A small Columbia University pilot is raising questions about whether repetitive transcranial magnetic stimulation, or rTMS, can be made more precise by using each patient’s structural brain scan to choose the treatment location.
Among 11 adults treated at a junction involving a region called parcel 46, six met the study’s definition of remission after a six-week course. The result comes from a medRxiv preprint that has not been peer reviewed, and the study was not designed to establish that parcel-guided treatment works better than standard approaches.
What the pilot found
The study enrolled adults with moderate to severe major depression that had not improved after at least two adequate antidepressant trials. Participants also had a baseline Montgomery-Åsberg Depression Rating Scale, or MADRS, score of at least 20.
Sixteen people began treatment and 15 completed it. The analysis described 11 participants who received stimulation at the junction of parcel 46 and neighboring parcels. Their depression scores fell by an average of 57.8%. Eight of the 11 participants met the study’s response threshold, defined as at least a 50% reduction in MADRS score. Six, or 54.5%, reached remission, defined as a final MADRS score of 10 or less.
Four other participants were treated at the middle of parcel 46. Their scores fell by an average of 38.2%; one responded and none reached remission. Because these groups were very small, the report did not include a statistical test of the difference between them. It also did not describe how participants were assigned to the two target locations.
How the target was selected
Researchers used structural MRI to reconstruct each participant’s cortical surface, fit atlas-based labels to it and identify the parcel junction for coil navigation. The treatment itself was standard 10-Hz rTMS, delivered in 27 sessions over six weeks.
Resting-state fMRI scans were collected before and after treatment for analysis, but the scans were not used to select the treatment target. Before treatment, the strength of the connection between parcel 46 and the subgenual cingulate was associated with who responded.
Participants who went into remission had near-zero baseline connectivity between those regions. People with stronger anti-correlation improved less. The researchers also reported that larger reductions in connectivity between the peak target region and the subgenual cingulate were associated with larger symptom improvements.
Why the result needs caution
This was an open-label pilot without a sham group or a scalp-targeted comparison. Participants knew they were receiving active treatment. As a result, the study cannot determine how much improvement came from the parcel-guided targeting itself, from rTMS more generally or from other influences such as expectations and the natural course of symptoms.
The study also did not directly compare parcel targeting with scalp-based or fMRI-guided targeting, and it did not include follow-up after the six-week treatment course. Three of the seven listed authors were affiliated with Soterix Medical, which makes the CloudTMS system used in the study.
The report described a connectivity-based response predictor with an area under the receiver operating characteristic curve of 0.88, with 75% sensitivity and 100% specificity at the best cutoff. The reported remission AUC was 1.0. Those figures come from a very small sample and should not be treated as evidence that the predictor is ready for routine use.
What it means for treatment
Standard rTMS is supported by sham-controlled trials, but parcel-guided rTMS remains experimental. The researchers stated that the connectivity predictor needs replication in prospective studies and existing treatment datasets before it can guide care.
Earlier and related findings cited in the report are also encouraging but do not settle the question. A 2020 parcel-guided pilot reportedly found responses in all 10 treated participants, with half reaching remission. Another randomized study found a larger median MADRS reduction with connectivity-guided accelerated TMS than with scalp-based F3 targeting, while a cited sham-controlled trial of Stanford neuromodulation therapy reported one-month remission in 50.0% of active-treatment participants and 20.8% of those receiving sham treatment.
Together, these results support further controlled research rather than a change in clinical practice. Larger randomized trials with sham or scalp-targeted comparison groups, transparent target assignment and longer follow-up will be needed to determine whether individualized parcel targeting adds meaningful benefit.
AI tools were used to assist with the preparation of this article.
Leave a Reply