Three white matter pathways separate durable response from persistent depression
In severe treatment-resistant depression, the difference between improvement and failure may come down to which neural connections an electrode actually reaches. Mount Sinai investigators mapped deep brain stimulation outcomes across the largest multi-cohort collection studied to date and linked lasting benefit to stimulation of three white matter pathways: the cingulum bundle, forceps minor, and the medial uncinate fasciculus/ventral amygdalofugal pathway.
The work sharpens a long-standing problem in psychiatric neurosurgery. Two patients can receive electrodes in nearly similar anatomical regions and still experience very different clinical courses because deep brain stimulation does not act on a single point of gray matter; it modulates circuits. In a disorder defined by dysregulated networks involving mood, salience, and cognitive control, those millimeters of difference can decide whether therapy engages the intended pathway or misses it entirely.
Network engagement, not just coordinates, drove durable benefit
The study drew on data from several cohorts, including the BROADEN trial and a University of Toronto group in which subcallosal cingulate deep brain stimulation had been developed. Investigators combined electrode models with brain connectivity maps to estimate which tracts were being stimulated. Across sites, sustained improvement aligned consistently with activation of the cingulum bundle, forceps minor, and medial uncinate fasciculus/ventral amygdalofugal pathways.
A striking pattern emerged in patients who failed to improve after two years: insufficient engagement of the left cingulum bundle. The finding offers a plausible mechanistic explanation for why some implants that appear anatomically acceptable never translate into meaningful symptom relief. If the stimulation field does not recruit the relevant fiber architecture, the circuit-level effect may be too weak or too incomplete to shift depressive physiology in a durable way.
The result is especially important because the cingulum bundle and adjacent tracts sit within broader limbic-prefrontal circuitry. Those connections help organize communication between regions involved in emotional appraisal, self-referential processing, and top-down regulation. When stimulation reaches the right bundles, it may help rebalance a network that has remained rigidly pathological despite medication, psychotherapy, and other interventions.
A single pathway miss may help explain nonresponse
Patients who remained depressed at two years shared one especially revealing feature: the left cingulum bundle was not adequately engaged. That pattern does not prove causality, and no attempt was made here to reposition electrodes and test whether response would follow. Even so, the association is difficult to dismiss because it appeared across study sites and across analytic approaches.
This is where connectome-based targeting becomes more than a technical refinement. Traditional DBS planning has relied heavily on anatomical landmarks, but a landmark alone does not guarantee that the operative electric field intersects the fibers most relevant to mood regulation. By using connectivity maps, surgeons can move from aiming at a region to aiming at a circuit. For a network disorder such as depression, that distinction can determine whether stimulation changes symptoms or leaves them largely untouched.
One sentence captures the clinical reality: the wrong millimeters can mean the wrong pathway.
Why connectome-based targeting is now moving into trials
The findings strengthen the rationale for individualized surgical planning in ongoing and future DBS studies. Connectome-based targeting has already helped shape the Abbott TRANSCEND randomized controlled trial, where individualized brain connectivity maps guide SCC DBS treatment. The logic is straightforward: if durable response depends on engaging specific white matter pathways, then patient selection, electrode placement, and even programming should be organized around those pathways rather than around anatomy alone.
Helen Mayberg emphasized that precise lead placement and confirmation that the intended target has been engaged are essential first steps in any randomized neuromodulation study. The statement reflects a practical reality for the field. DBS trials can fail to show benefit if the therapy is delivered to the wrong circuit, even when the surgical procedure itself is technically successful.
Brian Kopell framed the issue in similarly direct terms, pointing to the need to identify which connections within the target region must be activated. His point matters because depression is not a single-lesion disease. It is a circuit disorder, and circuit disorders are often sensitive to the spatial geometry of stimulation in ways that do not show up on routine imaging or crude anatomical planning.
Imaging limits remain a real constraint
The analysis was not without limitations. Individual diffusion MRI scans, which can better characterize each patient’s white matter architecture, were unavailable for all participants. As a result, much of the work relied on generalized connectivity maps rather than fully patient-specific models.
For the Emory cohort, investigators compared generalized maps with individualized diffusion imaging and found consistent results. That consistency adds confidence, but it does not eliminate the need for more precise imaging. Patient-specific data may prove especially useful when surgeons need to refine lead placement or adjust stimulation parameters after implantation.
The next step is not simply to confirm that these three tracts matter. It is to determine which biologic signatures identify the patients most likely to benefit and which stimulation settings optimize the effect once the target is reached. The study points toward amplitude, frequency, and pulse width as variables that may need to be tailored alongside electrode placement, especially if abnormal structural or functional states differ from one person to another.
The unresolved question is programming, not just placement
Ki Sueng Choi described the current findings as a first step toward sustained response and argued for more detailed models that can guide patient-specific stimulation settings. That is the correct direction for the field. Placement alone rarely explains the full therapeutic story, because a circuit that is partly engaged can still require parameter tuning to achieve a clinically meaningful effect.
For now, the strongest message is anatomical and functional at once: lasting improvement tracked with stimulation of three specific white matter pathways, and lack of left cingulum bundle engagement tracked with nonresponse at two years. The practical consequence is clear for teams developing DBS for depression. Surgical targeting, imaging strategy, and postoperative programming need to be designed around the circuit itself, not just the coordinates on a brain atlas.
The remaining gap is whether deliberately correcting pathway engagement in nonresponders will convert those disappointing outcomes into durable remission. Until that question is tested directly, the study defines the mechanism more precisely than it defines the treatment algorithm.
The research received support from the NIH BRAIN Initiative, the Hope for Depression Research Foundation, and the DANA Foundation. Abbott provided BROADEN trial data but did not design or analyze the new study.