationships they see via neural imaging.

Pariante agrees. “My feeling is that genetic studies alone will not be as strong, but if genetic studies are integrated with clinical features and other biological investigations, then we can create algorithms that would hopefully be much more powerful in predicting response to antidepressants.”

Insel describes three convoluted legs in the path to personalized medicine: discovery, development and dissemination. The discovery phase focuses on identifying a list of markers that might be useful, such as the SNP variant identified by Uhr. Then, the markers must be tested in clinical trials to determine their impact on clinical care. Dissemination requires regulatory approval, buy-in by insurance companies and other healthcare payers to cover the costs and, finally, integration into clinical practice.

In depression, Insel says, “We’re in the early innings of the discovery phase. We’re not far from having a list of possible markers, but we’re a long way from having something that would be ready for prime time.”