Abstract

Background:

Major depressive disorder (MDD) necessitates treatments that balance efficacy with tolerability. Electroconvulsive therapy (ECT) is highly effective but limited by cognitive side effects. Magnetic seizure therapy (MST), a more focal convulsive therapy, may offer a superior safety profile. This systematic review and meta-analysis directly compares the efficacy and cognitive safety of MST and ECT for MDD.

Methods:

We systematically searched PubMed, Embase, Cochrane CENTRAL, Web of Science, WanFang, and CNKI (inception to Nov 2025) for randomized and non-randomized controlled studies comparing MST and ECT in adults with MDD. Primary outcomes were antidepressant response and depression score changes. Secondary outcomes included cognitive function, reorientation time, and adverse events. Pooled effect estimates (RR, SMD, MD, OR) with 95% CIs were calculated.

Results:

13 studies (N = 607 participants) were included. MST and ECT demonstrated comparable clinical response (RR = 1.10, 95% CI: 0.94-1.27) and remission (RR = 1.04, 95% CI: 0.72-1.50) rates. Post-sensitivity analysis favored ECT for depression score change (SMD = 0.36, p=0.0001). MST was superior in preserving cognitive function (SMD = 1.19, p=0.005), enabling faster reorientation (MD=-16.72 min, p<0.00001), and reducing overall adverse event risk (OR = 0.23, p<0.00001), notably for memory loss, headache, and muscle pain. Seizure durations were shorter with MST.

Conclusions:

While MST and ECT had comparable response and remission rates, sensitivity analysis of depression score changes suggested potential superiority of ECT, warranting cautious interpretation. MST provided significantly better cognitive safety and tolerability, including fewer cognitive adverse events and faster reorientation. These findings support MST as a valuable alternative for MDD patients, especially when cognitive side effects are a primary concern.

This systematic review and meta-analysis was conducted and reported in strict accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines (23).

Systematic Review Registration:

https://www.crd.york.ac.uk/PROSPERO/view/CRD420261277041, identifier CRD420261277041.

1 Introduction

Major depressive disorder (MDD) is a leading cause of global disability and a critical public health challenge (1, 2). Over one in five individuals is projected to experience at least one depressive episode (3). A cross-national study reported that the lifetime prevalence of MDD is 7.5% in males and 13.6% in females, with projected cumulative risks by age 75 of 20.1% and 34.0%, respectively (4). In addition, the COVID-19 pandemic significantly increased the global burden of MDD, particularly among women and young adults (5). Beyond its profound personal suffering, MDD inflicts a severe societal and economic toll (1, 6). Alarmingly, a vast treatment gap persists; an analysis of 15 countries revealed that while 41.8% of individuals with 12-month MDD accessed mental health services, only 23.2% of these received minimally adequate treatment, culminating in an overall 90% gap in effective treatment coverage (7). This chasm is driven by a combination of service underutilization and deficiencies in treatment quality or adherence, highlighting the urgent, unmet need for accessible, tolerable, and highly effective therapeutic strategies (8). In this context, neuromodulation therapies have assumed an increasingly prominent role in the management of MDD (9).

Electroconvulsive therapy (ECT) is widely regarded as an effective acute intervention for MDD, particularly in patients with psychotic features, high suicidality, or profound functional impairment (10, 11). Robust evidence from randomized controlled trials and meta-analyses has consistently demonstrated its superior antidepressant efficacy compared with pharmacotherapy and sham treatments (10–12). However, the clinical applicability of ECT is substantially limited by concerns regarding cognitive adverse effects, including acute disorientation as well as anterograde and retrograde memory impairment (13, 14), which remain a major barrier to treatment acceptance despite technical refinements (15). Magnetic seizure therapy (MST) has emerged as a novel convulsive neuromodulation technique aiming to preserve the antidepressant efficacy of ECT while reducing cognitive burden through more focal stimulation and limited involvement of medial temporal structures (9, 16). The first deliberate seizure induction using MST was demonstrated in non−human primates (17) and subsequently in patients with major depression (18). Early studies suggested that MST may offer comparable antidepressant outcomes with a more favorable cognitive safety profile (16, 19, 20). Subsequently, larger randomized controlled trials, including a double-blind trial by Deng et al. (21) and the recent non−inferiority CREST-MST trial (22) randomizing 239 patients, have provided higher−quality evidence generally supporting these findings. Nevertheless, existing evidence still shows heterogeneity in stimulation parameters, comparator ECT protocols, and outcome measures, and a comprehensive systematic synthesis of all available controlled studies remains warranted to quantify the overall effect sizes and to explore sources of variability.

The present systematic review and meta-analysis aims to compare the antidepressant efficacy and cognitive safety of magnetic seizure therapy versus electroconvulsive therapy in patients with major depressive disorder. By integrating evidence from randomized controlled trials and controlled clinical studies, this study seeks to provide a clearer evidence base to inform clinical decision-making, guide future research priorities, and ultimately contribute to optimizing treatment pathways for patients with MDD.

2 Methods

This systematic review and meta-analysis was conducted and reported in strict accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines (23). The study protocol was prospectively registered on the International Prospective Register of Systematic Reviews (PROSPERO) under the registration number CRD420261277041.

2.1 Search strategy and information sources

A comprehensive and systematic literature search was performed to identify all relevant studies comparing Magnetic Seizure Therapy (MST) and Electroconvulsive Therapy (ECT) for Major Depressive Disorder (MDD). With the assistance of a senior research librarian, we developed search strategies tailored to each database’s syntax. In order to avoid missing any relevant literature as much as possible, the search strategy combined controlled vocabulary terms (e.g., MeSH in PubMed) and free-text keywords related to two core concepts: Magnetic Seizure Therapy (e.g., “Magnetic Seizure Therapy”, “MST”) and Major Depressive Disorder (e.g., “Depressive Disorder, Major”, “Depression”, “Treatment Resistant Depression”). We searched the following six electronic databases from their inception until November 4, 2025: PubMed/MEDLINE, Embase, Cochrane Central Register of Controlled Trials (CENTRAL), Web of Science Core Collection, WanFang and CNKI. No language or publication status restrictions were applied. The complete search history for each database was provided in Supplementary File 1. To ensure literature saturation, we also manually screened the reference lists of all included studies and relevant systematic reviews.

2.2 Eligibility criteria

Studies were selected based on the following predefined criteria: Population: Adult patients diagnosed with major depressive disorder according to standardized diagnostic criteria (e.g., DSM or ICD); Intervention: Magnetic seizure therapy; Comparator: Electroconvulsive therapy; Outcomes: At least one predefined outcome related to antidepressant efficacy [e.g., Hamilton Depression Rating Scale (HAMD), Montgomery-Asberg Depression Rating Scale (MADRS)] or cognitive safety [e.g., Montreal Cognitive Assessment (MoCA), Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) scores]; Study Design: Randomized controlled trials (RCTs), non-randomized controlled trials, and prospective or retrospective comparative cohort studies. Case reports, case series, reviews, conference abstracts without full data, and non-comparative studies were excluded.

2.3 Study selection and data extraction

All records identified through database searching were imported into Endnote (24) for deduplication and management. The selection process was conducted independently by two authors. First, titles and abstracts were screened against the eligibility criteria. Second, the full texts of potentially relevant articles were retrieved and assessed in detail. Any discrepancies at either stage were resolved through discussion or by consulting a third senior author. The reasons for excluding studies at the full-text stage were recorded. Extracted data included study characteristics (author, year, country, study design), participant demographics, treatment parameters, outcome measures, and follow-up duration. Detailed characteristics of the included studies are summarized in the study characteristics table.

2.4 Risk of bias assessment

The methodological quality of included randomized controlled trials was independently assessed by two authors using the Cochrane Risk of Bias 2 (ROB 2) tool (25), which evaluates bias across five domains: randomization process, deviations from intended interventions, missing outcome data, outcome measurement, and selection of the reported result. Overall risk of bias judgments were categorized as low risk, some concerns, or high risk. Any disagreements in assessment were resolved by consensus.

2.5 Statistical analysis

Meta-analyses were performed using Review Manager (RevMan) 5.4 (26) For continuous outcomes, when the measurement units of the outcome were entirely consistent and the same assessment tool was used across studies in a meta-analysis, the weighted mean difference (WMD) was preferentially selected. Conversely, when the measurement units differ or different assessment tools were used, the standardized mean difference (SMD) was the preferred choice. For the analysis of dichotomous variables, the risk ratio (RR) is used for comparing efficacy rates (e.g., response rates), whereas the odds ratio (OR) is typically chosen for the analysis of adverse events. All results obtained were reported with 95% confidence intervals (CI). For the statistical model, a fixed-effect model is employed when heterogeneity is low, while a random-effects model is adopted when heterogeneity is substantial. Heterogeneity among studies was determined by Q test and I2 statistics [Cochrane book 9.5.2 Identifying and measuring heterogeneity, 0%–40%: might not be important; 30%–60%: may represent moderate heterogeneity*; 50%–90%: may represent substantial heterogeneity*; 75%–100%: considerable heterogeneity*] (27). Sensitivity analyses were conducted by sequentially excluding individual studies to examine the robustness of the primary outcome. Publication bias was evaluated visually using funnel plots for the primary outcome.

3 Results3.1 Study selection

The PRISMA flow diagram (Figure 1) summarizes the study selection process. A systematic search of six electronic databases (WanFang, CNKI, Cochrane CENTRAL, Embase, PubMed, and Web of Science) yielded 2,991 records. An additional 10 records were identified through citation searching of included studies and relevant reviews. After removing duplicate records and ineligible publication types during the initial screening, a total of 1,605 records were screened by title and abstract. Of these, 1,594 records were excluded because they did not meet the eligibility criteria. The remaining 21 reports were sought for full-text retrieval, of which 1 could not be obtained. The full texts of the remaining 20 reports were assessed for eligibility. Among these, 5 were excluded due to mismatched outcome indicators, and 2 were excluded because it represented duplicate reporting of data from an already included study, as detailed in Supplementary file 2. Consequently, a total of 13 studies met all predefined eligibility criteria and were included in the meta-analysis.

Literature selection and inclusion process.

3.2 Characteristics of included studies

The meta-analysis included 13 studies (8 randomized controlled trials [RCTs] and 5 non-randomized studies) published between 2006 and 2025, involving a total of 607 participants. Among them, 297 patients were treated with MST and 310 with ECT. All studies enrolled adult patients diagnosed with a major depressive episode, most with treatment-resistant depression (TRD). Concomitant antidepressant use was common. The stimulation parameters for MST were highly consistent across studies, predominantly utilizing 100 Hz frequency and maximal output intensity targeted at the vertex. In contrast, ECT protocols were heterogeneous, employing varied electrode placements (right unilateral, bifrontal, bitemporal) and dosing methods. All studies assessed depression severity using the Hamilton Depression Rating Scale (HAMD), and most evaluated cognitive function. Detailed study characteristics are available in Table 1.

Study IDCountryStudy DesignPopulationDiagnostic CriteriaAntidepressantsMST GroupECT GroupOutcomes & ToolsAssessment TimepointsNAge (years)Duration of illnessCoil TypeTargetFrequency (Hz)IntensityPulses/DurationTreatment CourseNAge (years)Duration of illnessCharge (mC)Pulse WidthElectrode PlacementTreatment CourseWang 2025b (28)ChinaEvaluator-blinded RCTAdolescents (13–18 yrs) with TR-MDD or high suicide riskDSM-VYes (continued SSRI)60 (randomized), 45(completed)15.20 ± 1.7514.52 ± 13.06 (month)/Cz (10–20 system)100 Hz100% intensity8–10 s/session3 times/week for 12–16 sessions60 (randomized), 50 (completed)15.15 ± 1.6412.42 ± 13.28 (month)Age-based percentage method/Bilateral3 times/week for 12–16 sessionsBDI-II; MoCA; C-SSRS; CTCAE v5.0; Reorientation timeBaseline, Day 7 post-final treatment.Yang 2025 (29)ChinaRCTInpatients with MDE (MDD or bipolar disorder)DSM-VYes (SSRIs)2029 (21, 46)78 (36, 192) (month)//100 Hz100% output1st: 4s, +2s each session, max 10s3 times/week for 4 weeks (12 sessions)2031 (21, 55)48 (7, 102) (month)Determined by age (0.8 × age × 100%)1.0 ms/3 times/week for 4 weeks (12 sessions)HAMD-17; RBANS (total & factor scores)Pre- and post-treatment.Deng 2024 (21)USADouble-blind RCTAged 18-90, referred for ECT, MDE in MDD/bipolar, baseline HDRS-24≥18.DSM-IV-TR (SCID interview)No (tapered/washed out pre-treatment)35 (randomized)47.7 ± 15.6135.2 ± 208.3 (weeks)Double-cone circular coilVertex100 Hz100% max output10 s (titrated from 5s)3 times/week until remission criteria, plateau, or <25% improvement after 8 sessions. Avg: 9.0 sessions.38 (randomized)48.2 ± 12.8114.5 ± 129.4 (weeks)6 × seizure thresholdUltrabrief pulseRUL3 times/week until remission criteria, plateau, or <25% improvement after 8 sessions. Avg: 6.7 sessions.HDRS-24; IDS; CGI-I; GAF; MMSE; Columbia ECT Side Effects; Reorientation time; AMTBaseline, each treatment morning, 24–72 hrs post-final, follow-up (first 2 months: 2/month, then 1/month, up to 6 months).EI-Deeb 2020 (30)EgyptOpen-label RCTMDD patients, aged 18-65, with indication for convulsive therapy (e.g., suicidality, psychosis).DSM-IV-TRNo (drug-free ≥6 wks or naïve)3039.07 ± 12.857.73 ± 5.66 (month)Circular coilVertex100 Hz100% max output10 s5 sessions, 2 times/week30 (15 RUL, 15 BT)BT:38.80 ± 14.0
RUL:39.60 ± 12.32BT:6.00 ± 5.41
RUL:5.73 ± 4.03
(month)RUL: age%; BT: (age/2)%Brief (0.5 ms)RUL (15) & BT (15)5 sessions, 2 times/weekHAMD-21; BDI; TRO; Wechsler Memory Scale; Wisconsin Card Sorting Test; Columbia ECT Side Effects ScheduleBaseline, after sessions 1, 3, 5.Fitzgerald 2018 (20)AustraliaDouble-blind RCTPatients with TRD (Stage II), moderate-severe depressionDSM-IVYes, continued use allowed18 (final analysis)44.6 ± 14.822.7 ± 14.3 (years)Bilateral double-cone coilVertex100 Hz100% machine outputTitration: +2s, Treatment: threshold+4s, max 10sUp to 15 sessions, 3 times/week1947.2 ± 16.127.6 ± 14.4 (years)3 × threshold/RULUp to 15 sessions, 3 times/weekHAMD; IDS-C; QIDS; Cognitive tests (AMI, Digit Symbol, Stroop, etc.)Baseline, after sessions 6, 9, 12, 15, end of treatment.Polster 2015 (31)GermanyOpen-label RCTPatients with TRDDSM-IV (MDE)Yes (stable 1 month before & during)1043.7 ± 114.1 ± 4 (years)Twin coil (13 cm wide)Vertex100 Hz~4 T (coil surface)Avg 5–8 s10–12 sessions, 2 times/week1054.7 ± 133.1 ± 3 (years)5 × seizure thresholdBiphasic, square-wave, 0.5 msRUL10–12 sessions, 2 times/weekAcute memory tests (Delayed & Cued recall)Treatment vs. control days: morning (learning) & noon (recall).Kayser 2017 (32)GermanyOpen-label RCTPatients with TRD (MDD, BPI, BPII)DSM-IV-TRYes (stable ≥4 wks)10 (completed)45 ± 14/Twin coilCz (10–20)100 Hz100% max outputMax 800 pulses, 8 s8–12 sessions, 2 times/week10 (completed)55 ± 12/RUL: 6×threshold; BL: 3×threshold/RUL (9), Bifrontotemporal (1)8–12 sessions, 2 times/weekHRSD-28; Seizure characteristics (postictal suppression)Within 7 days pre- and post-treatment.Kayser 2011 (19)GermanyOpen-label RCTPatients with TRDDSM-IVYes (as add-on)1050.0 ± 11.912.7 ± 3.6 (month)Twin coilVertex100 Hz100% outputMax 600 pulses, 6 s12 sessions, 2 times/week1049.0 ± 11.013.3 ± 4.1 (month)3 × seizure thresholdBiphasic square-waveRUL12 sessions, 2 times/weekMADRS; HDRS-28; HAMA; BDI; SCL-90; Neuropsychological tests; Recovery/Orientation timeBaseline, post-treatment (1 month ±3 days after final), multiple cognitive assessments during treatment.Wang 2025a (33)ChinaControlledInpatients with TRD, aged 18–65ICD-11 MDD + TRD criteriaYes4228.6 ± 13.815.9 ± 13.2 (month)Double-cone coilVertex (10–20 system)100 Hz100% output8–10 s3 times/week for 12 sessions (up to 16)4430.7 ± 14.718.2 ± 13.4 (month)Initial: (age × 0.8)%; Max: (2 × age)%/Bitemporal3 times/week for 12 sessions (up to 16)HAMD; MoCA; Reorientation time; Length of stay; Columbia ECT Side Effects QuestionnaireBaseline (within 1 wk pre-treatment); Post-treatment (48–72 hrs after final session).Zhang 2020 (34)ChinaEvaluator-blinded, non-randomizedDepressive episode patients, aged 18–60DSM-IV-TRYes (SSRIs allowed, stable dose)1829.00 ± 8.323.77 ± 3.77 (years)Circular coilVertex100 Hz100% max outputMax 10 sSessions on days 1, 2, 3, 5, 7, 9 (6 sessions total)2732.78 ± 8.844.47 ± 5.42 (years)Determined by half-age method/BifrontalSessions on days 1, 2, 3, 5, 7, 9 (6 sessions total)HAMD-17; HAMA; RBANS; Response/Remission rates; Seizure duration; Recovery times (breathing, consciousness, orientation)Baseline, after 3rd session (Day 3), after 6th session (Day 9).White 2006 (35)USAOpen-label case-controlAdult patients with severe depression//1048 ± 4/Custom-modified Magstim device (16 booster modules)/50 Hz2 T (coil surface)8 s10–12 sessions over 3–4 weeks1049 ± 6/2.5 × seizure threshold0.5 msBitemporal10–12 sessions over 3–4 weeksHam-D; BIS; Reorientation time; Cardiovascular response; Seizure duration (EEG & motor)Baseline, post-treatment.Atluri 2018 (36)CanadaProspective observational75 TRD patients; 55 healthy controls.DSM-IV MDDYes, most on antidepressants or benzodiazepines.2446.8 ± 15.819.0 ± 12.0 (years)Twin Coil (MagVenture)Bifrontal (F3, F4; DMPFC)100 Hz (12), 60 Hz (1), 50 Hz (2), 25 Hz (9)Max E-field intensity/Up to 24 sessions or until remission, 2–3 times/week.2242.0 ± 13.420.3 ± 13.7 (years)/RUL ultrabrief pulse (RUL-UB) or Bitemporal brief pulse (BL)RUL or Bitemporal2–3 times/week until remission/plateau.HRSD-17; MoCA; BDI; MADRS; SSIWithin 1 week pre-treatment; within 2 weeks post-treatment.Soehle 2014 (37)GermanyProspective observationalPatients with TRDDSM-IV-TRYes1045 ± 14/Twin coil/100 Hz100% amplitudeMax 8 s~10–12 sessions, 2 times/week1055 ± 12///Right unilateral (9), Bilateral (1)~10–12 sessions, 2 times/weekBIS monitoring; Recovery times (breathing, eye-opening); Seizure duration (EEG & motor)Baseline, pre-treatment, within 10 min post-treatment.

Characteristics of included studies.

AMI, Autobiographical Memory Interview; BDI, Beck Depression Inventory; BIS, Bispectral Index; BL, Bitemporal; BT, Bitemporal; CGI-I, Clinical Global Impression-Improvement; CFQ, Cognitive Failures Questionnaire; C-SSRS, Columbia-Suicide Severity Rating Scale; CTCAE, Common Terminology Criteria for Adverse Events; DMPFC, Dorsomedial Prefrontal Cortex; ECT, Electroconvulsive Therapy; GAF, Global Assessment of Functioning; Ham-D/HAMD, Hamilton Depression Rating Scale; HDRS/HRSD, Hamilton Depression Rating Scale (item number specified); IDS, Inventory of Depressive Symptomatology; MADRS, Montgomery-Åsberg Depression Rating Scale; MDE, Major Depressive Episode; MDD, Major Depressive Disorder; MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment; MST, Magnetic Seizure Therapy; QIDS, Quick Inventory of Depressive Symptomatology; RCT, Randomized Controlled Trial; RBANS, Repeatable Battery for the Assessment of Neuropsychological Status; RUL, Right Unilateral; SSI, Scale for Suicidal Ideation; AMT, Autobiographical Memory Test; TRD, Treatment-Resistant Depression; TRO, Test of Reorientation.

3.3 Risk of bias assessment

A risk of bias assessment was performed on the 13 included studies (Figure 2). 3 studies were judged as “low risk” (20, 21, 28), 4 as “some concerns” (29, 31, 32, 36), and 6 as “high risk” (19, 30, 33–35, 37). Elevated risk in the “randomization process” domain was noted, attributable to the inclusion of 5 non-randomized studies (33–37) and 3 with unclear randomization (19, 30, 31). In addition, blinding of treating personnel was not feasible due to procedural differences. Participant blinding is achievable under anesthesia, but several included studies did not implement it adequately, introducing potential performance and detection bias. In the “measurement of the outcome” domain, four studies (28.6%) were at high risk and two (14.3%) raised some concerns, primarily because key outcome measures involved subjective assessment. Detailed assessments for each domain and study are provided in Supplementary File 3.

Risk of bias graph. (A) each risk of bias item for each included study. (B) each risk of bias item presented as percentages across all included studies.

3.4 Primary outcome

Meta-analysis of the change in depression scale scores across 11 studies showed no statistically significant difference between MST and ECT prior to sensitivity analysis (SMD: 0.16, 95% CI: -0.23 to 0.55, p = 0.42), though with high heterogeneity (I² = 79%) (Figure 3A). Subsequent sensitivity analysis identified one study (30) that reported an effect size strongly favoring MST (SMD = -1.52) and was an outlier within the overall dataset. The removal of this study significantly altered the pooled estimate. The analysis of the remaining 10 studies yielded a statistically significant result, with heterogeneity further reduced (I² = 8%), indicating superior antidepressant efficacy for ECT (SMD: 0.36, 95% CI: 0.18 to 0.55, p = 0.0001) (Figure 3B).

Forest plot of primary indicators. (A) Pooled results of the change in depression scale scores prior to sensitivity analysis. (B) Pooled results of the change in depression scale scores after sensitivity analysis.

A total of 8 studies involving 396 participants (MST: n=190; ECT: n=206) reported data on treatment response rates. Meta-analysis showed no statistically significant difference between the MST and ECT groups. The pooled risk ratio (RR) was 1.10 (95% CI: 0.94 to 1.27, p = 0.23), indicating comparable rates of clinical response. Heterogeneity among studies was low (I² = 20%) (Figure 4A