OBJECTIVE: To compare early response prediction across accelerated and standard intermittent theta-burst stimulation (iTBS) protocols for major depressive disorder, examining whether predictive performance differs when assessments are matched by cumulative pulse exposure versus calendar time. METHODS: Pooled analysis of two prospective iTBS trials conducted at Instituto de Psiquiatria, Hospital das Clínicas, Universidade de São Paulo: the open-label BRAEN-MAP trial (once-daily; N = 65; 20 sessions over 4 weeks) and the active arm of the triple-blind, sham-controlled randomized TTT trial (accelerated; N = 50; 45 sessions over 3 weeks). Assessments were matched by cumulative pulses (18,000 pulses), calendar time (Weeks 1 and 2), or session count (15 sessions). Early improvement was defined as ≥ 20% HDRS-17 reduction. Positive predictive values (PPV) and negative predictive values (NPV) were compared using bootstrap resampling. RESULTS: Response rates did not differ between trials (BRAEN-MAP: 55.4%; TTT: 54.0%; p = 0.95). For response prediction, PPV ranged from 66.7% to 75.0% across protocol-timepoint combinations, with no significant differences between protocols at pulse-matched (66.7% vs. 68.9%, p = 0.84), time-matched, or session-matched (66.7% vs. 68.8%, p = 0.86) comparisons (all bootstrap p > 0.43). NPV for non-response prediction ranged from 69.0% to 82.4%, also without significant between-protocol differences. Trajectory analyses demonstrated that early improvers maintained steeper improvement through endpoint. CONCLUSION: Early symptom improvement predicts iTBS response consistently across accelerated and standard protocols regardless of whether assessments are matched by cumulative pulses, calendar time, or number of sessions. These preliminary findings support early response assessment as a protocol-independent clinical tool, though the accuracy levels observed (PPV 67-75%) are insufficient to justify treatment discontinuation and may be most useful within adaptive, measurement-based care frameworks.
Suen et al. (Mon,) studied this question.