Confirmatory Efficacy and Safety Trial of Magnetic Seizure Therapy versus Right Unilateral Ultra-Brief Electroconvulsive Therapy in Depression: CREST–MST Non-inferiority Randomized Trial

Abstract

Background:

Magnetic Seizure therapy (MST) is an innovative convulsive therapy that is clinically beneficial for patients with depression and has fewer cognitive adverse effects. This trial aimed to confirm the efficacy, tolerability and cognitive safety of MST compared to right unilateral ultra-brief (RUL-UB) pulse width electroconvulsive therapy (ECT).

Methods:

This multi-site randomized, double blind, parallel-group non-inferiority trial was conducted at three academic centers across Canada and the United States. Participants aged 18 and older with major depressive disorder (MDD) received treatment until they achieved remission, dropped out, or up to 21 total treatments. MST was delivered using a twin coil in a midline frontal position. Remission of depression on the Hamilton Rating Scale for Depression-24 item (HRSD-24) and worsening of autobiographical memory on the autobiographical memory test (AMT) were the co-primary outcomes. The non-inferiority margin was 15% absolute difference in remission rates and AMT worsening was defined as 25% reduction, corresponding to one standard deviation reduction in performance. The trial is registered with ClinicalTrials.gov (NCT03191058). While people with lived experience were not directly involved; this study was informed by indirect feedback from prior trials and clinical discussions with those who declined ECT.

Findings:

Between June 26,2018 and March 1,2024, 239 participants (RUL-UB ECT: mean age 45·5 ±14·8, range 18–78, 79·8% Caucasian, 54·6% (N=65) female, 45·4% (N=54) male; MST: mean age 44·5 ±13·8, range 18–75, 85·5% Caucasian, 59·8% (N=70) female, 40·2%(N=47) male) were randomized across study sites. The difference in remission rates was 5·3% in favor of RUL-UB ECT (27·8%) compared to MST (22·5%) establishing non-inferiority of MST, p=0·048; 95% CI [−4·4%,14·9%]. RUL-UB ECT participants had a higher rate of worsening autobiographical memory (17·3%) compared to MST participants (2·7%), p<0·001. 12 participants in the RUL-UB ECT group and 3 in the MST group withdrew from treatment due to non-serious adverse events.

Interpretation:

MST demonstrated non-inferior efficacy relative to RUL-UB ECT in achieving remission of depression, and a more favorable cognitive safety profile. Collectively, the overall risk-benefit profile of MST supports its consideration as a first line convulsive therapy in MDD, particularly in those who refuse RUL-UB ECT.

Introduction

A large proportion of individuals with major depressive disorder (MDD) experience treatment-resistant depression (TRD) and have little to no benefit from multiple medication trials¹. Electroconvulsive therapy (ECT) is an effective treatment in patients with TRD², yet only 1% accept treatment,^3,4 due largely to negative perceptions and cognitive adverse effects⁵. In particular, ECT results in significant post-ictal disorientation, and anterograde and retrograde amnesia⁶. The extent of anterograde and retrograde amnesia remains a significant challenge for individuals and their families⁷. In fact, patients often refuse ECT due to the risk of cognitive adverse effects⁵. Efforts to lessen the cognitive adverse effects of ECT while retaining its efficacy have yielded treatment advancements such as right unilateral (RUL) and ultra brief pulse-width (RUL-UB) ECT². These advancements demonstrate clinical efficacy in treating depression and fewer cognitive adverse effects, compared to both conventional bilateral forms of ECT⁸ as well as brief pulse RUL ECT⁹. Nonetheless, the cognitive adverse effects associated with RUL and RUL-UB ECT continue be a barrier to treatment acceptance¹⁰.

Magnetic seizure therapy (MST) is an emerging form of convulsive therapy that also produces antidepressant effects by inducing a therapeutic seizure¹¹. However, MST differs from ECT as it uses a magnetic field¹² to produce a therapeutic seizure. This is a key advance as magnetic stimulation is not impeded or shunted by non-conducting material (e.g., the skull) and can achieve more focal brain activation compared to ECT¹³. In fact, compared to RUL-UB ECT, MST is 5–10 times more focal and spares medial temporal structures linked to the cognitive adverse effects of ECT¹⁴. Several studies of MST have found favorable antidepressant effects¹⁵, reductions in suicidal ideation¹⁶, and minimal cognitive adverse effects¹⁷. Importantly, a blinded randomized study that compared MST to RUL-UB ECT demonstrated similar efficacy in MDD and a superior cognitive safety profile for MST¹⁸. The sample size of that study, however, was not large enough to test non-inferiority.

The primary aim of the present study was to confirm the efficacy, tolerability and cognitive safety of MST to RUL-UB ECT in individuals with MDD through a non-inferiority clinical trial. We hypothesized that MST would be non-inferior to RUL-UB ECT in achieving remission of depressive symptoms while also demonstrating a superior cognitive safety profile, specifically in the domain of autobiographical memory. A secondary aim was to compare MST to RUL-UB ECT in relation to physical side effects and remission of suicidal ideation.

Methods

Study Design and Participants

The study was a randomized, double blind, parallel-group non-inferiority clinical trial with two treatment arms conducted at three academic medical centers: the Centre for Addiction and Mental Health (CAMH) in Toronto, University of Texas Southwestern (UTSW) Medical Center in Dallas, and University of California, San Diego (UCSD). Approval was received from the Research Ethics Boards of all participating sites, written informed consent was obtained from all study participants and the trial was registered with ClinicalTrials.Gov (ID: NCT03191058). Additionally, the trial was issued an Investigational Device Exemption (IDE; Reference Number G170127) by the U.S. FDA and an Investigational Testing Authorization (ITA; Reference Number 270547) by Health Canada. Although individuals with lived experience were not directly involved in the development of the protocol or in reporting the study findings, the research design was informed by insights from prior neurostimulation trials, clinical discussions with patients who declined ECT treatment at our centers, and the investigators’ extensive experience caring for individuals with depression. A detailed description of the study methodology has been previously published¹⁹.

Hospitalized and non-hospitalized individuals ages 18 or older were included if: voluntary and competent to consent to treatment and research procedures; had a MINI International Neuropsychiatric Interview diagnosis (MINI–6.0) diagnosis of non-psychotic MDD; had a baseline 24-item Hamilton Rating Scaled for Depression (HRSD-24) total score ≥21; considered to be appropriate to receive convulsive therapy by an ECT attending psychiatrist and consultant anesthesiologist; agreeable to maintain antidepressant treatment constant during the study intervention; able to adhere to the intervention schedule; met the MST safety criteria, as assessed by the TMS adult safety screen²⁰; provided a negative pregnancy test and agreed not to pursue pregnancy during trial participation.

In contrast, individuals were excluded based on: diagnosis of substance dependence or abuse within the past three months; concomitant major unstable medical illness; diagnosis of an obsessive compulsive disorder, post-traumatic stress disorder deemed to be primary; probable dementia or significant neurological disorder; a medical cause of a major depressive episode or significant cognitive impairment; an intracranial implant or any other metal object within the head, excluding the mouth; required a benzodiazepine with a dose >2 mg/day of lorazepam equivalent or any anticonvulsant; inability to complete neuropsychological tests; non-correctable clinically significant sensory impairment. Discontinuation and withdrawal criteria are included in the Supplementary Appendix.

Randomization and Blinding

Participants were randomized using permuted blocks of sizes 2 and 4. These were not divulged to the study team at any point during the trial. The study statistician prepared the randomization scheme. While the treatment team administering MST or ECT could not be blinded, all participants and the independent raters administering outcome assessments remained blind to treatment assignments throughout the study. Participants were prepared for both treatment procedures and both MST and ECT machines were triggered simultaneously at every treatment session to optimally blind the treatment condition. To assess blinding, participants and raters completed a conventional guess form after the participant’s first treatment.

Treatment Arms

MST was administered using the MagPro XP with a twin coil. Each circular coil has an inner diameter of 13 centimeters and the distance between the two coils was adjusted between 1 and 4 cm depending on the skull shape of each individual participant to ensure the closest contact with the skull. Stimulation was delivered over the frontal cortex at the midline position directly over the electrode FZ according to the international 10–20 system²¹. The current direction induced in the brain was anterior to posterior. MST was delivered using 100% machine output at 100 Hz with progressively escalating train durations, commencing at 2 seconds and increasing by 2 seconds with each subsequent stimulation until an adequate seizure occurred. During subsequent sessions, stimulation was delivered using a train duration that was four seconds longer than the threshold stimulation (maximum train of 10 seconds). RUL-UB ECT treatments were administered using the MECTA spECTrum 5000Q or the SigmaStim Sigma. ECT seizure threshold, treatment (six times threshold) and parameter adjustment (see Supplementary Appendix) at subsequent sessions were based on a standard published protocol²².

Outcomes

A description of the screening and baseline assessments are detailed in the protocol publication¹⁹. Clinical outcome measures were completed at baseline, after every three treatments (or every four treatments to facilitate treatment scheduling) immediately prior to the next treatment session, within four days of the last treatment session, and then six months post-treatment. Remission, defined as a HRSD-24 total score²³ of ≤10 and 60% reduction from baseline on two consecutive assessments, was the co-primary non-inferiority outcome measure. Remission from depression on the HRSD-24 was selected as the primary outcome measure to facilitate comparison with previous neurostimulation trials (including studies of electroconvulsive therapy and magnetic seizure therapy). The HRSD-24 is a well validated measure and has shown good internal consistency, inter-rater reliability, and test–retest reliability²³. The co-primary cognitive outcome measure was ≥25% worsening on the Autobiographical Memory Test (AMT)²⁴. We used the AMT, which is a well-recognized instrument in autobiographical memory research that has demonstrated sensitivity to ECT¹⁰.

Secondary outcome measures included response on the HRSD-24 (≥50% reduction in total score from baseline), the Scale for Suicidal Ideation (SSI)²⁵, Brief Symptom Inventory Anxiety section (BSI-A)²⁶, Clinical Global Impression Severity/Improvement Scale (CGI-S, CGI-I)²⁷, Quality of Life Enjoyment and Satisfaction Questionnaire (Q-LES-Q)²⁸, and the Columbia ECT Subjective Side Effects Schedule²⁹. Additional exploratory neurocognitive outcomes, which have been shown to be impacted by seizure therapies³⁰, were assessed using a comprehensive cognitive battery. This battery evaluated global cognitive function, attention, processing speed, verbal fluency, working memory, executive function, anterograde and retrograde memory, and included the following measures: the Montreal Cognitive Assessment (MoCA, version 7.1)³¹, the Delis-Kaplan Executive Function System (D-KEFS; ColorWord Interference, Tower, and Verbal Fluency Tests)³², the California Verbal Learning Test – Third Edition (CVLT-3)³³, the NIH Toolbox (Flanker Inhibitory Control and Attention Test, List Sorting Working Memory Test, Picture Sequency Memory Test)³⁴ and time to reorientation, which was recorded following each of the first three treatment sessions²². These assessments were administered at baseline and at the end of treatment or drop out and six months post-treatment. An exploratory 6-month naturalistic follow-up occurred; during this period there were no restrictions on the type of treatments participants could receive.

Statistical Analysis

The sample size calculation was based on non-inferiority trial calculations that were sufficiently large enough to minimize Type II error³⁵ and are consistent with previous large, multi-center ECT trials⁶. The total sample size was derived as a function of tolerance and power with a significance level of 0.05. A total sample size of 260 participants (130 per group) yielded 80% power to confirm a non-inferior absolute difference in HRSD-24 defined remission rates of 15% between the two study groups. This absolute difference was chosen on clinical grounds recognizing that many patients refuse ECT due to the risk of cognitive adverse effects and that lower remission rate would be acceptable to patients if the there was a clearly superior cognitive safety profile. This sample size also provided >95% power to detect a difference ≥ 25% change on the AMT total score, which corresponds to a meaningful change of a one standard deviation reduction in performance³⁶. The decision to go with a wider non-inferiority margin was informed by the additional stringency of a co-primary superiority outcome related to cognitive safety.

The co-primary analysis was carried out in two stages. The first stage tested the non-inferiority of MST compared to RUL-UB ECT on remission using a one-sided Z-test in difference of proportions, compared against the non-inferiority margin of 15%. The non-inferiority margin was informed based on discussions with many patients over the years who refused ECT and often opted for treatments that had 25 to 35% lower remission rates than the 50% remission rate that is used when informing patients of the benefits of ECT. For example, a patient who has failed three antidepressants could potentially consider ECT, but the vast majority would opt for another medication trial where remission rates are less than 20%³⁷. Since intention to treat (ITT) analysis introduces a conservatism that is undesirable for non-inferiority trials, a modified completer analysis in those participants that had an adequate course (i.e., 8 treatments or achieved remission) was the main analysis. An ITT analysis was conducted as a sensitivity analysis using a logistic regression for binary data where all randomized patients were included including those that dropped out early. The second stage was a superiority comparison on cognitive safety. The binary outcome was defined as a worsening from baseline of ≥25% on the AMT total score. This co-primary cognitive analysis was ITT and the hypothesis was tested with a chi-square test. If outcomes were missing in more than 5% of the patients in the cognitive ITT superiority analysis, an inverse probability weighting and multiple imputation approach was planned to assess and mitigate the effect of missing data.

Secondary analyses of efficacy included an assessment of non-inferiority of HRSD-24 response using the same 15% absolute margin. Relative to other treatment options a 15% absolute margin for response rates is similarly conservative to the margin for remission rates. As little data is known regarding SSI remission rates with ECT, we used a 30% relative difference to the ECT SSI remission rate to determine the non-inferiority margin for this analysis. Tolerability outcomes included unstructured subjective side effect reports, the structured Columbia ECT Subjective Side Effects Schedule²⁹, and time to reorientation and were analyzed with estimated marginal means and 95% confidence intervals. All additional cognitive measures were considered exploratory.

Results

From June 26, 2018 to March 1, 2024, 292 participants with MDD were enrolled; of whom 47 were ineligible or declined to participate. 239 participants were randomly assigned to receive treatment and 3 withdrew participation before receiving treatment. The modified ITT sample comprised all randomized participants that received treatment and included 119 participants in the RUL-UB ECT group and 117 in the MST group. The completer sample was comprised of all participants who completed 8 treatment sessions or met remission criteria and included 108 participants in the ECT group and 111 in the MST group. Table 1 provides the baseline characteristics of study participants, however in brief, the RUL-UB ECT group was comprised of 54·6% (N=65) female and 45·4% (N=54) males while the MST group was comprised of 59·8% (N=70) female and 40·2%(N=47) male. Refer to Figure 1 for the CONSORT diagram.

Table 1·.

Baseline Demographic and Clinical Characteristics, and Treatment Information

Participant Characteristics	Electroconvulsive Therapy (N=119)	Magnetic Seizure Therapy (N=117)
Age, mean (SD)	45·5 (14·8)	44·5 (13·8)
Female, No. (%)	65 (54·6)	70 (59·8)
Years Education, mean (SD)	16·9 (3·8)	16·6 (3·7)
Left-Handed, mean (SD)	13 (10·9)	17 (14·5)
Race, No. (%)
American Indian/Alaska native	0	0
Asian	7 (5·9)	12 (10·3)
Black or African American	4 (3·4)	2 (1·7)
White	95 (79·8)	100 (85·5)
More than one race	5 (4·2)	1 (0·9)
Other	7 (5·9)	1 (0·9)
Unknown or not reported	1 (0·8)	1 (0·9)
Baseline HRSD-24, mean (SD)	28·8 (5·4)	29·4 (5·0)
Baseline BSI, mean (SD)	9·6 (5·6)	9·0 (5·3)
Baseline SSI, mean (SD)	8·2 (7·4)	9·0 (6·8)
Baseline QLESQ, mean (SD)	32·9 (7·1)	33·1 (6·0)
Baseline AMT Total memories, mean (SD)	9·2 (1·24)	9·0 (1·8)
Baseline MoCA, mean (SD)	26·0 (2·8)	26·1 (2·5)
Baseline CGI-S, No. (%)
Not at all to mildly ill (1 – 3)	0	0
Moderately ill (4)	14 (11·8)	10 (8·6)
Markedly ill (5)	77 (64·7)	62 (53·0)
Severely ill (6)	28 (23·5)	44 (37·6)
Among the most extremely ill (7)	0	1 (0·9)
Episode Duration (months), No. (%)
1–6 months	22 (18·5)	23 (19·7)
7–12 months	18 (15·1)	22 (18·8)
13–24 months	26 (21·9)	25 (21·4)
More than 24 months	52 (42·9)	47 (40·2)
Missing	2 (0·02)	0
History of prior ECT, No. (%)	15 (12·6)	16 (13·7)
Anxiety Comorbidity, No. (%)	60 (50·4)	58 (49·6)
Benzodiazepine During Treatment, No. (%)	58 (48·7)	71 (60·1)
Inpatient at treatment start, No. (%)	18 (15·1)	23 (19·7)
Previous Antidepressant Treatment History, No. (%)
No trials	0	2 (1·7)
Only Inadequate Trials	3 (2·5)	1 (0·9)
One Failed Treatment	19 (16·0)	18 (15·4)
Two or more failed treatments	97 (81·5)	96 (82·1)
rTMS	39 (32·8)	41 (35·0)
Treatment Information
Number of Treatments per Participant, mean (SD)	11·4 (5·1)	14·2 (5·3)
Seizure Duration – Motor, mean (SD)	37·4 (10·9)	36·0 (11·7)
Seizure Duration – EEG, mean (SD)	50·5 (12·1)	47·5 (17·6)
Stimulus Charge, mean (SD)	194·5 (89·5)	··
Stimulus Duration, mean (SD)	··	7·9 (1·2)
Titration Threshold of Stimulus Charge	24·7 (9·2)	··
Titration Threshold of Stimulus Duration	··	2·8 (1·1)
Seizure Threshold Duration – Motor, mean (SD)	54·1 (24·4)	40·5 (27·9)
Seizure Threshold Duration – EEG, mean (SD)	72·8 (37·2)	53·7 (38·4)
Parameter Changes
Any Charge Increases in ECT, No. (%)	79 (66·4)	··
Any Pulse Increases in MST, No. (%)	··	87 (74·4)
>1 Charge Increases in ECT, No. (%)	47 (39·5)	··
>1 Pulse Increases in MST, No. (%)	··	40 (34·2)
Anesthesia or muscle relaxant dose, mg
Midazolam, mean (SD)	1·6 (0·5)	1·8 (0·5)
Etomidate, mean (SD)	18·4 (3·2)	18·2 (4·0)
Succinylcholine, mean (SD)	35·0 (17·4)	30·3 (14·9)
Methohexital, mean (SD)	74·1 (16·5)	70·6 (17·9)

AMT=Autobiographical Memory Test; ATHF=Antidepressant Treatment History Form; BSI=Brief Symptom Inventory Anxiety Subscale; CGI-S=Clinical Global Impression – Severity; HRSD-24=24-item Hamilton Rating Scale for Depression; SSI=Scale for Suicidal Ideation; QLESQ=Quality of Life Enjoyment and Satisfaction Questionnaire

Figure 1.

CONSORT diagram

HRSD-24 Remission rate:

In the completer sample, the observed absolute difference in remission rates between the two groups was 5·3% (27.8% in the RUL-UB ECT group, 22·5% in the MST group). The Z-test using a non-inferiority margin of 15% resulted in p=0·048, 95% CI [−4·4%, 14·9%]; therefore, non-inferiority was established. In the modified ITT sample, the observed difference in remission rates between the two groups was 3.8% (25·2% in the RUL-UB ECT group, 21·4% in the MST group). The Z-test using a non-inferiority margin of 15% resulted in p= 0·021, 95% CI [−5·2%, 12·9%]. Additional sensitivity analyses are included in the Supplementary Appendix. See Figure 2 for outcomes relative to the non-inferiority margin.

Figure 2.

Differences on Primary and Secondary Clinical Measures

Error bars indicate 2-sided 95% CIs. The blue dashed line at x = 0.15 indicates the noninferiority margin, the blue tinted region to the left of x = 0.15 indicates the zone of inferiority. Response and SSI rates were included as secondary analyses. RUL-UB ECT=Right Unilateral Ultra-Brief Electroconvulsive Therapy; MST=Magnetic Seizure Therapy

Cognition:

In the modified ITT sample, there was significantly greater worsening on the AMT in the ECT group compared to the MST group, p<0·001. In the ECT group, 19 (17·3%) participants showed a ≥ 25% decrease in AMT scores compared to 3 (2·7%) in the MST group. As there were only 5% missing data, multiple imputation was not conducted.

HRSD-24 Response rate:

The Z-test with a 15% non-inferiority margin demonstrated non-inferiority in response rates between the RUL-UB ECT and MST completer groups, p=0·021, 95% CI [−9·8%, 12·4%]. In the ECT group, 52 (48·1%) participants met criteria for response compared to 52 (46·8%) participants in the MST group.

SSI Remission:

In the completer sample, including only those participants who had a non-zero SSI score at baseline, the suicidality remission rates were 48·0% for the RUL-UB ECT group and 48·4% in the MST group. The non-inferiority margin based on a 30% relative difference from the incidence in the ECT group was determined to be 14% at the post-acute time point. Using this margin, the Z-test revealed that non-inferiority was established, p=0·033, 95% CI [−13·1%, 12·5%].

Secondary clinical outcomes including the CSSES, Q-LES-Q, BSI, and CGI are reported in Table 2 and the Supplementary Appendix. Additional exploratory cognitive outcomes including the MoCA, the comprehensive cognitive battery, and time to reorientation are reported in Table 3 and the Supplementary Appendix. Across most of the cognitive outcomes, scores significantly worsened following ECT and remained relatively the same or improved following MST.

Table 2·.

Secondary Clinical Outcomes (mITT Sample)

Assessment	Electroconvulsive Therapy (N=119)		Magnetic Seizure Therapy (N=117)		Statistic
	Baseline	End	Baseline	End	Baseline-Adjusted Estimatea	p-value	95% CIs
CSSES, mean (SD)
Physical	0·4 (0·5)	0·5 (0·5)	0·4 (0·5)	0·4 (0·5)	−0·134	0·018	−0.25 – −0.02
Cognitive	0·3 (0·6)	0·8 (0·8)	0·3 (0·6)	0·2 (0·4)	−0·588	<0·001	−0.75 – −0.42
Mood	0·4 (0·4)	0·6 (0·5)	0·4 (0·4)	0·6 (0·4)	−0·019	0·764	−0.14 – 0.10
Q-LES-Q, mean (SD)	32·9 (7·1)	42·5 (11·1)	33·1 (6·0)	41·3(11·7)	−1·253	0·376	−4.04 – 1.53
BSI-A, mean (SD)	9·6 (5·6)	6·5 (5·4)	9·0 (5·3)	5·2 (5·3)	−0·925	0·073	−1.94 – 0.09
CGI-I, No. (%)
Very much improved (1)	··	13 (11·6)	··	21 (18·4)	··	··	··
Much improved (2)	··	37 (33·0)	··	27 (23·7)	··	··	··
Minimally improved (3)	··	26 (23·2)	··	31 (27·2)	··	··	··
No change (4)	··	18 (16·1)	··	24 (21·1)	··	··	··
Minimally worse (5)	··	16 (14·3)	··	9 (7·9)	··	··	··
Much worse (6)	··	2 (1·8)	··	2 (1·8)	··	··	··

^aAll statistics are listed as baseline adjusted treatment effect on post-acute neurocognitive assessment in MST group compared to ECT group. The odds ratio estimate of achieving CGI remission (1·72, p=0·155, 95% CI [0·814, 3·631]) and CGI response (0·96, p=0·877, 95% CI [0·572, 1·611]) did not differ between groups. Data were missing from less than 5% of participants, therefore those with missing post baseline data were excluded. CSSES= Columbia ECT Subjective Side Effects Schedule; BSI= Brief Symptom Inventory Anxiety Subscale; CGI-I= Clinical Global Impression – Improvement; Q-LES-Q=Quality of Life Enjoyment and Satisfaction Questionnaire

Table 3·.

Secondary Neurocognitive Outcomes (mITT Sample)

Neurocognitive Assessment	Electroconvulsive Therapy (N=119)		Magnetic Seizure Therapy (N=117)		Statistic
	Baseline Mean (SD)	End Mean (SD)	Baseline Mean (SD)	End Mean (SD)	Baseline-Adjusted Estimatea	p-value	95% CIs
Global Cognitive Function
MoCA	26·0 (2·8)	25·0 (3·2)	26·1 (2·5)	26·8 3·0)	1·939	<0·001	1·30 – 2·58
Attention / Processing Speed
DKEFS CW Condition 1 (Color Naming) Completion Time Scaled Score	8·5 (3·0)	7·3 (3·3)	8·8 (3·2)	9·1 (2·8)	1·739	<0·001	1·12 – 2·36
DKEFS CW Condition 2 (Word Reading) Completion Time Scaled Score	9·2 (2·9)	7·5 (3·5)	9·5 (2·9)	9·6 (2·4)	2·181	<0·001	1·55 – 2·81
Verbal Fluency
DKEFS Category Fluency - Total Scaled Score	10·8 (3·7)	8·4 (3·9)	10·9 (3·2)	11·1 (3·3)	2·669	<0·001	1·95 −3·39
DKEFS Category Switching - Total Correct Scaled Score	11·2 (3·8)	9·0 (3·5)	11·6 (3·4)	12·0 (3·6)	2·847	<0·001	2·04 – 3·65
Learning and Memory
CVLT-3 Trials 1–5 Correct Index Score	97·9 (16·1)	101·6 (18·0)	98·0 (16·2)	111·6 (15·9)	9·815	<0·001	6·40 – 13·23
CVLT-3 Short-delay Free Recall Correct Standard Score	9·9 (3·2)	9·9 (3·7)	10·1 (3·5)	12·1 (3·1)	2·052	<0·001	1·35 – 2·76
CVLT-3 Long-delay Free Recall Correct Standard Score	9·7 (3·1)	9·4 (3·7)	9·7 (3·2)	11·7 (3·1)	2·291	<0·001	1·58 – 3·00
CVLT-3 Yes/no Recognition Discriminability Scaled Score CVYNRDSS	9·7 (2·9)	9·2 (3·3)	9·7 (3·1)	11·1 (2·9)	1·893	<0·001	1·24 – 2·55
NIHTCB Picture Sequence Memory Test - Corrected T-Score	49·7 (11·4)	50·0 (10·2)	49·7 (12·6)	54·3 (11·1)	4·012	0·028	0·45 – 7·57
Working Memory
NIHTCB List Sorting Working Memory Test - Corrected T-Score	52·6 (9·9)	51·6 (11·8)	48·6 (10·0)	53·1 (9·1)	3·555	0·109	−0·82 – 7·93
Executive Function
DKEFS Category Switching - Total Switching Accuracy Scaled Score	11·5 (3·5)	9·7 (3·4)	11·7 (3·3)	12·3 (3·3)	2·648	<0·001	1·88 – 3·42
DKEFS CW Condition 3 (Inhibition) Completion Time Scaled Score	9·7 (3·1)	9·4 (3·8)	10·4 (3·0)	11·2 (2·7)	1·428	<0·001	0·76 – 2·10
DKEFS CW Condition 4 (Inhibition/Switching) Completion Time Scaled Score	9·8 (3·0)	7·9 (3·6)	10·1 (3·0)	11·3 (2·4)	3·274	<0·001	2·58 – 3·97
DKEFS Tower Test - Total Achievement Scaled Score	10·0 (2·9)	10·7 (3·2)	10·7 (2·8)	11·4 (2·5)	0·069	0·906	−1·09 – 1·23
NIHTCB Flanker Test - Corrected T-Score	40·8 (12·2)	36·8 (9·7)	41·0 (10·2)	41·5 (8·8)	4·251	0·003	1·53 – 6·98
Time to Reorientation b	··	15·3 (7·5)	··	6·2 (4·2)	−9·139	<0·001	−11·81 - −6·47

^aStatistic listed as baseline adjusted treatment effect on post-acute neurocognitive assessment in MST group compared to ECT group;

^bTime to Reorientation is measured in minutes, and analyzed as the estimated treatment effect on the average of mean time to orientation of all items over available treatment sessions of MST compared to ECT. Time to reorientation was available in 81 participants. No correction for multiple comparisons was employed. The cognitive measures listed are adjusted, continuous variables. Data were missing from less than 5% of participants, therefore those with missing post baseline data were excluded. CVLT-3=California Verbal Learning Test-3^rd Edition; DKEFS=Delis Kaplan Executive Function System; MoCA=Montreal Cognitive Assessment; NIHTCB=National Institutes of Health (NIH) Toolbox Cognitive Battery

Adverse Events and Drop-outs:

In the modified ITT sample, 2658 adverse events (1447 in the RUL-UB ECT group and 1211 in the MST group) were reported, of which 13 were deemed serious (see Table 4). Among those who experienced an AE, the median [IQR] number of events per participant was nine [5–15]. The majority of reported AE’s were of mild severity (1217 (84·1%) in the ECT group and 1054 (87·0%) in the MST group). Amongst the AEs in the ECT group 812 (56·1%) were deemed possibly related and 322 (22·3%) probably related, while 649 (54·3%) were deemed possibly related and 181(14·9%) probably related in the MST group. During the acute treatment, two participants experienced serious adverse events (SAE’s) unrelated to the intervention and were hospitalized (one hospitalized twice) for worsening depression in the MST group and one participant experienced an SAE probably related to the intervention and was hospitalized for a mild aspiration pneumonitis in the ECT group. During acute treatment 9 participants withdrew due to tolerability and 12 withdrew due to non-serious AEs in the RUL-UB ECT group and 4 participants withdrew due to tolerability and 3 withdrew to non-serious AEs in the MST group (Figure 1). We were unable to elicit a seizure, even at the maximum number of pulses in one participant in the MST group. MD clinical withdrawal (28 in RUL-UB ECT; 29 in MST) and patient withdrawal due to lack of perceived benefit (13 in RUL-UB ECT; 10 in MST) during the acute treatment course were similar in the two groups (Figure 1). Of note, there were no treatment related mortalities or suicide attempt-related outcomes. Numerically, more SAE’s occurred in the MST group across the entire study period, most of these adverse events were unrelated to the study intervention (see Supplemental Appendix for more detail). Additional clinical, cognitive, and tolerability outcomes are reported in Table 4 and follow-up phase SAEs in the Supplementary Appendix.

Table 4·.

Type of patient reported adverse events

Adverse Event, No. (%)	Electroconvulsive Therapy		Magnetic Seizure Therapy		χ2	p-value
	Male (N = 54)	Female (N = 65)	Male (N = 47)	Female (N = 70)	By Treatment Group
Memory impairment	30 (55·6)	39 (60·0)	17 (36·2)	14 (20·0)	19·34	<0·001
Post-treatment breathing difficulty	2 (3·7)	2 (3·1)	10 (21·3)	19 (27·1)	18·91	<0·001
Aching/stiffness in muscles	36 (66·7)	43 (66·2)	31 (66·0)	44 (62·9)	0·14	0·713
Pain in jaw	8 (14·8)	20 (30·8)	4 (8·5)	6 (8·6)	7·71	0·005
Headache	43 (79·6)	57 (87·7)	32 (68·1)	51 (73·9)	2·23	0·135
Nausea	22 (40·7)	34 (52·3)	13 (27·7)	29 (41·4)	3·03	0·082
Confusion	10 (18·5)	12 (18·5)	3 (6·4)	3 (4·3)	4·83	0·028
Post-ictal agitationa	4 (7·4)	2 (3·1)	1 (2·1)	0 (0)	1·79	0·181

^aRecovery room staff reported

χ² analysis of the number of participants reported AEs between the ECT and MST treatment groups.

Blinding:

In the modified ITT group, Simple Blinding Index (SBI) was calculated for clinical raters, cognitive raters and participants according recent recommendations³⁸. SBI values range from −1 to 1, with a value of 0 implying perfect blinding. The SBI was −0·02 95% CI [−0·14, 0·10] for clinical raters, −0·23 95% CI [0·10, 0·35] for cognitive raters and 0·17 95% CI [0·04, 0·28] in the participants. The results suggest that blinding was well preserved.

Discussion

This confirmatory efficacy trial found that MST had a numerically lower remission rate that was non-inferior to RUL-UB ECT in achieving depression symptom remission based on the a-priori defined non-inferiority margin of 15%. Importantly, the co-primary outcome of cognitive safety was also achieved as participants in the MST group experienced significantly less clinically important worsening in autobiographical memory. Secondary depression outcome measures reinforced and were consistent with the primary non-inferiority finding including response on the HRSD-24 and remission of suicidal ideation. Furthermore, the secondary cognition measures also substantiated the primary cognitive safety outcome as participants in the MST group demonstrated better global cognition, verbal learning, verbal fluency, attention, and executive functioning after treatment and reported fewer subjective complaints of cognitive adverse effects. We used a conservative modified completer analysis that provides a more clinically interpretable estimate of outcomes following adequate exposure to treatment and better reflects real-world clinical practice where early remission and late course drop-outs are common. The non-inferiority finding is stronger with the more liberal ITT analysis, likely related to a higher number of early drop outs in the RUL-UB ECT group. Taken together, the findings from this large confirmatory efficacy trial substantiate the efficacy and tolerability of MST seen in multiple prior trials as MST was both non-inferior to RUL-UB ECT in relation to remission of depressive symptoms and more favorable to ECT in relation to cognitive safety.

The overall remission outcomes in this trial demonstrate a similar trend seen in more recent convulsive therapy trials in that remission rates (i.e., closer to 30%⁹) were lower than compared to earlier ECT trials, which reported RUL-UB remission rates of ~50%²². Several reasons may explain this lower-than-expected remission rate. This trial excluded patients with MDD and psychotic features, a group that often responds more favorably to ECT than those with non-psychotic depression³⁹. In addition, the remission criteria in this trial were stricter than in other trials⁸. Also, the average age in this trial was lower than in previous trials. ECT studies have consistently demonstrated that older age is associated with higher remission rates⁴⁰. There was a relatively high rate of benzodiazepine usage in this trial (54.6%). Benzodiazepine usage has been shown to negatively impact outcome with ECT⁴¹ and outcomes with non-convulsive TMS⁴²; thus, remission rates should be interpreted with caution due to the numeric differences in benzodiazepine use between the two treatment groups (refer to Table 1) and potential confounding impact of benzodiazepines within the study sample. However, attempts were made to mitigate the impact of benzodiazepine use by capping the maximum daily dose to 2mg lorazepam equivalents and participants were asked to hold their dose after 1700 the evenings prior to treatment. Additionally, the protocol did not allow for an electrode position switch to bitemporal in the ECT arm as is commonly done in clinical practice if patients are not responding after 6 to 8 treatments. As such, our findings may not generalize to bifrontal or bitemporal ECT. However, such protocol restrictions are consistent with trials that have evaluated the efficacy of RUL-UB parameters in ECT²². This protocol restriction likely contributed to the lower ECT remission rates but likely mitigated the cognitive adverse effects in the ECT arm as bitemporal treatments are associated with worse cognitive outcomes⁶. Additional research is warranted to understand clinical and cognitive outcomes when a patient starts with MST and switches to bitemporal or bifrontal ECT as is common practice in clinical practice with patients who start with RUL-UB ECT. It should also be noted that many participants in this trial also failed multiple antidepressant trials and a course of repetitive transcranial magnetic stimulation (rTMS), suggestive of a greater degree of treatment resistance compared to previous ECT trials and, as a corollary, lower efficacy outcomes⁴⁰. The participants in this trial were primarily outpatients which may have also impacted outcomes as they did not derive the additional benefits of an inpatient admission. Finally, remission may have been adversely affected due to the randomized and blinded study design, which may have selected for patients with low expectation for both interventions. Potential participants with high expectation for ECT declined participation and opted to receive clinical ECT and potential participants with high expectation for MST combined with a fear of the cognitive adverse effects of ECT similarly declined participation. This phenomenon is evident in other neurostimulation treatment trials where outcomes are lower than what is reported in real world registries⁴³. Future trials should carefully consider whether remission may be the most clinically meaningful primary outcome. Alternative primary or co-primary endpoint outcomes such as response, mean change in HRSD scores, or measures of quality of life or function, should be explored as they may yield more clinically meaningful and informative outcomes. Finally, while these predictive variables have been shown to impact ECT outcomes, further research is required to establish whether similar effects are observed with MST.

The cognitive outcomes in this trial are consistent with multiple prior reports on the cognitive safety profile of MST¹⁰. That is, a greater proportion of individuals in the MST group relative to the RUL-UB ECT group did not experience worsening in autobiographical memory. The cognitive safety of MST was also seen in other measures, including global cognitive function (MoCA), verbal learning, verbal fluency, executive functioning, unstructured subjective reports of cognitive concerns, and the structured CSSES cognitive domain. While the exploratory cognitive measures demonstrate statistically significant better outcomes in the MST group, these results should be interpreted with caution as the clinical relevance is unclear. Ultimately, the more favorable outcomes in terms of the cognitive safety of MST is a clinically meaningful and important distinction that may address a major reason for patients to decline a course convulsive therapy^5,44, thus potentially leading to improved treatment accessibility and clinical care.

ECT is one of the few treatments that can rapidly and reliably reduce suicidality and suicide death in MDD⁴⁵. The results of this trial suggest that the effect of MST on remission of suicidal ideation was non-inferior to ECT. When considered with MST’s more favorable cognitive safety profile, one of the main barriers preventing ECT from achieving more widespread use to treat suicidality and prevent death from suicide may be mitigated. This opens the possibility that MST may serve as a viable and effective intervention for reducing suicidality. Future studies should further explore MST as a treatment for suicidality in MDD and other primary mood and psychotic disorders.

In addition to the more favorable cognitive safety profile, MST also demonstrated a more favorable physical adverse effect profile and faster time to reorientation post-treatment. The rapid reorientation after the seizure in the MST group was likely related to the higher number of post stimulation breathing weakness reports. This suggests that practitioners need to take additional care in considering the initial paralytic dose and in the early post-ictal period to potentially add an additional small dose of anesthesia (i.e. methohexital) as some patients may have residual muscle weakness after the post ictal confusion resolves⁴⁶.

Despite the strengths of the study, several limitations should be acknowledged. While people with lived experience were not directly involved in the protocol development or reporting of the study findings, the research design was informed by insights from prior neurostimulation trials and clinical discussions with those who declined ECT treatment in our centers, and the investigators’ extensive experience caring for individuals with depression. Despite this being the largest trial to date comparing MST and ECT, there were challenges in reaching the intended sample size of 260 completers, and the trial was not powered to perform specific sex/gender analyses. The COVID-19 pandemic led to major challenges in the delivery of convulsive therapy⁴⁷ and a suspension to enrollment for approximately 6-months. Of note, the decision to prematurely terminate participant recruitment was reviewed and approved by the DSMB and the NIH. In addition, the social circumstances and isolation during and after the pandemic may have impacted depression outcomes. Further, the observed remission rates complicate the assessment of the non-inferiority margin as a more conservative margin was inconclusive with respect to non-inferiority at the current sample size and would require an impractically large sample to do so. However, the overall importance of the results is reinforced by non-inferiority on multiple secondary outcome measures of depression symptoms and superiority on all cognitive measures. The randomized and blinded design contributed to differences in the population enrolled in this trial that are likely not fully generalizable to the population of patients with depression that currently receive ECT as evidence by the younger age and absence of patients with psychotic depression. Nevertheless, the population in this trial is likely closely representative of the large population of patients with refractory depression that do not want to risk the cognitive adverse effects of ECT and refuse a course of treatment.

In summary, compared to RUL-UB ECT, this trial confirmed the non-inferior efficacy, based on the a-priori defined non-inferiority margin of 15%, similar tolerability, and more favorable cognitive safety profile of MST. Though remission rates in this trial were modest and lower than expected, these rates may have been related to the randomized design of the study, strict remission criteria, and cohort characteristics (e.g., degree of treatment resistance). Among the substantial population of patients who could benefit from convulsive therapy but currently refuse to consider the treatment, MST may provide an effective alternative approach, that aligns more closely with patient preference. The availability of MST will depend on regulatory approval across jurisdictions and the development of appropriate training programs to ensure practitioners are prepared to deliver this treatment. From a logistical standpoint, however, an MST device can be incorporated into existing ECT suites with only minor electrical modifications. Future MST studies are warranted to optimize the therapeutic efficacy, understand the impact of implementation into clinical practice, and examine the long-term outcomes.

Research In Context.

Evidence before this study

Evidence used to inform this trial was identified using a PubMed search from 2001 to September Week 1 2025 for reviews and clinical trials with the following terms: depression, electroconvulsive therapy, magnetic seizure therapy. Right unilateral ultra-brief (RUL-UB) pulse width electroconvulsive therapy (ECT) is often used first line in patients with severe depression. Several clinical trials have suggested that remission rates can be as high as 50 to 60% while a systematic review suggested that the remission rate may be closer to 30% in clinical practice. RUL-UB is frequently chosen as it has demonstrated a better cognitive adverse effect profile than other brief pulse and bilateral forms of ECT. Nonetheless, many patients who could benefit from a course of ECT refuse due to fears around cognitive adverse effects. Magnetic seizure therapy (MST) is a novel convulsive therapy that utilizes high energy and frequency repetitive transcranial magnetic stimulation (rTMS) to induce a therapeutic seizure. Multiple open label and randomized trials have suggested similar clinical effects on depression and cognitive safety of MST relative to RUL-UB ECT and other forms of ECT.

Added value of this study

This trial addresses the question of whether MST is non-inferior at achieving remission of depression symptoms and has a superior cognitive safety profile compared to the most cognitive sparing form of ECT, RUL-UB. To our knowledge, this is the largest randomized trial of convulsive therapy ever conducted and the first adequately powered non-inferiority trial to address the question of MST efficacy compared to a common first-line form of ECT (RUL-UB). Our data demonstrate statistical non-inferiority of MST in achieving remission of depression symptoms as well as secondary measures of depression symptomology, and more favorable cognitive safety on autobiographical memory, time to reorientation, global cognition, verbal memory, verbal fluency, and executive functioning.

Implications of all the available evidence

Using a 15% non-inferiority margin, MST demonstrates non-inferior ability to achieve meaningful reduction in depressive symptoms with a better cognitive safety profile relative to RUL-UB ECT, the most cognitive sparing form of ECT. Among the substantial population of patients who could benefit from convulsive therapy but currently refuse to consider the treatment, MST may provide an effective alternative approach.

Data sharing

Data will be available from the NIMH Data Archive one year after publication.