Brain-based markers may help personalize depression treatment

Irvine, Calif., April 24, 2025 — Finding the right antidepressant treatment can be a frustrating, time-consuming process – one that often requires individuals to endure weeks of ineffective medication before trying something new. Now, a new study supported by the National Institutes of Health offers hope for a more personalized approach.

Published in JAMA Network Open, the study reveals promising progress toward predicting how patients with major depressive disorder will respond to antidepressant medications using brain imaging and clinical data. The research demonstrated that brain connectivity patterns – specifically in the dorsal anterior cingulate cortex – could significantly improve predictions of treatment response across two large, independent clinical trials.

“In spite of the availability of several antidepressant treatments, including medications and psychotherapy, many individuals with depression have difficulties finding the treatment that works best for them,” said Diego Pizzagalli, the founding director of the Noel Drury, M.D. Institute for Translational Depression Discoveries at UC Irvine and Distinguished Professor in medicine, biological sciences and engineering. “As a result, for many, treatment follows a trial-and-error approach. Discovering brain-based markers predicting positive antidepressant response promises to allow a more personalized treatment and thereby speed up reduction of symptoms.”

Using machine learning models trained on clinical and neuroimaging data from more than 350 participants in two international trials – EMBARC in the U.S. and CANBIND-1 in Canada – the researchers evaluated whether their algorithms could reliably predict who would respond to common antidepressants like sertraline and escitalopram. They found that adding a brain connectivity marker to traditional clinical data (such as age, sex and baseline depression severity) significantly improved prediction performance across both studies.

“We identified a brain connectivity marker that was predictive of response to common antidepressants across two large-scale clinical trials in the U.S. and Canada,” said Peter Zhukovsky, a former postdoctoral fellow in Pizzagalli’s laboratory and now scientist in the Brain Health Imaging Centre at the Centre for Addiction and Mental Health and first author of the study. “The predictive performance of our algorithm was improved by the addition of the brain connectivity feature to clinical and demographic markers, reaching moderate levels. Our findings are promising for the search for biomarkers predicting depression response. We hope these efforts will help connect patients with treatments that are most likely to work for them.”

The study also tackled the often-overlooked challenge of generalizability – whether a prediction model developed in one trial will hold up in a completely separate population. That’s where this research stands out. Models trained on one trial performed surprisingly well when tested on another, highlighting the potential for broader real-world use.

“Data harmonization and building a large-scale database with different treatments is challenging,” said Zhukovsky. “However, we’re hopeful that cross-trial analyses such as the one we conducted in this project will advance precision medicine goals.”

The study’s implications are far-reaching. By developing biomarkers that are not limited to one treatment setting or population, researchers are laying the groundwork for clinical tools that could eventually match patients with effective treatments earlier, potentially reducing suffering and speeding recovery.

“We investigated biomarkers predicting antidepressant treatment response,” Zhukovsky said. “However, many other options are available for treating depression and if we can identify markers for specific treatments, then the resulting decision support tools could be tested in biomarker-guided clinical studies.”

As mental health disorders continue to rise globally, the need for faster, data-driven treatment approaches is more urgent than ever. The team’s findings underscore the promise of brain-based diagnostics to transform how depression is treated. But they also stress that more research is needed – larger trials, new treatment comparisons and real-world implementation studies – to bring these insights from the lab to the clinic. This line of work will be one of the key priorities within the recently launched Noel Drury, M.D. Institute for Translational Depression Discoveries at UC Irvine.

This study was made possible through a multi-institutional collaboration among researchers from McLean Hospital and Harvard Medical School; the University of Texas Southwestern Medical Center; the New York State Psychiatric Institute; Columbia University Vagelos College of Physicians and Surgeons; Stony Brook University; the University of Toronto; and the Centre for Depression and Suicide Studies at Unity Health Toronto. The EMBARC study was supported by grants from the NIH National Institute of Mental Health (U01MH092221 and U01MH092250). The CAN-BIND-1 trial was conducted in partnership with, and received financial support from, the Ontario Brain Institute and the Brain-CODE platform, with partial funding from the government of Ontario.