The Effects of Focal Electrically-Administered Seizure Therapy (FEAST) Compared to Ultrabrief Pulse Right Unilateral Electroconvulsive Therapy on Suicidal Ideation: A Two Site Clinical Trial

Abstract

Background:

Preliminary data suggest that Focal Electrically-Administered Seizure Therapy (FEAST) has antidepressant effects and less adverse cognitive effects than traditional forms of ECT. This study compared the impact of FEAST and ultrabrief pulse, right unilateral (UB-RUL) ECT on suicidal ideation.

Methods:

At two sites, patients in a major depressive episode were treated openly with FEAST or UB-RUL ECT, depending on their preference. The primary outcome measure was scores on the Scale for Suicide Ideation (SSI). Scores on the suicide item of the Hamilton Rating Scale for Depression (HRSD-SI) provided a secondary outcome measure

Results:

Thirty-nine patients were included in the intent-to-treat sample (FEAST N=20; UB-RUL ECT N=19). Scores on both the SSI and HRSD-SI were equivalently reduced with both interventions. Both responders and non-responders to the interventions showed substantial reductions in SSI and HRSD-SI scores, although the magnitude of improvement was greater among treatment responders.

Conclusions:

Although limited by the open label, non-randomized design, FEAST showed comparable effects on suicidal ideation when compared to routine use of UB-RUL ECT. These results are encouraging and support the need for further research and a non-inferiority trial.

Suicide is among the three causes of death that are increasing in the United States.^1,2 The increased suicide rate applies to all racial and ethnic groups and both sexes.^3,4 Depressive disorder is a leading risk factor for attempted and completed suicide.^5,6 Among those in a major depressive episode (MDE), approximately one-third of individuals present with treatment-resistant depression (TRD), as they are not responsive to standard therapeutics, including psychotherapy, psychopharmacology, and brain stimulation interventions.^7–9 Patients with TRD are at especially high risk for suicide,¹⁰ and manifest a greater number of suicidal thoughts and more suicide attempts.^11,12 In this context, ECT is often life-saving, as it commonly exerts marked therapeutic effects on suicidal ideation and intent,^13,14 as well as more broadly treating TRD and improving quality of life.¹¹ However, ECT may also result in adverse cognitive side effects,^15,16 which contributes to the reluctance of some patients to consider this intervention, resulting in underutilization¹⁷ or long treatment delays.^18–20

Preliminary evidence suggests that Focal Electrically-Administered Seizure Therapy (FEAST), a new form of ECT, retains efficacy, but has less cognitive side effects compared to traditional forms of ECT, including titrated, ultrabrief pulse, right unilateral (UB-RUL) ECT, as suggested by three clinical trials.^21–23 Nahas et al. in 2013 published an open-label single arm trial of seventeen unmedicated patients in a depressive episode who were administered FEAST until patients clinically responded.²¹ Time to reorientation (TRO), a predictor of long-term cognitive effects, was assessed at each treatment. After a median course of ten FEAST sessions, full reorientation (4 of 5 items) was obtained in 5.5 ± 6.4 min (median = 3.6), suggestive of a favorable cognitive profile. In another single arm open-label trial by Sahlem et al., 20 patients with depression were administered FEAST, TRO (4 of 5 items) was achieved in 4.4 ± 3.0 min (median, 4.5 minutes), with TRO timed from eyes opening on command. In other studies patients treated with FEAST showed minimal change on the Columbia University Autobiographical Memory Interview: Short Form (CUAMI-SF), a measure of retrograde amnesia for autobiographical information. This study further suggested a favorable cognitive profile.²²

A third clinical trial was reported by Sahlem et al. in 2020. This open label, non-randomized study included 39 patients with MDE, with 20 patients electing treatment with FEAST and 19 electing UB-RUL ECT.²³ This larger trial found numeric superiority for FEAST compared to RUL-UBP ECT in average TRO (FEAST: 6.6 ± 5.0 min; RUL-UBP ECT: 8.8 ± 5.8 min; Cohens d = 0.41), and on the CUAMI-SF consistency score (FEAST: 69.2 ± 14.2%; RUL-UBP ECT: 63.9 ± 9.9%; Cohens d = 0.43). However, the findings were not statistically significant.

FEAST differs from traditional ECT in three respects. First, current in FEAST is unidirectional. In traditional ECT, using bidirectional stimulation, the direction of current reverses every other pulse. With unidirectional stimulation, current flows from one electrode to the other electrode in a consistent direction. Second, FEAST uses a novel electrode geometry, with a large posterior electrode and a relatively small anterior electrode. Third, FEAST uses a novel electrode placement, with the right-sided large posterior electrode placed anterior to the vertex and the small anterior electrode placed just above the middle of the right eyebrow. These innovations were introduced to improve spatial targeting of the electric field with ECT. This instantiation of FEAST attempts to maximize stimulation and trigger seizures from right orbitofrontal cortex, while minimizing stimulation of medial temporal regions. While the behavioral effects of FEAST have been encouraging,^21–23 the impact of FEAST on suicidal ideation have yet to be examined. In a multi-site, open label, non-randomized study, we compared FEAST with routine UB-RUL ECT in their effects on suicidal ideation.

MATERIALS AND METHODS

Study Design

The study took place at two enrolling sites: the Medical University of South Carolina (MUSC), Charleston, SC, and the Medical College of Georgia at Augusta University (AU), Augusta, GA. Institutional Review Board (IRB) study approval was obtained at each site. The study was also performed with an Investigational Device Exemption (IDE: G150034) from the US Food and Drug Administration (FDA) since the FEAST device is considered investigational. The study was pre-registered (ClinicalTrials.gov Identifier: NCT02535572) and conducted in accordance to the Declaration of Helsinki code of ethics.

Participants were recruited from those referred to the ECT clinical programs at each site. Potential participants were given a full explanation of the rationale for the study, the nature of the interventions and assessments, and the likely risks and benefits. Participants elected to receive either FEAST or UB-RUL ECT, depending on personal preference. All participants provided written, informed consent. Thus, this study used a two-site, open label, non-randomized, parallel group design. The major findings regarding antidepressant efficacy and cognitive outcomes have been reported.²³

Participants

The Structured Clinical Interview for DSM-5 (SCID)²⁴ was conducted with each participant to derive psychiatric diagnoses using the Diagnostic and Statistical Manual, 5^th Edition (DSM-5) criteria.²⁵ Participants met criteria for MDE either in the context of major depressive disorder (MDD) or bipolar disorder (BD). They also were required to have baseline HRSD₂₄ scores of 21 or greater and age between 18 and 90 years.

Potential participants were excluded if they had a history of dementia, non-mood related psychosis or rapid-cycling BD. Patients were also excluded if they had moderate or severe alcohol/substance use disorder within the past three months, were enrolled in another study using an investigational device at the time of initial screening, had an acute course of ECT in the past 6 months, were unable to discontinue lithium, anticonvulsant, or stimulant medications, or had an active and unstable neurologic or medical condition.

Assessments

Suicidal ideation was assessed with the Beck Scale for Suicide Ideation (SSI) which is a 19-item questionnaire, administered by a trained research team member, with each item rated on a 3-point scale (0–2) to assess the severity of suicidal thoughts, intent, and plans.^26,27 Specifically, the SSI examines the duration and frequency of suicidal ideation, the sense of control over an attempt, the number of deterrents, and the amount of planning involved in a contemplated attempt. The SSI was administered at baseline and at the end of the acute ECT course. The SSI has strong internal reliability (reported α coefficients between 0.90 and 0.97).^28,29 The SSI also has reasonable convergent, discriminative and predictive validity.^28,30–32 The 24-item Hamilton Rating Scale for Depression (HRSD₂₄)³³ was the primary measure of depression severity. Scores on the “suicide” item on this scale (HRSD-SI) were examined as a secondary measure of effects on suicidal ideation.

Treatment Procedures

Participants were withdrawn from lithium, stimulants, and anticonvulsants at least 24 hours prior to the first FEAST or UB-RUL ECT session. Benzodiazepines were restricted to the equivalent of lorazepam 3 mg/d or less.³⁴ Other psychotropics, including antidepressant medications, were allowed and kept unchanged in dosage until completion of assessments following the acute ECT course.

Participants were oxygenated (100%, positive pressure) prior to anesthesia induction and until the return of spontaneous respiration. Glycopyrrolate (0.2–0.4 mg IV) was administered routinely at the first treatment session. For all sessions during the course, standardized anesthesia procedures included induction with methohexital (0.75–1.0 mg/kg) and muscle relaxation with succinylcholine (0.75–1.0 mg/kg).

FEAST was administered using an investigational MECTA spECTrum 5000Q device, similar to the standard commercial device, but with the capability of delivering unidirectional stimulation (rather than the bidirectional pulses in conventional devices). In this study, the direction of the current flow was from the posterior to the anterior electrode. To achieve unidirectional stimulation, half the pulses were rectified.

The anterior FEAST electrode was circular, with a 1.25-inch diameter. It was placed such that the electrode’s lower boundary was just above the center of the right eyebrow. An adherent disposable locator, slightly wider than the anterior FEAST electrode, marked the exact position and prevented seepage of electrolyte from extending the effective stimulation area. The larger posterior electrode was oblong in shape (2 × 3 inch) and curved to better couple with the curvature of the scalp. This electrode was positioned so that its medial boundary was tangential to the line connecting the nasion and inion, and the lateral portion of the electrode extending over the right hemisphere, while the posterior boundary of the posterior electrode was 1 inch anterior to the vertex.²² Like the UB-RUL ECT electrodes, the FEAST electrodes were stainless steel and threaded onto remote-treat, hand-held, electrode assemblies.

UB-RUL ECT was administered with a standard MECTA spECTrum 5000Q device (MECTA Corporation, Tualatin, OR). UB-RUL ECT used 2-inch diameter circular, stainless steel electrodes positioned at the standard d’Elia placement, and also threaded onto remote-treat, hand-held, electrode assemblies.³⁵

Seizure threshold was quantified in the first titration session using the empirical titration method.³⁶ The same titration schedule was used for FEAST and UB-RUL ECT. At all sessions, pulse width was fixed at 0.3 ms and current (pulse amplitude) was fixed at 800 mA. Train duration, and if needed, pulse frequency, were adjusted to increase dosage. With a traditional device, pulse frequency is assessed in terms of the number of bidirectional pulse pairs per second (Hz). This convention was maintained with FEAST, so that a frequency of 20 Hz on the FEAST device corresponded to 40 unidirectional pulses per second, and to 20 bidirectional pulse pairs on the standard, UB-RUL ECT device.

After the first titration session, participants were treated with a charge six times of the seizure threshold.^37–40 Patients who did not achieve 40% improvement or more after six FEAST sessions were offered the option of increasing the stimulus dosage to nine times the seizure threshold or switching to RUL-UP ECT. For patients who did not achieve 40% improvement or more after six sessions of RUL-UP ECT were offered a dosage increase to nine times the seizure threshold.

Electroencephalography (EEG) was monitored with left and right prefrontal leads each referenced to the ipsilateral mastoid process. A motor or EEG seizure duration of at least 20 seconds was required. Seizures of inadequate duration were followed 60 s later by restimulation at higher intensity. Observation of ictal motor manifestations was facilitated by use of the cuff technique to block the distribution of succinylcholine in the right foot.⁴¹

Statistical Analyses

Analyses were conducted on an intent-to-treat (ITT) sample and in a Completer sample. Participants in the ITT sample had at least one UB-RUL ECT or FEAST session and at least one subsequent assessment. The Completer sample comprised those patients who met remission criteria (final HRSD₂₄ score ≤ 10) or had at least 8 treatments. In this study the classification of remission was used to only define the Completer sample. The data were visualized and assessed for normality. Major deviation from normal distributions was not observed, and sample characteristics were reported as mean (standard deviation) for continuous variables and as percentages for categorical variables.

The primary outcome of interest was SSI score following the acute treatment course in the ITT sample. An analysis of covariance (ANCOVA) was conducted on these scores with treatment group (FEAST vs. UB-RUL ECT), site, and their interaction as between-subject terms, and baseline SSI score as a covariate. To enhance sensitivity to potential treatment group effects, an expanded ANCOVA was also performed, with sex, age, and baseline HRSD₂₄ score serving as additional covariates. The single HRSD-SI item provided a secondary outcome measure. The same reduced and expanded ANCOVA models used for SSI analyses were applied to the post-treatment HRSD-SI scores.

Responders to UB-RUL ECT or FEAST were defined as patients with a 50% or greater decrease in post-treatment HRSD₂₄ scores relative to baseline. To examine the impact of clinical response on change in suicidal ideation, a repeated measures ANOVA was conducted on baseline and post-treatment SSI scores, with response status, treatment condition (UB-RUL ECT vs. FEAST) and their interaction as between-subject variables. The same repeated measures ANOVA model as applied to the baseline and post-treatment HRSD-SI scores.

All statistical tests were two-tailed, with alpha level of 0.05 denoting statistical significance. The statistical analyses were performed using JMP Pro 14.3.0 (SAS Institute, Inc., Cary, North Carolina, USA).

RESULTS

Demographic and Clinical Characteristics

Thirty-nine participants comprised the ITT sample: 20 in the FEAST group and 19 in the UB-RUL ECT group. Average age was 44.2 (SD=14.5) years [FEAST: 45.2 (SD=12.7) years; UB-RUL ECT: 43.2 (SD=16.4) years]. The sample consisted of 76.9% women [FEAST: 70.0% women; UB-RUL ECT: 84.2% women]. In the ITT sample, 89.7% of participants were Caucasian and 10.3% were African Americans (AA), with no difference in the distribution between the treatment groups. The treatment groups also did not differ in years of education, employment status, or marital status.²³

The average HRSD₂₄ score at baseline was 33.3 (SD = 6.7), indicating severe depressive symptomatology, with no difference between the treatment groups [FEAST: 34.3 (SD=6.8); UB-RUL ECT: 32.2 (SD=6.6)]. Similarly, average Inventory of Depressive Symptomatology Self‐Report (IDS-SR) and Clinical Global Impression – Severity scale (CGI-S) baseline scores indicated marked symptom severity, with no treatment group differences. However, only 10.3% of the ITT sample (2 participants in each treatment group) met criteria for psychotic depression. The treatment groups were imbalanced in the representation of bipolar disorder which was more common in the UB-RUL group (52.6%) than the FEAST group (5.0%), χ²(1) = 10.92, P = 0.001. In contrast, the treatment groups did not differ in average duration of the depressive episode [85.7 (SD=119.6) weeks] or in the number of failed adequate antidepressant treatment trials in the current episode [2.7 (SD=1.8)]. The majority of all participants (84.6%) were treated with one or more antidepressant medications while receiving FEAST (85.0%) or UB-RUL ECT (84.2%).

Three patients in the FEAST group, and two in the UB-RUL ECT group discontinued treatment before session 8 or before reaching remission and were dropped from the Completer sample. As the findings were consistent across the ITT and Completer samples, only the results from the ITT analyses are reported. Other details regarding recruitment and retention are described elsewhere, along with FEAST/RUL ECT stimulus parameters and seizure characteristics.²³

Baseline Suicidal Ideation

Table 1 presents descriptive statistics on suicide ideation scores at baseline and endpoint for the FEAST and UB-RUL ECT groups. At baseline, some degree of suicidal ideation was characteristic of virtually all participants. In the total sample, the percentage of patients with baseline scores above zero was 89.7% for the SSI total score and 92.3% for the HRSD-SI item. Total baseline SSI and HRSD₂₄ scores were only modestly correlated, r(37) = 0.37, P = 0.02, while the baseline HRSD-SI item had stronger relations with both the total SSI score, r(37) = 0.57, P = 0.0002 and the total HRSD₂₄ score, r(37) = 0.53, P = 0.0005. The treatment groups did not differ in total baseline SSI, t(37) = 0.66, P = 0.52, HRSD₂₄, t(37) = 0.91, P = 0.37, and HRSD-SI, t(37) = 0.16, P = 0.87.

Table 1.

Descriptive statistics on suicide ideation scores at baseline and following treatment with FEAST or Ultrabrief Pulse Right Unilateral ECT

	FEAST					Ultrabrief Pulse Right Unilateral ECT
	N	Mean	SD	Median	Score > 0 (%)	N	Mean	SD	Median	Score > 0 (%)
Beck Scale for Suicide Ideation (SSI)
Baseline 1	20	11.80	7.12	12.00	85.00	19	13.32	7.33	12.00	94.74
Baseline 2	16	11.69	7.80	12.50	81.25	14	12.57	7.84	13.50	92.86
Endpoint	16	3.75	5.39	1.50	53.33	14	4.00	4.37	3.50	46.67
Hamilton Rating Scale for Depression Suicide Item (HRSD-SI)
Baseline	20	2.00	0.97	2.00	90.00	19	2.05	1.03	2.00	89.47
Endpoint	20	0.5	0.18	0.00	35.00	19	0.37	0.60	0.00	31.58

Baseline 1 contains all 39 patients in the study intent-to-treat sample. Baseline 2 contians only those patints who contributed SSI scores at post-treatment. There were no missing data for the HRSD-SI.

Change in Suicidal Ideation

Nine patients did not complete the post-treatment SSI: 5 of these patients were treated with UB-RUL ECT and 4 patients were treated with FEAST. There was no data loss on the HRSD-SI item. As seen in Table 1 and Figure 1 there was a profound reduction in suicide ideation scores following both interventions. Across the sample, the average SSI score dropped by 68.0%, paired-t(29) = 5.91, P < 0.0001, and the average HRSD-SI item score dropped by 78.5%, paired-t(38) = 8.53, P < 0.0001. As seen in Figure 1, 4 of 30 patients showed a small increase in SSI score from baseline to post-treatment, while this was true of only 1 of 39 patients on the HRSD-SI item score.

Figure 1.

Change in Scores on the Beck Scale for Suicide Ideation and the Hamilton Rating Scale for Depression Suicide Item for Patients Treated with Focal Electrically-Administered Seizure Therapy (FEAST) and Ultrabrief Pulse Right Unilateral (UB-RUL) ECT

The reduced ANCOVA on SSI scores following treatment did not yield a main effect of treatment group, F(1, 25) = 0.03, P = 0.87. The only significant effect in this analysis was that of the baseline SSI score, F(1, 25) = 5.11, P = 0.03. The expanded ANCOVA on post-treatment SSI scores also did not yield any effect involving treatment group. The reduced ANCOVA on post-treatment HRSD-SI scores did not yield any significant effects. Likewise, in the expanded model, no effect was significant.

Clinical Response and Change in Suicidal Ideation

The repeated measures ANOVA on SSI scores yielded trends for a main effect of time, F(1, 26) = 3.75, P = 0.06, and an interaction between response status and time, F(1, 26) = 4.07, P = 0.05. As seen in Figure 2, both responders, t(19) = 5.87, P <0.0001, and non-responders, t(9) = 2.27, P < 0.05, showed significant post-treatment decreases in SSI scores, but the magnitude tended to be greater among responders. The same repeated measures ANOVA applied to HRSD-SI item scores yielded a main effect for time, F(1, 35) = 20.69, P < 0.0001, and a trend for an interaction between response status and time, F(1, 35) = #.20, P = 0.08. As seen in Figure 2, HRSD-SI scores decreased markedly in both responders, t(23) = 7.96, P < 0.0001, and nonresponses, t(14) = 5.55, P <0.0001, although the magnitude of the decrease tended to be greater in the responder group. In these analyses, there were no effects involving treatment group (i.e., UB-RUL ECT vs. FEAST).

Figure 2.

Scores on the Beck Scale for Suicidal Ideation (SSI) and the Hamilton Rating Scale for Depression-Suicide Item (HRSD-SI) Before and Following the ECT Course in Treatment Responders and Non-responders.

DISCUSSION

This study compared FEAST and UB-RUL ECT on their impact on suicidal ideation. Both treatments produced a profound reduction in SSI and HRSD-SI scores, indicating substantially reduced suicidal ideation. There was no difference between FEAST and standard UB-RUL ECT in the magnitude or prevalence of this effect. Substantially reduced suicidal ideation has been observed following both bitemporal (BT) and right unilateral ECT¹³, but had never been examined before for the investigational FEAST technique.

Clinical trials, especially in the past 3 decades, have examined the effect of BT and RUL ECT on suicidal ideation. All the suicidal ideation outcome assessments of the studies were secondary outcomes of studies of different interventions that included ECT (e.g., different ECT techniques or ECT versus pharmacotherapy).

Our findings regarding the effects of UB-RUL ECT and FEAST on suicidal ideation are consistent with prior work. For instance, Prudic and Sackeim¹³ pooled data from 2 double-blinded trials (N=148),^42,43 In both trials, patients were randomized to BT or RUL ECT using brief pulse stimulation. Across patients, the extent of decrease in the HRSD-SI item score was greater than the average change of all other HRSD₂₄ items. Similarly, of the 24 items, the degree of change in HRSD-SI item scores over the ECT course was the fifth largest, even though a substantial number of patients had HRSD-SI scores of 0 at baseline. Both ECT responders and non-responders showed improvement in HRSD-SI scores, with responders showing a decrease from a mean baseline of score of 1.8 to a post-ECT score of 0.1. ECT non-responders had mean scores of 1.6 and 0.9 at the comparable time points. Thus, this study found that the degree of improvement in suicidal ideation was greater following ECT than other depressive symptoms, and that substantial improvement in suicidal ideation was observed even among those considered as “nonresponders”.

In a similar study, Kellner et al.⁴⁴ examined the effects of an acute ECT course on HRSD-SI item scores in a Consortium for Research in ECT (CORE) clinical trial using BT brief pulse ECT. Of 444 participants, 131 patients (29.5%) with high expressed suicidal intent (baseline scores of 3 or 4 on the HRSD-SI), 106 patients (80.9%) had scores of 0 following the ECT course. Resolution of suicidal ideation on the HRSD-SI was 61.1% after 6 sessions, and 76.3% after 9 ECT sessions. Fink⁴⁵ pooled these data with those of a second CORE study⁴⁶ (total N=746) and reported similar results; the largest studies in the literature examined scores on the single HRSD-SI to quantify the impact of ECT on suicidal ideation. We found only a moderate correlation between baseline scores on this item and scores on the SSI. Scores on the HRSD-SI can only range from 0 to 4. The SSI provides a more comprehensive assessment of suicidal ideation, has strong psychometric properties, and evidence supporting its validity. In this study, the findings were consistent across the SSI and HRSD-SI. There was no indication that FEAST and UB-RUL ECT differed in efficacy in reducing suicidal ideation. Anti-suicidal effects were observed with both interventions. While this improvement was especially marked in ECT responders, it was also substantial in ECT non-responders.

This study is limited in several respects. Assignment to the treatment conditions was based on patient preference and the trial was open-label and non-randomized. The sample size is modest and more definitive research is certainly needed. In addition, suicidal ideation was not the primary outcome of this study, and thus the sample included patients with and without suicidal ideation at baseline.

Nevertheless, the therapeutic effects on suicidal ideation observed here across the treatment conditions and assessment tools were consistent with the acute anti-suicidal effects observed in large trials using bitemporal or RUL brief pulse ECT. The findings of this study suggest that the acute anti-suicidal effects of ECT are also observed with the newer interventions, UB-RUL ECT and FEAST, which are likely the most benign forms of ECT in terms of adverse cognitive effects.

While there is broad consensus that ECT has marked, acute anti-suicidal effects,^13,17 there is less certainty regarding its long-term impact on suicide. There is consistent evidence that ECT reduces all-cause mortality, linked to its impact on medical co-morbidities,^13,47–51 but, in contrast, there is little consistent evidence of a beneficial or detrimental effect on the long-term rate of suicide.^50,52–58 Part of the difficulty in establishing a long-term effect may be methodological. High levels of suicidal ideation and intent and lack of response to alternative interventions are often key factors in referral to ECT, complicating comparison of trajectories with other antidepressant treatments. This is especially problematic when suicide may occur years after the intervention being examined was terminated with multiple intervening treatments used prophylactically or treat recurrence. Regardless, the findings of this study suggest that the anti-suicidal effect observed with brief pulse ECT also applies to UB-RUL ECT and FEAST.