Accelerated Intermittent Theta-Burst Stimulation (aiTBS) Improves Emotional and Spatial Working Memory Across Treatment Phases in Treatment-Resistant Depression (TRD)
DOI: Pending DOI
Synopsis
Depression is often associated with cognitive impairments, including working memory function. Fifteen patients with treatment-resistant depression (TRD) underwent accelerated intermittent theta-burst stimulation (aiTBS) in two treatment phases: an intensive week (8 sessions/day, 5 days/week) and a three-week maintenance phase (3 sessions/day, 3 days/week). The EMOBACK (emotional n-back) and CANTAB tasks, along with the Beck Depression Inventory (BDI-II), were assessed at baseline (T1), post-intensive week (T2), post-treatment completion (T3), and at one-month follow-up (T4). EMOBACK accuracy (p = .015) and CANTAB between-errors (p = .015) peaked at T3, suggesting that cognitive gains may accumulate during the maintenance phase. BDI-II scores decreased significantly (all p-values ≤ .003), and responders (≥ 50% reduction) showed lower reaction time to positive words.
Background
Working memory deficits are common in depression, often persist after mood symptoms improve, and are associated with poorer functional outcomes and treatment response [1, 2, 3, 4, 5]. Accelerated intermittent theta-burst stimulation (aiTBS) has emerged as a promising treatment protocol for TRD, with rapid antidepressant effects delivered through high-density daily sessions [6]. However, its cognitive effects remain poorly understood. We investigated whether cognitive changes emerge during the intensive phase of aiTBS or require cumulative stimulation during a lower-frequency maintenance phase.
Methods
Fifteen patients (9 women; age [mean ± SD] 45.2 ± 11.8 years) with TRD (baseline BDI-II, mean ± SD = 40.4 ± 6.7) completed a single-arm, four-timepoint longitudinal aiTBS study targeting the left dorsolateral prefrontal cortex (dlPFC). All met TRD criteria (≥ 2 adequate antidepressant treatment trials) and remained on stable medication. Participants received 67 aiTBS sessions targeting the left dlPFC at 90% resting motor threshold: 40 sessions during an intensive 5-day phase (8/day) followed by 27 sessions during a 3-week maintenance phase (3/day, 3 days/week). Assessments were conducted at T1 (baseline), T2 (post-intensive week), T3 (post-treatment completion), and T4 (1-month follow-up), including two working memory tasks (EMOBACK [7, 8], a 3-back working memory task with emotional words; CANTAB spatial working memory task – SWM) and a depression severity scale (BDI-II). Primary cognitive outcomes were emotional working memory accuracy, indexed by EMOBACK proportion correct (all stimuli), and spatial working memory accuracy, indexed by CANTAB SWM between-error. EMOBACK reaction time composite and valence-specific accuracy and reaction time indices (negative, neutral, positive) alongside with SWM strategy were analyzed exploratorily. Linear mixed-effects models controlling for age and sex were used. Bonferroni correction was applied to primary cognitive outcomes.
Results
aiTBS was associated with sustained antidepressant effects, measured by the BDI-II, with significant decrease from baseline to all follow-ups: T2 (β = -10.6, p = .003, d = 1.39), T3 (β = -17.2, p < .001, d = 1.80), T4 (β = -16.2, p < .001, d = 1.39). Response rates (≥ 50% reduction) increased from 6.7% at T2 to 28.6% at T3 and 33.3% at T4.
For the primary EMOBACK accuracy outcome, it significantly increased at the end of the treatment, T3 (β = 0.046, p= .015, d = 0.71). No significant changes were observed at T2 (p = .176) or T4 (p = .866) for overall accuracy. Exploratory analysis showed that reaction time also improved, decreasing from baseline to the end of treatment across all words (β = -102.20, p = .003, d = 0.86). Additionally, valence-specific words analysis indicated that EMOBACK proportion correct for neutral words significantly improved at T3 (β = 0.049, p= .017, d = 0.59), in parallel with its reaction time (β = -88.39, p = .017, d = 0.59). Furthermore, reaction time for negative words started to decrease at the end of the intensive week and remained faster at all post-baseline assessments (T2 – T4; all p-values ≤ .012 and d ≥ 0.72) while the changes at the end of the treatment from baseline for positive words were nominally significant but did not survive correction (β = -105.68, p= .029, d = 0.78).
For spatial working memory, SWM between-errors showed a significant reduction from T1 to T3 (β = -6.10, p = .015, d = 0.69), without significant changes at T2 (p = .135) or T4 (p = .196). CANTAB strategy scores also peaked at the end of treatment (β = -1.537, p = 0.012, d = 0.66), with nominally significant (uncorrected) changes already starting at T2 (β = -1.267, p = .035, d = 0.51) but no further improvement at T4 (p = . 142).
Exploratory responder analyses suggested that greater antidepressant response was associated with larger reductions in EMOBACK reaction times, particularly for positive words (Wald χ²(3) = 7.92, p = .047), whereas spatial working memory changes were less clearly related to clinical improvement.
Discussion
aiTBS was associated with sustained antidepressant improvement across all post-baseline assessments, alongside gains in both emotional and spatial working memory. Remarkably, both emotional and spatial working memory accuracy followed a convergent pattern, peaking at treatment completion with medium effect sizes and showing no significant changes at T2 or T4, suggesting that cumulative stimulation across intensive and maintenance phases may be required for accuracy gains, regardless of working memory domain. By contrast, emotional processing speed of negative words improved already after the intensive week and remained enhanced at follow-up, proposing a different course for speed- and accuracy-related effects. Additionally, parallel improvements in EMOBACK accuracy and reaction time suggest a genuine enhancement in task performance rather than a speed-accuracy tradeoff. In exploratory analyses, BDI-II responders showed greater improvement in EMOBACK reaction time for positive stimuli, suggesting that emotional working memory may be more closely associated to clinical response than spatial strategy performance. Together, these preliminary findings support the feasibility of targeting working memory in TRD with aiTBS and motivate controlled trials to clarify phase-specific dosing, cognitive-affective mechanisms, and links between symptom remission and cognitive changes.
Disclosures
No conflicts of interest to declare.
References
[1] Bortolato, B et al. (2016). Cognitive remission: a novel objective for the treatment of major depression? BMC Medicine, 14:9. doi: 10.1186/s12916-016-0560-3
[2] Kriesche, D et al. (2023). Neurocognitive deficits in depression: a systematic review of cognitive impairment in the acute and remitted state. Eur Arch Psychiatry Clin Neurosci, 273: 1105 - 1128. doi: 10.1007/s00406-022-01479-5
[3] Perini, G et al. (2019). Cognitive impairment in depression: recent advances and novel treatments. Neuropsychiatr Dis and Treat, 15: 1249 - 1258. doi: 10.2147/NDT.S199746
[4] Lam, RW et al. (2014). Cognitive dysfunction in major depressive disorder: effects on psychosocial functioning and implications for treatment. Can J Psychiatry, 59(12): 649 - 654. doi: 10.1177/070674371405901206
[5] Rock, PL et al. (2014). Cognitive impairment in depression: a systematic review and meta-analysis. Psychological Medicine, 44: 2029 - 2040. doi: 10.1017/S0033291713002535
[6] Cole, EJ et al. (2022). Stanford Neuromodulation Therapy (SNT): a double-blind randomized controlled trial. Am J Psychiatry, 179(2): 132 – 141. doi: 10.1176/appi.ajp.2021.20101429
[7] Grimm, S et al. (2012). Neural mechanisms underlying the integration of emotion and working memory. NeuroImage, 61(4): 1188 – 1194. doi: 10.1016/j.neuroimage.2012.04.004
[8] Weigand, A et al. (2013). Lateralized effects of prefrontal repetitive transcranial magnetic stimulation on emotional working memory. Exp Brain Res, 227(1). doi: 10.1007/s00221-013-3483-7