Therapeutic Neurostimulation in Obsessive-Compulsive and Related Disorders: A Systematic Review

Invasive and noninvasive neurostimulation therapies for obsessive-compulsive and related disorders (OCRD) were systematically reviewed with the aim of assessing clinical characteristics, methodologies, neuroanatomical substrates, and varied stimulation parameters. Previous reviews have focused on a narrow scope, statistical rather than clinical significance, grouped together heterogenous protocols, and proposed inconclusive outcomes and directions. Herein, a comprehensive and transdiagnostic evaluation of all clinically relevant determinants is presented with translational clinical recommendations and novel response rates. Electroconvulsive therapy (ECT) studies were limited in number and quality but demonstrated greater efficacy than previously identified. Targeting the pre-SMA/SMA is recommended for transcranial direct current stimulation (tDCS) and transcranial magnetic stimulation (TMS). TMS yielded superior outcomes, although polarity findings were conflicting, and refinement of frontal/cognitive control protocols may optimize outcomes. For both techniques, standardization of polarity, more treatment sessions (20), and targeting multiple structures are encouraged. A deep brain stimulation (DBS) ‘sweet spot’ of the striatum for OCD was proposed, and CBT is strongly encouraged. Tourette’s patients showed less variance and reliance on treatment optimization. Several DBS targets achieved consistent, rapid, and sustained clinical response. Analysis of fiber connectivity, as opposed to precise neural regions, should be implemented for target selection. Standardization of protocols is necessary to achieve translational outcomes.


Introduction
Obsessive-compulsive disorder (OCD) is characterized by distressing thoughts (obsessions) and repetitive mental or behavioral acts (compulsions). The most recent edition of the Diagnostic and Statistical Manual (DSM) for Mental Disorders [1] broadened OCD and related disorders to encompass: body dysmorphic disorder (BDD), i.e., a preoccupation with perceived flaws of defects in physical appearance or body build with repetitive behaviors or mental acts in response to these; hoarding disorder (HD), i.e., persistent difficulty and distress associated with discarding possessions, with excessive accumulation and/or acquisition of the latter; trichotillomania, i.e., recurrent, obsessive hair-pulling; and excoriation disorder, i.e., recurrent skin-picking with resultant skin lesions. Whilst Tourette Syndrome (TS) is classified as a tic rather than an OCRD in the DSM-5, significant obsessive-compulsive symptomatology is present in almost half of affected people [2]. Moreover, there is considerable genetic contribution in obsessive-compulsive symptomatology, Tourette's Syndrome, and hoarding, with covariation of phenotypes ranging from 50.4% to 61.1% [3]. The syndromal aggregation of disorders in the DSM-5 is based on a natural history of symptom similarities outweighing differences.
There is a substantial need for novel effective treatment interventions for OCD and related disorders. Even when the best available treatments are applied, approximately 30-40% of people experiencing OCD do not respond adequately [4], and 10% remain severely afflicted [5,6]. Treatment-refractoriness in these disorders is variably defined, but usually centers on suboptimal response to adequate therapeutic trials of serotonergic reuptake inhibitor antidepressants (selective serotonin reuptake inhibitors, (SSRIs) or clomipramine), as well as evidence-based psychological treatments such as cognitivebehavioral therapy (CBT) undertaken by experienced clinicians.

Psychopathology
Despite inconsistencies in structural and functional neuroimaging findings, hyperactivity of cortico-striato-thalamo-cortical (CSTC) circuits involving a central role of the basal ganglia is the prevailing neurobiological model of OCD pathogenesis [7][8][9]. Further explication of this model includes the orbitofrontal cortex (OFC), that integrates limbic and emotional information into behavioral responses, and the anterior cingulate cortex (ACC), that is integrally involved in motivated behaviors. Prefrontal goal-orientated behaviors and reward processing are thought to be modulated by the Nucleus Accumbens (NAc), which is at the motor-limbic interface of cortico-subcortical circuits. The NAc may also integrate contextual information from the hippocampus and emotional information from the amygdala. In OCD, there is hyperactivity of the caudate nuclei and OFC, as evidenced by increased metabolism in PET scanning [10]. Impairments in the neuropsychological function of executive control, memory, visuospatial abilities, attention, and processing speed are prominent in OCD and related to frontostriatal dysfunction [11]. Despite avoidance of anxiety provoking stimuli and engagement in compulsive rituals, classical conditioning mechanisms can maintain or exacerbate obsessive-compulsive symptomology [12]. Thus, OCD has a complex clinical profile involving bio-psycho-social determinants. Moreover, the central role of the prefrontal cortex in OCD has led to this region being identified both as a functional neuroimaging marker of the disorder and a therapeutic target [13].
Whilst altered functioning of the CSTC circuits and the ACC have been consistently highlighted in OCD, frontostriatal hyperactivity may also be associated with obsessive thoughts and compulsive behaviors in BDD [14]. The ACC is also involved in the neurocircuitry of BDD, along with right OFC, thalami, hippocampi, and amygdalae [15]. Unlike OCD, occipital hypofunction may contribute to core psychopathology in BDD [14]. There is also preliminary evidence of reduced local amygdala connectivity in BDD [16]. TS also involves altered frontostriatal circuitry [17,18], but predominantly involving the motor and premotor cortices and dorsal striatum [17], potentially explaining why motor tics are more common in TS than the other OCRD. Similarly, trichotillomania and skin picking disorder involve altered frontal-striatal circuits and decreased prefrontal control, as well as motor and reward regions [19].

Neurostimulation
Given that a significant proportion of people with OCRD are refractory to extant evidence-based treatments, the etiology and neurocircuitry of OCD and related disorders support consideration of neurostimulation as a means of targeting particular areas of the brain via activation or inhibition, and, in turn, modulation of neurocircuitry and neuroplasticity farther afield [20]. Furthermore, neurostimulation has the potential to personalize treatments based on patient-specific neural substrates and the complex interactions among different neural circuits.
In light of the evolving knowledge of neurocircuitry in OCRD and the ongoing treatment gap for these disorders, we systematically reviewed neurostimulation modalities in these disorders, and herein address their current and potential role in the continuum of care. We specifically included OCD, BDD, trichotillomania, excoriation disorder, HD, and TS in order to test for between-disorder similarities and differences. Reviewed data were collated with the aims of (1) reviewing the clinical significance of neurostimulation therapies for OCD and OCRD; (2) making clinical recommendations for optimal stimulation protocols; (3) reporting on limitations and strengths of current neurostimulation protocols in order to (4) identify future directions for greater consistency and validity of treatment effects. Figure 1 shows a graphical abstract in the form of a flow diagram of the techniques, conditions, and neurostimulation targets identified. A brief summary for each technique, and all investigations is provided.

Protocol
The Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement was followed for the review process [21]. The protocol was registered with the international prospective register of systematic reviews (PROSPERO) registration number: CRD42020171054.

Search
The databases PubMED and CINAHL were searched with the following terms: (obsessive compulsive disorder OR body dysmorph* OR hoarding disorder OR skin picking disorder OR excoriation disorder OR trichotillomania OR hair-pulling OR Tourette*) AND (neurostimulation OR deep brain stimulation (DBS) OR electroconvulsive therapy (ECT) OR transcranial direct current stimulation tDCS OR vagus nerve stimulation (VNS) OR transcranial magnetic stimulation (TMS)). The search was conducted with no specified start date up until 1 January 2020. Citations of identified articles and recent systematic reviews [22][23][24][25][26][27][28][29][30][31][32] were checked for additional articles.

Inclusion and Exclusion Criteria
Inclusion criteria: (1) Investigation of any neurostimulation intervention (DBS, ECT, TMS, tDCS) in patients with a primary diagnosis of OCD or OCRD (BDD, trichotillomania, excoriation disorder, HD) or TS, with no adjunct treatment except pharmacology. (2) Assessment at pre-and post-treatment using a standardized outcome.
(3) English language literature. (4) Peer-reviewed article with primary data including randomized control trials (RCTs), open-label (OL) trials, multisite studies, case studies and letters to the editor.
If stimulation parameters, diagnostic method, or response criteria were not stated, or a retrospective design was implemented, this was not an exclusion but a limitation, and was considered in the quality assessment.
Exclusion criteria: (1) Investigation of patients that did not have a primary diagnosis of OCD or OCRD, or the primary diagnosis was unclear. (2) Comorbid severe psychiatric condition including schizophrenia, catatonia, bipolar or psychosis. Common comorbidities that did not warrant exclusion included major depressive disorder (MDD), anxiety, attention-deficit-hyperactivity-disorder (ADHD), and personality disorder.  As per the PRISMA method, a total of 1756 records were identified, and an additional 68 were identified through reference lists and previous review articles. After duplicates were removed, 886 records were screened by title and abstract by one author (T.P. or N.A.); 254 were deemed eligible. These 254 full text articles were reviewed for eligibility by two authors (T.P. or N.A.), and 153 were included in the final qualitative synthesis, at which stage 101 were excluded (see Supplementary File S1, S1 PRISMA diagram). Any discrepancies were resolved by a third author (S.R.). Across each neurostimulation technique, insufficient or no protocols were comparable, and thus, quantitative synthesis was not appropriate.

Data Extraction
The following data were extracted from the selected articles: clinical characteristics, demographics, methodology, stimulation protocols, outcome measures, clinical and statistical significance, and adverse events. We also provided commentary on the limitations and strengths of each study.

Risk of Bias Assessment
Items from the Cochrane Collaboration Risk of Bias (RoB) tool [33] that were relevant were used to evaluate risk of bias, and are presented in Supplementary File S2, S2 Risk of Bias Table. 2.6. Quality Assessment No quality assessment tool previously implemented in a neurostimulation review that was appropriate for the current protocol was identified. The Oxford Quality Scoring System [34] and the Cochrane Grading of Recommendations Assessment, Development and Evaluation (GRADE) tool [35] were thus adapted to develop a tool in line with the current protocol, variables, and outcomes, and are presented in Supplementary File S3, S3 Quality assessment table.

Reporting of Data
The primary clinical outcomes were reported and discussed for each neurostimulation technique. The primary outcome for OCD articles was the Yale Brown Obsessive Compulsive Scale (YBOCS), which is a clinician rated scale that provides a symptom severity score for obsessions (0-20), compulsions (0-20), and an overall score (0-40). The primary outcome for TS was the Yale Global Tic Severity Scale (YGTSS), which is a semi-structured interview that assesses motor tics, phonic tics, total impairment, and total severity. The YGTSS provides a tic severity score (motor tic severity + phonic tic severity, 0-50), and/or a global severity score (tic severity+ tic impairment, 0-100). The primary outcome for BDD was the Body Dysmorphic Disorder-YBOCS (BDD-YBOCS), adapted from the YBOCS scale to assess preoccupation, insight, and avoidance, as well as obsessions and compulsions; it generates an overall score (0-48). The primary outcome for skin picking was the Neurotic Excoriation-YBOCS (NE-YBOCS), which provides a score for thoughts and behavior related to tics (0-20) and a total score (0-40). The primary outcome for hoarding disorder was the Saving Inventory-Revised (SI-R), which assess difficulty discarding, clutter and acquisition; it generates sub-scores and an overall score (0-92). All scales assess symptoms in the previous 7 days, or 7-10 days in the case of the YGTSS.
Findings were considered in the context of clinical rather than statistical significance, as the two often do not correspond and the former is more sensitive and ecologically valid. Pallanti defines full treatment response as ≥35% reduction in YBOCS, and a partial treatment response as ≥25% reduction in YBOCS [36]. If a different treatment response criterion was reported within articles, then this was used to report responders; if not, the widely accepted criterion was used [36]. In the case of DBS, a full response is most often categorized as ≥40% reduction in YBOCS, which was implemented in the absence of a defined treatment response.
Results were categorized for each technique, then condition and target, followed by a discussion, before reporting results on the consecutive technique. Prior to reporting the results of each technique, a synopsis on the proposed mechanisms of action, important considerations for stimulation protocols, and methodological limitations were described. This structure was implemented in order to present the data coherently, and for ease of interpretation, i.e., to present important theoretical and technical insights; to discuss parameters of stimulation protocols; to allow for transdiagnostic interpretation; and to account for disparities and limitations in methodology which were specific to each technique. There were a range of targeted regions in the TMS and DBS studies; thus, the results were discussed in the context of each target as well as common themes identified for these techniques. Lastly, a general discussion provides clinical recommendations across all techniques and conditions within the context of CSTC circuitry. Due to the breadth of conditions and techniques, an in-depth analysis of secondary clinical outcomes was beyond the scope of the review. Rather, notable findings of depression and anxiety in relation to neurostimulation-mediated change in obsessive-compulsive (OC) symptoms were reported.

Results and Discussion
A brief report of cohort studies is provided herein. Refer to the supplementary results for an in-depth report of all protocols for tDCS, TMS and DBS, including methodology aspects (i.e., blinding, placebo effects, treatment dose and titration), stimulation parameters, anatomy, and individual response patterns, which are integral to the discussion.

ECT Results
ECT involves the application of an electric charge via electrodes placed on the scalp to induce transient generalized seizures for therapeutic benefit [37]. Beyond the induction of generalized seizures, ECT does not target specific areas of the brain.
Thirty-six articles were screened for eligibility, and 10 were included in the final synthesis: 6 for OCD and 4 for TS. Of the 26 articles excluded, 15 lacked a standardized outcome measure of primary symptoms; in eight, the primary diagnosis was not an OCRD or this was unclear; in a further two, both limitations were present; finally, in one investigation two neurostimulation therapies were applied. Of the included articles, OCD articles consisted of one retrospective review of medical records, four case series, and one case report, whilst TS articles consisted of four case reports. No RCT or prospective investigation of a cohort with greater than five patients was identified (across included and excluded articles), and most articles reported on a single patient. Further, clinicaltrials.gov was searched for potential RCTs that had not yet been published or identified, but no results were obtained. The final sample therefore included 46 OCD and four TS patients. Table 1 shows summary results of ECT investigations for OCD and Table 2 shows summary results of ECT investigations for TS.
Quantitative follow-up outcomes were reported in 43% of case studies (6/14 patients). In one study [41], three out of five patients were responders following treatment; only one remained a responder at 3-and 6-month follow-up, according to the clinical global impression (CGI) scale (not YBOCS). Another patient relapsed to baseline twice following consecutive cycles of ECT [42].
Depression symptoms showed clinically significant improvements of 48-62% in two investigations [38,40], and were not reported in the other articles.

ECT Results for TS
Improvement rates between 83-100% were reported; thus, all four patients achieved clinical response [44][45][46][47]. Qualitative follow-up outcomes were reported in three cases: all experienced complete remission with no relapse up to 8 months following treatment [44][45][46], yet no studies reported quantitative (i.e., YGTSS) outcomes at follow-up. Depression outcomes were not reported, despite two patients reportedly experiencing comorbid depression.

ECT Discussion
We identified variable response rates to ECT in OCD patients, with a lack of quantitative evidence of long-term efficacy. In contrast, TS patients consistently showed substantial response and remission rates, even though only four patients were reported on. The risk of bias was low for all articles (S2); the quality assessment rated no studies as good, seven as moderate, and three as poor (S3).
Quality ratings were impacted by a lack of RCTs and cohort studies, poor reporting of clinical demographics and stimulation parameters, and a lack of quantitative followup. Transient effects of general fatigue and short-term memory loss were reported for some patients.
OCD patients had heterogeneous symptom profiles, including obsessions of sexuality, persecution, checking, cleaning, slowness, contamination, and pathological doubt. Illness duration varied from abrupt onset of 2.5 months to 13 years, and treated individuals were between 18-47 years of age, (demographics missing for some patients). TS patients had similar complex motor and vocal tics, coprolalia, and self-injurious behaviors among other impairments; illness duration was 8-30 years, and age at ECT was between 18-36 years.   mA, milliamps; mC, milicoulombs; OCB, obsessive compulsive behavior; Rx, patients continued taking prescribed medication; SIB, self-injurious behavior; y = years; + = clinically significant change from baseline; = criterion applies; = not reported.

Pattern of Response
Within OCD studies, 79% of case reports were responders, with change between 43-95%; and within a single cohort, 42% change was achieved. Previous reviews have reported 'positive response' rates of 60% [26] and 73% [24] for ECT in OCD but acknowledged that only 17% of studies employed standardized assessments. There was a lack of evidence beyond subjective accounts of long-term efficacy. It was stated that patients remained 'symptom free' or 'maintained response' up to 4 years following treatment, in the absence of quantitative evidence. In line with the current findings, previous reviews also demonstrated a lack of standardized follow-up outcomes, which were often not reported (50-66%). When follow-up was reported, relapse or deterioration occurred in 35-55% of cases [24,26]. This highlights the vagueness and subjectivity of clinical findings, as a 'positive response' could indicate that the patient subjectively reported they felt better, or it could mean the patient had a 95% improvement assessed by a clinical tool; these disparate scenarios cannot be grouped and interpreted as proportional.
The only cohort study [38] reported that improvement was maintained in 42% of patients' post-treatment to 35% at 12-month follow-up; only 16% (5/32) received maintenance ECT due to deterioration, which indicates a greater level of sustained efficacy than reported in case reports. Within the cohort study, all patients resumed CBT, which may have maintained the therapeutic effect of ECT. The primary outcome was the Maudsley Obsessional Compulsive Scale (MOCI), as opposed to the YBOCS. Nevertheless, the cohort was highly treatment resistant and chronic, with an average disease history of 27 years, and seven medication trials. Thus, with adequate clinical follow up and adjunct treatment, highly resistant OCD patients may achieve long-term symptom improvement from ECT.
TS patients responded consistently to ECT (100% responded and 75% reached remission), and there was no reported deterioration following treatment cessation. However, this conclusion is in the absence of follow-up standardized assessments and is based on only four patients.
There were no predictors of treatment response due to inconsistencies in OCD patients and a ceiling effect of clinical efficacy in TS patients. Although ECT had a differential effect on depression based on response, comorbid depression was not correlated with YBOCS response [38].

Treatment Dose
A treatment-efficacy dose relationship could not be estimated due to the very small sample size, as well as the fact that some studies did not specify doses. The number of sessions in the OCD and TS cohort varied from 3-14 and 6-37 sessions, respectively. Treatment duration was not always specified for OCD patients, and was estimated at a maximum of 5 weeks, whereas TS patients underwent treatment for 5 weeks to 5 years. In the cohort study, the range of sessions administered was not described, only the average of 3.5 treatments. In the TS studies, 50% included maintenance ECT, and in one study, this was extended for 5 years. The patient with comorbid TS and OCD completed only six sessions and experienced complete remission of OCD and almost complete remission of TS.
OCD patients had a more variable clinical response to ECT, but responders required fewer treatments. Whilst TS patients consistently responded well to ECT, they required longer treatment regimes. Although there was no relapse reported in the TS studies, it is unclear whether maintenance ECT was administered due to clinical deterioration or was predefined.

tDCS Results
tDCS generates a week, direct current to the scalp between two electrodes and induces polarity dependent changes in the resting membrane potential at a subthreshold level [48]. The polarity effects of tDCS on neural elements are postulated to have an anodal-excitation and cathodal-inhibition (AeCi) relationship. Theories on the mechanisms of action are based on investigations at low doses (~10 min, 1 milliampere, mA) over the motor cortex in healthy individuals. Polarity effects are likely to differ with consecutive applications of relatively high doses (>1 mA, >10 min) on brain regions involved in cognitive and emotional processes, in which transmitter availability and brain function are altered in psychiatric disorders.
Theoretically, tDCS is capable of inducing lasting behavioral changes mediated through synaptic plasticity mechanisms of long-term potentiation (LTP) and long-term depression (LTD); this notion is supported by behavioral, neurochemical [49], and neuroimaging evidence [50]. Neuromodulation is not restricted to the region under the electrodes, and widespread cortical and subcortical modulation can occur [50]. Modulation of discrete cortical excitation and diffuse network connectivity is of particular importance in targeting OCRDs, which are underpinned by hyperactive nodes and network imbalances. tDCS therapy presumably restores pathological excitatory/inhibitory imbalances.
Twenty-six articles were screened for eligibility, with 19 being included in the final synthesis: 15 for OCD and 4 for TS treatment. Four articles were excluded due to a lack of standardized assessment of primary symptoms, two because the primary diagnosis was not an OCRD or was unclear, and one because immediate post-treatment outcomes were not reported. Of the included articles, OCD investigations consisted of five case reports, five open-label trials, three double blind RCTs (two sham controlled), two case series, and one double blind case report. TS investigations consisted of two case series/pilot data of a double blind RCT, one case series, and one case report. The final sample included 148 OCD patients and 8 TS patients. Tables 3 and 4 show summary results of tDCS investigations for OCD, and Table 5 shows summary results of tDCS investigations for TS.
DLPFC investigations included an open-label trial and two case studies. Improvement of 65% occurred on a cohort level from cathodal tDCS of bilateral DLPFC [55]. Dual targeting of the right OFC (cathodal) and left DLPFC (anodal) led to 23% improvement [52].
OFC investigations included one RCT, two open-label trials and two case studies. Open-and closed-label cathodal tDCS of the left OFC led to 5% and 20% improvement, respectively [51].
Open-label transcranial alternating current stimulation (tACS) over the bilateral frontotemporal region led to 52% improvement, 86% full response, and 100% partial response [54].   Cathodal tDCS over the left DLPFC had no effect on OC symptoms. Depression and anxiety symptoms were in the moderate range at baseline and respectively improved by 34% and 17.8% after active; and improved by 14% and declined by 12% following sham. Following a 2 week wash out, 10 sessions of inhibitory rTMS of primary motor cortex had no effect on OC symptoms within a sham-controlled context. Anxiety and depression increased by 30% and 12% respectively after active TMS.  GAD, generalized anxiety disorders; SAD, social anxiety disorder; y = years; + = clinically significant change from baseline; = criterion applies; = not reported.

tDCS Results for TS
TS investigations included four case reports of cathodal tDCS over motor regions. tDCS of pre-SMA or SMA led to improvements between 20-43% [67]. Stimulation of both the pre-SMA and SMA led to incongruent outcomes, with improvement in the YGTSS but dramatic increase in tic counts [68]. tDCS of the primary motor cortex led to 11-30% improvement [66].

tDCS Discussion
The risk of bias was low for 15 articles and medium for four (S2); the quality assessments rated six as good, 13 as moderate, and one as poor (S3). Quality assessments highlighted a lack of RCTs with a sham control (only two studies met criteria), a failure to report the number of clinically significant responders (four met criteria), and a lack of follow-up outcomes (10 met criteria). Statistical significance was often the focus of the results, which has limited clinical relevance; for example, mean change as little as 4.7% in 20 patients was sufficient to yield statistical significance [56], yet arguably, this does not reflect clinical significance.
OCD patients had failed between one and five medication trials, and case studies exemplified higher treatment resistance in relation to cohort studies. Several comorbid mood disorders were reported and did not appear to interfere with efficacy. Illness duration in the case studies ranged between 1-23 years of illness, and type of obsessions varied (predominantly of contamination, symmetry, aggression, sexuality, and guilt). TS patients had disparate baseline severity, and illness duration ranged between 4-45 years. Information on symptom domains and the definition of treatment resistance was lacking.
When reported, all except for six OCD and two TS patients were taking prescribed medication; this did not appear to impact treatment response, although SSRIs can enhance and prolong the facilitation effect of anodal tDCS [69]. Depression symptoms improved by 10-87% and anxiety symptoms by 17-100% in OCD patients and were not reported in TS patients.
OCD protocols targeted the pre-SMA, SMA, DLPFC and OFC; yet the return electrode placement-which determines the distribution of current flow and is a critical determinant of network effects [70]-was variable. Stimulation involved 2 mA for 20 min, except for two studies in which it was 30 min [55,71], across 8-20 sessions with varying polarities and hemispheres. Protocols for TS articles targeted different motor regions including the pre-SMA, SMA and motor cortex. The return location and polarity was consistent. Stimulation was applied between 1.4-2 mA for 20 or 30 min across 10-18 sessions.

OCD, Pre-SMA and SMA Targets
Within-group and within-patient analysis showed a superiority of cathodal stimulation [53,63], yet randomized controlled and open-label studies revealed only modest effects of cathodal tDCS with 17-29% symptom improvement [53,57,58,63]. Anodal stimulation was directly investigated only in case reports: three patients responded with symptom change of 40-70% [61,64]. Benefit from anodal tDCS over the pre-SMA/SMA occurred after twice daily sessions rather than daily sessions. Polarity comparison studies showed superiority of daily cathodal sessions [53,63]. Thus, a perceived superiority of cathodal stimulation may have biased previous reviews to conclude ineffectiveness of anodal stimulation [72].
Cathodal and anodal polarity after-effects are linear for intensities up to 1 mA and durations up to 13 min; however, opposing polarity effects may not be apparent beyond these parameters. Cathodal and anodal stimulation at 2 mA can enhance cortical excitability for 120 and 60 min, respectively, returning to baseline after 6-8 h [73]. The timing between consecutive tDCS sessions is also an important determinant of the cumulative polarity effect. A second application of tDCS during the after-effects of the first can enhance and prolong the polarity effects on cortical excitability, whereas a second application outside the after-effects of the first can attenuate or abolish the polarity effects [74,75]. In turn, applications within this review of 2 mA cathodal tDCS may lead to cortical excitability that is enhanced by a second daily session. Cathodal SMA stimulation may be theoretically optimal [76], although twice daily anodal stimulation may cause greater inhibition and appears to be therapeutically optimal.
Follow-up data was lacking in many studies. In those four patients for whom followup was performed, one responder (anodal stimulation) further improved at one week follow-up [61], whilst one partial responder (cathodal stimulation) further improved at three-month follow-up [53]. Two patients who had not responded to cathodal stimulation did show improvement six months later [65].
The pre-SMA/SMA is involved in mediating important cognitive functions, particularly response inhibition, monitoring of conflicting information, and the control of internally triggered movements [77]. SMA dysfunction in OCD leads to reduced inhibition of the striatum, resulting in striatal hyperactivity, manifesting as inhibitory dysfunction related to intrusive ideas and ritualistic behaviors [78]. Modelling of a montage over the pre-SMA is predicted to modulate the medial PFC to the striatum [76], inclusive of the critical cortical and subcortical regions implicated in OCD pathophysiology. Thus, establishing the optimal stimulation montage for the pre-SMA/SMA can potentially alleviate hyperactivity in several regions and suppress intrusive thoughts and behaviors involved in OCD.

OCD, DLPFC Target
Investigations of DLPFC implemented disparate montages, studies failed to report clinically significant responders, and there were no randomized or controlled protocols. Thus, conclusions that can be drawn are limited. The open-label cohort study implemented an extensive stimulation montage (bilateral DLPFC cathode contacts and three anode/return contacts), with 65% symptom improvement, rising to 81.5% at three-month follow-up [55]. Case reports did not demonstrate efficacy to anodal stimulation of the left and cathodal stimulation of the right DLPFC [71], or cathodal stimulation of the left DLPFC [59].
There was insufficient evidence to establish the treatment parameters (i.e., intensity, duration, number of sessions or electrodes) that are the most important determinants of clinical outcomes. Across investigations of the DLPFC, 10 sessions were insufficient to induce symptom change [59], whilst 15 or 20 sessions appeared to be sufficient to reach a partial response but not to maintain efficacy [52,71]; however, 15 sessions with increased stimulation electrodes, and thus area of current, were shown to be highly effective [55].
Thus, it appears that at least 15 sessions of cathodal DLPFC stimulation should be applied to achieve a clinical response, and greater stimulation intensity and current distribution may be required for response to be optimized and maintained.

OCD, OFC Target
Post-treatment efficacy was comparable between open-and closed-label conditions [51,56], in which 22% symptom improvement occurred, and 33% response achieved. However, the open-label trial provided valuable insight that efficacy was greatest at three-month follow-up. Cathodal tDCS of the left OFC and anodal tDCS of the right OFC showed a delayed partial response in one patient [60] and a dramatic clinical response in another [62]. Two patterns emerged: (1) greater response to cathodal tDCS of the left OFC was associated with less treatment resistance [56]; and (2) all investigations showed further improvement at follow-up, identifying a delayed and prolonged pattern of response.

Transcranial Alternating Current
Klimke (2016) demonstrated a dramatic clinical response in six out of seven OCD patients using bilateral fronto-temporal tACS [54]. Further, the results were achieved with a fraction of the stimulation intensity typically employed in tDCS protocols (0.65 mA vs. 2 mA). Rather than the direct current used in tDCS, tACS produces sine-wave stimulation at a frequency of interest. It is theorized that tACS leads to manipulation and entrainment of intrinsic cortical oscillations in a frequency-specific manner [79]. It is known that phase and frequency are fundamental parameters of neuronal function, and thus LTP; also, endogenous oscillations can entrain to extrinsic rhythmic stimuli [80]. Thus, tACS has relevance for implicit learning, and follow-up data would be valuable to establish a possible learning effect. Applications of tACS are limited and novel, and sham controlled investigations of the current montage are warranted.

TS, Motor Targets
Four case studies of cathodal tDCS of motor regions yielded mixed outcomes. The pre-SMA and SMA showed superiority, but even so, only two of six patients achieved a response, with overall improvements ranging between 6-41%. Follow-up data were available for three patients and showed that response was maintained for up to six months. Cathodal tDCS of the motor cortex in two patients did not lead to clinical response; however, the number of sessions was likely too low (five) to lead to behavioral change. The case studies that achieved higher clinical efficacy [25,67] implemented daily cathodal tDCS sessions of the pre-SMA/SMA at the relatively low stimulation dosage of 1.4 mA.
SMA activity is positively correlated with tic symptom severity in TS, and thus, is implicated as a neural substrate [18]; yet distinction between an initiation and suppressive role has not been established. It is concluded that a low dose (~1.4 mA) of daily cathodal tDCS over the pre-SMA and SMA leads to clinical response and maintained efficacy in case studies in TS; however, randomized controlled evidence is required.

TMS Results
TMS produces a rapidly changing magnetic field perpendicular to the plane of the coil that penetrates the skull and induces an electric field perpendicular to the magnetic field [81]. In turn, axons within underlying cortical and subcortical regions depolarize and activate neural circuits whilst interacting with spontaneous oscillatory rhythms [82]. Low frequency (LF, 1 Hz) and high frequency (HF, ≥5 Hz) rTMS respectively cause inhibition and facilitation of axons within the motor cortex and are assumed to have the same effect on nonmotor functions [83]. The sustained effects likely depend on NMDA and GABA receptors, with LTP and LTD, although the precise mechanisms are unknown [84]. The effects of rTMS depend inter alia on the orientation of the coil, the stimulation protocol (frequency, intensity, and pulse parameters), the resting brain state, and the axonal threshold of activation. Modulation may involve local and distant, excitatory and/or inhibitory axons. Low stimulation intensities will activate a limited number of neurons, activation is more likely if neurons are close to the firing threshold, and currents parallel to the axon are more likely to reach activation [85].
Novel devices and stimulation protocols have been implemented into clinical research to achieve focal targeting. Neuronavigation implements imaging to localize a withinsubject target in order to overcome differences in anatomy. Figure-eight coils concentrate current by two-fold in the mid-point of the two loops, compared to circular coils. Deep TMS coils (double cone or H coils) allow a slow fall-off of the intensity of the magnetic field, thus achieving greater strength and depth of penetration [81]. Alpha frequency rTMS is based on the notion that various mental disorders express abnormal EEG power band frequencies, and alpha (α) has shown to be abnormal in prefrontal and temporal lobes in OCD [86]. Lastly, theta burst stimulation (TBS) is typically applied for shorter periods (1-3 min) at lower intensities and can induce rapid and long lasting inhibitory (continuous TBS) or excitatory (intermittent TBS) cortical changes [81].
The neural effects of rTMS are accompanied by sensory effects, predominantly a loud clicking noise and skin sensations. This poses difficulties with blinding. Successful blinding does not guarantee placebo effects are controlled for; the sham condition should fulfil certain technical aspects to minimize disparities across conditions. An ideal placebo condition should (1) not result in cortical stimulation, (2) produce sensory sensations identical to those of real stimulation, and (3) be positioned the same way as the active coil [87]. There is a trade-off between the absence of neural effects and presence of somatic effects. In the articles reviewed here, sham methods were assessed through the RoB assessment (S2).
Sixty-seven articles were screened for eligibility, with 48 being included in the final synthesis. Of these, 36 were for OCD, seven for TS, three for comorbid TS and OCD, one for skin picking, and one for hoarding disorder. Eighteen articles were excluded due to a lack of standardized assessment of primary symptoms (n = 6), adjunct behavioral therapy (n = 5), lack of original data (n = 3), the primary diagnosis not being an OCRD or being unclear (n = 2), methodological concerns with the stimulation protocol (n = 1), and outcomes not being reported directly from the patient (n = 1). Of the excluded articles, there were six RCTs, four open-label trials, four case reports, two retrospective studies, one case series, and one review article. Of the included articles, OCD investigations consisted of 21 RCTs (18 double blinded single-site, two multisite double blinded, one single blinded), four open-label trials, five case reports, two investigations as standard clinical care, one single blind nonrandomized trial, two retrospective reports, and one case series. TS investigations consisted of three RCTs (one each multisite double blinded, single-site double blinded, and single blinded), one pilot RCT, three open-label trials, and one case series. In addition, there was one open-label trial, one case report, and one case series for patients with comorbid OCD and TS. The skin picking investigation was a double-blinded RCT, and the hoarding disorder investigation was a case report. The final sample included 952 OCD patients, 92 TS patients, 14 skin picking patients, three patients with comorbid OCD and TS, and one hoarding disorder patient. Tables 6 and 7 show summary results of rTMS investigations for OCD, Tables 8 and 9 show summary results of rTMS investigations for TS. Tables 10 and 11 show summary results of rTMS investigations for skin picking disorder and HD.
Stimulation of the pre-SMA and SMA was investigated in four RCTs (one multisite), one randomized open-label trial, three open-label trials, and four case studies; all employed LF rTMS. Closed-and open-label investigation achieved symptom improvement between 23-49%, and 1 trial led to 8% change [96,107,109,113] increasing the treatment sessions by two-fold changed improvements from 25% to 49% [96]. TMS targeting of the SMA for OCD symptoms and right DLPFC for comorbid MDD led to a high response rate of 83% and 42% improvement [111]. TBS treatment yielded 13% improvement [117].
Investigations of the OFC included two RCTs and a retrospective study, all using LF rTMS. Symptom improvements of 19%-24% were achieved and maintained at 4-10 weeks [94,115], but not within a trial using deep TMS [101]. When patients were prescribed LF rTMS of the SMA or OFC based on their preference, 27% improvement was reached overall [118].       A, active; dmPFC, dorsomedial prefrontal cortex; Hz, Herts; L, left; mPFC, medial prefrontal cortex; RMT, resting motor threshold; R, right; Rx, patients continued taking prescribed medication; S, sham; TAU, treatment as usual; + = clinically significant change from baseline; * = statistically significant change from baseline; # = statistically significant change compared to the control condition; = criterion applies; = not reported.ˆ= Seo et al., 2016 [110] reported YBOCS score change not percentage change;~= outcomes were not reported and inferred from graphical reporting. cTBS, continuous theta burst stimulation; Hz, Herts; iTBS, intermittent theta burst stimulation; P, participant; RMT, resting motor threshold; Rx, patients continued taking prescribed medication; y, years; + = clinically significant change from baseline; * = statistically significant change from baseline; = criterion applies; = not reported.  A, active; AMT, active motor threshold; cTBS, continuous theta burst stimulation; Hz, Herts; L, left; RMT, resting motor threshold; R, right; Rx, patients continued taking prescribed medication; S, sham; * = statistically significant change from baseline; = criterion applies; = not reported. Hz, Herts; P, Participants; MOVES, motor tic, obsessions and compulsions, vocal tic evaluation survey; RMT, resting motor threshold; Rx, patients continued taking prescribed medication; SIB, self-injurious behavior; + = clinically significant change from baseline; = criterion applies; = not reported.  LF and HF rTMS of other prefrontal regions were investigated in two RCTs (one multisite), two open-label trials, and one case study. Deep TMS of the medial PFC and ACC led to 44% and 45% improvement across a pilot and multisite study [114,116]. LF and HF rTMS of the medial PFC led to comparable outcomes of 39% and 40% improvement, respectively [105,106].

TMS Results for TS
Investigations for pre-SMA and SMA included two RCTs, three open-label trials and three case studies, all of which implemented LF rTMS. Closed-label conditions led to response in 33% and 50% when implementing continuous TBS [128,129], while open-label trials led to 34%, 30% and 5% improvement [126,127,130]. Additional treatment led to greater efficacy, and when patients were followed up, outcomes were maintained at 3 and 6 months.
Two small RCTs targeted the premotor and motor cortex, which led to no change and 29% improvement, respectively [124,125].

Other Conditions
A small trial of LF rTMS of the pre-SMA for skin picking led to a 36% improvement, followed by a 70-110% decline [133]. In one study, hoarding disorder patients achieved 30% improvements from LF rTMS of the right DLPFC [134].

TMS Discussion
For the TMS studies reported here, the risk of bias was low for 20, medium for 21, and high for seven studies (S2); the quality assessments classified 28 articles as good, 19 as moderate, and one as poor (S3). Within the quality assessment, the main methodological limitation was lack of follow-up (19 articles met the criteria), as well as a lack of controls and adequate blinding in a number of studies. The lack of follow-up outcomes is an important limitation, as rTMS effects are often slow and progressive [135], and the onset and duration of the effect is critical for insight into clinical care. Of the included articles, 43% (20/49) reported dropouts: out of the entire sample, 3.2% (31) of patients withdrew during treatment and 1.4% (14) did so prior to treatment. The only serious adverse event reported was the onset of manic symptoms in an OCD patient with comorbid bipolar disorder [98].
rTMS stimulation protocols contain numerous complex parameters, and no two studies presented comparable methodologies and stimulation protocols. When grouping each stimulation parameter within ranges (e.g., 10-20 sessions, 1200-1500 pulses etc.), there were no more than three studies with similar protocols for pre-SMA and DLPFC targets, which was insufficient for a meta-analysis. Stimulation protocols for OCD varied in the number of sessions (10-30 acute sessions, up to 44 for second phase protocols), pulses (750-3000, 600 for TBS), intensity (80-120%), frequency (1 Hz, 10 Hz, 20 Hz, 30 Hz, 50 Hz, α), coil type (figure-eight, deep TMS, one circular coil protocol), and targeting method (standardized coordinates, individualized neuronavigation). Protocols for TS also varied in session number (range 2-20), pulses (900-1800, 600 for TBS), intensity (80-110%), frequency (1 Hz, one 30 Hz protocol) and coil type (Figure-eight, one deep TMS protocol), and targeting method (standardized coordinates, one individualized neuronavigation protocol). The DLPFC, pre-SMA/SMA, OFC, mPFC and ACC were targeted for OCD; the pre-SMA/SMA and the motor cortex were targeted for TS. The pre-SMA was targeted for skin picking, and the right DLPFC for a HD patient.
Blinding is a perennial problem in TMS, and only six studies (13% of the total) assessed the effectiveness of blinding [92,93,99,114,116,129]. Three did not report the sham method [89,110,128]. It cannot be established whether protocols failed to assess the effectiveness of sham, or whether ineffective blinding was simply not reported. Studies implemented the following sham method: A sham coil placed on the same site as the active condition and an active coil placed 0.6-1 m away from the scalp [92,102].
The tilt method produces the same sound and generally 25-50% of the electrical field compared to active TMS [136,137], thus likely producing some somatic sensations. The magnitude of electrical field that penetrates the scalp varies greatly as a function of the direction of the tilt and whether one or two wings are touching the scalp [136,137]; yet these aspects were not reported. A sham coil yields negligible stimulation to the cortex (<3%), and produces the same sound as active TMS, yet lacks somatic sensations. The mechanisms of method 3 (active and sham coil) are not discussed in the literature, and thus, the limitations of this method are unknown. A sham coil combined with surface electrodes that induces somatic sensations is considered the sham gold standard [138], but none of the studies reviewed here implemented this method [87].

OCD, DLPFC Bilateral DLPFC
Within-group examination of laterality revealed a slight clinical superiority of HF rTMS of the right DLPFC immediately post-treatment that become a clear superiority at one-month follow-up [89]. Alpha (α) guided rTMS of bilateral DLPFC led to response in 36% of patients' post-treatment, with minimal placebo effect, yet response was not maintained in three patients one week later [102].

Left DLPFC
Similar clinical response was achieved with LF (24%) and HF (30%) rTMS; however, comparable placebo effects occurred [91,92]. One study reported post-treatment effects and showed further improvement at two weeks, yet sham benefit was greater than active [91]. A response rate of 55% was achieved when HF rTMS was combined with SSRI treatment, and the placebo effect was marginal [95]. Although all RCTs included medicated patients, controlling for medication-specifically SSRIs-may optimize the aftereffects of HF rTMS. Indeed, SSRIs can enhance and prolong excitatory neuromodulation [69]. In addition, an 'interactive' model of rTMS mechanisms proposes that rTMS may not necessarily restore specific functions due to limited after effects, but rather, may "allow the brain to restore itself" [85]. rTMS may modulate a certain level of pathological activity and prime the neural system to become more adaptable to adjunct therapy and further excitatory modulations.
Analysis of symptom subdomains provided further insight. For HF rTMS, only leftsided application led to significant improvements in obsessions but not compulsions, but when controlling for depression, the effect held true for the left and right conditions together, but not for the left condition alone [89,92]. This finding suggests that the therapeutic effects of HF rTMS of the left DLPFC are driven by changes in depression, and that rTMS treatment of the left DLPFC alone is insufficient to produce clinical benefit for OC symptoms.

Right DLPFC
Mean symptom improvements were moderate from rTMS of the right DLPFC in randomized controlled contexts, with response rates of 20-50% for LF and 31-54% for HF protocols. Although response rates were similar for low and high frequency trials, direct comparison of frequency demonstrated within-patient superiority of LF at post-treatment that increased in magnitude three months later [103]. Placebo effects were relatively low, with the exception of one LF [93] and one HF trial [97], in which active and control groups showed comparable outcomes. Follow-up reports consistently demonstrated continued improvements four and six weeks after HF treatment [89,97,99], maintained at three months from LF but not HF treatment [103].

DLPFC General
Various DLPFC montages reported a larger antidepressant and anxiolytic effect compared to the anti-obsessive-compulsive effect [91][92][93]97,99,102,110,111]. This is compatible with a recent meta-analysis [139]. It is well established that excitatory stimulation of the left DLPFC or inhibitory stimulation of the right DLPFC has an antidepressant effect, and that both montages are an effective treatment for depression [140]. It could be stated that DLPFC applications have a strong antidepressant effect irrespective of laterality, frequency, and pathological state. Here, this effect most commonly occurred with LF (inhibitory) rTMS of the right DLPFC. The premise of targeting the DLPFC is also related to consistent beneficial clinical effects in numerous psychiatric conditions, playing a global role in decision-making and emotional regulation, rather than showing a specialized correlation to OCD pathophysiology. LF and HF rTMS of the right DLPFC led to similar outcomes immediately following treatment, which is likely driven by a primary antidepressant effect and secondary effect on obsessions and compulsions. Yet, LF rTMS of the right DLPFC led to more prolonged benefits, possibly due to a stronger antidepressant effect.
No predictors of response were identified [110,111]. Across studies, greater treatment dose (i.e., sessions and pulses) did not relate to greater efficacy; in fact, optimal outcomes were achieved with just 10 sessions and ≤1000 pulses per day [104,111]. This is contrary to the treatment efficacy-dose relationship of rTMS DLPFC application for depression [141] and highlights that stimulation protocols should not be extrapolated across diagnostic groups.
When post-treatment outcomes were reported, symptom suppression was always maintained or improved, which highlights a possible learning effect. Considering the fact that DLPFC applications may not directly modulate obsessions and compulsions, and greater dosage did not achieve greater response, a treatment regime that administers a moderate treatment dose combined with behavioral therapy that directly targets obsessions and compulsions may achieve optimal clinical efficacy across several global symptoms.

OCD, Pre-SMA/SMA
Apart from the study of Pelissolo (2016), which reported a greater placebo effect than treatment effect, LF rTMS of the pre-SMA/SMA achieved responses in between 22-80% of patients across five RCTs, an open-label randomized trial and one retrospective report. Combined targeting of both OCD and MDD symptoms (pre-SMA and right DLPFC) showed an impressive response rate of 83% [111]. TBS of pre-SMA achieved a response in almost a third of patients, but the placebo effect was greater than the treatment effect [117]. Follow-up outcomes were reported in two studies that showed a slight improvement at two-week follow-up [107] or decline at three-month follow-up [100]. It was stated that responders were maintained at three-month follow-up from LF rTMS of SMA, but outcomes were not reported [90]. No discernible differences were identified between pre-SMA and SMA applications, and as such, outcomes were discussed jointly herein.
There was a treatment dose-efficacy association that became evident only after 20 sessions, with sustained improvements between 20-44 sessions and no reported adverse events. Within the cohort of Mantovani (2010), a small number of patients entered a second phase of 20 additional sessions, and response was achieved in all. Hawken (2016) administered 25 treatments and titrated down the regime, resulting in a response rate of 80% that was maintained six weeks later. In contrast, Gomes and colleagues found that 10 sessions of the same protocol did not maintain efficacy at three-month follow-up [100]. Furthermore, 24 maintenance sessions following 20 acute sessions enhanced improvements from 35% to 65% in a single patient [123]. Hyperactivity in the pre-SMA, particularly during inhibitory behavior, is a potential endophenotype of OCD [142]. Individuals with greater hyperactive CSTC likely require additional stimulation to regulate pathological activity and reach a threshold that alters behavior. Thus, 20 or more LF rTMS sessions of the pre-SMA may be required to reach response and maintain efficacy.
Insufficient and inconsistent information prevented us from identifying a relationship between rTMS effects for comorbid symptoms. Although articles did not report or demonstrate evidence that depression hindered OCD outcomes, a large group level analysis identified comorbid depression as a potential negative prognostic marker [118]. Notably, tailoring targets for both OCD and MDD symptoms may optimize outcomes for both conditions [111]. Psychiatric comorbidities other than MDD were reported in two studies, and it cannot be determined whether they impacted outcomes [108,112]. Mostly anxiety and depression outcomes showed a lack of correlation with OCD (YBOCS) outcomes [90,96] although OCD and anxiety symptoms did correlate within one RCT that showed high efficacy [100].
It can be concluded that LF rTMS of the pre-SMA/SMA can achieve high clinical efficacy for OCD, as confirmed by several randomized controlled investigations and clinical case series. If the most recent RCTs are considered (excluding that of Mantovani et al., 2010), response rates were consistently high (42-80%). It is possible that the methods and expertise of rTMS applications became refined over time, and recent trials represent a realistic response rate for pre-SMA/SMA applications. When comorbidities are present, additional targeting of comorbid symptom domains appears to have additional benefits. Longer-term treatment (≥20 sessions) with slow tapering off may be required for some patients to achieve a response and should be considered in stimulation protocols.

OCD, OFC
Response rates between 25-44% were achieved from LF rTMS of left OFC, and efficacy was maintained at four-and eight-weeks following treatment [94,115]. A RCT showed greatest improvement in the active group at the two-week FU (21.4%), which was maintained at eight weeks FU (19.2%) but showed some loss of gains at 12-week FU (15.5%). LF deep TMS of right OFC led to 19% change, which was not maintained at one-month follow-up [101].
The study of Nauczyciel (2014) showed decreased metabolism in bilateral orbitofrontal lobes (pretreatment) to a greater extent in the (targeted) right hemisphere. Additionally, improvements in YBOCS were correlated with a decrease in right OFC activity. Despite limited response to ten sessions of LF deep rTMS of the right OFC, sufficient modulation occurred to induce changes in resting state activity. Within-group analysis showed a slight superiority of LF rTMS of the left OFC over LF rTMS of the SMA [118].
Although only four studies were identified, the number of sessions had a progressive positive relationship with efficacy, and baseline severity scores were relatively high compared to most trials targeting other brain regions. Two trials [115,118] showed efficacy (43-44% response) only after having extended their treatment protocols to 20 sessions. The two trials that achieved less positive change [94,101] used 10 or 15 sessions, and their patients also had high baseline YBOCS scores (~32 points).
Given the central role of a hyperactive OFC in OCD pathophysiology, it is reasonable to assume that less severe patients with hyperactivity of the OFC may respond more rapidly to rTMS, whilst more severe patients may require additional treatment sessions. In support of this notion, the number of previous medication trials (which is related to treatment duration or severity) was significantly lower in TMS responders in the trial of Kumar and colleagues [115]. A separate study found that, compared to responders, nonresponders had a significantly greater duration of illness, higher baseline YBOCS, were more likely have comorbid MDD, and had failed more medication trials [118]. These data suggest that effects of OFC rTMS are related to several aspects of illness severity and comorbidities.

OCD, Other Prefrontal Targets
Investigations of HF rTMS of mPFC and ACC led to 44% and 45% response with marginal placebo effects within reviewed RCTs [114,116]; a 50% response was found when targeting the dmPFC alone [106]. These findings gain support from a large multisite trial, in which efficacy was maintained four weeks later [116]. Dunlop (2016) administered an additional 10 sessions to nonresponders, and 50% of them achieved response based on a high criterion (50% improvement). Also, LF rTMS of the medial PFC alone achieved 40% mean improvement [105]. Baseline clinical characteristics did not predict YBOCS outcomes [105,106].
Targeting the medial PFC led to consistent efficacy, despite heterogeneous stimulation protocols (10-30 sessions, 1200-3000 pulses), varying symptom severity (mean baseline YBOCS of 22.8-30.5), and complex psychiatric comorbidities (MDD, bipolar, anorexia nervosa, bulimia nervosa, PTSD, TS). It should be noted that all studies employed novel techniques, including neuronavigation and symptom provocation. Symptom provocation activates CSTC circuits [143] and is implemented on the basis that subsequent facilitatory stimulation may have additive effects.

TS, Motor Targets
LF rTMS of the SMA showed 33-71% response and 17-34% change within open-and closed-label trials, except in a single pilot study, in which it did not achieve efficacy [130]. Theta burst stimulation was associated with a high response rate, yet the placebo effect was greater than the treatment effect [128]. One investigation showed no effect of targeting the pre-SMA using deep TMS [130]. Studies of LF rTMS of the premotor cortex or pre-SMA did not result in symptom improvement, suggesting too low a 'dose' of TMS for efficacy.
Notably, when efficacy was achieved, symptoms remained suppressed in the long term, i.e., from 3-6-months [126,127,131]. A previous meta-analysis contended that the daily and total dose of TMS was not related to clinical outcomes [27], but a treatment threshold likely exists, as two sessions was insufficient, and some patients required maintenance therapy (>15 sessions) for response to be achieved or maintained [129,131]. There was heterogeneity in baseline severity (YGTSS between 20.6-64.7 on average, and 37-85 within case studies), although severity was not correlated with outcomes.
LF rTMS of the SMA was able to achieve high efficacy and long-term outcomes in a subgroup of TS patients and is recommended for clinical use. Further investigations into predictors of response are required, although response does not appear to be related to symptom severity.

Other Conditions
LF rTMS of the pre-SMA led to a transient effect for skin picking, followed by deterioration [133]. One HD patient experienced benefit from LF rTMS of right DLPFC, which was associated with increased functional connectivity between the right DLPFC and vmPFC [134]. The data are too sparse to allow conclusions to be drawn.

Polarity Dependent Effects
Low and high frequency protocols showed counterintuitive clinical outcomes. Theoretically, LF rTMS protocols should be optimal in suppressing the frontal-striatal hyperactivity that underlies OCRD. However, both LF and HF stimulation were effective for OCD, and HF stimulation was particularly effective for mPFC targeted alone or in combination with ACC. The mPFC and ACC play central roles in monitoring internal states and motivational drives and integrating emotionally salient information [144]. Although the mPFC and ACC are involved in the dorsal and affective CTSC loops, hyperactivity is not consistently associated with obsessions and compulsions. The favorable facilitatory effect on these regions may be due to their broad role in cognitive control and emotional regulation, which are supposedly enhanced in an obsessive-compulsive state.

Novel Techniques
Novel techniques including deep coil, neuronavigation, and TBS were implemented in a few of the studies reviewed here [105,106,109,114,116,117,121,130]. Three out of five deep TMS studies showed efficacious outcomes and a low placebo effect for mPFC and ACC [105,114,116], while two showed marginal treatment effects for right OFC and SMA [101,130]. Deep TMS coils were developed to overcome the limited penetration depth of 1.5-2.5 cm from figure-eight coils to target deep neural structures, i.e., around 6 cm in depth [145]. In any neuromodulation scenario, there is a trade-off between focality and depth of penetration, owing to an exponential decay of the signal as a function of distance from the stimulation discharge [146]. The protocols that did not achieve efficacy from deep coils (yet efficacy was shown with figure-eight coils), targeted largely superficial (SMA) or deep structures (OFC), whereas efficacy was achieved by targeting medial structures (mPFC, ACC). Therefore, deep rTMS current may be too diffuse for regions in either close or far proximity and is better suited for use on regions of medial proximity.
Three studies administered continuous (inhibitory) theta burst stimulation. Although moderate treatment effects were found, placebo effects were comparable to [128] or even larger than [117] treatment effects. Stimulation protocols using TBS may require refinement or induce nontherapeutic oscillations for OCRD.
Neuronavigation techniques most often showed efficacious outcomes in four out of six studies [96,105,106,134]; however, superiority over standardized targeting has not been established for OCD.
Dual targeting may be more effective than single targeting. Donse (2017) targeted the SMA and right DLPFC (inhibitory rTMS), while Carmi (2018; 2019) targeted the mPFC and ACC (facilitatory rTMS); both achieved high efficacy. Importantly, the former montage was applied to a heterogeneous clinical group, and hence, is representative of OCD populations. Additionally, the latter montage was supported by a large multisite RCT. Although Kang (2009) did not achieve efficacy by targeting the SMA and right DLPFC, a much higher number of pulses was delivered (2400 vs. 1000) compared to those applied in the study by Donse (2017). Therefore, dual targeting should be further explored with consistent stimulation protocols that have already shown effectiveness.

DBS Results
DBS involves the placement of electrodes in the brain. These are stimulated by a battery-powered stimulator usually placed under the clavicle and produce a biphasic and high frequency electrical current that travels in and out of neurological substrates (cells, axons, dendrites, and glial cells), resulting in a small electric field of around 2.5-5 mm within deep neural structures. Stimulation of a specific target has widespread effects on neural circuits, depending on, among other factors, the orientation and size of activated nerve fibers and the cytoarchitectural organization of innervated neural populations [147].
The most well-established theory on the mechanisms of DBS action postulates that a 'functional lesion' occurs when the stimulation frequency is around twice the firing frequency of the neurons; as such, a circuit is 'captured', resulting in local inhibition [148]. Recent theories have been expanded to consider network models, and have recognized that local excitation, suppression of pathological firing, and plasticity mechanisms may also underpin the therapeutic effects of DBS [149].
Determining the 'sweet spot' of stimulation is complex. Potential sources of mechanical, technical, or human error related to surgery and postoperative programming can confound patient outcomes. Unlike applications for movement disorders, in which response is almost immediate and overtly observed, DBS therapy for psychiatric conditions often has a prolonged onset and more variable response.
Appropriate patient selection, accurate placement of the electrode, and effective programming are the major factors that contribute to DBS outcomes [150]. Programming is the only modifiable factor of therapy once the leads have been implanted and is of particular importance when electrodes are placed at the border of the targeted structure or have been misplaced. The overall aim of programming is to optimize clinical benefit, avoid side effects, and minimize current consumption [151]. Stimulation amplitude (constant or cyclic current or voltage) is the amount of stimulation delivered to the neural tissue and is the most adjusted parameter. The location and number of active contacts (or configuration) changes the volume of tissue activated (VTA). The monopolar configuration has a spherical current field, bipolar has a narrow oval current field, and multipolar has two current fields [147]. High frequency DBS is effective due to the time-locking parameters of axons, and 130 Hz is common based on a trade-off between efficacy and battery consumption. Pulse width variation in increments of 60µs is common, as it is the minimum pulse duration needed to initiate an action potential in a myelinated axon [151]. There are generally three programming phases. The initial visit is to screen for the optimal contact, often combined with anatomical (neuroimaging) information and intra-operative test stimulation observations. The early optimization phase aims to optimize the stimulation parameters and medication dose, which commonly involves titrating the stimulation amplitude. Then, patients are monitored around once a year, to check for unexpected worsening, battery consumption and troubleshooting.
Investigations of ALIC DBS encompassed two RCTs, one trial with a staggered switch on, one long-term follow-up report, and two case reports. Closed-label investigations led to 20% and 43% improvement [153,157]; and long-term (1-9 years) treatment led to 43-67% improvement [153,159,160]. The ALIC was also targeted in the cohort of Mantione et al., (2014) through shifts in targeting, and achieved 43% improvement at 1 year [220].
Investigations into VC/VS DBS involved one RCT, one open-label trial with a longterm follow-up report, and three case reports. The RCT originally implanted the ALIC [153], and implemented a posterior shift in target to the VC/VS. A larger cohort from the same site as  achieved 42% mean improvement from closed-label treatment, and at three-year follow up, 39% symptom improvement was maintained [156].
Investigations of amSTN DBS involved a multisite RCT that resulted in 25% median improvement, and 51% mean improvement was reached at four-year follow up [158].

DBS for TS
One clinical care study and four case reports implanted different targets within the same cohort, or implanted patients with two targets. Stimulation of the thalamus, GPi, or both targets led to 45%, 32-78%, and 36-60%, respectively [186,204].
Thalamic DBS was investigated in two RCTs, three open-label trials (two with followup reports), two retrospective reports, and five case studies; outcomes of the closed-label phase of one RCT is not reported here due to the limited treatment duration of 7 days per condition. Closed-label treatment achieved 40% improvement, while open-label led to 19%, 44% and 55% improvements [179,183,187,191]. Sustained improvements were achieved across several cohorts with 30-73% change from 1-6 years of therapy [180][181][182][183]188,192,196].

DBS Discussion
The RoB assessment rated 46 articles as low risk, 16 as medium risk, and nine as high risk (S2). The quality assessment rated 19 articles as good, 44 as moderate and eight as poor (S3). Only 11 out of 71 articles were RCTs, and 35 were case reports, which meant a randomized control aspect and group level analysis was not present in almost half of the patients included here. Furthermore, only half (36) of the articles reported on more than one time-point, which limits interpretations regarding the duration and pattern of response. Within the bias assessment, there were multiple deviations from the intended protocol, including DBS explants or switch off, and closed-label conditions ending early. It was reported that 18 (11.7%) OCD patients and 12 (6.8%) TS patients had their devices switched off or explanted due to limited/no efficacy or even worsening in some instances; a further three (1.7%) TS patients underwent repositioning. Also, five RCTs had patients that ended the closed-label phase early. It is possible that not all cases of device switch off, explant, or repositioning were captured.
Adverse events included transient psychiatric symptoms, particularly hypomania, increased anxiety, deterioration of mood and suicidal thoughts, which were generally resolved with programming adjustments. There were seven suicide attempts, and one completed suicide [157]. Battery depletion was rarely reported on but seemed to occur between 5-22 months in OCD cohorts [153,154,170] and was reported to occur at 24-months for one TS patient [210].

OCD, NAc Target
Modest and consistent change was achieved from NAc DBS with 21-33% improvements from open-label treatment. Yet Barcia (2019) demonstrated that high efficacy of closed-label DBS (85% full response, 100% partial response) is achievable [165]. This was attributed to optimization of the stimulation contact, and thus, stimulation of the ideal anatomical structure on a patient-specific basis. The protocol implemented a three-month condition for each contact and reported outcomes on the best one. Although other investigations may trial each contact in a monopolar review or exploratory programming across several days, this is likely insufficient to determine the true therapeutic effect [222]. Thus, optimization of programming appears to influence efficacy.
Further supporting the findings, the RCT that showed minimal improvements applied predefined and global stimulation parameters for the closed-label phase [161]. Although refining the stimulation parameters prior to the closed-label conditions has implications for blinding, applying the same therapeutic settings to diverse neurobiological patient profiles that have electrodes implanted in slightly different anatomical positions limits therapeutic benefits. Further, the importance of optimizing therapy within RCTs is rarely discussed and is proposed herein as a critical factor for outcomes. It should also be considered that Huff (2010) applied unilateral DBS, which may have also affected outcomes.
In a different protocol, eight months of open-label DBS led to 25% initial improvement. Following 24 weeks of adjunct CBT, this increased to 46%, and a subsequent closedlabel phase (with CBT) of active DBS led to 1.9% deterioration while sham led to 45% deterioration-reaching baseline severity [162]. This protocol highlights the strength of DBS therapy, such that open-label effects can be maintained during closed-label conditions with CBT, yet CBT alone was not sufficient to maintain previous DBS therapeutic effects. This protocol also demonstrated the importance of optimizing therapies in a staged manner. Programming was assessed fortnightly across an eight-month period and adjusted if necessary. Once stimulation optimization had been achieved, CBT was introduced. This allowed an extended period for DBS response and a transition to CBT once patients were receptive to behavioral therapy.
Long-term outcomes were modest, with a maximum of 33% improvement at 12 months or last (8-54 months) follow-up [164,167]. The case study that experienced deterioration had comorbid mild TS, which may have hindered efficacy [169].

OCD, ALIC Target
Three study sites showed efficacious long-term outcomes from ALIC DBS, such that 58-67% of participants achieved response at one year, which was maintained at 6-9 years [153,159,160,220]. Substantial change was achieved within three-months [153,178], with subsequent gradual improvement.
However, in comparison to these studies, the cohort of Abelson (2005) did not show comparable efficacy within both open-and closed-label phases. The RCT that achieved high response levels implemented an extensive programming regime that assessed all parameters and contact configurations across weeks to months prior to blinded phases [153]. The study of Abelson (2005), which showed inferior results to other RCTs, conducted exploratory programming across just 3-8 days before the blinded phases. This study was a pilot and supposedly involved the first implants at that site; thus, limited experience with DBS patients may have contributed to moderate outcomes. Additionally, limited programming likely hindered patients reaching response.
It is well established that the duration of clinical response to DBS is slow within psychiatric conditions in comparison to movement disorders [222,223]. Differing behavioral effects can occur across minutes to months, which influences the ability to optimize stimulation. The number of stimulation combinations is vast and the assessment of efficacy of each is laborious.
ALIC DBS responses appeared to rely on programming adjustments, particularly stimulation amplitude. High intensities up to 8.5 V or 10.5 V were applied for chronic ALIC DBS across all trials with the exception of the most recent investigation [220], and the patient that declined between 4-12 months [177] had a lower stimulation intensity (2-4 V). A case series showed a ceiling effect of clinical response at six months which coincided with optimization of stimulation [178]. Also, it was reported that one patient did not reach response until eight months, coinciding with a second contact being activated [159]. Fayad (2016) showed that responders were maintained between one to 6-9 years with comparable stimulation parameters [158]. Thus, ALIC DBS may require particularly extensive programming, requiring amplitude titration across 6-12 months, at which point stimulation parameters and clinical efficacy should stabilize.

OCD, VC/VS Target
Two RCTs (one with an absence of sham) showed rapid improvements in blinded phases (42% and 53% mean improvement) [156,168], and further improvements in the longer term, with ultimate response rates of 62-83%. The greatest magnitude of response was achieved within three months, although optimization of treatment, including CBT, further enhanced DBS outcomes. Luyten (2016) reported outcomes depending on the location of the chronic active contact; BNST DBS led to superior outcomes with 50% improvement and 80% response, compared to ALIC DBS, that led to 22% improvement and 17% response. Activation of both the BNST and ALIC with multipolar configuration had a mean improvement of 66% and 100% response rate but was trialed in just three patients. Many fibers from the PFC-including the ACC-transverse through the ALIC and are part of the ventral capsule (VS) of the VC/VS; also, VS and NAc are terms which are used interchangeably to refer to a confluence between the putamen and caudate [224]. Thus, stimulation of the VC/VS likely modulates the ALIC and/or closely connected structures. Stimulation of both BNST and ALIC through monopolar stimulation causes two adjacent current spreads, which is also likely to reach neighboring structures. Therefore, active contacts within the ALIC may cause a current spread that overlaps to other regions, and optimization of stimulation within this region may rely on a diffuse current spread over neighboring regions.
Controlled comparison of amSTN and VC/VS showed the superiority of VC/VS, with 83% response in patients who had been implanted with leads in both targets; stimulation of both slightly increased efficacy to 60% (from 53% from VC/VS alone), but no further responders were achieved [168]. Owing to the heterogeneous surgical and clinical practices across sites, there is great value in assessing targets within the same center. Variance in clinical practices that may confound patient outcomes are minimized in this approach. Tyagi (2019) implemented within-patient comparison of two targets, which further minimized variance in individual clinical and anatomical characteristics.
There is within-patient evidence that targeting the VC/VS is superior to amSTN DBS, and efficacious outcomes from VC/VS DBS are likely attributed to the stimulation of additional structures-BNST and ALIC.
Randomized controlled outcomes were similar across different programming protocols. An extensive programming phase was implemented across several months prior to the closed-label phase in the multisite cohort study of Luyten and colleagues [156]. The smaller cohort of Tyagi et al. (2019) employed two weeks of programming adjustments prior to each phase, but there was no sham control [168]. These studies suggest benefits from extensive programming for VC/VS DBS prior to closed-label phases, with enhancement of long-term outcomes.
The Luyten (2016) cohort showed that the majority of improvements were achieved within three months, but further improvements were seen even years after implantation (e.g., at four years, there was a 66% improvement).

OCD, amSTN Target
One cohort showed progressive and high efficacy through targeting amSTN: 51% at four years, with 75% being full responders [13]. An earlier report from this cohort recognized errors in targeting, as four leads missed the target, and 9/33 active contacts were not within the STN [158]. However, the follow-up report noted that both responders and nonresponders had leads placed within the target [13]. Another patient showed a similar pattern with 29% and 92% change at 3 and 36 months, respectively [215]. Tygai (2019) implanted patients with both amSTN and VC/VS leads, which are discussed above (within VC/VS); briefly, closed-label outcomes of amSTN DBS achieved 45% improvement and 50% response, and VC/VS DBS led to greater efficacy. Owing to consistent slow response from targeting the amSTN, a closed-label period of three months is likely insufficient to achieve full response for amSTN DBS therapy. This delayed pattern of response should be considered in clinical care and research methodologies.

OCD, BNST Target
Across three cohorts, BNST DBS achieved 43% improvement during closed-label conditions [225], 39% at six months of open-label treatment [163], and 24% at long-term (8-54 months) follow-up [164]. Although the cohort of Luyten et al. (2016) initially targeted the VC/VS, long-term outcomes were reported for each location of the active contact-ALIC or BNST. At last follow-up (54-171 months), BNST DBS led to 80% response, and the majority of the cohort (15/24) had active contacts within the BNST (outcomes discussed above in ALIC). Although direct targeting of BNST was limited, incidental investigations though target shifts provided valuable insight and robust evidence of high efficacy.

OCD, Other Targets
A recent pilot study of ITP DBS showed efficacy in five patients [166], and a separate study found that superior-lateral MFB DBS led to efficacy in two patients [173]. Thalamic target DBS was not effective for OCD in two patients who had previously received NAc DBS; depression ratings worsened [175].

Optimized Localization
Across studies, OCD outcomes were variable and dependent on fine-tuning of stimulation location and other parameters. Most of the OCD investigations targeted the striatal regions (ALIC, VC/VS, BNST, NAc), and there was robust evidence to support targeting dorsal and posterior regions. In one trial [165], YBOCS improvement was significantly greater in caudate (E2,3) compared to NAc (E0,1) contacts. Caudate contacts were within the ALIC and likely also stimulated the VS, yet the NAc was still reported as the anatomical target. The other RCT that targeted the NAc did not reach efficacy and applied active contacts within the NAc with fixed stimulation parameters [161]. Patients receiving NAc DBS often had chronic activation of multiple contacts, likely leading to the stimulation of other regions outside the NAc itself [161,167]. Yet, when the optimal location across the NAc trajectory was determined (within and outside the NAc), multipolar stimulation was not required to reach high efficacy [165]. Although studies commonly targeted the NAc, stimulation parameters showed that stimulating the NAc alone led to suboptimal therapy.
The patients reported by Mantione (2014) were included in the largest cohort study thus far reported, comprising 70 patients [226]. Although the report was outside the cutoff date for the current review, we provide a summary of the main findings. The initial 16 patients reported by Mantione (2014) were targeted with two contacts within the NAc, but it was later revealed that the position of the electrode shifted so that just one contact was within the NAc, and three contacts were within the ventral ALIC. The patients all had chronic active contacts within the ventral ALIC (not the NAc), which corresponds with the ventral capsule of the VC/VS, and mean symptom (YBOCS) improvement of 40% and response in 52% were achieved. Therefore, there is evidence from several investigations at several sites that ALIC stimulation is superior to NAc stimulation.
Investigations of other striatal regions identified even more precise optimal target. A large multisite cohort originally targeted the ALIC with the most ventral contact in the NAc [153]. Again, the target shifted, i.e., posterior, ventral, and medial to the VC/VS, and at the junction of the anterior capsule, anterior commissure, and posterior ventral striatum, with the most ventral contact in the BNST [155,156]. The shift had a dramatic effect on outcomes, lifting an initial 33% response rate to 78% and 75% in the latter cohorts with the optimized target. The target was originally reported as the ALIC [153], then the VC/VS as the optimized target [154,155], and finally, as the BNST or ALIC-depending on the chronic active contact [156]. Most patients were receiving chronic BNST stimulation (n = 15), while some received ALIC stimulation (n = 6). A similar pattern occurred in a separate cohort, in which the original target was the ALIC [159] but shifted to the VC/VS in a follow-up report [160]. Importantly, through extensive localization refinement, regions posterior and medial to the ALIC (including the VC/VS and BNST) demonstrated superior outcomes. This effect is consistent across and within cohorts, and within patients.
Further support for this notion can be found in the extremely high stimulation amplitudes for ALIC DBS (up to 10.5 V), which indicates a large VTA is required for efficacy, and likely involves stimulation of several neighboring regions. Activation of multiple contacts was often required for NAc DBS [161,167], which creates several overlapping VTAs and indicates stimulation of superior structures.
Across all published studies, the anatomical target, and often the chronic active contact, were specified (E0-E3); yet the precise positioning of the active contact(s) in relation to the target was rarely reported. Owing to all the possible sources of positioning errors and differences in anatomy, patients will be implanted at slightly different locations, as determined by postoperative imaging. It cannot be assumed that leads are placed in a comparable trajectory across patients. Depending on the chronic contact(s) within a single lead trajectory, several different regions may be activated.
Reporting of the precise anatomical localization of the active contact and stimulation parameters (including configuration) is required for progress to occur the field, and these aspects of therapy are critical for refining DBS therapy.

Optimized Stimulation Parameters
Although implementing predefined stimulation parameters during closed-label phases is advantageous for blinding, it likely limits efficacy. Across targets, it was identified that a lack of programming was a major determinant of suboptimal therapy [157,161]. Trials that had an extensive optimization phase in the weeks prior to closed-label conditions all achieved high efficacy [153,156,165,168]. The extent of programming for one RCT was not clear [158].
The largest reported cohort of DBS for OCD [226] included an extensive optimization phase with assessments every two weeks, which likely contributed to efficacious outcomes. Refinement of suboptimal stimulation by highly experienced clinicians allows nonresponders to reach efficacy at all points of clinical management, even three years after surgery [227,228]. The fine-tuning of programming can be burdensome; it is reliant on the expertise of the clinician and the fluctuating state of the patient. Yet, it is contended that a 'one size fits all' approach is inappropriate within this context of a complex therapy and pathophysiology.

Functional Connectivity Insight
Recent neuromodulation perspectives have shifted away from focal stimulation of brain nuclei, focusing instead on the modulation of distributed brain networks through analyses of connectomics [229][230][231]. The concept of 'circuitopathies' is not new, yet recently enhanced MRI capabilities have made it possible to identify white matter tracts and connectivity pathways. Fiber tracking analysis has consistently shown that activation of fibers from the target nuclei to the PFC (medial, lateral, dorsolateral) is correlated with good response across different OCD DBS targets and cohorts [221,229,232]. This indicates that multiple targets modulate a shared network that similarly affects OC symptoms. Indeed, the connectomic approach was able to explain 40% of variance, thus presenting itself as a promising biomarker [221].
Furthermore, the cortico-thalamo-basal ganglia network has recurrent excitatory and inhibitory loops, and different DBS targets have similar and unique connectivity patterns. The striatal target of the ventral ALIC and VS involves fibers in the OFC and ACC. The BNST contains fibers from the PFC to thalamus, and likely involves the modulation of ALIC fibers. The VS contains a complex mixture of myelinated fibers from the OFC, ACC, amygdala, and BNST among other connections, and stimulation of the VS. will likely modulate ALIC fibers. The medial STN receives fibers from the PFC, ACC, and other prefrontal regions, depending on the functional subdivision, and stimulation will likely have a secondary effect on the VS. [224]. Thus, the ALIC, VS and STN are topographically organized to receive OFC and ACC innervations, connecting to distinct subcortical pathways. Therefore, it is proposed that different OCD DBS targets modulate a shared subcortical-prefrontal network that can similarly affect OC symptoms. Also, DBS targets likely activate a specialized circuit that affects comorbid symptoms (mood, anxiety, reward, avoidance) or cognitive functions (inhibition, memory, attention) to varying degrees.
This notion is further supported by diffusion tractography analysis, which was shown to be able to predict clinical efficacy [165,168,220]. Tractography of the optimal contact within amSTN and VC/VS leads showed that both had connections to the OFC, and additional distinct tracts [168]. amSTN and VC/VS had comparable effects on OC symptoms, but divergent outcomes on cognition and mood. Fibers of the amSTN connect to the dorsal ACC, DLPFC, and medial forebrain bundle, and stimulation was associated with change in cognitive flexibility, while stimulation of the VC/VS (which is connected to the mediodorsal thalamus, amygdala, hypothalamus and habenula) resulted in changes in mood. Even within a single target (ALIC), the anatomical stimulation site cannot predict clinical response, while fiber connectivity can [220]. To elaborate, clinical response was correlated with active contacts which were closer to one fiber branch (superior-lateral MFB) than another (anterior thalamic radiation), yet the anatomical location of the electrode in a standardized space did not predict response.
Insight from connectome modelling elucidates the mechanisms underlying neuromodulation on a network level and makes it possible to target neural networks in order to predict DBS response, heralding a more objective and personalized treatment approach. Tractography analysis can also benefit non-responders by repositioning the electrode based on predictive fibers or selecting the stimulation site to prevent numerous time-consuming programming trials. Although limitations exist regarding the use of fiber tracking to target DBS, there are significant advantages in targeting symptom specific (not disease specific) circuits depending on the patient profile [229,230]. Future work should incorporate this approach throughout the pre-and post-operative stages of targeting and localizing DBS electrodes.

Other Clinical Management Considerations
DBS-mediated effects on distinct obsessive and compulsive symptom domains were reported in some trials [13,161,162,164,233,234]. Improvements were comparable for both domains, except for one amSTN trial, that had a 76% improvement in obsessions and 55% improvement in compulsions [233].
Two trials showed CBT could further enhance open-label treatment within a short period [162,168], and sometimes it was reported that CBT was initiated or resumed after 3-12 months of open-label therapy [154,172,220]. It was proposed that DBS may be able to break the association between stimuli and obsessions (i.e., anxiety), and that a second break in the association between obsessions and ritualistic behaviors (i.e., inhibition) occurs through CBT [162]. Although this did not translate to quantitative outcomes within the review, further investigation is warranted.

TS, Thalamus Target
Closed-and open-label investigations of thalamic stimulation showed high efficacy in TS, although reports of the former were limited. One RCT achieved 67% response; eight follow-up reports across six cohorts showed sustained long-term efficacy, with 30-73% improvements and 60-100% response across 1-6 years of therapy, excluding a trial of cycling therapy, that alternates between off and on periods of therapy.
Clear efficacy was achieved in all reports (≥60% response) except for one small open-label trial [187,188]. Delayed switch on implemented in the protocol did not have a statistical effect on outcomes. However, this was the only study to implement scheduled cycling, which alternated between periods of stimulation on and off to account for the intermittent nature of tics and varying symptom profiles. Although scheduled cycling may benefit battery life, it was shown to be inferior to constant DBS of the CM thalamus.
Progressive improvements appeared within CM thalamic applications from three months to six years across three sites [181][182][183]187,188], but a cohort that received ventral anterior and ventrolateral thalamic DBS reached ay ceiling effect at six months [191].

TS, Globus Pallidus Internus Target
Open-label and case studies showed consistent therapeutic responses from GPi DBS in TS, while randomized controlled investigations showed only modest outcomes (10-22% improvement). Long-term open-label outcomes across four cohorts achieved 41-63% mean improvements (50-75% response rate) from 12 to 46 months of treatment.
Open-label therapy [184,185] demonstrated that response was largely achievable within just one month, with subsequent gradual improvement over the ensuing eight to 46 months. The majority of trials included several follow-ups, which all showed continued improvement from up to four years of therapy [184,185,189,190,193,194].
Although most trials targeted the anterior (limbic) subdivision of the GPi, similar response rates were achieved from posterior (69%) and anterior (71-75%) GPi DBS. Case reports showed mean improvements of 52% (11-93%) for posterior GPi DBS [45,197,205,211] and 68% (20-95%) for anterior GPi DBS [205,209,212,213]. Also, outcomes from a case series [205] that implanted both targets suggested superiority for the anterior GPi. It appears that targeting the limbic functional division is standard practice, showing a slight superiority, while the motor division is targeted in patients with more complex or motor predominant symptoms, including self-injurious behavior and comorbid dystonia [235].

TS, Other Targets
One patient was implanted per GPe, ALIC, STN, NAc targets. Although response was achieved in all except ALIC DBS, conclusions cannot be drawn from such limited evidence. Further, NAc DBS required activations of all four contacts with a stimulation intensity of 7 V, thus posing limitations on battery duration and indicating possible suboptimal placement [200].

Optimized Localization
For TS, the thalamus and GPi were predominantly targeted, and both achieved efficacy. Blinded outcomes were greater for thalamic DBS over GPi DBS, although RCT investigations were limited for both targets. Long-term response rates showed superiority for thalamic DBS (60-100%) over GPi DBS (50-75%). Within-patient comparisons of targets were limited: GPi showed superior outcomes in three patients, whilst in another, CM-PF thalamus was found to be slightly superior [199,204].
Like this review, a database registry study of 185 TS patients showed comparable outcomes between CM thalamic DBS and GPi DBS, but superiority of the anterior GPi over the posterior GPi [236]. A previous review of the VTA across several sites showed that the most stimulated region of the GPi was the amGPi [237]. Further, the anterior GPi has higher connectivity to regions that are positive predictors of response compared to the posterior GPi [235]. Therefore, clinical and connectivity evidence favors the anterior-limbic functional region of the GPi unless the patient profile is suited to an alternative region.

Optimized Stimulation Parameters
The extent of programming optimization prior to closed-label conditions did not appear to affect outcomes. The trial that achieved high efficacy implemented three weeks of programming prior to blinding [183], whereas the other two trials that achieved moderate outcomes implemented one week [190] or one month [193] of programming. It should be noted that during open-label treatment, recurrence of symptoms occurred in some patients, indicating that long-term monitoring and adjustments may be necessary [180,209].
Open-label programming varied from four weeks to 12 months. Although such disparities in clinical management may theoretically impact outcomes, no clear relationship was identified between programming within open-and closed-label conditions and clinical outcomes. Also, TS outcomes had greater consistency compared to OCD outcomes.

Other Clinical Management Considerations
Apart from one study that reported a relationship between lower baseline disease severity and better outcomes [191], we found no reports supporting a relationship between baseline severity and clinical response to DBS in TS. Some patients with very severe symptoms achieved a dramatic response, while some with lesser severity exhibited treatment resistance. A more striking predictor of response for TS was that younger age (<20) was associated with clinical response for thalamic and GPi DBS, with improvements of between 37-100% [184,186,189,192]. A previous review also found younger age and lower disease severity at implantation to be associated with better outcomes [238]. However, a separate review found that median time to response was not impacted by age, despite the fact that lower age at implantation was associated with higher baseline YGTSS [237]. Thus, younger age at implantation may be predictive of response when baseline severity is low.
Owing to the intermittent nature of tics, and evidence that tic-generating networks function at varying time points [239], it has been contended that continuous DBS stimulation may not be necessary for tic relief. However, cycling of DBS has been associated with inferior outcomes in comparison with continuous DBS [187,188].
A number of case reports have suggested that comorbid OCD is a positive prognostic factor for DBS in TS. Stimulation of the GPi in comorbid patients achieved improvements of 67%, 85%, 90%, 94%, 94% and 95% [201,212,213]. When targeting the thalamus, comorbid patients achieved individual improvements of 60%, 69%, 80%, 82%, 83% [206][207][208]. However, the effect of comorbid OCD was not addressed in clinical trials, even though comorbidity was common. DBS modulates several networks that underlie tic and obsessivecompulsive behaviors, as well as mood and cognition; thus, it is likely that alleviating one aspect of impairment will have complimentary effects on other aspects. Indeed, several TS patients with comorbid OCD/obsessive-compulsive behavior achieved suppression of both conditions from DBS targeted for TS [200,201,207,209,211,213,217].
DBS treatment for TS showed a more consistent pattern of response than OCD applications, and minimal or no placebo effect from sham [183,190]. Across all investigations, just three study sites did not achieve efficacy [45,187,188,205]. Despite the consistency of response identified in this review, a previous in-depth analysis showed large variability in YGTSS change, i.e., 46.7 ± 29.7, and a YBOCS change of 21.1 ± 52.9 from DBS therapy for TS [237]. Thus, whilst clinical response was consistent across TS investigations, the variance in change is large, indicating that the identification of prognostic factors may allow patients to achieve even higher levels of efficacy.

ECT
A recently published expert report on new developments in evidence-based management of OCD [240] recommends ECT only for acute treatment of comorbid conditions (e.g., depression, psychosis). Currently, ECT is usually considered for OCD only after a number of other treatment interventions, i.e., as a last resort, when rapid improvements are necessary, or if a life-threatening psychiatric state is present [37]. Previous systematic reviews of ECT for OCRD have concluded that there is a lack of unequivocal evidence to support the efficacy of ECT. The current review found greater response rates than previous reviews [24,26] that adopted more lenient definitions of response, and included a greater spectrum of obsessive-compulsive conditions.
The current study identified response rates of 79% and 100% from ECT in OCD and TS cases, respectively. Although investigations involved a small number of patients, and there were no randomized or sham controlled investigations, the magnitude of effect was large considering the patients' level of severity and treatment resistance. Yet, without randomized placebo-controlled trials, valid recommendations cannot be made. The current review implemented stricter inclusion criteria than previous reviews, and necessitated standardized assessments, which resulted in the inclusion of fewer articles. Pooling together heterogeneous samples with biased methods and reporting may have previously obscured clinical interpretations. Cohort studies with standardized assessments following treatment and improved reporting of clinical characteristics are necessary to establish more objective guidelines regarding the potential value of ECT in OCRD.

tDCS
In agreement with previous systematic reviews and expert opinions [23,241,242], we found tDCS outcomes for OCRD to be modest and heterogeneous. This has previously been attributed to differences in stimulation protocols and clinical characteristics of patients. Our findings accord with those of Jacobson et al. (2012), i.e., that the observed heterogeneity of outcomes in part reflects the diversity of polarity effects [243]. Cathodal stimulation was the most often applied polarity in tDCS investigations, based on the assumption of inhibitory modulation. Yet, this may not be the true mechanism of action. Insight into the precise effects of consecutive applications of a relatively high dose (2 mA) tDCS on local cortical excitation and diffuse connectivity in psychiatric states is necessitated to understand behavioral changes and refine stimulation protocols.
Recently, da Silva et al. (2019) modelled the spatial distribution of electrical fields of the montages associated with tDCS in OCD patients [241]. Two prominent diffusion patterns were identified: (1) electrical fields focused within different PFC regions [52,62,64,71], and (2) electrical fields diffused across regions within and outside the PFC [51,53,55,59,60,63,65]. Consideration of these diffusion patterns within the context of the response patterns identified here showed that montages with a focused current spread within the PFC resulted in immediate symptom improvement, although not always meeting response criteria. An antidepressant effect was also common. In contrast, montages with diffuse modulation across the brain tended to show a delayed onset of response, in which symptom improvement was minimal immediately following treatment, but improved at follow-up. Thus, the distribution of tDCS neuromodulation may influence the pattern (onset and duration) of response, and other factors (i.e., stimulation dosage and clinical characteristics) may influence whether a particular patient responds to tDCS. This has implications for individualized therapy, notably for scenarios that necessitate immediate symptom suppression.
Stimulation dose (but not frequency) had a positive relationship with improvements with tDCS, in OCD patients. Protocols did not plan more than 20 sessions, but extended therapy led to further improvement in nonresponders [57]. This highlights the notion that a subgroup of patients may respond at a slower rate and require more than 20 sessions.
Protocols should also consider that by increasing the regions targeted, a rapid response may be achieved within a few sessions [55,61,64]. Finally, despite the small number of studies, tACS was shown to be able to induce changes with minimal stimulation dose [54].
Recent reviews of tDCS in OCD have failed to reach definitive conclusions or propose clinical recommendations [23,72,242,244], in part reflecting the fact that there were no RCTs included in previous systematic reviews [23,242]. A review of previous case studies of tDCS for TS proposed the pre-SMA as an optimal target, but no further recommendations were reached [25]. Previously, neither response rates per target nor patterns of response have been established; rather, these methods rely on 'improvement' or statistical change on the primary outcome measure, which limits clinical interpretations. From the studies reviewed here, we conclude that for OCD patients (1) twice daily anodal tDCS of the pre-SMA/SMA can achieve response in at least a third of individuals; (2) predictors of response should be investigated, (3) high dose tDCS of the DLPFC warrants randomized controlled investigations into optimal dosage and laterality; and (4) daily cathodal tDCS of the left OFC should be investigated in patients without a high level of treatment resistance, implementing an extended treatment regime and long-term follow-up. For TS patients, daily cathodal tDCS of the pre-SMA/SMA at low stimulation intensity (~1.4 mA) warrants further investigation.
There are several limitations that cannot be controlled across studies, including disparities in medication, resting brain state, clinical characteristics (level of treatment resistance, illness duration, and symptom profile), precision of targeting, and multiple and varied stimulation parameters (polarity, stimulation intensity, duration, and frequency, and return electrode placement). Controlled methodology, standardized protocols, and understanding of cortical aftereffects are required to establish more robust clinical recommendations.

TMS
A previous meta-analysis [245] concluded that the greatest treatment effects from TMS for OCD were associated with targeting the right DLPFC, then bilateral DLPFC, and then left DLPFC; investigations for OFC and SMA were too sparse to include a ranking. Further, no differences between HF and LF protocols were identified, and intensities of 100% RMT were favorable. More recently, Rehn (2018) concluded that LF rTMS of the SMA was optimal for OCD, but that bilateral DLPFC and right DLPFC targets were also effective [139]. Treatment effects were maximal at 12 weeks, and greater for anxiety and depressive symptoms than primary OCD symptoms.
For TS, previous reviews have identified younger age and comorbid ADHD/OCD as positive prognostic factors, and the SMA as a favorable target [27,82]. Yet, a recent meta-analysis concluded that rTMS is no more effective than placebo for the treatment of TS [27].
The current review agrees with previous reviews and provides further insights. It is concluded that LF rTMS of the pre-SMA/SMA yields the highest response in up to 80% of OCD patients, and up to 70% in TS patients, and that increased treatment sessions should be investigated for maximal response.
For OCD patients, LF rTMS of the right DLPFC carries sustained efficacy. Outcomes on OC symptoms are likely mediated through antidepressant and anxiolytic effects. Recent investigations have highlighted the potential for dual targeting, activating/priming neural systems prior to stimulation, and HF stimulation. Specifically, HF rTMS of the mPFC and ACC yielded robust findings that should be subjected to long-term follow-up. Also, regions that are not directly implicated in OCD pathophysiology (DLPFC, mPFC, ACC) but which have a more global function in cognitive and emotional control consistently showed high efficacy. Yet, an optimal frequency was not identified, as efficacy was achieved from both excitatory and inhibitory rTMS. Regions directly implicated in OCD pathophysiology (pre-SMA/SMA, OFC) may require a relatively high number of treatment sessions (>20) to normalize hyperactivity and thus alter behavioral manifestations.
No clear predictors of treatment response were identified for rTMS in OCD. It is proposed that neuroimaging should be implemented for the purpose of identifying abnormalities in functional connectivity rather than in precise regions. This approach may yield more optimal patient identification and target selection. Further, considering the importance of the resting brain state on rTMS outcomes [85], careful consideration of the state of patients prior to and after treatment may reduce response variability and improve outcomes. Studies generally have not reported methods to control for this effect. For example, priming the pathological brain state [114,116] and controlling pharmacological interventions [95], which both alter cortical excitability, provided consistent outcomes. Novel techniques (neuronavigation, deep TMS and TBS) did not show clear superiority over conventional protocols, but symptom provocation during rTMS appeared to enhance outcomes in OCD.
For TS patients, long-term efficacy for tic suppression is achievable with LF rTMS of the SMA, and thus, is recommended for clinical practice. Longer therapeutic trials should be considered for initial nonresponders. Predictors of response were not identified in TS, in part due to a lack of consistent protocols and randomized controlled investigations. Tic generation is underpinned by hyperactive motor pathways and hypoactive control pathways of the CSTC loops [18]. Like OCD applications, LF rTMS of the pre-SMA/SMA and HF rTMS of cognitive control regions may be effective in TS, but these procedures have not been subjected to clinical trials. Investigations into rTMS for other OCRD were very limited, and conclusions cannot be drawn.

DBS
One previous systematic review contended that there is insufficient evidence to conclude the existence of an optimal DBS target for OCD [31], while another proposed the NAc as the optimal target [29]. Similarly, for TS, no superior target has been determined, although the inferiority of NAc/ALIC DBS [238] and comparable efficacy of the thalamus and GPi DBS have been argued in international database studies [236,237]. Experts in the field [31,246,247] have proposed that a multisite registry is necessary for enhanced DBS management and the standardization of protocols. Such TS registries [236,248] have already allowed progress to be made in the research and clinical care discussed herein [235,237].
The current review identified novel findings that depart to some extent from previous reviews. For OCD treatment, NAc DBS led to modest changes, with 50-70% reaching partial response from 12 months of therapy and around 30% improvement over longer periods. Optimization of therapy through contact selection and adjunct behavioral therapy can enhance NAc DBS efficacy and achieve 85% response. ALIC DBS led to rapid and dramatic efficacy with further changes seen at long-term follow-up, with response in at least 60% of patients being achieved. Varying patterns of response were observed from ALIC DBS, which appears to rely on stimulation intensity and fine-tuning of programming. VC/VS DBS can achieve high efficacy and shows superiority over other striatal regions and the amSTN. Rapid improvements occurred in closed-label conditions from VC/VS DBS, and response continued to improve, with long-term response rates of 80% or more. amSTN investigations were limited, but 75% response was achieved from long-term treatment, supporting further applications for OCD. Direct investigations of BNST are recommended, as there is evidence for superiority over other targets (ALIC, NAc) and long-term efficacy, with up to 80% response.
There is extensive behavioral and programming evidence that the VC/VS and BNST encompass the DBS 'sweet spot' of the striatum; as such, these regions have been identified as optimal targets for OCD therapy. Extensive programming prior to closed-label conditions is critical to achieve full DBS effects. Also, CBT is likely to optimize DBS effects and should be incorporated into care.
For TS treatment, a smaller number of patients have been implanted than for OCD, but with less variance in targets, with thalamus and GPi being consistently chosen, even though the targeted subregions did show variation across studies. Continuous CM thalamic DBS led to consistent and sustained efficacy for tic suppression, with 60-100% response rates. Response usually occurred within months, with continued improvements for up to six years following implantation. Despite limited efficacy from blinded treatment, there is robust open-label support for GPi DBS in TS. Effects occured rapidly (1-3 months) with gradual improvement, such that years after surgery, response rates of 50-88% can be achieved. Efficacy from affective and motor divisions of the GPi were consistent, but there was greater evidence to support targeting of the anterior-limbic functional division in the long-term treatment of TS.
Treatment for TS appeared to be less reliant on programming optimization compared to OCD patients; however, this does not exclude the necessity to optimize programming for TS, or refute the hypothesis that greater efficacy may be achievable by doing so. Furthermore, TS patients tend to achieve a more rapid response than those with OCD, and show continued improvement following years of therapy. Younger age appears to be a predictor of DBS response for TS. For TS patients with comorbid OCD, stimulation of TS targets will likely achieve symptom suppression for both conditions. DBS mediates change in obsessive-compulsive behavior through modulation of a shared network, which is consistent across targets and conditions. In comparison, DBS effects on tics occur through distinct networks depending on the stimulated region. Connectivity analysis, but not anatomical location, was sufficient to predict response for OCD and TS. There was an absence of behavioral or clinical predictors of response. Connectome modelling is a data-driven yet complex approach to identify biomarkers. Thus, fiber connectivity analysis is an important evolving field that should be considered in anatomical targeting and postoperative management.
The complexities of DBS-mediated recovery and the effects on different determinants of functioning are not fully captured within quantitative outcome measures [222,249,250]. Some patients do not achieve response on the primary outcome variable, yet choose to maintain stimulation due to improved anxiety, depression, and/or quality of life. Other patients achieve response yet struggle with newfound realizations and the 'burden of normality'. Future reviews may consider a more integrated clinical and global functioning picture; however, this is beyond the scope of the current review.
Across all techniques, there was a scarcity of research for OCRDs (other than OCD and TS), and thus, conclusions could not be made for the application of neurostimulation therapy across all OCRDs.  Data Availability Statement: Data is contained within the article (see summary tables), extracted data from articles can be made available upon request.

Conflicts of Interest:
The authors declare no conflict of interest.