Teasing out Depression’s Genetic Pathways


by Carl Sherman

July 7, 2010

Even the best antidepressants don’t help all people with depression—less than half get better with the first drug they take—and no one knows just why. Is the answer in our genes? The hope of matching medication to the individual without the frustrations of trial and error has been a motivating force behind the emerging science of pharmacogenetics.

In the past few years, gene-seekers received a huge gift of data: findings from three large clinical trials that compared people who responded to an antidepressant with those who didn’t.  

“These trials collected DNA on large numbers of patients, which made it possible to use newer genome-wide association study (GWAS) techniques,” said Anil Malhotra of Albert Einstein College of Medicine, program chairman of a recent Pharmacogenetics in Psychiatry Conference held in New York. With GWAS, researchers can scan DNA samples for an array of the most common gene variants, also called single nucleotide polymorphisms (SNPs).

The largest of these trials,  Sequenced Treatment Alternatives to Relieve Depression (STAR*D), followed nearly 3,000 people with depression who were first treated with citalopram,  a widely used selective serotonin reuptake inhibitor (SSRI) that works by increasing the neurotransmitter serotonin in the brain.

In one study based on STAR*D data and reported Jan. 15 in Biological Psychiatry, researchers analyzed DNA from 883 people who did well on citalopram (743 achieved full remission from depression) and 608 people who didn’t, comparing  the frequency of nearly 500,000 SNPs.

The results were “not quite as productive as we had hoped,” says researcher Steven Hamilton, of University of California, San Francisco. No single gene showed the striking association with drug response researchers were looking for. But when they analyzed the 25 SNPs where the association was strongest, “some of the genes made sense biologically,” says Hamilton.

These included one SNP for a protein involved in circadian rhythm (which is disturbed in depression and altered by antidepressants) and others for proteins that regulate the passage of sodium and calcium through the neuron.

While still far from being keys that unlock antidepressant response, these findings might guide researchers exploring the underlying biological processes, leading to other genes where the association is more powerful or suggesting new approaches to treatment, he says.

A more fundamental problem, highlighted by the study, is that “each individual gene probably explains such a small amount of the overall effect,” Hamilton says. “If a gene alters drug response by just a tiny bit, it won’t be very useful to identify it.”

As increasing amounts of genetic data become available, he says, the field seems to be moving toward “a computational approach, to identify groups of genes that work in concert.” By looking at the interactions between 10 or even 100 genes, it may be possible to piece together genetic profiles that predict who will respond to a drug and who will not.      

Not by genes alone

 But genes are at best only part of the depression story. Environmental factors such as stress also play a role, and the interaction between them may provide a stronger clue to drug response than either alone.

  Robert Keers and colleagues at King’s College London analyzed data from the Genome-based Therapeutic Drugs for Depression research program (GENDEP), which involved 811 people with depression who were given one of two drugs: the SSRI escitalopram or an older antidepressant, nortriptyline, which principally affects a different neurotransmitter, noradrenaline.

GENDEP collected not only DNA samples but also information on participants’ lives, including the stressful events—such as divorce, illness or death in the family, or job loss— they had experienced in the six months before treatment.

In analyzing DNA, the researchers paid particular attention to the gene for the serotonin transporter. Earlier studies had shown that people with the “short” variant of this gene tend to fare poorly on antidepressants compared to those with the “long” variant. Carriers of the short gene also seem more likely to become depressed after stressful events.

They found that people who had the short variant had a worse response to escitalopram—but only if they had also experienced one or more recent stressful events. Without the stress, the short gene made no difference. And for those who were given nortriptyline, the gene-stress interaction didn’t matter at all. The research team reported their findings March 9 in The Pharmacogenomics Journal 

“Our findings highlight the idea that response to an antidepressant is multifactorial—that you can’t just look at genes and events in isolation, but have to put these things together to make a predictive model,” says Keers.

It may be that by considering genes and recent history together, physicians might sort out different types of depression, he says. People with the short transporter gene who became depressed after stress might be suffering from “reactive depression” (which is less likely to improve with drugs), while those with the long gene were having “endogenous”—more biological—episodes, which tend to respond better.  

Looking at interactions is “a very reasonable approach,” says Malhotra. “But interaction studies take a lot of power—extremely large samples— because there are so many environmental factors you could study, as well as so many genes.”

Since one of the difficulties in studying depression is its heterogeneity—it’s probably not just one disorder, but several, with differing biology—“trying to categorize depressives into subgroups based on stressful life experiences, as they did here, is certainly valuable,” Malhotra adds. “But you would have to assess stress and other environmental factors very carefully and rigorously.”

Malhotra suggests that future clinical trial research may rely more on biomarkers of brain activity, such as neurohormones, brain imaging, and neurocognitive tests, both to define subcategories of depression and to assess drug response. “It might be useful to measure cortisol, which tends to go up during stressful periods, for example.”  

Behind the genes

Taking another tack, some researchers look beyond the genes associated with drug response to the neurobiological mechanisms they regulate.

One such gene, identified several years ago, produces an important serotonin receptor (5HT1A) in the brain. Researchers have found that people with the “C” variant of this gene tend to respond better to antidepressants than those with the “G” variant. They proposed that the G gene expressed higher numbers of autoreceptors—receptors on serotonin-secreting neurons that act as negative feedback to curtail neurotransmitter release as concentration builds in the synapse. Autoreceptors are a safety valve, but also limit the effect of SSRIs.

What made the connection between autoreceptors and drug response hard to study was that the same gene also appears to regulate the post-synaptic receptors that transmit serotonin signals from nerve to nerve, so standard experimental techniques like knocking out the gene altogether would have too broad an effect.

René Hen and his colleagues at Columbia University found an ingenious solution: They bred mice with the gene variant that produced high numbers of serotonin receptors, and developed a way to turn  the gene off that reduced autoreceptor levels without affecting post-synaptic receptors.

They found that mice in whom the gene was left on (high-1A), didn’t respond to the SSRI fluoxetine. When the researchers turned the gene off (low-1A), the mice responded strongly to the drug. They reported theif findings in the Jan. 14 issue of Neuron

Animal studies “complement human genetic studies,” Hen says. “While the genetics of depression are complicated because the effect of any one gene is tempered by others, as well as by the environment, with mice we can study a single gene against a constant genetic and environmental background.”

If further research confirms these findings, doctors may learn to identify those likely to respond to SSRIs by measuring 5HT1A autoreceptors. While this might involve genetic analysis, “we could also look at PET images of the receptor itself, or use other bioassays,” Hen says.

Paul Albert of the Ottawa Hospital Research Institute, University of Ottawa, one of the researchers who helped identify the gene, saysthat “what Dr. Hen has published gives some real validation to what we proposed—that changes in levels of serotonin autoreceptors could be a risk factor for depression and also affect response to SSRIs.”

“It’s only one piece of the puzzle, for sure,” Albert says. But the fact that a relatively small effect—a reduction of autoreceptors by 30 percent—caused such a strong change in antidepressant response suggests it is a promising target for pharmacological research, he says.  

Surveying pharmacogenetic antidepressant research overall, Anil Malhotra concedes that it hasn’t lived up to early expectations. But he remains “reasonably hopeful” of further progress.

“In reality, the field has only been around for 10 years,” he says. “It’s still very early on.”