Notes on Emotion Regulation and rTMS for GAD

Introduction

• Generalized Anxiety Disorder (GAD) is characterized by difficulties in identifying and regulating emotional experiences.
• The emotion regulation model of GAD proposes that these deficits lead to excessive emotional arousal and maladaptive emotion regulation strategies, maintaining symptoms.
• Patients with GAD struggle with emotion intensity, labeling, expression, acceptance, and modulation.
• Cognitive-behavioral and acceptance-based therapies target emotion regulation deficits in GAD treatment.
• This study investigates whether neuromodulation can improve emotion regulation deficits in GAD patients.

Neural Circuits and Emotion Regulation

• Neuroimaging studies suggest emotion regulation is supported by a neural circuit including fronto-limbic regions.
• Abnormalities in this circuitry are found in GAD patients.
• Patients with GAD show stronger connectivity between limbic and prefrontal regions during worry induction.
• They are unable to inhibit worry-related neural activity after induction tasks.
• During emotion modulation tasks (reappraisal, suppression), GAD patients show hypoactivation of prefrontal and anterior cingulate cortex.
• They also exhibit less connectivity between the medial prefrontal cortex and prefrontal/limbic regions (DLPFC, insula).
• In emotional conflict tasks, GAD patients fail to engage the anterior cingulate cortex.

Neuromodulation and rTMS

• Repetitive transcranial magnetic stimulation (rTMS) is effective for treating depression.
• Anxiety symptoms may improve in depressed patients receiving neuromodulation.
• Uncontrolled research indicates DLPFC neuromodulation improves anxiety in GAD patients.
• A randomized controlled trial (RCT) showed active DLPFC neuromodulation was superior to sham for improving GAD symptoms.
• Some research suggests pharmacotherapy improves DLPFC connectivity during emotion regulation in GAD patients.
• This study reports secondary analyses from an RCT comparing active versus sham rTMS of the right DLPFC in GAD patients.
• Participants completed the Difficulties in Emotion Regulation Scale (DERS) at pretreatment, posttreatment, and 3-month follow-up.
• The DERS contains six subscales assessing different facets of emotion regulation.
• The hypothesis was that patients receiving active rTMS would report more improvements in emotion regulation (DERS total score) and that improvement in emotion regulation would be associated with treatment response.
• Exploratory analyses were conducted on DERS subscales to determine the pervasiveness or specificity of treatment effects.

Material and Methods

• Participants were recruited from an outpatient clinic and the community for an RCT comparing active to sham rTMS in adult outpatients diagnosed with GAD.
• Thirty-four participants enrolled, but twenty-five participants (n = 13 active; n = 12 sham) were included in data analyses.
• Participants in both groups reported a mean age of 44 (active M = 44.00, SD = 11.95; sham M = 44.58, SD = 14.75) and were predominantly women (active = 11/13, 84.6%; sham = 8/12, 66.7%).
• Of these participants, six discontinued treatment prematurely (n = 4 active, n = 2 sham), and 1 (sham) was lost to follow-up.
• Inclusion criteria included age ≥18, principal or co-principal GAD, Clinical Global Impression-Severity rating ≥4, Hamilton Anxiety Rating Scale ≥18, and 17-item Hamilton Rating Scale for Depression ≤17.
• Exclusion criteria included neurological disorder, unstable medical illness, contraindications for MRI/rTMS, current posttraumatic stress disorder, substance use disorder, lifetime bipolar, psychotic, developmental, or obsessive-compulsive disorder, concurrent psychotherapy, or psychiatric instability.
• Concurrent pharmacotherapy was stabilized prior to study entry.

Measures

• Study inclusion criteria were assessed using the Mini International Neuropsychiatric Interview, Clinical Global Impression-Severity scale, Hamilton Anxiety Rating Scale, and 17-item Hamilton Rating Scale for Depression.
• The Clinical Global Impression-Improvement scale (CGI-I) was used as the measure of treatment response, taking into account improvements in overall emotional symptoms and functional impairment.
• Emotion regulation was assessed using the Difficulties in Emotion Regulation Scale (DERS), a 36-item self-report measure with six subscales: Non-Acceptance, Goals, Impulse Control, Awareness, Strategies, and Clarity.
• Higher scores on the DERS indicate more severe difficulties with emotion regulation.
• Previous research has found the DERS to demonstrate adequate construct validity, good test-retest reliability, and high internal consistency.

rTMS Protocol

• Participants randomized to active treatment received 30 daily sessions (5 days/week) of low-frequency rTMS stimulation to the right DLPFC (MNI coordinates: x = 42, y = 36, z = 32) using the NeuroStar TMS Therapy System.
• Right-side stimulation of the DLPFC was chosen given evidence that emotion regulation processes are lateralized to the right side.
• Right-sided DLFPC rTMS has also been associated with improvements in anxiety symptoms, including in patients with GAD, and may be superior to high-frequency, left-sided stimulation for treating anxiety symptoms.
• Stimulation parameters (1 Hz, 900 pulses/session, 90% resting motor threshold) were chosen to be the same as those used in a previous open trial of rTMS for GAD.
• The right DLPFC point for stimulation was identified using structural MRI scan and located using a frameless stereotactic neuronavigation system.
• Procedures were similar for those participants receiving sham with the exception that a sham coil (Neuronetics XPLOR coil) was used.

Procedure

• Written informed consent was obtained from study participants prior to entry.
• Procedures were approved by the Hartford Hospital Institutional Review Board and the study was listed on clinicaltrials.gov (Registration Number: NCT01659736).
• The randomization schedule was developed using a computerized random number generator.
• Participants were randomized to condition at the first treatment session.
• Study eligibility was determined by a licensed clinical psychologist.
• The clinical psychologist, who was blind to treatment condition, also completed the CGI-I at posttreatment and 3-month follow-up.
• The DERS was completed by participants at pretreatment, posttreatment, and 3-month follow-up.
• The clinician and participants independently completed a self-report measure to assess the blind.

Data Analytic Plan

• Data distributions of continuous variables were checked for normality using normal probability plots prior to statistical analysis.
• Outcome analyses were conducted using the intent-to-treat (ITT) sample with a linear mixed effects model with treatment condition (active rTMS vs. sham) as a between-subject factor, time (pretreatment, posttreatment, follow-up) as a within-subject factor and the interaction between treatment condition and time for each outcome.
• The best-fitting variance-covariance structure for each outcome was selected using Schwartz’ Bayesian Information Criterion (BIC).
• Least square means and standard errors were plotted for the ITT sample and least square mean comparison post-hoc tests were performed to explain significant effects in the models.
• Linear contrast of least square mean effect sizes with 95% confidence intervals were computed to illustrate the effect size of group differences in emotion regulation changes over time.

Post-hoc analyses controlling for pretreatment scores

• Given results of baseline differences in treatment conditions, post-hoc analyses were conducted on the sample of treatment completers (n = 9 participants in active rTMS, n = 9 in sham owing to missing DERS data for one participant at posttreatment) to investigate the nature of changes in emotion regulation over time when controlling for pretreatment scores.
• Percentage of change from (1) pretreatment to posttreatment and (2) pretreatment to follow-up was determined for all measures demonstrating a significant treatment condition by time interaction in linear mixed effects models.
• Effect size of group differences on the DERS percent of change was determined using Cohen’s d, and interpreted as 0.20 = small, 0.50 = medium and 0.80 = large.
• Next, analysis of covariance, controlling for pretreatment scores, was computed comparing the active versus sham treatment completers.

Associations with global treatment response

• Lastly, associations between emotion regulation improvements (as measured by percentage change at posttreatment and follow-up) and treatment outcome (as measured by CGI-I) were determined using Spearman’s rho correlations (this analysis was used because of small sample size).
• To ensure that outliers would not bias correlations all variables used in these analyses were checked for outliers (defined as a value falling outside the range of the mean score plus or minus 2.5 standard deviations).

Results

DERS total score

• There was a statistically significant treatment condition by time interaction, with significant improvements occurring only for those participants receiving active rTMS.
• The improvement in the active rTMS condition was statistically significant at both posttreatment and follow-up compared to baseline.
• However, the mean DERS total score was statistically significantly different between the two treatment conditions only at pretreatment [F (1, 29.2) = 6.26, p = 0.02; partial eta^2 = 0.249], with the active rTMS group scoring significantly higher than the sham group.
• Linear contrast of least square mean effect sizes with 95%confidence intervals of group differences in emotion regulation changes over time showed large between-group differences in symptom change over time for the total score at both posttreatment and 3-month follow-up.

DERS subscales

• A statistically significant treatment condition by time interaction was found for both the DERS goals and impulse control subscales.
• In both cases there was a statistically significant improvement in the active rTMS group, but not in the sham group.
• The improvement in the active rTMS group was statistically significant at both posttreatment and follow-up compared to baseline.
• However, the treatment conditions differed significantly only at pretreatment [goals F (1, 31.5) = 12.13, p = 0.002, partial eta^2 = 0.414; impulse control F (1, 28.8) = 7.21, p = 0.01, partial eta^2 = 0.239], with the active rTMS group scoring significantly higher than the sham group.
• There were no statistically significant treatment condition by time interaction effects for the DERS non-acceptance, lack of awareness, emotional clarity, or strategies subscales.

Post-hoc analyses controlling for pretreatment scores

• ANCOVA results indicated a significant effect of treatment condition [F (1, 17) = 6.62, p = 0.021; partial eta^2 = 0.306] with larger change occurring in the active (M = 0.18, SD = 0.15) than in the sham group (M = −0.01, SD = 0.13) at posttreatment while controlling for pretreatment DERS total score.
• However, the active (M = 0.16, SD = 0.17) and sham (M = 0.03, SD = 0.16) groups no longer differed significantly at 3-month follow-up [F (1, 17) = 2.62, p = 0.126; partial eta^2 = 0.149].
• The between-group effect size was large at post- treatment (d = 1.35) and moderate to large at 3-month follow-up (d = 0.78).
• ANCOVA performed on percent change of the DERS impulsivity scale (again controlling for pretreatment scores) indicated larger improvements in the active (M = 0.29, SD = 0.21) than sham (M = −0.06, SD = 0.21) groups at posttreatment [F (1, 17) = 8.55, p = 0.01; partial eta^2 = 0.363].
• Significant differences tended to be maintained over the 3-month follow-up [F (1, 17) = 4.56, p = 0.05; partial eta^2 = 0.233], again with active (M = 0.25, SD = 0.20) demonstrating larger changes than sham (M = −0.06, SD = 0.30) over time.
• Between-group effect sizes were large for percentage of change in the DERS impulsivity scale at both posttreatment (d = 1.67) and follow-up (d = 1.21).
• ANCOVA performed on percent change of the DERS goals scale (again controlling for pretreatment scores) indicated no significant differences between active and sham at either posttreatment [Active: M = 0.16, SD = 0.22; Sham: M = −0.10, SD = 0.20; F (1, 17) = 3.07, p = 0.100; partial eta^2 = 0.170] or 3-month follow- up [Active: M = 0.16, SD = 0.27; Sham: M = −0.03, SD = 0.25; F (1, 17) = 1.19, p = 0.292; partial eta^2 = 0.074].
• Between-group effect sizes were large for percentage of change in the DERS goals scale at posttreatment (d = 1.24) and moderate at follow-up (d = 0.50).

Associations with global treatment response

• Improvements in self-reported emotion regulation difficulties as measured by post-treatment percentage change in the DERS total score were significantly associated with better clinician-rated improvements in global severity and impairment at posttreatment [rs (18) = -0.46, p = 0.045] and at 3-month follow-up [rs (16) = -0.71, p = 0.002].
• When examining the treatment conditions separately, the association remained strong only in the active rTMS condition at posttreatment [posttreatment rs (8) = -0.78, p = 0.01; 3-month follow-up rs (8) = -0.62, p = 0.07], and not in sham at either timepoint [posttreatment rs (8) = 0.41, p = 0.27; 3-month follow-up rs (7) = 0.13, p = 0.98].
• Improvements in the DERS goals subscale were also associated with global clinical improvements at posttreatment [rs (18) = -0.47, p = 0.04], and 3-month follow-up [rs (16) = -0.54, p = 0.03].
• However, this association held only in the active rTMS treatment condition at posttreatment [active rTMS rs (8) = -0.76, p = 0.02; sham rs (8) = 0.02, p = 0.93] and not at 3-month follow-up [active rTMS rs (8) = -0.36, p = 0.34; sham rs (7) = -0.24, p = 0.56].

Discussion

• This study indicated that rTMS of the right DLPFC improves emotion regulation as measured by the DERS total score in patients with GAD from pre-to-posttreatment and 3- month follow-up.
• Analyses indicated that the percentage of improvement in DERS total score was associated with clinician ratings of global clinical outcome.
• Supplementary analyses indicated that subscales measuring goal-directed behaviors and impulse control demonstrated the largest and statistically significant rTMS treatment effects, although only the goals subscale was associated with clinical outcome.
• Current results are consistent with previous research suggesting that DLPFC neuromodulation alters emotion-related cognitive processes and decision making.
• The degree of connectivity between DLPFC and VMPFC is related to treatment outcome of rTMS in a variety of psychiatric conditions.
• This study is the first to investigate the effect of neuromodulation treatment on emotion regulation in a sample of patients with GAD.
• Results lend further support to the clinical effects of DLPFC neuromodulation in patients with GAD and indicate the importance of future research to investigate the efficacy and mechanisms of neuromodulation in larger and better-matched samples of patients with GAD.
• It is a limitation that emotion regulation assessment was conducted using a self-report measure, and it will be important for future research on neuromodulation effects in GAD to include measures of biological and/or behavioral indices of emotion regulation as well.