Two recent comparison studies now indicate that once patients have failed only two antidepressant trials, TMS is significantly more likely to be effective compared with a third medication trial.

TMS is a cutting-edge therapy for treatment resistant major depression. It involves using an external device that generates a magnetic field which activates the neurons in a particular region of the brain that helps regulate mood. It is FDA cleared, safe and effective.

Up until now, one of the biggest unanswered treatment questions was, what is the point at which TMS becomes more effective than trying one more medication?  It is a very important question, because we see people who are referred for TMS who have already tried and failed sometimes 10 or more medications.  Yet science gave us only some hints, nothing really authoritative.

Until this month, the best we could do was to compare disparate studies. The Star*D trial, which analyzed response rates to a series of antidepressant trials in a well-defined sequence, revealed that the remission rate dropped off from about 28% for the first medication trial to 7% for those who had already failed three, while intolerance to medications increased.  Meanwhile, the published remission rates for TMS across multiple studies generally centered around 25%.  Cramming all these studies together suggested that TMS and medications were about even at two prior medication failures, and at three TMS becomes the more effective option.  But comparing different studies to one another is a less scientifically valid method than one head-to-head study.  However, head-to-head studies, particularly in the United States, are extremely rare, and I personally never expected to see one.

Fortunately I was wrong.  In the September 2024 American Journal of Psychiatry (1), the premier psychiatric journal, there was a randomized, prospective study out of the Netherlands comparing patients who had failed at least two antidepressant trials with TMS versus another medication trial.  

You may skip this paragraph if you want, but here is a deeper dive into the methodology. The study involved 89 participants who had failed at least two medication trials.  Three-quarters of them had failed only two. They were randomized to either another medication trial or TMS.  Of the 89, 76 patients completed the study. One patient dropped out from TMS and two from medications.  As it was a head-to-head study, it was unblinded; there was no placebo arm.  Inclusion criteria were moderate to severe depression (HAM-D score of greater than 16) and a current episode duration of less than two years.  For the TMS arm, the left dorsolateral prefrontal cortex was targeted via the Beam-FT3 method, and motor threshold was redetermine at the beginning of every week.  rTMS was delivered at 10 Hz, 120% of motor threshold and 3000 pulses per session.  This is all very consistent with standard United States protocols. However, they only performed 25 sessions, whereas we typically do 36, and the literature suggests that even more than 36 might produce better results. So if anything, the TMS arm of the study was disadvantaged by this design.  All participants received weekly psychotherapy.  For the medication arm, patients were either switched from their current antidepressant to a tricyclic, or their current medications were augmented with lithium or a second-generation antipsychotic.  If neither of these options were medically suitable, a different antidepressant could be prescribed.  The article does not say, but I presume the choice between these options was up to the clinician.

The results were striking.  TMS showed more than twice the decrease in depression 

asymptoms.  The mean decrease in HAMD score for TMS was 10.02, versus 4.19 for medications.  The response rate was 37.5% versus 14.6%, and remission rate was 27.1% versus 4.9% respectively.  P values were 0.01 or less.  The number needed to treat was 4.4 for response and 4.5 for remission.  This indicates a high degree of statistical significance and a large effect size.  TMS was superior in reducing anhedonia and anxiety, whereas there were no differences in rumination, cognitive reactivity or sleep problems.