EMBARC

EMBARC is an NIMH-funded randomized clinical trial built to find clinical and biological markers that predict who improves on an antidepressant. Adults with major depressive disorder were randomized to sertraline or placebo for eight weeks; those who did not respond were switched under double blind to bupropion (after sertraline) or sertraline (after placebo) for eight more weeks. MRI, EEG, behavioral tasks and blood samples were collected before treatment and again after one week, so that both baseline predictors and early changes can be related to outcome. A small healthy control group was tested to measure test-retest reliability and comparability of the markers across sites. Data are held in NIMH Data Archive collection 2199.

Composition

The design paper reports 309 randomized participants with MDD, recruited from August 2011 to December 2015, and 40 healthy controls (10 per clinical site, 12 of whom were also tested at a second site). Patients were outpatients aged 18 to 65 with early-onset, chronic or recurrent MDD and no psychosis or bipolar disorder. Their mean age is 37.3 years and 57% are female. Clinical sites were UT Southwestern, Columbia University, Massachusetts General Hospital and the University of Michigan. The NDA collection shares 336 subjects, 330 of them with imaging data.

Acquisition

All four sites scanned on 3T systems. The protocol covers a 3D T1-weighted structural scan, an emotional conflict task and a reward task in fMRI, resting-state fMRI, pseudo-continuous arterial spin labeling and diffusion tensor imaging. Every sequence except DTI was repeated at week 1. At baseline, 305 patients had anatomical MRI, 300 fMRI and 230 DTI. Resting and evoked-potential EEG were recorded at the same visits.

Annotations

There are no image labels. Each subject carries a SCID diagnosis, treatment arm, repeated Hamilton depression ratings and many clinical scales, which define response and remission outcomes.

Known limitations

The sample is restricted to early-onset or recurrent, chronic MDD, which limits how well findings transfer to other patients. Scanner models differ between sites and vendors are not reported in the design paper. The NDA collection description still names the originally planned citalopram, bupropion and CBT arms with 675 patients, which does not match the trial as run.