Predictive biomarkers could ease the exhaustive trial-and-error of antidepressants

This post originally appeared on StatNews.

When a patient is suffering from depression and considering medication, practically all physicians have the same go-to treatment: a selective serotonin reuptake inhibitor. Patients start on a low dose and slowly increase it. It may take weeks for the drug to work, if it works. If not, a cycle begins. Wean off the SSRI, wean onto a new medication. All the while, the patient must manage depressive symptoms along with any side effects of a medication, which, counterintuitively, can include suicidal thoughts.

This trial-and-error approach can exhaust and discourage patients, and too many failed trials could lead some to stop seeking treatment altogether.

“Every failure that they have, it’s a huge setback,” said Diego Pizzagalli, director of the Center for Depression, Anxiety and Stress Research at McLean Hospital in Belmont, Mass.


In a new study that’s now recruiting patients, Pizzagalli and his team are attempting to use MRI scans and other technology to identify biomarkers in the brain’s reward system that may help predict which of two antidepressants will work best for patients with depression highlighted by anhedonia, or an inability to experience pleasure, which Pizzagalli calls a “cardinal symptom” of depression.

While medical fields like oncology have long used predictive biomarkers to develop therapies, it’s a clinical approach that’s been elusive in psychiatry.


“We’ve always been, in some respect, very envious of that approach,” Pizzagalli said. “I’ve been doing this for 20 years and I think we have never been as close.”

Pizzagalli and his team at McLean have been international leaders in identifying biomarkers for anhedonia, with few peers focusing on other aspects of depression.

“They’re doing a wonderful job of developing an approach where you can treat one of these types of depression that is not responding to the current standard antidepressant agents,” said Leanne Williams, who leads similar prospective research using MRI to find predictive biomarkers of cognitive-based subtypes of depression at the Stanford Center for Precision Mental Health and Wellness.

Pizzagalli’s team is using MRI scans to assess the reward system in someone’s brain while in a resting state; the researchers are also assigning volunteers certain computer tasks. Participants are shown certain “stimuli” on the computer, asked to make certain decisions, given rewards, and then tested again. In the same way that an oncologist might take an image of a tumor, then perform additional tests to confirm the results, the behavioral task acts as a back-up to the prediction made based on the MRI.

A previous study by Pizzagalli’s group analyzed patients who took sertraline (an SSRI) and bupropion (an atypical antidepressant that boosts dopamine and norepinephrine). The researchers found that stronger connections between two specific nodes in the brain’s reward system indicated a response to the atypical antidepressant, as opposed to the SSRI. This was reinforced by the behavioral task, which showed that a higher sensitivity to reward also indicated a better response to the atypical depressant.

In the new trial, participants will do the same scan and tasks, then go through eight weeks of treatment. Some people will receive their “intended” antidepressant — the one that’s aligned with the prediction based on their biomarkers — and others will not. Pizzagalli’s team will assess whether participants who were given their intended treatment show more improvement than those who were not.

If the researchers can successfully predict which of the two antidepressants will work for people, it could, finally, be a major step toward much-needed clinical action for patients.

Marin Moore is a 22-year-old public school teacher in Virginia, who went to college in Boston and was part of a separate trial run by Pizzagalli. Diagnosed with depression at age 16, Moore has been on and off of various medications throughout her life. She was off meds in her senior year of college when she began experiencing a depressive episode. She knew antidepressants might help, but she didn’t want to go through the trouble of finding one.

“The process of finding that right dosage takes months, and that time when you’re not taking any enjoyment, when you’re not able to focus, you’re not able to really be a person — it disrupts your life,” Moore said.

Pizzagalli’s long-term vision is to develop actionable steps to predict which antidepressant will be most effective for a patient. Williams is optimistic that the research could lead to clinical action. “I don’t think that we are that far away from it being possible,” she said.

But despite the abundance of research on biomarkers, implementation is hard without any previous framework for using predictive technology in clinical psychiatry settings. Prospective studies like the one at McLean take a lot of resources and cost a lot of money. Not every physician has access to an MRI machine, and it’s unclear whether every insurance company would cover the high cost of the scans.

“A lot of decision-making about treatment is driven by economics,” said Andrew Leuchter, a physician and mood disorders researcher at the University of California, Los Angeles, who focuses on treatment-resistant depression.

And, of course, biomarkers can’t predict everything.

When she was first diagnosed, Moore was able to find an antidepressant that worked for her fairly quickly, as her mother had previously taken the medication successfully. But what neither she nor her psychiatrist could predict were the side effects — intense nausea on one medication, then incidents where she completely lost her vision on another.

“It was such an overwhelming, time-consuming, and sometimes physically painful process to get through that I would rather find ways to cope with my depression — like really bad depression — than trying to go back on medication,” Moore said.

In research so far, it seems that when a treatment is a “match,” side effects are also lessened, said Williams. But they can still occur, and experts hope that future technology will be able to accurately predict physical side effects as well. For now, there may be other treatments that would work better for Moore, but there are no shortcuts to finding them.

In light of the new study at McLean, experts focused on precision psychiatry are hopeful. They imagine a future where simple scans can save patients months or years of trial-and-error experimentation, and instead lead them on the first try to one most likely to help them.

Get your daily dose of health and medicine every weekday with STAT’s free newsletter Morning Rounds. Sign up here.

This post originally appeared on StatNews.