Key Points
Question
What is the therapeutic efficacy of 77.5-Hz, 15-milliampere high-intensity transcranial alternating current stimulation (HI-tACS) in ameliorating the core cognitive-emotional deficits in adolescents with major depressive disorder and comorbid nonsuicidal self-injury?
Findings
In this randomized clinical trial enrolling 60 adolescents, the twice-daily 15-day HI-tACS intervention was associated with significant improvements in emotion regulation, reduced attentional bias toward NSSI-related cues, and alleviated depressive symptoms compared with sham stimulation.
Meaning
The findings of this proof-of-concept trial indicate that HI-tACS targets the underlying pathophysiological mechanisms of nonsuicidal self-injury, providing a promising, rapidly acting, and clinically feasible neuromodulation strategy for this underserved, high-risk patient population.
This randomized clinical trial evaluates the efficacy and safety of high-intensity transcranial alternating current stimulation (HI-tACS) in targeting core deficits of emotion dysregulation and attentional bias in adolescents with major depressive disorder and nonsuicidal self-injury.
Abstract
Importance
Nonsuicidal self-injury (NSSI)—a major threat to adolescents with major depressive disorder (MDD), substantially raising hospitalization and suicidality risk—stems from core deficits in emotion regulation and attentional bias toward self-injurious cues, yet current interventions are limited by slow onset and poor accessibility.
Objective
To evaluate the efficacy and safety of high-intensity transcranial alternating current stimulation (HI-tACS), delivered at 77.5 Hz and 15 milliamperes, in targeting core deficits of emotion dysregulation and attentional bias in adolescents with MDD and NSSI.
Design, Setting, and Participants
This double-blind, sham-controlled randomized clinical trial was conducted at a psychiatric hospital in Western China. Eligible adolescents aged 12 to 18 years diagnosed with comorbid MDD and NSSI were enrolled between October 2025 and January 2026. Data were analyzed in February 2026.
Interventions
Participants were randomly assigned (1:1) to active or sham treatment groups. Each participant received treatment twice daily for 3 weeks, totaling 30 sessions.
Main Outcomes and Measures
The primary outcome was the change in total score on the Difficulties in Emotion Regulation Scale after the intervention. Secondary outcomes included changes in the attentional bias index (via a dot-probe task), scores on the Emotion Regulation Questionnaire, and the Hamilton Depression Rating Scale.
Results
Of 60 enrolled adolescents (mean [SD] age, 14.22 [1.11] years; 48 [80%] female), baseline characteristics were well balanced between the active (30 participants) and sham (30 participants) groups. In the intention-to-treat analysis, emotion regulation scores declined significantly from baseline to posttreatment in the active HI-tACS group (121.38 [95% CI, 119.62 to 123.14] vs 108.15 [95% CI, 106.30 to 109.99]; P < .001) but not in the sham group (119.74 [95% CI, 118.09 to 121.39] vs 117.99 [95% CI, 116.27 to 119.72]). Attentional bias toward NSSI cues also decreased more with active treatment (F = 10.34; P = .002; ηp2 = 0.169). Depression scores also decreased significantly more in the active treatment group than the sham group (estimated mean difference, −2.78 [95% CI, −4.12 to −1.44]; P < .001). Adverse event rates were comparable between active and sham groups (8 participants [29.6%] vs 3 participants [10.7%] reporting ≥1 adverse event).
Conclusions and Relevance
In this randomized clinical trial of 60 adolescents with MDD and comorbid NSSI, HI-tACS was an effective and well-tolerated intervention that simultaneously improved core emotional and cognitive deficits in this population.
Trial Registration
chictr.org.cn Identifier: ChiCTR2500110319
Introduction
Nonsuicidal self-injury (NSSI) is defined as deliberate, repetitive self-inflicted bodily harm. It is intended to temporarily relieve overwhelming emotional distress.1 This behavior represents a dysfunctional emotion regulation strategy.2 Clinically, NSSI is particularly prevalent among adolescents, with a significantly higher incidence in females than in males.3,4,5 In adolescents with major depressive disorder (MDD), comorbid NSSI markedly increases the risk of hospitalization, emergency visits, and future suicide.6,7,8 This high-risk behavior highlights an urgent need for feasible, effective treatments for NSSI in adolescents with depression. However, current clinical practice faces a dual limitation: first-line pharmacological treatments show limited efficacy,9 while effective psychotherapies are often inaccessible due to their lengthy duration and resource intensity.10,11
Deficits in emotion regulation play a central role in NSSI development and maintenance and act as both key predisposing risk factors and the core maintaining mechanisms within the behavioral addiction framework.12,13 Individuals use self-injury to temporarily relieve emotional pain, engaging in a negative reinforcement process that strengthens the behavior and induces addiction-like features.14,15 This behavioral mechanism aligns with well-documented neural circuit dysfunctions: individuals with NSSI display weakened top-down regulation from the prefrontal cortex (particularly the dorsolateral and ventrolateral regions) over limbic structures, such as the amygdala, as well as enhanced bottom-up emotional responses in regions like the anterior cingulate cortex.16,17,18 These neural dysfunctions are reflected in 2 measurable core behavioral phenotypes: difficulties in emotion regulation and excessive attentional bias toward self-injury-related cues.12,19
Notably, emotion regulation difficulties and related cognitive processing bias do not exist independently. Instead, they form a mutually reinforcing pathological unit.20,21 Individuals with poor emotion regulation are more easily influenced by negative environmental cues (eg, NSSI-related cues), and struggle to disengage their attention.19 This entrenched attentional pattern amplifies emotional salience, increases subjective distress, and uses cognitive resources that could support adaptive emotion regulation strategies.12 Thus, attentional bias toward NSSI cues is both a direct cognitive manifestation of dysfunctional neural circuits and a major process that sustains and worsens emotion dysregulation. These factors are critical targets for new interventions.
Transcranial alternating current stimulation (tACS) offers a promising method to test whether neuromodulation may alleviate NSSI by improving emotion regulation and attentional mechanisms. Recent studies show that high-intensity tACS (HI-tACS) at 15 milliamperes (mA) generates an electric field that modulates key nodes in the prefrontal-limbic circuit.22,23 Gamma-frequency (77.5 Hz) HI-tACS is a promising approach. Gamma oscillations are closely linked to higher-order cognitive functions, such as attentional control and cognitive flexibility, and may form the rhythmic basis for prefrontal cortex top-downregulation.24,25 We hypothesize that frequency-specific entrainment of the prefrontal cortex by HI-tACS enhances synchronized activity of prefrontal neuronal populations. This may strengthen inhibitory control over limbic structures and optimize attentional resource allocation. Consequently, this randomized clinical trial directly tests whether 77.5 Hz HI-tACS targeting the prefrontal cortex can simultaneously improve emotion regulation and attentional bias in adolescents with MDD and NSSI while evaluating the intervention’s clinical potential and safety.
Methods
Study Design
This double-blind, sham-controlled randomized clinical trial was conducted at the Third People’s Hospital of Lanzhou, China. It was part of a larger clinical project. The hospital’s ethics committee approved the protocol. All eligible adolescents provided informed assent, and their legal guardians gave written consent. The formal intervention ran from October 2025 to January 2026. The full trial protocol and statistical analysis plan are provided in Supplement 1. This study was reported following the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline.
Participant Recruitment and Screening
Participants were recruited from the hospital’s psychiatric inpatient and outpatient departments. Inclusion criteria were age 12 to 18 years (this range coincides with both the peak prevalence of NSSI and the most pronounced manifestation of the cognitive-emotional deficits targeted by HI-tACS); current major depressive episode confirmed by the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders (Fifth Edition) (DSM-5); at least 3 NSSI episodes meeting DSM-5 criteria in the month before enrollment, with no current suicidal intent (assessed via clinical interview by an independent psychiatrist per DSM-5 diagnostic criteria for suicidal intent); total score at least 18 on the 17-item Hamilton Depression Rating Scale (HAMD-17); and at least 1 legal guardian to supervise and ensure follow-up. Exclusion criteria were current substance use disorder (DSM-5 criteria); comorbid bipolar disorder, schizophrenia spectrum disorder, or other primary psychotic disorder (confirmed via the Structured Clinical Interview for DSM-5); current manic episode, assessed as score more than 12 using the Young Mania Rating Scale to quantify manic symptom severity; known organic brain disease or significant neurological disorder; tACS contraindications; personal or family history of epilepsy; daily lorazepam equivalents more than 2 mg or monoamine oxidase inhibitor use within 2 weeks before enrollment; structured psychological treatment (eg, dialectical behavior therapy, cognitive behavioral therapy) within the past 3 months; prior electroconvulsive therapy; serious or unstable medical conditions (as determined by the investigator to interfere with assessments or increase risk); and pregnancy or lactation. Participants were withdrawn if their antidepressant regimen changed unexpectedly, they had serious adverse events, or they showed poor adherence.
Randomization and Blinding
A centralized randomization procedure with concealed allocation was used. A block-randomized sequence (SAS version 9.4 [SAS Institute]) assigned participants to active or sham groups in a 1:1 ratio. Treatment operators knew the group assignment only for accurate intervention delivery. They did not participate in later assessments or data processing. Clinical evaluators, data managers, and statisticians remained blinded. Blinding integrity was assessed after the intervention: participants attempted to guess their group assignment, and responses were compared between groups.
Procedures
All participants underwent 30 stimulation sessions, twice daily (40-minute sessions) at fixed times, Monday to Friday, over 3 weeks. The active group received 77.5 Hz, 15 mA HI-tACS; the control group received sham stimulation via identical devices with no active current delivery.
Stimulation was delivered using a tACS device (Nexalin Technology) approved by the US Food and Drug Administration and Chinese National Medical Products Administration according to procedures described in previous studies.23,25 Trained nurses positioned electrodes over the forehead-to-mastoid regions per the international 10-20 system: a 4.45 × 9.53 cm electrode on the forehead (Fp1, Fpz, and Fp2) and a 3.18 × 3.81 cm cathodes over bilateral mastoids. The stimulation current ramped up from 1 mA to 15 mA in 2 mA increments, with the following parameters: square waveform, 77.5 Hz frequency, 15 mA intensity (distributed between the frontal and mastoid electrodes), 180-second ramp-up, and 12-second ramp-down. The sham group used an identical apparatus but with zero current output.
To isolate neuromodulation effects, all participants maintained a stable, guideline-concordant MDD pharmacotherapy regimen (selective serotonin reuptake inhibitors or serotonin-norepinephrine reuptake inhibitors, with adjunctive antipsychotics as needed) throughout the trial. Baseline antipsychotic and antidepressant doses were converted to chlorpromazine and fluoxetine equivalents, respectively, using the defined daily dose method.26,27 Doses were standardized and comparable between groups at baseline.
Assessments
The primary outcome was the Difficulties in Emotion Regulation Scale (DERS),28 a psychometrically robust tool widely used in adolescent neuromodulation and psychological intervention research.29,30 Assessments were conducted at baseline, postintervention, and 1-month follow-up. The DERS ranges from 36 to 180, with higher score indicating more severe difficulty. Secondary outcomes were attentional bias at baseline and postintervention, assessed via the same dot-probe task used in Goreis et al19; habitual reappraisal and expressive suppression, assessed using the Emotion Regulation Questionnaire (ERQ; including the 6-item cognitive reappraisal subdomain [range, 6-42] and 4-item expressive suppression subdomain [range, 4-28]; higher scores indicate more frequent use of the strategy)31,32,33; and clinical symptom improvement, assessed using HAMD-17 (range, 0-52; higher score indicates more severe depressive symptoms).34 Detailed information on the assessment instrument is provided in the eMethods in Supplement 2. All self-report scales showed good internal consistency across time points (DERS: Cronbach α = 0.82-0.94; ERQ: Cronbach α = 0.78-0.88; Cronbach α = HAMD-17: 0.80-0.85).
Suicide Risk Assessment and Safety Monitoring
Suicide risk was assessed at screening by an independent psychiatrist; participants with active suicidal intent were excluded. During the 3-week intervention, trained nursing staff conducted daily in-person monitoring for emergent suicidal ideation, with a predefined safety protocol mandating immediate psychiatric evaluation and emergency referral if needed. Adverse events were prospectively collected after each session using a standardized adverse event record form; stimulation-related pain intensity was quantified via a visual analog scale after each session (range, 0-10; with higher score indicating more pain).
Sample Size
Previous studies reported a large effect size for HI-tACS (effect size >1).35 Sample size was estimated using PASS 2021 software (NCSS), with a conservative medium-to-large effect size (Cohen d = 0.8), 2-sided α = .05, and 80% power, yielding a minimum per-group sample size of 25. Accounting for an expected 15% dropout rate (consistent with similar intervention studies in hospitalized adolescents), per-group enrollment was increased to 30, for a total planned sample of 60 participants (1:1 randomization).
Statistical Analysis
Baseline between-group comparisons were conducted by variable type: independent-sample t test (normally distributed continuous variables), Mann-Whitney U test (nonnormally distributed variables), and χ2 test (or Fisher exact test for expected frequencies <5; categorical variables). Primary and secondary outcome analyses followed the intention-to-treat (ITT) principle. For the primary outcome (DERS total score), the model included fixed effects for group, assessment time (baseline, postintervention, 1-month follow-up), and the group-by-time interaction, along with adjustments for age, sex, and baseline scores. For 2-time point behavioral tasks (eg, attentional bias), a 2-factor repeated-measures analysis of variance was used. Other secondary outcomes (emotion regulation strategies, depressive symptoms) used the same modeling framework (baseline, postintervention, 4-week follow-up). The group-by-time interaction was the primary indicator of treatment efficacy; Bonferroni-corrected post hoc comparisons were performed if the interaction was significant.
Supplementary analyses included per-protocol (PP) analysis; sensitivity analyses (restricted to female participants); Spearman rank correlation tests (associations between core mechanism improvements and clinical outcomes); 2 single-mediator models (5000 bootstraps, bias-corrected accelerated CIs) were conducted, with treatment assignment as the independent variable and change scores (posttreatment minus baseline) for mediators and outcome (change in HAMD-17). A 2-sided P < .05 was statistically significant. Data are presented as means and SDs or means and 95% CIs. Data were analyzed using R version 4.5.1 (R Project for Statistical Computing) in February 2026.
Results
Participants
Of 81 screened patients, 60 (mean [SD] age, 14.22 [1.11] years; 48 [80%] female) were enrolled and randomized 1:1 to active HI-tACS (30 participants) or sham (30 participants) (Figure), with balanced baseline characteristics (Table 1). All 60 participants were included in the ITT analysis. At baseline, DERS total scores were 121.38 (95% CI, 119.62 to 123.14) in the active group and 119.74 (95% CI, 118.09 to 121.39) in the sham group. During the intervention, 3 active group participants (10.0%) and 2 sham group participants (6.7%) discontinued (withdrawal of consent: 2 active group participants; 1 sham group participant; nonadherence: 1 participant per group); no discontinuations were due to adverse events. Posttreatment, 2 sham group participants (6.7%) were lost to follow-up (none in the active group). Discontinuation rates did not differ significantly between groups. Treatment completion was achieved by 27 active group participants (90.0%) and 28 sham group participants (93.3%), forming the PP populations.
Figure. Flowchart of Study Enrollment.
HI-tACS indicates high-intensity transcranial alternating current stimulation; ITT, intention-to-treat; PP, per-protocol.
Table 1. Baseline Demographic and Clinical Characteristics of Randomized Participants.
| Characteristic | Mean (SD) | ||
|---|---|---|---|
| Treatment group | Total (N = 60) | ||
| HI-tACS (n = 30) | Sham (n = 30) | ||
| Demographics | |||
| Sex, No. (%) | |||
| Male | 5 (16.7) | 7 (23.3) | 12 (20.0) |
| Female | 25 (83.3) | 23 (76.7) | 48 (80.0) |
| Age, y | 14.47 (1.01) | 13.97 (1.19) | 14.22 (1.11) |
| Only child, No. (%) | 12 (40.0) | 14 (46.7) | 26 (43.3) |
| Urban residence, No. (%) | 13 (43.3) | 11 (36.7) | 24 (40.0) |
| Live on campus, No. (%) | 11 (36.7) | 14 (46.7) | 25 (41.7) |
| Clinical characteristics | |||
| First hospitalization, No. (%) | 15 (50.0) | 13 (43.3) | 28 (46.7) |
| NSSI duration, y | 1.83 (0.70) | 2.03 (0.81) | 1.93 (0.76) |
| Hospitalization duration, d | 8.30 (2.79) | 7.67 (3.55) | 7.98 (3.18) |
| Chlorpromazine equivalent, mg/d | 95.70 (67.31) | 93.27 (68.32) | 94.48 (67.53) |
| Fluoxetine equivalent, mg/d | 34.53 (14.61) | 33.17 (12.85) | 33.85 (13.70) |
| DSHI-AR | 26.70 (15.48) | 29.30 (15.46) | 28.00 (15.44) |
| Suicide attempts, No. | 0.87 (0.90) | 0.70 (0.75) | 0.78 (0.83) |
| BIS-11 | 74.87 (6.55) | 73.43 (5.92) | 74.15 (6.22) |
| HAMD-17 | 25.67 (4.04) | 26.63 (3.19) | 26.15 (3.64) |
| HAMA | 22.97 (3.60) | 23.50 (3.42) | 23.23 (3.50) |
| DERS subscales | |||
| Nonacceptance | 20.40 (3.86) | 19.17 (3.24) | 19.78 (3.57) |
| Goals | 19.80 (3.60) | 20.93 (3.43) | 20.37 (3.53) |
| Impulse control | 21.67 (4.44) | 19.97 (4.41) | 20.82 (4.48) |
| Awareness | 17.33 (2.70) | 18.60 (3.45) | 17.97 (3.12) |
| Strategies | 23.87 (4.22) | 21.87 (4.81) | 22.87 (4.58) |
| Clarity | 19.73 (3.28) | 18.37 (3.37) | 19.05 (3.35) |
| Total | 122.80 (9.70) | 118.70 (12.13) | 120.75 (11.00) |
| ERQ subscales | |||
| Reappraisal | 29.77 (5.64) | 28.07 (4.60) | 28.92 (5.17) |
| Suppression | 19.10 (4.03) | 18.80 (3.62) | 18.95 (3.81) |
Abbreviations: BIS-11, Barratt Impulsiveness Scale-11; DERS, Difficulties in Emotion Regulation Scale; DSHI-AR Deliberate Self-Harm Inventory–Adolescent Revised Version; ERQ, Emotion Regulation Questionnaire; HAMA, Hamilton Anxiety Rating Scale; HAMD-17, 17-item Hamilton Depression Rating Scale; HI-tACS, high-intensity transcranial alternating current stimulation; NSSI, nonsuicidal self-harm.
Primary Outcome
ITT analysis revealed a significant time-by-group interaction for DERS total score (F = 32.79, P < .001). The active group had significantly lower mean DERS scores than the sham group at postintervention (108.15 [95% CI, 106.30 to 109.99] vs 117.99 [95% CI, 116.27 to 119.72]; P < .001; Cohen d = 0.89) and follow-up (109.19 [95% CI, 107.34 to 111.03] vs 117.73 [95% CI, 115.94 to 119.53]; P < .001; Cohen d = 0.74) (Table 2). DERS subscale analyses showed improvements were primarily driven by 2 core dimensions: difficulties engaging in goal-directed behavior (interaction F = 15.51; P < .001) and difficulties with impulse control (interaction F = 45.28; P < .001), with the active group scoring significantly lower at postintervention and follow-up (Table 2).
Table 2. Intention-to-Treat Analyses of Primary and Secondary Measurementsa.
| Outcomes | Active group | Sham group | Estimated mean difference (95% CI) | P value | P value for interaction | |||
|---|---|---|---|---|---|---|---|---|
| No. | Adjusted mean (95% CI) | No. | Adjusted mean (95% CI) | |||||
| DERS: total score | ||||||||
| Baseline | 30 | 121.38 (119.62 to 123.14) | 30 | 119.74 (118.09 to 121.39) | NA | NA | <.001 | |
| Posttreatment | 27 | 108.15 (106.30 to 109.99) | 28 | 117.99 (116.27 to 119.72) | −9.85 (−12.29 to −7.40) | <.001 | ||
| Follow-up | 27 | 109.19 (107.34 to 111.03) | 26 | 117.73 (115.94 to 119.53) | −8.55 (−11.03 to −6.06) | <.001 | ||
| DERS: nonacceptance | ||||||||
| Baseline | 30 | 19.79 (19.20 to 20.39) | 30 | 19.56 (19.01 to 20.12) | NA | NA | .96 | |
| Posttreatment | 27 | 19.20 (18.58 to 19.82) | 28 | 19.07 (18.49 to 19.65) | 0.13 (−0.68 to 0.95) | .75 | ||
| Follow-up | 27 | 18.94 (18.32 to 19.56) | 26 | 18.68 (18.09 to 19.26) | 0.26 (−0.55 to 1.08) | .53 | ||
| DERS: goals | ||||||||
| Baseline | 30 | 20.32 (19.52 to 21.12) | 30 | 20.41 (19.66 to 21.17) | NA | NA | <.001 | |
| Posttreatment | 27 | 16.06 (15.22 to 16.90) | 28 | 19.68 (18.89 to 20.47) | −3.62 (−4.73 to −2.51) | <.001 | ||
| Follow-up | 27 | 16.84 (16.00 to 17.68) | 26 | 19.97 (19.18 to 20.75) | −3.13 (−4.24 to −2.02) | <.001 | ||
| DERS: impulse control | ||||||||
| Baseline | 30 | 21.12 (20.25 to 22.00) | 30 | 20.44 (19.60 to 21.27) | NA | NA | <.001 | |
| Posttreatment | 27 | 15.39 (14.47 to 16.31) | 28 | 20.80 (19.92 to 21.67) | −5.41 (−6.65 to −4.17) | <.001 | ||
| Follow-up | 27 | 15.54 (14.62 to 16.45) | 26 | 20.80 (19.92 to 21.67) | −5.26 (−6.50 to −4.02) | <.001 | ||
| DERS: awareness | ||||||||
| Baseline | 30 | 17.98 (17.35 to 18.62) | 30 | 18.25 (17.63 to 18.87) | NA | NA | .56 | |
| Posttreatment | 27 | 17.29 (16.63 to 17.96) | 28 | 18.04 (17.39 to 18.68) | −0.75 (−1.64 to 0.15) | .10 | ||
| Follow-up | 27 | 17.59 (16.92 to 18.25) | 26 | 17.75 (17.11 to 18.40) | −0.16 (−1.06 to 0.73) | .72 | ||
| DERS: strategies | ||||||||
| Baseline | 30 | 23.05 (22.36 to 23.74) | 30 | 22.67 (22.02 to 23.32) | NA | NA | .29 | |
| Posttreatment | 27 | 21.97 (21.24 to 22.69) | 28 | 22.53 (21.85 to 23.21) | −0.57 (−1.53 to 0.40) | .25 | ||
| Follow-up | 27 | 22.08 (21.35 to 22.80) | 26 | 21.96 (21.28 to 22.64) | 0.12 (−0.85 to 1.08) | .81 | ||
| DERS: clarity | ||||||||
| Baseline | 30 | 19.08 (18.47 to 19.68) | 30 | 18.57 (18.01 to 19.13) | NA | NA | .79 | |
| Posttreatment | 27 | 18.22 (17.60 to 18.85) | 28 | 18.07 (17.48 to 18.65) | 0.16 (−0.68 to 0.99) | .71 | ||
| Follow-up | 27 | 18.19 (17.56 to 18.82) | 26 | 17.89 (17.31 to 18.48) | 0.30 (−0.53 to 1.13) | .49 | ||
| ERQ: reappraisal | ||||||||
| Baseline | 30 | 27.34 (25.92 to 28.75) | 30 | 26.91 (25.58 to 28.24) | NA | NA | .01 | |
| Posttreatment | 27 | 32.01 (30.53 to 33.48) | 28 | 28.10 (26.71 to 29.48) | 3.91 (1.96 to 5.86) | <.001 | ||
| Follow-up | 27 | 31.12 (29.64 to 32.60) | 26 | 28.08 (26.64 to 29.52) | 3.04 (1.07 to 5.02) | <.01 | ||
| ERQ: suppression | ||||||||
| Baseline | 30 | 19.01 (18.36 to 19.65) | 30 | 18.79 (18.18 to 19.40) | NA | NA | .81 | |
| Posttreatment | 27 | 18.68 (18.00 to 19.35) | 28 | 18.08 (17.45 to 18.72) | 0.59 (−0.30 to 1.48) | .19 | ||
| Follow-up | 27 | 18.79 (18.11 to 19.46) | 26 | 18.39 (17.73 to 19.05) | 0.39 (−0.51 to 1.30) | .39 | ||
| HAMD-17 | ||||||||
| Baseline | 30 | 25.40 (24.44 to 26.36) | 30 | 26.18 (25.27 to 27.10) | NA | NA | .01 | |
| Posttreatment | 27 | 15.78 (14.77 to 16.78) | 28 | 18.56 (17.60 to 19.51) | −2.78 (−4.12 to −1.44) | <.001 | ||
| Follow-up | 27 | 16.15 (15.14 to 17.15) | 26 | 19.45 (18.45 to 20.44) | −3.30 (−4.65 to −1.95) | <.001 | ||
Abbreviations: DERS, Difficulties in Emotion Regulation Scale; ERQ, Emotion Regulation Questionnaire; HAMD-17, Hamilton Depression Rating Scale–17 item version; NA, not applicable.
Random intercept mixed models were used including time, group, and group-by-time interaction as fixed effects after adjusting for age, sex, duration of nonsuicidal self-injury, baseline HAMD-17, fluoxetine equivalent, and relevant baseline outcome measurements. Post hoc multiple comparisons were Bonferroni-corrected.
Secondary Outcomes
Adaptive Emotion Regulation Strategies
A significant time-by-group interaction was observed for reappraisal (F = 3.93; P = .01), with a greater increase in the active group (P < .001; Cohen d = 0.85). No significant between-group differences were found for expressive suppression (Table 2).
Attention to Threat Stimuli
Significant time-by-group interactions were observed for attentional bias (F = 10.34; P = .002; ηp2 = 0.17), difficulty in disengagement from threat (F = 13.67; P = .001, ηp2 = 0.21), and threat avoidance (F = 8.10; P = .006; ηp2 = 0.14); the interaction for initial orienting was not significant (F = 3.07; P = .09; ηp2 = 0.06). No significant main effect of time was observed for any index. A simple effect analysis showed that the active group had greater reductions in attentional bias, improved threat disengagement, and increased threat avoidance than the sham group (Table 3; eTable 1 in Supplement 2).
Table 3. Between-Group Comparisons of Dot-Probe Task Indicesa.
| Characteristic | Mean (SD) | t Test | P value | |
|---|---|---|---|---|
| Active group | Sham group | |||
| Baseline | ||||
| Attention bias | 4.84 (19.00) | −1.89 (10.94) | 1.57 | .12 |
| Initial orienting | 0.18 (20.47) | 4.95 (11.35) | −1.04 | .30 |
| Difficulty disengaging | 1.00 (12.23) | −5.54 (14.23) | 1.80 | .08 |
| Threat avoidance | −3.42 (19.32) | 2.27 (10.86) | −1.32 | .19 |
| Posttreatment | ||||
| Attention bias | −6.31 (12.49) | 4.02 (12.36) | −3.02 | .004 |
| Initial orienting | 6.42 (10.31) | 1.33 (13.38) | 1.55 | .13 |
| Difficulty disengaging | −9.38 (11.86) | 1.64 (13.58) | −3.15 | .003 |
| Threat avoidance | 6.60 (12.33) | −2.87 (12.69) | 2.76 | .008 |
Two-factor repeated measures analysis of variance. All independent-sample t tests were performed under the assumption of equal variances, as Levene tests for equality of variances were not significant for all indices.
Clinical Symptoms
A significant time-by-group interaction was found for depressive symptoms (HAMD-17: F = 4.30; P = .02). While symptom scores decreased significantly over time in both groups, the reduction was greater in the active group than the control group (estimated mean difference, −2.78 [95% CI, −4.12 to −1.44]; P < .001; Cohen d = 1.15), with lower mean scores at follow-up (P < .001; Cohen d = 1.54) (Table 2).
Exploratory Analysis
Correlation analysis found that improvements in emotion regulation were significantly associated with reductions in depressive symptoms (r = 0.74; P < .001) and attentional bias difficulty (r = −0.61; P = .002) (eTable 2 in Supplement 2). Reduced disengagement difficulty was strongly correlated with decreased depressive symptoms (r = 0.73; P < .001). Increased reappraisal use correlated with improved threat avoidance (r = 0.63; P = .002), and improved threat avoidance was significantly negatively correlated with improved emotion regulation (r = −0.77; P < .001). Post hoc exploratory mediation analyses showed neither emotion regulation change (indirect effect = –1.02 [95% CI, –2.87 to 0.56]) nor attentional bias change (indirect effect = –0.73 [95% CI, –2.11 to 0.38]) significantly mediated the treatment effect on depressive symptoms.
Supplementary Analyses
PP analysis results were largely consistent with ITT primary and secondary outcomes (eTable 3 in Supplement 2). Sensitivity analyses restricted to female participants yielded results similar to the primary analysis (eTable 4 in Supplement 2).
Blinding Integrity
A total of 55 participants attempted to guess their treatment allocation (eTable 5 in Supplement 2): 18 active group participants (66.7%) and 13 sham group participants (46.4%) believed they received active treatment; 5 active group (18.5%) and 6 sham group (21.4%) participants believed they received sham treatment; the remainder were uncertain. Fisher exact test showed no significant between-group difference in perceived allocation (P = .24).
Safety
Of 27 participants in each group reporting on adverse events, 8 active group participants (29.6%) and 3 sham group participants (10.7%) reported at least 1 adverse event (Fisher exact test: P = .10). Headache was the most common (3 active group participants [11.1%]; 1 sham group participant [3.6%]), with other events including sleepiness (2 active group participants), stimulation site tenderness (2 active group participants; 1 sham group participant), and dizziness (1 participant from each group). No participants reported phosphenes or other perceptual phenomena. All adverse events were mild (grade 1 using the Common Terminology Criteria for Adverse Events version 5.0) and resolved spontaneously without intervention; no participants discontinued treatment due to adverse events (eTable 6 in Supplement 2).
Discussion
To our knowledge, this double-blind, sham-controlled randomized clinical trial provides the first evidence that 77.5 Hz HI-tACS targeting the prefrontal-limbic circuit is safe and effective for adolescents with MDD and NSSI, simultaneously improving emotion regulation, reducing NSSI-related attentional bias, and alleviating depressive symptoms. These findings establish proof of concept that noninvasive neuromodulation can concurrently target NSSI’s core cognitive-emotional mechanisms, supporting mechanism-guided, transdiagnostic interventions for this high-risk population.
Clinically, while emotion regulation and dialectical behavior therapies are effective, their lengthy protocols hinder implementation.36 In contrast, the 3-week HI-tACS protocol shows potential for rapid onset of effects, offering a standardized clinical option. Notably, HI-tACS may have a distinct mechanism of action from intermittent theta-burst stimulation, for which the efficacy in reducing self-injury ideation is attributed to long-term potentiation-like synaptic plasticity,34 whereas HI-tACS may optimize prefrontal-limbic neural oscillatory synchrony, as supported by simultaneous improvements in emotion regulation and attentional bias.
Our proposed spatial-frequency synergy framework unifies these observations. Spatially, 15 mA HI-tACS modulates both prefrontal control regions and deep limbic response nodes.23 Frequency-wise, 77.5 Hz gamma stimulation targets cognitive control-related neural oscillations.37,38 We propose that HI-tACS synchronizes dysregulated gamma activity via neural entrainment, enhancing prefrontal efficiency in emotion regulation and impulse inhibition—consistent with observed improvements in impulse control and attentional disengagement. Building on this framework and consistent with prior work,20,21 as well as the strong correlation we observed between increased threat avoidance and improved emotion regulation, the concurrent improvements in both mechanisms support a bidirectional feedback model: improved attentional control reduces NSSI cue salience, diminishing reliance on maladaptive emotion regulation; conversely, enhanced emotion regulation frees cognitive resources for flexible attentional disengagement. This reciprocal interaction creates a positive feedback loop that amplifies therapeutic effects, although our single posttreatment assessment prevents us from establishing definitive temporal precedence, and future longitudinal studies are needed to clarify these directional relationships.
Beyond its implications for NSSI, this randomized clinical trial advances the field of precision psychiatry by targeting core cognitive-emotional deficits, rather than focusing solely on depressive symptomology. The spatial-frequency synergy framework proposed herein may be adapted to target other transdiagnostic mechanisms, such as theta-frequency stimulation for cognitive control deficits in attention-deficit/hyperactivity disorder and gamma-frequency stimulation for social cognitive impairments in autism spectrum disorder. Additionally, this study provides a template for mechanism-based trial design, wherein mechanistic outcomes are incorporated as primary targets alongside clinical outcomes, thereby enabling smaller, more focused trials that directly test mechanistic hypotheses.
The mediation findings were not statistically significant and should be interpreted with caution. First, this trial was powered for the primary outcome, leaving the mediation analysis underpowered to detect small to medium indirect effects. Second, our single-mediator approach could not capture the synergistic effect of the mutually reinforcing emotion regulation and attentional bias mechanisms, as our core hypothesis posits. Third, the synchronous assessment of mediators and depressive changes lacked the temporal order required to establish causal mediation effects.
The numerically higher adverse event rate in the active group compared with the sham group (8 participants [29.6%] vs 3 participants [10.7%]) reflects expected active neuromodulation effects, consistent with prior HI-tACS studies.23,24 All events were mild and self-limited. The number needed to harm for any adverse event was 5.3, while the number needed to treat for depressive symptom response (≥50% HAMD-17 reduction) was 3.2. Given the mild, transient adverse events and substantial clinical benefit for this high-risk, treatment-limited population, the risk-benefit profile is favorable. Pretreatment counseling about mild adverse events (eg, headache) may improve adherence.
Limitations
This study has some limitations. First, we lacked direct neuroimaging evidence for neural oscillation modulation; future research should incorporate electroencephalogram or functional magnetic resonance imaging to assess prefrontal gamma oscillation power and prefrontal-amygdala functional connectivity. Second, our single-center design limits generalizability; multicenter studies with diverse populations are needed to validate findings. Third, the 1-month follow-up, designed to assess early mechanistic changes, limits evaluation of long-term efficacy, durability, and outcomes like NSSI recurrence. The short-term improvements observed in this study do not equate to sustained long-term benefit, and longer follow-up studies are required.
Conclusions
The findings of this randomized clinical trial suggest that 77.5 Hz HI-tACS targeting the prefrontal-limbic circuit is a promising intervention for rapidly addressing NSSI’s core pathology via spatial-frequency synergy. This intervention offers a new clinical option for adolescent NSSI and establishes a behavioral evidence base for mechanism-guided noninvasive neuromodulation. Future multicenter, multimodal neuroimaging studies will confirm efficacy and identify biomarkers of treatment response.
Trial Protocol and Statistical Analysis Plan
eMethods.
eTable 1. Effects of time and group on dot-probe task indices
eTable 2. Spearman correlations among changes in clinical and cognitive measures following Hi-tACS intervention.
eTable 3. PP analyses of primary and secondary measurements.
eTable 4. Sensitivity analysis: efficacy outcomes in the female subgroup
eTable 5. Blinding assessment
eTable 6. Adverse events reported during the trial
Data Sharing Statement
References
- 1.Lengel GJ, Ammerman BA, Washburn JJ. NSSI in the DSM-5. In: Lloyd-Richardson EE, Baetens I, Whitlock JL, eds. The Oxford Handbook of Nonsuicidal Self-Injury. Oxford University Press; 2023:7-23. doi: 10.1093/oxfordhb/9780197611272.013.2 [DOI] [Google Scholar]
- 2.Kuehn KS, Dora J, Harned MS, et al. A meta-analysis on the affect regulation function of real-time self-injurious thoughts and behaviours. Nat Hum Behav. 2022;6(7):964-974. doi: 10.1038/s41562-022-01340-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.De Luca L, Pastore M, Palladino BE, Reime B, Warth P, Menesini E. The development of non-suicidal self-injury (NSSI) during adolescence: a systematic review and bayesian meta-analysis. J Affect Disord. 2023;339:648-659. doi: 10.1016/j.jad.2023.07.091 [DOI] [PubMed] [Google Scholar]
- 4.Moloney F, Amini J, Sinyor M, Schaffer A, Lanctôt KL, Mitchell RHB. Sex differences in the global prevalence of nonsuicidal self-injury in adolescents: a meta-analysis. JAMA Netw Open. 2024;7(6):e2415436. doi: 10.1001/jamanetworkopen.2024.15436 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Zhang P, Ouyang L, Liang M, et al. A cross-sectional epidemiological study of non-suicidal self-injury prevalence in Chinese psychiatric patients. Nat Ment Health. 2023;1(4):266-272. doi: 10.1038/s44220-023-00050-y [DOI] [Google Scholar]
- 6.Hamza CA, Willoughby T, Heffer T. Impulsivity and nonsuicidal self-injury: A review and meta-analysis. Clin Psychol Rev. 2015;38:13-24. doi: 10.1016/j.cpr.2015.02.010 [DOI] [PubMed] [Google Scholar]
- 7.Tuisku V, Kiviruusu O, Pelkonen M, Karlsson L, Strandholm T, Marttunen M. Depressed adolescents as young adults—predictors of suicide attempt and non-suicidal self-injury during an 8-year follow-up. J Affect Disord. 2014;152-154:313-319. doi: 10.1016/j.jad.2013.09.031 [DOI] [PubMed] [Google Scholar]
- 8.Victor SE, Klonsky ED. Correlates of suicide attempts among self-injurers: a meta-analysis. Clin Psychol Rev. 2014;34(4):282-297. doi: 10.1016/j.cpr.2014.03.005 [DOI] [PubMed] [Google Scholar]
- 9.Huang X, Harris LM, Funsch KM, Fox KR, Ribeiro JD. Efficacy of psychotropic medications on suicide and self-injury: a meta-analysis of randomized controlled trials. Transl Psychiatry. 2022;12(1):400. doi: 10.1038/s41398-022-02173-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Turner BJ, Austin SB, Chapman AL. Treating nonsuicidal self-injury: a systematic review of psychological and pharmacological interventions. Can J Psychiatry. 2014;59(11):576-585. doi: 10.1177/070674371405901103 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Fonagy P, Luyten P. Socioeconomic and sociocultural factors affecting access to psychotherapies: the way forward. World Psychiatry. 2021;20(3):315-316. doi: 10.1002/wps.20911 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Brausch AM, Clapham RB, Littlefield AK. Identifying specific emotion regulation deficits that associate with nonsuicidal self-injury and suicide ideation in adolescents. J Youth Adolesc. 2022;51(3):556-569. doi: 10.1007/s10964-021-01525-w [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Poon JA, López R Jr, Marie-Shea L, Liu RT. Longitudinal relations between childhood maltreatment, emotion regulation difficulties, and suicidal ideation and non-suicidal self-injury: an 18-month investigation of psychiatrically hospitalized adolescents. Res Child Adolesc Psychopathol. 2023;51(9):1315-1326. doi: 10.1007/s10802-023-01067-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Liu J, Gao Y, Liang C, Liu X. The potential addictive mechanism involved in repetitive nonsuicidal self-injury: the roles of emotion dysregulation and impulsivity in adolescents. J Behav Addict. 2022;11(4):953-962. doi: 10.1556/2006.2022.00077 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Himelein-Wachowiak M, Giorgi S, Kwarteng A, et al. Getting “clean” from nonsuicidal self-injury: experiences of addiction on the subreddit r/selfharm. J Behav Addict. 2022;11(1):128-139. doi: 10.1556/2006.2022.00005 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Cummings LR, Mattfeld AT, Pettit JW, McMakin DL. Viewing nonsuicidal self-injury in adolescence through a developmental neuroscience lens: the impact of neural sensitivity to socioaffective pain and reward. Clin Psychol Sci. 2021;9(5):767-790. doi: 10.1177/2167702621989323 [DOI] [Google Scholar]
- 17.Lang AN, Zhong Y, Lei W, et al. Neural mechanism of non-adaptive cognitive emotion regulation in patients with non-suicidal self-injury. Compr Psychiatry. 2024;133:152487. doi: 10.1016/j.comppsych.2024.152487 [DOI] [PubMed] [Google Scholar]
- 18.Wu D, Wang S, Wang H, Zhang J, Tian Y. Neural mechanisms and neuromodulation therapies for non-suicidal self-injury. Mol Psychiatry. 2025;30(11):5513-5522. doi: 10.1038/s41380-025-03125-7 [DOI] [PubMed] [Google Scholar]
- 19.Goreis A, Pfeffer B, Hajek Gross C, et al. Attentional biases and nonsuicidal self-injury urges in adolescents. JAMA Netw Open. 2024;7(7):e2422892. doi: 10.1001/jamanetworkopen.2024.22892 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Joormann J, Quinn ME. Cognitive processes and emotion regulation in depression. Depress Anxiety. 2014;31(4):308-315. doi: 10.1002/da.22264 [DOI] [PubMed] [Google Scholar]
- 21.Todd RM, Cunningham WA, Anderson AK, Thompson E. Affect-biased attention as emotion regulation. Trends Cogn Sci. 2012;16(7):365-372. doi: 10.1016/j.tics.2012.06.003 [DOI] [PubMed] [Google Scholar]
- 22.Shan Y, Wang H, Yang Y, et al. Evidence of a large current of transcranial alternating current stimulation directly to deep brain regions. Mol Psychiatry. 2023;28(12):5402-5410. doi: 10.1038/s41380-023-02150-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Wang J, Zhao W, Wang H, et al. Brain-wide activation involved in 15 mA transcranial alternating current stimulation in patients with first-episode major depressive disorder. Gen Psychiatr. 2024;37(2):e101338. doi: 10.1136/gpsych-2023-101338 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Fu Z, Tian J, Kang S, et al. Efficacy and safety of transcranial alternating current stimulation in adults with attention deficit hyperactivity disorder: a double-blind randomized sham-controlled trial. Mol Psychiatry. 2026;31(5):2675-2684. doi: 10.1038/s41380-025-03407-0 [DOI] [PubMed] [Google Scholar]
- 25.You Z, He Z, Zou L, Chen S, Tang J. High intensity transcranial altering current stimulation for cognitive dysfunction in schizophrenia: a pilot trial. Brain Stimul. 2025;18(2):119-121. doi: 10.1016/j.brs.2025.01.021 [DOI] [PubMed] [Google Scholar]
- 26.Furukawa TA, Cipriani A, Cowen PJ, Leucht S, Egger M, Salanti G. Optimal dose of selective serotonin reuptake inhibitors, venlafaxine, and mirtazapine in major depression: a systematic review and dose-response meta-analysis. Lancet Psychiatry. 2019;6(7):601-609. doi: 10.1016/S2215-0366(19)30217-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Leucht S, Samara M, Heres S, Davis JM. Dose equivalents for antipsychotic drugs: the DDD method. Schizophr Bull. 2016;42(Suppl 1)(suppl 1):S90-S94. doi: 10.1093/schbul/sbv167 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Gratz KL, Roemer L. Multidimensional assessment of emotion regulation and dysregulation: development, factor structure, and initial validation of the Difficulties in Emotion Regulation Scale. J Psychopathol Behav Assess. 2004;26(1):41-54. doi: 10.1023/B:JOBA.0000007455.08539.94 [DOI] [Google Scholar]
- 29.Diefenbach GJ, Assaf M, Goethe JW, Gueorguieva R, Tolin DF. Improvements in emotion regulation following repetitive transcranial magnetic stimulation for generalized anxiety disorder. J Anxiety Disord. 2016;43:1-7. doi: 10.1016/j.janxdis.2016.07.002 [DOI] [PubMed] [Google Scholar]
- 30.Moltrecht B, Deighton J, Patalay P, Edbrooke-Childs J. Effectiveness of current psychological interventions to improve emotion regulation in youth: a meta-analysis. Eur Child Adolesc Psychiatry. 2021;30(6):829-848. doi: 10.1007/s00787-020-01498-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.He Z, Liu Z, Zhao J, Elliott R, Zhang D. Improving emotion regulation of social exclusion in depression-prone individuals: a tDCS study targeting right VLPFC. Psychol Med. 2020;50(16):2768-2779. doi: 10.1017/S0033291719002915 [DOI] [PubMed] [Google Scholar]
- 32.Li Y, Li S, Tang Y, Hao S, Zhang D. Causal evidence for the role of prefrontal theta oscillations in emotion regulation using neurofeedback training. Neuroimage. 2025;320:121457. doi: 10.1016/j.neuroimage.2025.121457 [DOI] [PubMed] [Google Scholar]
- 33.Tang Y, Mo L, Peng Z, Li Y, Zhang D. Causal enhancement of cognitive reappraisal through synchronized dorsolateral and ventrolateral prefrontal cortex activity. Emotion. 2025;25(6):1418-1428. doi: 10.1037/emo0001507 [DOI] [PubMed] [Google Scholar]
- 34.Qin Y, Chen H, Liu F, et al. Intermittent theta burst stimulation for non-suicidal self-injury in adolescents with major depressive disorder: a randomized, sham-controlled trial. Mol Psychiatry. 2026;31(2):860-868. doi: 10.1038/s41380-025-03183-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Wang H, Wang K, Xue Q, et al. Transcranial alternating current stimulation for treating depression: a randomized controlled trial. Brain. 2022;145(1):83-91. doi: 10.1093/brain/awab252 [DOI] [PubMed] [Google Scholar]
- 36.Bjureberg J, Sahlin H, Hellner C, et al. Emotion regulation individual therapy for adolescents with nonsuicidal self-injury disorder: a feasibility study. BMC Psychiatry. 2017;17(1):411. doi: 10.1186/s12888-017-1527-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Popov T, Steffen A, Weisz N, Miller GA, Rockstroh B. Cross-frequency dynamics of neuromagnetic oscillatory activity: two mechanisms of emotion regulation. Psychophysiology. 2012;49(12):1545-1557. doi: 10.1111/j.1469-8986.2012.01484.x [DOI] [PubMed] [Google Scholar]
- 38.Sun Y, Giacobbe P, Tang CW, et al. Deep brain stimulation modulates gamma oscillations and theta–gamma coupling in treatment resistant depression. Brain Stimul. 2015;8(6):1033-1042. doi: 10.1016/j.brs.2015.06.010 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Trial Protocol and Statistical Analysis Plan
eMethods.
eTable 1. Effects of time and group on dot-probe task indices
eTable 2. Spearman correlations among changes in clinical and cognitive measures following Hi-tACS intervention.
eTable 3. PP analyses of primary and secondary measurements.
eTable 4. Sensitivity analysis: efficacy outcomes in the female subgroup
eTable 5. Blinding assessment
eTable 6. Adverse events reported during the trial
Data Sharing Statement
