When to hold and when to fold: Early prediction of nonresponse to transcranial magnetic stimulation in major depressive disorder

Background: Determining when to recommend a change in treatment regimen due to insufficient improvement is a common challenge in therapeutics. Methods: In a sample of 7215 patients with major depressive disorder treated with transcranial magnetic stimulation (TMS) and with PHQ-9 scores before, during and after the course, 3 groups were identified based on number of acute course sessions: exactly 36 sessions (N = 3591), more than 36 sessions (N = 975), and less than 36 sessions (N = 2649). Two techniques were used to determine thresholds for percentage change in PHQ-9 scores at assessments after 10, 20, and 30 sessions that optimized prediction of endpoint response status: the Youden index and fixing the false positive rate at 10%. Positive and negative predictive values were calculated to assess the accuracy of identifying final nonresponders and responders, respectively. Results: There was greater accuracy in predicting final response than nonresponse, especially in the groups that had at least 36 sessions. Substantial proportions of patients with low levels of early improvement were classified as responders at the end of treatment. Limitations: The findings should be validated with clinician ratings using a more comprehensive depression severity scale. Conclusions: Manifesting clinical improvement early in the TMS course is strongly predictive of final status as a responder, while lack of early improvement is a relatively poor indicator of final nonresponse status. The predictive value of lack of early symptomatic improvement is too low to make reliable recommendations regarding changes in treatment regimen.