Efficacy of Repetitive Transcranial Magnetic Stimulation on Comorbid Anxiety and Depression Symptoms in Obsessive-Compulsive Disorder: A Meta-Analysis of Randomized Sham-Controlled Trials

Abstract

Objective

To systematically evaluate the efficacy of repetitive transcranial magnetic stimulation (rTMS) in reducing comorbid anxiety and depressive symptoms in patients with obsessive-compulsive disorder (OCD).

Methods

Three electronic databases were searched for randomized, sham-controlled clinical trials evaluating rTMS for the treatment of OCD. Hedge's g was calculated as the effect size for anxiety/depression symptom severity (primary outcome) and OCD severity (secondary outcome). Subgroup analyses and meta-regression analyses were carried out to evaluate the most promising target and whether a reduction in OCD severity moderates the change in anxiety or depression scores.

Results

Twenty studies (n = 688) were included in the meta-analysis. rTMS had small-medium effect size on OCD (Hedge's g = 0.43; 95% confidence interval [CI]: [0.20, 0.65]; P < 0.001), anxiety (Hedge's g = 0.3; 95% CI: [0.11, 0.48]; P = 0.001) and depression (Hedge's g = 0.24; 95% CI: [0.07, 0.40]; P = 0.003) symptoms. Subgroup analysis showed that protocols targeting dorsolateral prefrontal cortex (DLPFC) were effective for 3 outcome measures. The change in anxiety, but not depression severity, was moderated by a change in OCD symptom scores. However, the findings are uncertain as a majority of the studies had some concerns or a high risk of bias.

Conclusions

Active rTMS protocol targeting DLPFC is effective in reducing the comorbid anxiety/depression symptoms along with OCD severity. The antidepressant effect is not moderated by the anti-obsessive effect of rTMS.

Introduction

Obsessive-compulsive disorder (OCD) is a chronic, disabling neuropsychiatric disorder with a lifetime prevalence of 1.3%.^1,2 OCD patients have a heterogeneous clinical profile in terms of symptom presentation, comorbidities, or treatment response. ³ The literature suggests a high prevalence of comorbid anxiety and depressive symptoms in OCD. Around 50% of patients with OCD have lifetime comorbidity of major depressive disorder (MDD), while more than 25% of them suffer from various anxiety disorders. ⁴ The presence of these comorbid conditions is associated with distinct clinical characteristics, increased severity of OCD, suicidality, greater disability, poorer quality of life and inadequate response to treatment.^5–8 Hence, it is imperative to incorporate treatment of anxiety/depressive symptoms in the management of OCD patients.

There are similarities in the treatment of OCD, anxiety and depressive disorders; selective serotonin reuptake inhibitors (SSRI) and cognitive-behaviour therapy (CBT) are helpful for OCD as well as anxiety/depressive disorders. However, there are notable differences in implementation of these interventions. For example, SSRIs are prescribed at a higher dose for anti-obsessive action and CBT techniques such as exposure and response prevention are somewhat specific to OCD. Nevertheless, shared genetic underpinnings and neurobiological mechanisms (involvement of amygdala, mesolimbic, and other areas of frontostriatal circuitry) suggest the possibility of a common neurobiological substrate for these apparently divergent symptoms. ⁹ The efficacy of anti-obsessive interventions for comorbid anxiety/depressive symptoms can be inferred from clinical trials, which evaluate the severity of the concomitant anxiety/depressive symptoms as secondary outcome measures. Further, it is not evident from the clinical studies whether the improvement in depression and anxiety is driven by improvement in OCD symptoms or if the treatments have independent effects. Also, the efficacy of various anti-obsessional treatments on comorbid anxiety and depression may vary. For example, there is evidence that CBT for OCD with exposure & response prevention may also decrease comorbid depressive symptoms. ¹⁰

Neuromodulatory interventions, such as repetitive transcranial magnetic stimulation (rTMS) aim to modulate activity in specific neural circuits. Multiple meta-analyses indicate that rTMS is helpful in treatment-resistant OCD.^11–13 rTMS protocols for OCD target different cortical regions including supplementary motor area (SMA), dorsolateral prefrontal cortex (DLPFC), orbitofrontal cortex (OFC) and dorsomedial prefrontal cortex (DMPFC). rTMS has also been found to be effective in the treatment of depressive or anxiety disorders.^14,15 rTMS targeting DLPFC is approved by the Food and Drug Association for the management of MDD. ¹⁶ Recent evidence also indicates the potential antidepressant effects of protocols modulating the DMPFC and OFC.^17,18 Furthermore, rTMS protocols targeting the prefrontal cortex have also been found to be helpful in anxiety disorders. ¹⁹ Along with differences in cortical targets, there are some variations in the protocols employed, for example stimulatory (high-frequency rTMS and intermittent theta burst stimulation) or inhibitory protocols (low-frequency rTMS and continuous theta burst stimulation).

As rTMS is a focal intervention, it is not clear whether treatments targeting OCD may have an impact on comorbidities too. While open-label studies have shown that rTMS over DLPFC/SMA/DMPFC decreases anxiety and depressive symptoms, ²⁰ sham-controlled studies have not been consistent.^21–23 The limited sample sizes of the individual studies might have decreased the power to find statistically significant differences. Hence, we conducted the current meta-analysis of randomized sham-controlled trials to examine whether rTMS for OCD decreases concomitant anxiety and depressive symptoms too. We conducted subgroup analyses to evaluate the most promising target for decreasing these symptoms. Further, we examined whether rTMS has independent anti-anxiety and anti-depressive effects in this population by conducting a meta-regression to evaluate whether the decrease in anxiety/depressive symptoms is moderated by anti-obsessive effects.

Methods

The meta-analysis was conducted in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. ²⁴

Search Strategy

A preliminary literature search was carried out on PubMed, PsycINFO, and The Cochrane Central Register of Controlled Trials (CENTRAL) up to July 2021, with the search terms “obsessive compulsive disorder” OR “OCD” OR “obsessions” OR “compulsions” AND “transcranial magnetic stimulation” OR “TMS” OR “magnetic stimulation” OR “rTMS.” Additional studies were searched manually from the reference list of relevant articles and previous meta-analyses.^11,12

Study Selection

The studies published in English, based on the following criteria were included: (1) participants: subjects diagnosed with a primary diagnosis of OCD without comorbid psychosis; (2) intervention: rTMS provided either as monotherapy or add-on treatment for obsessive-compulsive symptoms with a minimum of 5 sessions of rTMS; (3) comparison: studies comparing active rTMS protocols to sham rTMS; (4) outcome: studies which have reported clinician or self-rated measures of OCD along with either anxiety or depression symptom severity, both at baseline and post-intervention; and (5) double-blinded or single-blinded randomized controlled trials (RCTs) with a minimum sample size of 10.

Data Extraction and Coding

Two independent investigators (NST and PV) reviewed the final articles and the following variables were collected: (1) patient characteristics: number of participants, mean age, male:female ratio, comorbidity rate of depressive anxiety disorders, presence of treatment resistance, duration of illness, age of onset, concomitant treatment details; (2) rTMS parameters: type of intervention—high frequency, low frequency or theta burst stimulation, stimulation frequency, cortical targets, number of sessions, total pulses; (3) comparison: sham strategy; (4) outcome: baseline and post-treatment mean OCD, anxiety, depression severity scores and standard deviations in both arms. When both clinician-rated and patient-rated outcomes were reported, the analysis was conducted on clinician-rated measures; (5) study design: sample size of individual arms, dropouts, single- or double-blinded strategies and other methodological aspects that might contribute to bias as mentioned below.

Risk of Bias Assessment

A risk of bias assessment was conducted based on the recommendations of the updated version of the Cochrane risk-of-bias tool for randomized trials (RoB 2). ²⁵ The tool evaluated the following domains: randomization process; effect of assigned intervention; missing outcome data; measurement of outcome; and selection of reported result.

Data Synthesis

All analyses were performed using the meta package in R (v4.17-0). ²⁶ Any study with anxiety/depressive outcomes along with OCD measures was included for the quantitative synthesis. We employed the pre–post differences in each arm of studies for calculating the effect size; where it was not available, we calculated them in accordance with the Cochrane Handbook 6.2. ²⁷ For each study, Hedge's g was calculated as a measure of standardized mean difference, as the effect size measure, and this was meta-analysed using the random-effects model, with the DerSimionian Laird Inverse variance weighted method. For each outcome, sensitivity analyses were carried out by excluding studies with a high risk of bias (see below). The heterogeneity between studies was calculated using Cochran's Q statistic and the I² statistic. We used the following cut-offs for interpreting the I² values: 0%-25% mild, 25%-50% moderate, >50% high degree of heterogeneity. We then performed subgroup analyses by cortical target (DLPFC, SMA, DLPFC & SMA, DMPFC, OFC) and used the QM statistic to estimate the differences between targets. We performed two independent meta-regression analyses with the effect size of Yale-Brown Obsessive Compulsive Scale (Y-BOCS) reduction as the moderator and effect size on anxiety and depression as dependent variables. Contour-enhanced funnel plots with Egger's regression tests were done to assess publication bias.

Results

Study Selection and Characteristics

Twenty studies (Figure 1) met the eligibility criteria for inclusion in the meta-analysis.^{22,23,28–45} All studies reported Y-BOCS symptom severity as the primary outcome measure for OCD symptoms. Of these, 18 studies reported depression severity (n = 593) and 17 studies reported anxiety severity (n = 534) as a secondary outcome variable. Hamilton Anxiety Rating Scale (HAM-A) or Brief Anxiety Scale or Beck's Anxiety Inventory or State-Trait Anxiety Inventory were used for evaluation of anxiety severity. Hamilton Depression Rating Scale (HAM-D) or Montgomery-Asberg Depression Rating Scale (MADRS) were employed for evaluating the severity of depression. The final analysis had a total of 356 and 312 participants from active and sham arms, respectively. The mean age of the participants included was 34.21 (31.78-36.64) and 47% (41%-54%) were females. The demographic, baseline clinical measures and intervention characteristics of the included studies are summarized in Supplementary Table 1.

Figure 1.

Flow diagram of the literature search and inclusion of studies.

Risk of Bias Assessment

Most of the studies had some concerns of overall bias due to bias in randomization, handling of missing data or allocation concealment. There was a significant baseline difference in the Y-BOCS scores between active and sham arms in 3 studies.^22,29,44 Similarly, significant baseline differences between active and sham arms on measures of HAM-A and HAM-D scores were noted in 1 study. ⁴⁴ One study had a high risk of bias due to the non-blinding of the assessors. ³³ Four studies had a high risk of bias due to exclusion of subjects with missing data from the analysis.^35,40,43,45 The risk of bias assessment is summarized in Supplementary Figure 3.

Meta-Analysis of OCD Symptoms

The pooled estimate of Hedge's g of the 20 studies included was 0.43 (95% confidence interval [CI]: [0.20, 0.65]; z = 3.68; P < 0.001), with evidence for moderate heterogeneity (Q = 40.14; P = 0.003; I² = 52.7%). SMDs for the subgroup analysis were as follows: DLPFC: 0.57 (95% CI: [0.27, 0.87]); DLPFC & SMA: 0.08 (95% CI: [−0.79, 0.96]); SMA: 0.61 (95% CI: [−0.08, 1.29]); DMPFC: 0.29 (95% CI: [−0.12, 0.69]); OFC: 0.10 (95% CI: [−0.44, 0.65]) (Figure 2). There was no evidence for asymmetry in the funnel plot (Figure 3a), and Egger's test (β = -0.18; standard error [SE] = 0.47; t = 1.26; P = 0.22, $\tau$ ² = 2.05).

Figure 2.

Forest plot using random effects model for evaluating the effectiveness of active and sham rTMS on obsessive-compulsive symptom severity measures.

Note. DLPFC = dorsolateral prefrontal cortex; SMA = supplementary motor area; OFC = orbitofrontal cortex; DMPFC = dorsomedial prefrontal cortex.

Figure 3.

Contour-enhanced funnel plots for Yale-Brown obsessive-compulsive scale (Y-BOCS), anxiety and depression outcomes.

Meta-Analysis of Anxiety Symptoms

A total of 17 studies were included in the meta-analysis of evaluating the effectiveness of rTMS in relieving anxiety symptoms among OCD subjects and the pooled Hedge's g was 0.3 (95% CI: [0.11, 0.48]; z = 3.33; P = 0.001). The overall heterogeneity among included studies was Q = 16.97, I² = 5.7%, P = 0.39. Estimates based on the subgroup analysis were as follows: DLPFC: 0.44 (95% CI: [0.06, 0.81]); DLPFC & SMA: −0.15 (95% CI: [−1.03, 0.73]); SMA: 0.25 (95% CI: [−0.04, 0.53]); OFC: 0.11 (95% CI: [−0.30, 0.53]) (Figure 4). The funnel plot (Figure 3b) and Egger's regression test (β = 1; SE = 0.63; t = −1.14; P = 0.27, $\tau$ ² = 1.04) was not suggestive of publication bias.

Figure 4.

Forest plot using random effects model for evaluating the effectiveness of active and sham rTMS on anxiety severity measures.

Note. DLPFC = dorsolateral prefrontal cortex; SMA = supplementary motor area; OFC = orbitofrontal cortex.

Meta-Analysis of Depression Symptoms

Eighteen studies reported depression measures along with Y-BOCS. The overall pooled Hedge's g of the depression measures (HAM-D or MADRS) was 0.24 (95% CI: [0.08, 0.40]; z = 2.9; P = −0.003). The overall heterogeneity was mild [Q = 15.6; I² = 0%; P = 0.58]. The effect sizes for individual cortical targets were as follows: DLPFC: 0.31 [95% CI: (0.02, 0.61)]; DLPFC & SMA: 0.13 [95% CI: (−0.75, 1.01)]; SMA: 0.32 (95% CI: [−0.01, 0.64]); DMPFC: 0.0 (95% CI: [−0.4, 0.4]); OFC: 0.26 (95% CI: [−0.24, 0.75]) (Figure 5). The funnel plot (Figure 3c) and Egger's regression test (β = −0.24; SE = 0.32; t = 1.54; P = 0.14, $\tau$ ² = 0.82) were not suggestive of publication bias.

Figure 5.

Forest plot using random effects model for evaluating the effectiveness of active and sham rTMS on depression severity measures.

Note. DLPFC = dorsolateral prefrontal cortex; SMA = supplementary motor area; OFC = orbitofrontal cortex; DMPFC = dorsomedial prefrontal cortex; rTMS = repetitive transcranial magnetic stimulation.

Meta-Regression Analysis

We performed a meta-regression to estimate the effect of reduction in OCD symptoms on the improvement of anxiety or depressive symptoms. The effect size of Y-BOCS reduction moderated the effect size of anxiety score reduction significantly (β = 0.33; SE = 0.13; z = 2.43; 95% CI: [0.06, 0.59]; Q_M [degrees of freedom = 1] = 5.91; P = 0.01) (Supplementary Figure 1). However, the effect size of Y-BOCS score reduction did not significantly influence the effect size of depression scores (β = 0.14; SE = 0.13; z = 1.00; 95% CI: [−0.13, 0.40]; Q_M [df = 1] = 1.01; P = 0.31) (Supplementary Figure 2).

Sensitivity Analysis

The sensitivity analyses for each outcome measure were done by excluding the studies with a high risk of bias. While the pooled effect sizes of Y-BOCS scores with 13 studies (n = 414) was 0.54 (95% CI: [0.18, 0.89]) (Supplementary Figure 4), the effect size was not statistically significant for anxiety (Hedge's g = 0.27; 95% CI: [−0.04, 0.58]) and depressive symptoms score (Hedge's g = 0.16; 95% CI: [−0.05, 0.38]) with only 10 studies (n = 280) and 11 studies (n = 339) included, respectively (Supplementary Figure 5 and 6).

Discussion

To our knowledge, this is the first meta-analysis to explore the effect of rTMS on the comorbid anxiety and depressive symptoms in OCD. We found that both, comorbid anxiety (Hedge's g = 0.3) and depressive symptoms (Hedge's g = 0.24) decreased with small to medium effect sizes following active as compared to sham rTMS. Although significant heterogeneity was not observed for either analysis, subgroup analysis showed that rTMS targeting DLPFC was the only protocol that was significantly more effective than sham stimulation for decreasing both anxiety and depressive symptoms.

These findings are clinically relevant as concomitant anxiety/depression is extremely common in patients with OCD. ⁴ Most studies included in the current analysis have reported the anxiety and depressive symptom severity but not the proportion of subjects with diagnosable anxiety/depressive disorders. Hence, it is unclear whether the reported anxiety and depressive symptoms represent comorbid diagnosable disorders or subsyndromal symptoms. As subjects with severe depressive episodes were excluded in many studies, the effect of rTMS on these patients cannot be inferred from the current analysis. The mean baseline scores suggest considerable heterogeneity with scores ranging from mild to moderate severity for both anxiety and depression (Supplementary Table 1). Although post-intervention mean scores suggest a decrease in severity, the effect sizes were modest. Hence, some patients might require additional treatment for anxiety/mood symptoms. Since studies used different scales for assessing anxiety/depression severity, we were unable to assess whether the mean baseline score predicted improvement in symptoms post-intervention.

With respect to OCD symptoms, our findings are in line with other meta-analyses, showing that active rTMS is more effective than sham rTMS (Hedge's g = 0.43). Nevertheless, the overall effect size estimate was lower than that reported in earlier studies (Hedge's g = 0.64). ¹² This could be explained by the more restricted selection of studies in the current analysis as we focused only on studies that reported concomitant anxiety/depressive symptoms as outcome measures. Subgroup analysis conducted to explain the significant heterogeneity in OCD symptoms showed that DLPFC was the only target that was found to be significantly superior to sham stimulation (Hedge's g = 0.57). However, as there were few studies targeting DMPFC and OFC, the subgroup analysis may be underpowered for these targets. DLPFC, with its pivotal role in the regulation of executive functioning, plays an important role in the pathogenesis of OCD symptoms and is thus the most common target in rTMS studies. ⁴⁶

Recent meta-analyses have consistently demonstrated the effectiveness of targeting right or left or bilateral DLPFC for rTMS in OCD.^11,12 A recent network meta-analysis also showed that protocols targeting DLPFC were among the highest ranked treatments for relieving OCD symptoms. ⁴⁷ We found that stimulation of DLPFC was uniformly effective across symptom categories of anxiety (Hedge's g = 0.44), and depression (Hedge's g = 0.31) as well. rTMS targeting DLPFC, through either stimulation of left DLPFC or inhibition of right DLPFC is a well-established treatment for MDD. ¹⁴ rTMS targeting DLPFC has also been found to be effective in the treatment of various anxiety disorders and anxiety symptoms in patients with depressive disorder.^19,48

These findings add to the hypothesis of a probable shared anatomical substrate of various neuropsychiatric disorders such as OCD, generalized anxiety disorder, and MDD. Neuroimaging studies show that patients with these disorders have altered connectivity between various cortical areas (prefrontal cortex, orbitofrontal or cingulate cortex) and subcortical structures (amygdala and insula), which can manifest clinically as deficits in emotional regulation, error monitoring or fear conditioning.^49,50 However, neurobiological heterogeneity has been observed within individual disorders. For example, biotypes of MDD based on the dysfunction in resting-state activity in fronto-limbic structures were noted to predict the response to rTMS. ⁵¹

Meta-regression showed that a reduction in Y-BOCS scores could moderate the reduction in anxiety symptoms but not depression symptoms. Therefore, it is possible to assume that active rTMS may have a direct effect on depressive symptoms independent of its effect on the severity of OCD. In contrast, the anti-anxiety property of rTMS was not independent of its anti-obsessional effect possibly because the anxiety may be largely due to OCD and improvement in OCD may have resulted in improvement in secondary anxiety as well. ⁷ Thus, the anxiolytic effect of rTMS can at least partly be explained by the reduction in OCD severity.

Limitations

The findings should be interpreted cautiously due to the following limitations. Most of the studies included in the analysis have not reported the proportion of comorbid MDD or anxiety disorders in the participants. Hence, it is unclear whether the patients suffered from clinically significant symptoms. The severity of anxiety and depressive symptoms were collected as secondary outcome measures. Further, many of the individual RCTs did not show significant differences in the primary outcome measures between the groups (Figures 4 and 5), possibly because of their modest sample sizes. A large proportion of the included studies had some concerns about the risk of bias assessment. The statistical significance of the effect sizes of anxiety and depression measures was not significant after removing studies with a high risk of bias. However, this could also be explained by a lack of power as suggested by the wider confidence intervals in the sensitivity analyses due to a further reduction in sample size. The study was not prospectively registered. The data for the current meta-analysis was collected as a part of a network meta-analysis comparing different rTMS protocols for OCD, which has been registered in PROSPERO (ID: CRD42020164487 dated 28/04/2020). In addition, most of the studies were conducted on treatment-resistant subjects and hence the findings may be more relevant to this population.

Conclusion

The current meta-analysis suggests that active rTMS protocols targeting DLPFC are superior to sham stimulation in improving comorbid anxiety, and depression symptoms in patients with OCD. Furthermore, the antidepressant effect of rTMS might be unrelated to OCD symptom improvement, unlike anxiety symptoms. However, the findings have to be interpreted cautiously due to the possible risk of bias and the modest sample size of the individual studies. Larger sham-controlled trials in OCD patients with comorbid anxiety or depressive disorders are required to substantiate the clinical significance of these findings. Clinical trials comparing rTMS protocols targeting various cortical targets and correlating the effect of various outcomes with neuroimaging/electrophysiological markers will further enhance our understanding of the neurobiological basis of OCD and its comorbidities.