Combined transcranial direct current and repetitive transcranial magnetic stimulation for adolescent non-suicidal self-injury

Abstract

BACKGROUND

Non-suicidal self-injury (NSSI) is prevalent among adolescents and associated with increased suicide risk. Current psychotherapy has limited efficacy, and no approved pharmacotherapy exists. Combined transcranial direct current stimulation (tDCS) and repetitive transcranial magnetic stimulation (rTMS) targeting prefrontal-limbic dysfunction may offer a promising treatment approach, but evidence in adolescent NSSI is lacking.

AIM

To retrospectively analyze the clinical efficacy of tDCS combined with rTMS in treating NSSI in adolescents, and to explore factors influencing treatment outcomes.

METHODS

A retrospective study design was adopted. Clinical data from 256 adolescent NSSI patients treated at the psychology department of a tertiary hospital between December 2024 to December 2025 were collected. Based on treatment regimens, patients were divided into the combination group (tDCS + rTMS + psychotherapy, n = 86), the rTMS group (rTMS + psychotherapy, n = 92), and the control group (psychotherapy alone, n = 78). Changes in Self-Harm Inventory (SHI), Beck Depression Inventory-II (BDI-II), Beck Anxiety Inventory (BAI), Barratt Impulsiveness Scale-11 (BIS-11), and Difficulties in Emotion Regulation Scale (DERS) scores before and after treatment were compared among the three groups. Univariate and multivariate logistic regression analyses were used to identify factors associated with treatment response, and multiple linear regression was used to analyze factors influencing SHI score improvement.

RESULTS

After 6 weeks of treatment, SHI, BDI-II, BAI, BIS-11, and DERS scores significantly decreased in all three groups compared to baseline (P < 0.01), with the combination group showing greater improvement than the rTMS and control groups (P < 0.05). The overall response rate was 88.4% in the combination group, significantly higher than the rTMS group (72.8%) and control group (55.1%) (P < 0.01). Multivariate logistic regression showed that treatment regimen (combination therapy), baseline DERS score, illness duration, and comorbid depression were independent factors associated with treatment response (P < 0.05). Multiple linear regression showed that treatment regimen, baseline SHI score, baseline BIS-11 score, and childhood trauma history were independent predictors of SHI score improvement (P < 0.05).

CONCLUSION

tDCS combined with rTMS shows significant efficacy in treating adolescent NSSI. Patients with lower levels of emotion regulation difficulties, shorter illness duration, and comorbid depression respond better to combination therapy. Treatment regimen, baseline self-injury severity, and impulsivity level are important predictors of treatment outcomes.

Key Words: Transcranial direct current stimulation; Repetitive transcranial magnetic stimulation; Non-suicidal self-injury; Adolescents; Influencing factors; Logistic regression; Retrospective study

Core Tip: Non-suicidal self-injury in adolescents remains difficult to treat, and evidence for neuromodulation is limited. This retrospective study demonstrates that combined transcranial direct current stimulation and repetitive transcranial magnetic stimulation (rTMS) is associated with greater improvement in self-injury behaviors, emotional regulation, impulsivity, and affective symptoms compared with rTMS alone or psychotherapy. Moreover, baseline emotion regulation difficulty, illness duration, comorbid depression, and childhood trauma history were identified as important factors influencing treatment response, providing clinically relevant guidance for individualized intervention strategies.

INTRODUCTION

It is described as non-suicidal self-injury (NSSI) involving deliberate destruction of body tissue pierce without suicide intention mainly demonstrated by cutting, burning, hitting or scratching and other[1]. NSSI has been nominated as a condition for further research in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), underscoring the clinical seriousness associated with this construct. Epidemiological research suggests that lifetime prevalence of NSSI among adolescents is 17%-18% in population-based samples, and between 40%-60% in clinical samples[2]. The short-term prevalence of NSSI among Chinese adolescents is around 22.4%-27.4%, and has been increasing on a yearly basis. Nonsuicidal self-injury NSSI is a major issue in adolescence, with first onset generally occurring at age 12-14 and peaks around ages 15-16[3,4].

NSSI is characterized as being non-suicidal in nature which makes it a strong predictor of suicidal ideation and behavior. Studies indicate a 3-4 fold increased odds of later suicide attempts among those with a history of NSSI compared to peers without[5]. The reasons of NSSI are closely related to some psychiatric disorders such as depression, anxiety disorders (variation of mood-state), borderline personality disorder, and post-traumatic stress disorder, which have a significant impact on adolescents mental health condition and social functioning[6]. NSSI By contrast, recurrent NSSI places a heavy burden on victims and their families and society as it also results in tissue injury, infection(s), permanent scarring, and even accidental death[6]. Hence, it is of great importance to explore the effective treatment methods of NSSI for preventing adolescent suicide and promoting mental health.

NSSI is a multidimensional behavior, which incorporates biological, psychological, and sociological aspects of its pathogenic mechanisms. In terms of biology, it had been seen that NSSI patients showed frontal lobe hypofunctioning and hyperfunctioning of the amygdala, change in serotonergic system activity and dysregulation of hypothalamic-pituitary-adrenal (HPA) axis[7,8]. Neuroimaging studies show that NSSI patients have aberrant prefrontal-limbic functional connectivity during the performance of emotional processing tasks, characterized by impaired regulation of the limbic system by broader regions of the prefrontal cortex[9]. Emotion regulation difficulty is a central feature of NSSI diagnoses and patients tend to struggle with emotion recognition, have low emotional expression, and poorly controlled impulses[10]. Functional analysis suggests that NSSI is one of the strategies used for emotion regulation (negative emotions relief) and self-punishment functions[11]. Sociological factors associated with the NSSI included childhood trauma, adverse family environment, peer problems and school bullying[12].

The first-line treatments for NSSI mentioned currently mainly comprise psychotherapeutic approaches, including cognitive behavioral therapy (CBT), dialectical behavior therapy (DBT) and emotion regulation group therapy[13]. Conversely, psychotherapy is much slower in onset (often taking weeks to months to be effective) and many patients may fare poorly with this treatment modality[14]. With respect to pharmacotherapy, since there are currently no FDA-approved medications for the treatment of NSSI[15] It is hence of great value to investigate new treatment approaches for patients suffering from NSSI.

Cross-sectional studies have shown the effects of transcranial magnetic stimulation (TMS) and transcranial direct current stimulation (tDCS), non-invasive brain stimulation techniques that have become widespread to treat psychiatric disorders like depression or anxiety. Repetitive TMS (rTMS) applied over the dorsolateral prefrontal cortex (DLPFC) with high-frequency stimulation increases cortical excitability in this region, thereby reportedly targeting both executive control and emotion regulation[16]. Anodal tDCS is thought to increase cortical excitability; thus, the prefrontal cortex (PFC) cannot be considered a specific target but shows promise in promoting cognitive control and decreasing impulsivity[17]. Over the last years, a combination of different neuromodulation techniques has been the subject of interest research-wise due to potential synergy effects[18]. The rationale of the combination tDCS/rTMS is based on a hypothesis of complementary biophysical mechanisms. synchronized low direct in tDCS hyperpolarizes resting membrane potential to increase cortical excitability and prime neural circuits that may potentiate the rate of subsequent rTMS effects through decreased neuronal resistance (the disturbance relies on the relatively strong portion of direction from it). rTMS, by contrast, generates action potentials via time-varying magnetic fields, therefore manifesting more immediate and localized effects on synaptic plasticity. To maximize the effect, tDCS may influence rTMS effects when delivered sequentially where tDCS is applied 30 minutes before rTMS so that the network primed with increased cortical excitability from tDCS could produce higher efficacy of subsequent stimulation with rTMS. The mechanistic framework hierarchizes our protocol design to the prefrontall-imbic dysregulation model of NSSI as taxing dorsolateralprefrontal cortex to improve top-down regulatory control of limbic region functions.

However, there are currently few studies on tDCS combined with rTMS for treating adolescent NSSI, and systematic analysis of factors influencing treatment outcomes is lacking. Based on this, the present study adopted a retrospective design to analyze the clinical efficacy of tDCS combined with rTMS in treating adolescent NSSI and explore factors associated with treatment outcomes through multivariate regression analysis, providing evidence for clinical treatment selection and efficacy prediction.

MATERIALS AND METHODS

Study design and data source

A retrospective study design was adopted. Clinical data from adolescent NSSI patients treated at our hospital’s psychology outpatient or inpatient department between December 2024 to December 2025 were collected. Data were obtained from the hospital’s electronic medical record system, including patient demographics, diagnostic information, treatment regimens, scale assessment results, and laboratory test results. This study was approved by the hospital ethics committee, ethics approval No. 2025-143-01(z), with informed consent waived due to the retrospective study design.

Inclusion and exclusion criteria

Inclusion criteria: (1) Meeting DSM-5 NSSI research diagnostic criteria; (2) Age 12-18 years; (3) Received at least 6 weeks of standardized treatment; and (4) Complete scale assessment data before and after treatment.

Exclusion criteria: (1) Comorbid schizophrenia, bipolar disorder manic episode, or organic mental disorders; (2) History of epilepsy; (3) Treatment interruption exceeding 1 week; and (4) Incomplete medical records.

Grouping method

Patients were grouped according to the treatment regimen actually received: (1) Combination group: TDCS + rTMS + psychotherapy; (2) rTMS group: RTMS + psychotherapy; and (3) Control group: Psychotherapy alone. The psychotherapy regimen was identical across all three groups, including individual CBT (twice weekly) and DBT skills training (once weekly). rTMS treatment protocol: High-frequency (10 Hz) stimulation of the left DLPFC, 80% motor threshold, 2000 pulses per session, once daily, 5 times per week. tDCS treatment protocol: Anode placed over the left DLPFC, cathode over the right supraorbital region, 1.5 mA, 20 minutes per session, once daily, 5 times per week, administered 30 minutes before rTMS. Treatment duration was 6 weeks for all groups. The selection of stimulation parameters was based on established neurophysiological evidence. For rTMS, we selected 10 Hz high-frequency stimulation based on prior depression literature demonstrating that this frequency optimally enhances DLPFC excitability and produces reliable antidepressant effects. For tDCS, the 1.5 mA intensity and 20-minute duration were chosen to balance efficacy and tolerability in adolescents, consistent with safety guidelines demonstrating that 1-2 mA produces robust cortical modulation while maintaining safety in pediatric populations. The 30-minute interval between tDCS and rTMS was selected based on neurophysiological evidence demonstrating that tDCS-induced cortical excitability changes persist for 30-90 minutes post-stimulation and peak around 20-40 minutes, establishing the temporal window for potential synergistic effects.

Outcome measures

(1) General information: Gender, age, illness duration, NSSI onset age, primary self-injury method, self-injury frequency, comorbidities, family history, childhood trauma history, previous treatment history, etc.; (2) Scale assessments: Self-Harm Inventory (SHI), Beck Depression Inventory-II (BDI-II), Beck Anxiety Inventory (BAI), Barratt Impulsiveness Scale-11 (BIS-11), Difficulties in Emotion Regulation Scale (DERS), Childhood Trauma Questionnaire; and (3) Pre-post treatment changes: Improvement scores (pre-treatment score minus post-treatment score) and percentage reduction for each scale.

Efficacy assessment criteria

Based on SHI score reduction rate. Recovery: Reduction rate ≥ 75%; marked improvement: 50% ≤ reduction rate < 75%; improvement: 25% ≤ reduction rate < 50%; no response: Reduction rate < 25%. Overall response rate = (recovery + marked improvement + improvement) cases/total cases × 100%.

Statistical analysis

Data analysis was performed using SPSS 26.0 and R 4.2.0 software. Continuous variables with normal distribution were expressed as mean ± SD, with one-way ANOVA for three-group comparisons and LSD-t test for pairwise comparisons; non-normally distributed data were expressed as M (P25, P75) using Kruskal-Wallis H test. Categorical variables were expressed as n (%) using χ² test. Univariate and multivariate logistic regression analyses were performed with treatment response as the dependent variable; multiple linear regression analysis was performed with SHI improvement score as the dependent variable. Receiver operating characteristic (ROC) curves were used to evaluate prediction model performance. P < 0.05 was considered statistically significant.

RESULTS

General characteristics of study subjects

A total of 256 adolescent NSSI patients were included: 86 in the combination group, 92 in the rTMS group, and 78 in the control group. There were no statistically significant differences among the three groups in general demographic characteristics including gender, age, education level, and family structure (P > 0.05). There were also no significant differences in illness duration, NSSI onset age, primary self-injury method, comorbidities, childhood trauma history, and baseline scale scores (P > 0.05), indicating comparability (Table 1).

Table 1 Comparison of baseline characteristics among three groups, n (%)/mean ± SD.

Variable	Combination group (n = 86)	rTMS group (n = 92)	Control group (n = 78)	F/χ2	P value
Gender (male/female)	22/64	25/67	18/60	0.486	0.784
Age (years)	15.4 ± 1.7	15.2 ± 1.8	15.6 ± 1.6	1.125	0.326
Illness duration (months)	19.2 ± 9.6	18.5 ± 10.2	20.1 ± 9.8	0.625	0.536
NSSI onset age (years)	13.1 ± 1.4	13.3 ± 1.5	13.0 ± 1.6	0.862	0.424
Self-injury frequency (times/month)	8.5 ± 4.2	8.2 ± 4.5	8.8 ± 4.0	0.456	0.635
Primary self-injury method				0.562	0.755
Cutting	62 (72.1)	68 (73.9)	55 (70.5)
Other	24 (27.9)	24 (26.1)	23 (29.5)
Comorbid depression	52 (60.5)	54 (58.7)	45 (57.7)	0.142	0.932
Comorbid anxiety disorder	35 (40.7)	38 (41.3)	30 (38.5)	0.152	0.927
Childhood trauma history	48 (55.8)	50 (54.3)	42 (53.8)	0.068	0.967
Previous psychotherapy history	28 (32.6)	30 (32.6)	24 (30.8)	0.076	0.963
SHI score (points)	29.5 ± 7.2	28.8 ± 6.8	29.2 ± 7.0	0.225	0.799
BDI-II score (points)	27.2 ± 8.5	26.5 ± 8.2	27.0 ± 8.8	0.168	0.845
BAI score (points)	25.6 ± 7.8	24.8 ± 7.5	25.2 ± 8.0	0.245	0.783
BIS-11 score (points)	73.5 ± 11.2	72.8 ± 10.8	74.2 ± 11.5	0.358	0.699
DERS score (points)	110.5 ± 19.2	108.6 ± 18.5	111.2 ± 20.1	0.425	0.654
CTQ score (points)	48.6 ± 12.5	47.8 ± 11.8	49.2 ± 12.2	0.298	0.743

CTQ: Childhood Trauma Questionnaire; rTMS: Repetitive transcranial magnetic stimulation; SHI: Self-Harm Inventory; BIS-11: Barratt Impulsiveness Scale-11; BDI-II: Beck Depression Inventory-II; DERS: Difficulties in Emotion Regulation Scale; BAI: Beck Anxiety Inventory; NSSI: Non-suicidal self-injury.

Comparison of scale scores before and after treatment among three groups

After 6 weeks of treatment, SHI, BDI-II, BAI, BIS-11, and DERS scores significantly decreased in all three groups compared to baseline (P < 0.01). Inter-group comparison showed that the combination group had significantly greater improvement than the rTMS and control groups (P < 0.05 or P < 0.01), and the rTMS group showed greater improvement than the control group (P < 0.05) (Table 2).

Table 2 Comparison of scale scores before and after treatment among three groups (mean ± SD, points).

Measure	Time	Combination group (n = 86)	rTMS group (n = 92)	Control group (n = 78)	F value	P value
SHI	Pre-treatment	29.5 ± 7.2	28.8 ± 6.8	29.2 ± 7.0	0.225	0.799
	Post-treatment	10.8 ± 4.8a	15.2 ± 5.5a	19.6 ± 6.2a	52.36	< 0.001
	Improvement	18.7 ± 5.2	13.6 ± 4.8	9.6 ± 4.5	68.52	< 0.001
BDI-II	Pre-treatment	27.2 ± 8.5	26.5 ± 8.2	27.0 ± 8.8	0.168	0.845
	Post-treatment	12.5 ± 5.6a	16.8 ± 6.2a	20.5 ± 7.0a	35.62	< 0.001
	Improvement	14.7 ± 4.8	9.7 ± 4.2	6.5 ± 3.8	72.86	< 0.001
BAI	Pre-treatment	25.6 ± 7.8	24.8 ± 7.5	25.2 ± 8.0	0.245	0.783
	Post-treatment	11.2 ± 4.8a	15.5 ± 5.6a	18.8 ± 6.2a	38.56	< 0.001
	Improvement	14.4 ± 4.5	9.3 ± 4.0	6.4 ± 3.6	78.25	< 0.001
BIS-11	Pre-treatment	73.5 ± 11.2	72.8 ± 10.8	74.2 ± 11.5	0.358	0.699
	Post-treatment	54.2 ± 9.5a	60.5 ± 10.2a	66.8 ± 11.0a	32.15	< 0.001
	Improvement	19.3 ± 5.8	12.3 ± 5.2	7.4 ± 4.5	95.62	< 0.001
DERS	Pre-treatment	110.5 ± 19.2	108.6 ± 18.5	111.2 ± 20.1	0.425	0.654
	Post-treatment	70.5 ± 14.2a	82.6 ± 15.8a	95.5 ± 17.2a	48.26	< 0.001
	Improvement	40.0 ± 10.5	26.0 ± 9.2	15.7 ± 8.5	125.8	< 0.001

^aP < 0.01 vs pre-treatment within group.

rTMS: Repetitive transcranial magnetic stimulation; SHI: Self-Harm Inventory; BIS-11: Barratt Impulsiveness Scale-11; BDI-II: Beck Depression Inventory-II; DERS: Difficulties in Emotion Regulation Scale; BAI: Beck Anxiety Inventory.

Clinical efficacy comparison among three groups

After 6 weeks of treatment, the overall response rate was 88.4% (76/86) in the combination group, 72.8% (67/92) in the rTMS group, and 55.1% (43/78) in the control group, with statistically significant differences among groups (χ² = 22.56, P < 0.001). Pairwise comparison showed that the combination group had significantly higher response rate than both the rTMS and control groups (P < 0.05), and the rTMS group was significantly higher than the control group (P < 0.05) (Table 3).

Table 3 Clinical efficacy comparison among three groups, n (%).

Group	Cases	Recovery	Marked improvement	Improvement	No response	Overall response rate (%)
Combination group	86	28 (32.6)	32 (37.2)	16 (18.6)	10 (11.6)	88.4a,b
rTMS group	92	18 (19.6)	26 (28.3)	23 (25.0)	25 (27.2)	72.8b
Control group	78	8 (10.3)	15 (19.2)	20 (25.6)	35 (44.9)	55.1

^aP < 0.05 vs repetitive transcranial magnetic stimulation group.

^bP < 0.05 vs control group.

rTMS: Repetitive transcranial magnetic stimulation.

Univariate logistic regression analysis of treatment response

With treatment response (response = 1, no response = 0) as the dependent variable, univariate logistic regression analysis was performed on potential influencing factors. Results showed that treatment regimen, illness duration, comorbid depression, baseline DERS score, baseline BIS-11 score, and childhood trauma history were associated with treatment response (P < 0.1) (Table 4).

Table 4 Univariate logistic regression analysis of treatment response.

Variable	B	SE	OR (95%CI)	Wald	P value
Gender (male = 1)	-0.125	0.285	0.88 (0.50-1.54)	0.192	0.661
Age	0.086	0.092	1.09 (0.91-1.30)	0.875	0.350
Illness duration	-0.042	0.018	0.96 (0.93-0.99)	5.445	0.020
Treatment regimen (control as reference)				22.56	< 0.001
rTMS group	0.785	0.312	2.19 (1.19-4.04)	6.325	0.012
Combination group	1.652	0.365	5.22 (2.55-10.68)	20.48	< 0.001
Comorbid depression (yes = 1)	0.625	0.268	1.87 (1.10-3.16)	5.445	0.020
Comorbid anxiety disorder (yes = 1)	0.312	0.265	1.37 (0.81-2.30)	1.385	0.239
Childhood trauma history (yes = 1)	-0.485	0.262	0.62 (0.37-1.03)	3.426	0.064
Baseline SHI score	0.025	0.022	1.03 (0.98-1.07)	1.292	0.256
Baseline BDI-II score	0.018	0.018	1.02 (0.98-1.05)	1.000	0.317
Baseline BIS-11 score	-0.028	0.014	0.97 (0.95-1.00)	4.000	0.046
Baseline DERS score	-0.022	0.008	0.98(0.96-0.99)	7.563	0.006

rTMS: Repetitive transcranial magnetic stimulation; SHI: Self-Harm Inventory; BIS-11: Barratt Impulsiveness Scale-11; BDI-II: Beck Depression Inventory-II; DERS: Difficulties in Emotion Regulation Scale; OR: Odds ratio; CI: Confidence interval.

Multivariate logistic regression analysis of treatment response

Variables with P < 0.1 in univariate analysis were included in the multivariate logistic regression model. Results showed that treatment regimen [combination vs control, odds ratio (OR) = 4.86, 95% confidence interval (CI): 2.28-10.36], baseline DERS score (OR = 0.98, 95%CI: 0.96-0.99), illness duration (OR = 0.96, 95%CI: 0.93-0.99), and comorbid depression (OR = 1.92, 95%CI: 1.08-3.42) were independent factors associated with treatment response (P < 0.05) (Table 5).

Table 5 Multivariate logistic regression analysis of treatment response.

Variable	B	SE	OR (95%CI)	Wald	P value
Treatment regimen (control as reference)				19.86	< 0.001
rTMS group	0.695	0.325	2.00 (1.06-3.79)	4.572	0.033
Combination group	1.581	0.388	4.86 (2.28-10.36)	16.62	< 0.001
Illness duration	-0.038	0.019	0.96 (0.93-0.99)	4.000	0.046
Comorbid depression	0.652	0.295	1.92 (1.08-3.42)	4.882	0.027
Baseline DERS score	-0.020	0.009	0.98 (0.96-0.99)	4.938	0.026

Model χ² = 48.56, P < 0.001. Hosmer-Lemeshow test χ² = 6.825, P = 0.556. Nagelkerke R² = 0.256. rTMS: Repetitive transcranial magnetic stimulation; DERS: Difficulties in Emotion Regulation Scale; OR: Odds ratio; CI: Confidence interval.

Multiple linear regression analysis of SHI score improvement

With SHI score improvement as the dependent variable and potential influencing factors as independent variables, multiple linear regression analysis (stepwise method) was performed. Results showed that treatment regimen, baseline SHI score, baseline BIS-11 score, and childhood trauma history were independent predictors of SHI score improvement (P < 0.05). The adjusted R² = 0.485, indicating that these factors explained 48.5% of the variance in SHI score improvement (Table 6).

Table 6 Multiple linear regression analysis of Self-Harm Inventory score improvement.

Variable	B	SE	β	t	P value
Constant	2.856	2.125	-	1.344	0.180
Treatment regimen (control as reference)
rTMS group	3.856	0.825	0.285	4.674	< 0.001
Combination group	8.562	0.862	0.612	9.932	< 0.001
Baseline SHI score	0.325	0.058	0.286	5.603	< 0.001
Baseline BIS-11 score	0.125	0.038	0.185	3.289	0.001
Childhood trauma history (yes = 1)	-2.156	0.752	-0.156	-2.867	0.005

F = 48.56, P < 0.001. Adjusted R² = 0.485. rTMS: Repetitive transcranial magnetic stimulation; SHI: Self-Harm Inventory; BIS-11: Barratt Impulsiveness Scale-11.

ROC curve analysis of prediction model

A prediction model for treatment response was constructed based on the multivariate logistic regression model, and ROC curve was plotted. Results showed that the area under the curve (AUC) was 0.782 (95%CI: 0.724-0.840), with sensitivity of 72.6%, specificity of 71.4%, and Youden index of 0.440, indicating good predictive performance of the model (Figure 1).

Figure 1 Receiver operating characteristic curve analysis of the prediction model for treatment response. The area under the curve (AUC) was 0.782 (95% confidence interval: 0.724-0.840). At the optimal cutoff point (marked by the solid circle), the sensitivity was 72.6%, specificity was 71.4%, and Youden index was 0.440. The dashed diagonal line represents the reference line (AUC = 0.5). AUC: Area under the curve; CI: Confidence interval.

Adverse events

Adverse events occurred in 12 cases (14.0%) in the combination group, including mild headache in 5 cases, scalp tingling in 4 cases, and transient dizziness in 3 cases; adverse events occurred in 8 cases (8.7%) in the rTMS group, including mild headache in 4 cases, scalp discomfort in 3 cases, and transient nausea in 1 case; no treatment-related adverse events occurred in the control group. There was no statistically significant difference in adverse event rates between the combination group and rTMS group (χ² = 1.326, P = 0.250). However, adverse event rates were significantly higher in both the combination group (14.0% vs 0%, χ² = 11.52, P < 0.001) and rTMS group (8.7% vs 0%, χ² = 7.08, P = 0.008) compared to the control group. All adverse events were mild and resolved spontaneously without special treatment (Figure 2).

Figure 2 Adverse events in the three treatment groups. A: Overall adverse event rates among the combination group (14.0%, 12/86), repetitive transcranial magnetic stimulation (rTMS) group (8.7%, 8/92), and control group (0%, 0/78). There was no statistically significant difference between the combination group and rTMS group (χ² = 1.326, P = 0.250); B: Distribution of specific adverse event types. In the combination group: Mild headache (n = 5), scalp tingling (n = 4), and transient dizziness (n = 3). In the rTMS group: Mild headache (n = 4), scalp discomfort (n = 3), and transient nausea (n = 1). All adverse events were mild and self-limiting. rTMS: Repetitive transcranial magnetic stimulation.

DISCUSSION

This study employed a retrospective design to systematically analyze the clinical efficacy of tDCS combined with rTMS in treating adolescent NSSI and its influencing factors. Results showed that the combination treatment group achieved an overall response rate of 88.4%, which was associated with higher rates than the rTMS alone group (72.8%) and psychotherapy alone group (55.1%). Additionally, combination therapy demonstrated clear advantages in improving self-injury behaviors, depression and anxiety symptoms, impulsivity, and emotion regulation. Multivariate regression analysis further revealed the independent effects of treatment regimen, emotion regulation difficulty level, illness duration, and comorbid depression on treatment outcomes, providing important evidence for clinical treatment selection and efficacy prediction.

This suggests that tDCS or rTMS treatment of the adolescent NSSI performed combined is significantly more effective than single treatment methods. The association here may arise due to the synergistic effects of both techniques on the prefrontal cortex from a neurobiological mechanism standpoint. Nonetheless, the authors of these studies note that true causality is not established by these associations and that unmeasured confounding could be partly responsible for the differences in groups observed. The PFC, especially the DLPFC, is critical for executive function, cognitive control, emotion regulation and impulse inhibition[19]. Functional neuroimaging studies reveal DLPFC hypoactivation and aberrant prefrontal-limbic functional connectivity in NSSI patients, manifested as the decreased prefrontal regulation of emotional brain areas such as amygdala[20]. rTMS increases excitability of the DLPFC and plasticity of synapses by inducing cortical neuronal depolarization using a high frequency pulsed magnetic field, resulting in improvement of prefrontal executive control function[21]. tDCS produces prolonged low-level direct current, regulating neuronal resting membrane potential; with anodal stimulation decreasing neuronal activation threshold and generating fairly persistent enhancement of cortical excitability[22].

From the synergistic mechanisms point of view, tDCS prior to rTMS treatment might function as a “preconditioning” on DLPFC more easily activated state, thus enhancing therapeutic effects exerted by subsequent rTMS. Neuroimaging studies demonstrating that tDCS can adjust functional connectivity of brain regions and change neural network activity patterns support this hypothesis[23].

Additionally, the two techniques act on the nervous system through different biophysical mechanisms-rTMS primarily induces action potentials while tDCS primarily modulates resting membrane potential-and this mechanistic difference may be the basis for producing synergistic effects[24]. Future studies can directly observe the regulatory effects of combination treatment on prefrontal-limbic functional connectivity through task-state functional magnetic resonance imaging (fMRI) and other techniques to further elucidate the synergistic mechanisms.

Multivariate logistic regression analysis showed that baseline DERS score was an independent predictor of treatment response, with patients having lower DERS scores (less severe emotion regulation difficulties) responding better to treatment. This finding is consistent with functional analysis theory of NSSI. Studies show that approximately 90% of NSSI episodes are related to negative emotion regulation, with self-injury often serving as an “emergency” strategy for rapidly relieving emotional distress[25]. Patients with less severe emotion regulation difficulties may retain more adaptive emotion regulation strategies and can therefore more quickly establish and employ healthy emotion regulation methods to replace self-injury after neuromodulation treatment enhances prefrontal function. Conversely, patients with more severe emotion regulation difficulties may require longer treatment duration or more intensive treatment to achieve similar effects. This suggests that in clinical practice, patients with more severe emotion regulation difficulties may need extended treatment courses or combined intensive emotion regulation skills training.

Another important predictor was the duration of illness, which was inversely related to treatment response. This might have to do with the neuroplasticity time window. The adolescent brain remains fully in a developmental phase with increased plasticity, and early reintervention may more readily modify aberrant neural circuitry[26]. Second, patients with a longer illness duration may have developed more entrenched self-injury habits and neural representations that take longer to modify[27]. These results underscore the urgency of addressing NSSI as early as possible. Strengthening clinical screening of high-risk adolescents and initiating standardized treatment as soon as NSSI behavior is detected are necessary for achieving optimal long-term treatment outcomes. School mental health workers and primary care physicians can play important roles in early identification of NSSI.

Interestingly, comorbid depression was a positive predictor of treatment response. This seemingly contradictory finding may be explained as follows: First, rTMS and tDCS have demonstrated clear therapeutic effects on depression; patients with comorbid depression may simultaneously benefit from antidepressant effects during neuromodulation treatment, and improvement in depressive symptoms may indirectly promote NSSI reduction[28]. Second, NSSI in depressed patients may be more related to emotion regulation difficulties rather than reflecting more severe personality pathology, thus responding better to neuromodulation treatment targeting the prefrontal cortex[29]. Third, patients with comorbid depression may have higher motivation for help-seeking and better treatment adherence. To clarify, the OR of 1.92 for comorbid depression represents the odds of treatment response for patients with vs without comorbid depression, controlling for treatment group, illness duration, and baseline DERS scores. This effect holds across all treatment conditions and should not be interpreted as depression causing better response, but rather as comorbid depression identifying a subgroup with characteristics predicting favorable outcomes. This indicates that comorbid depression should not be considered a negative predictor for candidates undergoing neuromodulation treatment from these findings and may instead predict improvement when receiving this form of therapy.

Multiple linear regression analysis additionally showed the factors influencing the improvement of SHI score. Besides treatment regimen, baseline SHI score and baseline BIS-11 score were positive predictors (neither the SHI nor the BIS-11 predicted 8-week outcome; this means that patients who exhibited more severe self-injury behaviors and those with higher levels of impulsivity exhibited a greater magnitude of clinical improvement after treatment. This could be indicative of a statistical “ceiling effect” where less entrenched symptoms at baseline leave more opportunity for benefits to manifest. However, it may also mean that treatment with neuromodulation possesses specific therapeutic significance for patients with severe types of NSSI[30] that is, patients in whom traditional psychotherapy offers little benefit.

Mechanistically, patients with high impulsivity may have more pronounced weakness in prefrontal inhibitory function, and therefore respond more markedly to neuromodulation treatment that enhances prefrontal function.

Childhood trauma history was a negative predictor of SHI score improvement, with patients having childhood trauma history showing smaller improvement after treatment. This finding is consistent with previous research. Childhood trauma can lead to lasting neurobiological changes, including HPA axis dysfunction, amygdala hyperreactivity, and abnormal prefrontal development, which may reduce responsiveness to neuromodulation treatment[31]. Additionally, childhood trauma can lead to more complex psychopathology, such as dissociative symptoms and abnormal relationship patterns, which pure neuromodulation treatment may have difficulty fully addressing[32]. For patients with childhood trauma history, combining trauma-focused psychotherapy (such as trauma-focused CBT, eye movement desensitization and reprocessing, etc.) may be needed to achieve better outcomes. Meanwhile, treatment teams should appropriately adjust expectations for these patients to avoid prematurely terminating treatment due to outcomes not meeting expectations.

The prediction model constructed in this study had an AUC of 0.782, indicating good predictive performance. This model can support clinical decisions by helping identify patient populations most likely to benefit from combination therapy. However, the validation of this model needs to be confirmed in other cohorts outside of our own. In clinical practice, treatment regimen selection should extensively account for the individual circumstances and preference of the patients. In these patients predicted to respond poorly, neuromodulation treatment should not be written off and several approaches can allow for re-evaluation of efficacy through modification of the treatment parameters, extension of the treatment course, or combination with additional interventions.

Regarding safety, the adverse event rate in the combination group (14.0%) was not significantly different from the rTMS alone group (8.7%), and all adverse events were mild and self-limiting. This is consistent with previous systematic reviews on the safety of tDCS and rTMS[33]. Adolescent patients showed good tolerance to brain stimulation treatment, with no serious adverse events such as seizures occurring. These results provide safety evidence for clinical application of tDCS combined with rTMS in treating adolescent NSSI. While active treatments produced more adverse events than psychotherapy alone (as expected given brain stimulation’s physical effects), all adverse events were mild, self-limiting, and comparable to those reported in established tDCS/rTMS safety literature. The absence of serious adverse events and the lack of significant difference between combination and rTMS-only groups supports the safety and tolerability of the combined protocol. However, given that long-term effects of brain stimulation techniques in adolescents are not yet fully understood, strict adherence to indications, enhanced monitoring during treatment, and long-term follow-up observation should be maintained in clinical application.

This study has the following limitations: (1) The retrospective study design has inherent limitations including selection bias and information bias, and results interpretation requires caution; critically, causality cannot be inferred from these observational data, and the observed associations should be interpreted as hypothesis-generating rather than definitive evidence of treatment superiority; (2) Treatment allocation was non-randomized, and although baseline characteristics did not significantly differ among the three groups, unmeasured confounding factors may still exist; specific unmeasured confounders that may have influenced treatment allocation and outcomes include: Clinician bias in treatment assignment, as more severe or treatment-resistant cases may have preferentially received combination therapy; family dynamics and support systems; (3) No sham stimulation control group was set up, and placebo effects cannot be completely excluded; (4) Limited follow-up duration, unable to observe long-term efficacy and relapse rates; and (5) No neuroimaging examinations were conducted, unable to directly observe brain function changes.

Furthermore, incorporating biomarker exploration represents a critical future direction. EEG measures such as frontal alpha asymmetry and event-related potentials during emotion regulation tasks, along with fMRI connectivity analyses examining prefrontal-amygdala functional connectivity, could help elucidate mechanisms and develop predictive models. Such neuroimaging biomarkers could help identify patients most likely to respond before treatment initiation, enabling more personalized intervention selection. Longitudinal neuroimaging could also reveal whether observed clinical improvements correspond to normalization of prefrontal-limbic circuitry, directly testing our mechanistic hypotheses.

CONCLUSION

tDCS combined with rTMS treatment of adolescent NSSI is safer and more effective than single treatment methods in reducing NSSI frequency/severity, depressive/anxiety symptoms, and impulsivity levels; and improving emotion regulation ability. Multivariate regression analysis revealed combination treatment regimen, lower difficulty in emotion regulation level, shorter duration of illness and comorbid depression to be independent predictors of treatment response; while treatment regimen, baseline severity of self-injury, impulsivity and childhood trauma history were independent predictors for improvement in SHI score. Significance Although cohort studies find that concurrent treatment of adolescent NSSI and depression leads to reduced rates in the latter during follow-up, this study use evidence-based modelling to determine which patients would benefit most from combination therapy.