Individualized fMRI-Guided TMS Target Speeds Antidepressant Response in Resistant Depression
- Pilot blind controlled trial from Moscow (Research Center of Neurology and the Serbsky National Medical Research Center of Psychiatry and Narcology), N = 30 patients with a pharmacotherapy-resistant depressive episode, split into a 17-patient individualized-target group and a 13-patient standard-target group.
- Both groups received 20 sessions of high-frequency repetitive TMS; one group was stimulated at a left dorsolateral prefrontal cortex (DLPFC) point chosen from each patient's own resting-state fMRI connectivity map (strongest negative connectivity with the subgenual cingulate cortex), the other at the standard anatomical target 5 cm anterior to the motor hand area.
- Beck Depression Inventory (BDI) scores fell significantly by session 10 in the individualized-target group (p=0.0071) but not yet in the standard-target group (p=0.0547); by session 20 both groups were significant (p=0.00037 and p=0.0039), and the final between-group comparison found no significant difference (p=0.8886). Responder rates (≥50% BDI drop) were 47.1% versus 38.5% (p=0.72).
- Both protocols were described as satisfactorily tolerated, with transient adverse events (dizziness, headache, facial muscle contraction during sessions; headache and mood changes within 24 hours).
Choosing where to place the TMS coil is one of the first decisions in any protocol, and it has historically been made the same way for almost every patient – measure 5 cm forward from the spot that makes the thumb twitch, and stimulate there. A 2019 pilot trial from two Moscow research centers asked whether replacing that fixed rule with a target derived from each patient's own brain-connectivity data changes anything about the outcome.
Where the Coil Points First
Standard rTMS protocols for depression place the coil over the left DLPFC using the "5-cm rule": a fixed distance from the motor cortex, the same for every patient regardless of individual anatomy. Neuronavigation based on resting-state fMRI replaces that fixed rule with a target computed from each patient's own connectivity map – in this trial, the point within the left DLPFC showing the strongest negative functional connectivity, a method more commonly used in research settings than in routine practice.
Testing an Individualized Target Against the Standard One
The trial enrolled 30 adults with a depressive episode resistant to pharmacotherapy and split them into a 17-patient individualized-target group and a 13-patient standard-target group in a blind, pseudo-randomized design. One group received rTMS at the individually computed DLPFC target; the other at the standard 5-cm-rule target. Both groups underwent the same 20-session course, with the Beck Depression Inventory as the primary outcome and the mental-health subscale of the SF-36 as a secondary measure, assessed at baseline, after 10 sessions, and after 20.
A Faster Curve, Not a Taller One
The individualized-target group reached a significant BDI reduction after 10 sessions (p=0.0071); the standard-target group had not yet reached significance at that point (p=0.0547). By session 20 both groups were significant (p=0.00037 and p=0.0039), and when the authors compared the two groups directly, the eventual size of the BDI reduction did not differ significantly between them (p=0.8886) – nor did the share of patients meeting formal response criteria, a BDI drop of at least half (47.1% versus 38.5%, p=0.72). The SF-36 mental-health subscale showed the same pattern: significant by session 10 in the individualized-target group (p=0.0465) and not yet in the standard-target group (p=0.5195), both significant by session 20. Tolerability was similar in both arms, with side effects limited to transient dizziness, headache, and facial muscle contraction during stimulation, plus headache and mood changes reported within 24 hours.
The Limits of a Thirty-Patient Pilot
Thirteen and seventeen patients per arm is enough to describe a trend, not to rule one out, and the authors present the work explicitly as a search for a workable algorithm rather than a finished protocol, attributing the lack of a final between-group difference to the trial's small size and low statistical power. Computing an individual connectivity-based target also requires resting-state fMRI and image-guided neuronavigation – equipment and expertise that a standard "5-cm rule" session does not. The openly available Russian-language full text reports the p-values above but not the exact post-treatment BDI point scores or their confidence intervals, which appear only as a figure, not as a table.
A personalized fMRI target got patients to a significant improvement on session 10; the standard target got them to the same statistical finish line on session 20 – and by the end, the two lines met.
The trial was published in 2019 and has not been superseded by a larger Russian replication located during this search. Thirteen and seventeen patients per arm is a pilot-scale sample; the authors themselves frame the results as preliminary and attribute the null final comparison to low power. The full text reports p-values but not exact post-treatment BDI point values or confidence intervals in tabulated form – so the exact magnitude of the effect, as opposed to its direction, timing and statistical significance, cannot be independently confirmed from the accessible record.