Introduction

Major depressive disorder (MDD) is a chronic and disabling condition associated with significant morbidity, with an estimated lifetime prevalence of 14.6 and 11.1% in high- and low-to-middle-income countries, respectively.1, 2 Standard antidepressant drugs are thought to primarily act inhibiting or otherwise modulating monoamine neurotransmission.3, 4 Only approximately a third of patients with MDD achieve remission after an adequate trial with a first-line antidepressant agent.5, 6 The failure to respond to one or more adequate antidepressant trials (that is, with adequate doses and duration) indicates the presence of treatment-resistant depression (TRD), although the definition for TRD has varied across trials.7, 8, 9 Moreover, the use of first-line antidepressants is associated with safety and tolerability concerns.10 TRD is associated with elevated health-care costs, morbidity, reduced quality of life and work productivity, and thus meaningfully contributes to the overall burden of MDD.11 Therefore, the search for mechanistically novel therapeutic options for TRD is currently a research priority.12 In last decade, accumulating evidence indicates that ketamine is efficacious and may provide rapid antidepressant effects for patients with TRD.13, 14 Nevertheless, its long-term efficacy remains unclear, and benefits should be weighed against untoward effects including but not limited to dissociative effects, potential for abuse and deleterious cognitive side effects at higher or repeated doses.15 Given the significant public health impact of TRD, and the limited effectiveness of available psychological and pharmacological treatments for chronic TRD patients, the field has witnessed an increasing interest in exploring the therapeutic potential of non-pharmacological interventions like repetitive transcranial magnetic stimulation, transcranial direct current stimulation, vagus nerve stimulation, epidural cortical stimulation, electroconvulsive therapy (ECT) and deep brain stimulation (DBS) as therapeutic options for TRD.16, 17, 18, 19, 20 Herein, we provide an integrative review of preclinical and clinical studies that have assessed DBS, a relatively recent neuromodulatory treatment modality, within different neuroanatomical targets as a putative treatment for TRD. Details of the search strategy and criteria for selection of references are provided in the supporting online material.

Deep brain stimulation

In DBS surgery, the electrode is stereotactically implanted into specific neuroanatomical targets where stimulation is provided via a pacemaker-like stimulator device that delivers continuous electrical stimulation.21 A schematic representation of the apparatus is provided in Figure 1. Benabid and Pollak pioneered modern DBS over ablative surgery for the treatment of movement disorders by targeting the thalamic nucleus ventralis intermedius, globus pallidus internus and subthalamic nucleus (STN).22, 23 Because of the tremendous clinical success of DBS as a treatment for movement disorders and reported concurrent beneficial effects on neuropsychiatric manifestations, this neuromodulatory approach has also been explored as a possible treatment for many mental disorders including obsessive–compulsive disorder and intractable depression.24, 25, 26 Interestingly, ketamine’s rapid antidepressant and anti-anhedonic effects are associated with alterations in glucose metabolism in brain structures, that are also serving as potential targets for DBS, like the habenula, insula, prefrontal cortex (PFC) and anterior cingulate cortex in patients with TRD.27, 28 Despite the incomplete understanding of the underlying mechanisms of action (MOA) involved in the therapeutic response to DBS among patients with TRD,29 several brain targets have been tested, and thus DBS has evolved to become a promising strategy for the management of TRD.3, 24, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39

Figure 1
figure 1

Schematic representation of six DBS targets tested for the management of TRD. A quadripolar DBS electrode is implanted into the selected brain targets, which is connected with long connecting lead extension wires to the pacemaker-like device (pulse generator) that is mounted under the skin of the chest. The name of pioneering academic or research institutions that have tested each brain target are provided. DSB, deep brain stimulation; DR, dorsal raphe; LC, locus coeruleus; TRD, treatment-resistant depression; VTA, ventral tegmental area.

PowerPoint slide

Full size image

Clinical and preclinical outcomes

Clinical studies have assessed putative therapeutic effects of DBS in participants with TRD across several major brain targets namely Brodmann area 25 or subcallosal cingulate gyrus (SCG), nucleus accumbens (NAc), ventral capsule/ventral striatum (VC/VS) or ventral part of anterior limb of the internal capsule (vALIC), medial forebrain bundle (MFB), lateral habenular complex (LHb), and inferior thalamic peduncle (ITP). The names of the respective pioneering institutions that conducted DBS manipulations across these brain targets for the treatment of TRD are presented in Figure 1. The ideal settings for achieving optimum antidepressant effects in humans remain unclear, although it is worthy to note that therapeutic effects may vary as a function of respective DBS targets and stimulation parameters, and also according to clinical characteristics of individual patients.40 The exploratory meta-analysis conducted by Smith41 suggests that active DBS applied to some of the above brain targets could be 71% more efficacious than sham treatment (summary effect size: 1.71; 95% confidence interval: 1.47–1.96) for TRD. However, only eight studies were available, and effect sizes could not be separated according to specific brain targets. Detailed information pertaining to clinical and preclinical central nervous system targets chosen for DBS as a treatment for depression is provided in the following sections, and is briefly summarized in Tables 1 and 2.

Table 1 Summary of clinical trials and case reports of DBS applied to various brain targets for the management of TRD
Full size table
Table 2 Application of DBS in the various neuroanatomical targets using in vivo preclinical models
Full size table

Details of neuroanatomical substrates tested across preclinical DBS studies for depression-like phenotypes are diagrammatically depicted in Figure 2. As presented in Table 2, stimulation parameters markedly varied across preclinical investigations. Yet, settings between 60 and 130 Hz for frequency, 60 and 200 μs for pulse width and 50 and 300 μA for amplitude exhibited promising antidepressant-like effects across various DBS targets. Preclinical studies have also pointed that different stimulation parameters and neuroanatomical locations may influence antidepressant-related effects. However, at least in part because of the complex and multifactorial pathophysiology of human depression, currently no animal model meets all validity criteria (including predictive validities),42, 43 and hence limitations of preclinical models should be considered when inferences pertaining to depression in humans are made.

Figure 2
figure 2

Diagrammatic representation of DBS brain targets that have been tested in preclinical investigations. The outline of the sagittal diagram through the rat brain indicates the vmPFC, NAc, STN, MFB and LHb, which are not in scale. As denoted, monopolar or bipolar electrodes are stereotactically implanted in any one of the brain targets. At the time of stimulation, implanted electrodes are connected with long connecting wires to external stimulators, which delivers electric current as per protocol. Important functional and biochemical changes following DBS applied to each brain target are highlighted. 3V, third ventricle; 4V, fourth ventricle; 5-HT, 5-Hydroxytryptamine; AMPA, α-Amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid; BDNF, brain-derived neurotrophic factor; CC, corpus callosum; CREB, c-AMP-response element-binding; DA, dopamine; DBS, deep brain stimulation; DLP, depression-like phenotype; DRN, dorsal raphe nucleus; LHb, lateral habenula; MFB, medial forebrain bundle; mPFC, medial prefrontal cortex; mTOR, mammalian target of rapamycin; NA, noradrenaline; NAc, nucleus accumbens; OFC, orbitofrontal cortex; STN, subthalamic nucleus; trkB, tropomyosin receptor kinase B.

PowerPoint slide

Full size image

DBS of the SCG

Clinical studies

The SCG may have a pivotal role in the regulation of sadness and negative emotions occurring in both depressed and healthy subjects.24, 44, 45, 46 Clinical outcomes for this DBS target are summarized in Table 1. Mayberg and coworkers24 initially reported that four out of six patients with TRD achieved antidepressant response after 6 months of open-label SCG-DBS. Afterwards, Lozano et al.47 reported that 40% of participants with TRD (n=20) achieved response after 1 week of stimulation, whereas 60 and 55% of patients met response criteria at 6 and 12 months, respectively, in an open-label trial that tested SCG-DBS. A few case reports also demonstrated efficacy for this target in the TRD patients.48, 49, 50, 51, 52 Long-term outcomes of SCG-DBS for the aforementioned open-label trial were subsequently reported, with 55–60 and 64.3% response rates after 1–3 and 3-6 years’ follow-up visits, respectively,53 and four participants either committed or attempted suicide over the course of study, although because of uncontrolled design of this study it could not be established whether this serious adverse effect was related to DBS or to illness evolution per se. Holtzheimer et al.54 replicated those findings in an uncontrolled study involving 17 participants with TRD after SCG-DBS. This trial reported 43.6–70.1% response rates (assessed with the Hamilton depression rating scale) following 24-week, 1-year and 2-year follow-up visits. Although this study was uncontrolled, it should be noted that the blinded discontinuation of SCG-DBS resulted in the full relapse of depressive episodes in all three patients in which this was attempted, whereas depressive symptoms improved once stimulation was reinstated.54 Furthermore, these robust therapeutic benefits and brain metabolic changes have been further replicated in a preliminary uncontrolled observation;55, 56 seven out of eight TRD patients achieved antidepressant response after SCG-DBS at a 6-month follow-up visit. In a more recent randomized, double-blind, sham-controlled crossover trial Puigdemont et al.57 reported reversal of depressive scores after active SCG-DBS application in five participants with TRD,57 and long-term high frequency stimulation (HFS) exhibited better antidepressant response.58

Berlim et al.59 conducted an exploratory meta-analysis of four observational studies, and verified that response and remission rates following SCG-DBS were 36.6 and 16.7%, 53.9 and 24.1%, and 39.9 and 26.3% at 3-, 6- and 12-month follow-up periods, respectively. Moreover, open-label studies reported remission rates ranging from 29 to 58% after chronic stimulation of the SCG for up to 12–36 months,60, 61, 62 although evidence suggests that longer pulse durations could influence pathways farther from the SCG, and activation of the SCG–NAc network may contribute to antidepressant response after SCG-DBS.31 Patient-specific tractography modeling provided relevant insights for the identification of electrode location and critical neuronal tracts that could mediate antidepressant responses to SCG-DBS, and may also mitigate inter-individual variability in the direct effects of stimulation on brain circuitry.45, 63, 64 Using tractography-based surgical targeting, Riva-Posse et al.45 demonstrated 72.7% (8 of 11 participants) and 81.8% (9 of 11 participants) response rates at 6 months and 1 year after SCG-DBS, respectively, whereas six patients were in remission at both time points.

SCG-DBS may also improve memory as well as executive and motor functioning in participants with TRD, without meaningful adverse effects upon cognitive measures.65, 66, 67, 68, 69, 70, 71 Moreover, patients with TRD may exhibit transient emotional hypersensitivity and a predictable worsening of depressive symptom scores at the initial phases of SCG-DBS,55 thus long-term treatment could be critical for central nervous system remodeling and neuroplasticity, and hence to therapeutic benefits.72 In SCG-DBS-operated participants with TRD, the most frequently observed surgery-associated adverse events were hardware-related (11.4%), suicidality (9.3%)73 and risk of partial seizures.74 In addition, a multicenter, randomized controlled trial (RCT) of SCG-DBS for TRD (BROADEN study) was prematurely interrupted based on results of an interim futility analysis (St Jude Medical Clinical Study).38, 75 Recently, Holtzheimer et al.75 published results of multisite, randomized, double-blind, sham-controlled SCG-DBS study in 90 TRD participants. They did not observe statistically significant effects in the primary efficacy outcome between the stimulation (20%) and control group (17%), similarly in a double-blind study of eight participants Merkl et al.76 reported no difference between active versus sham group. However, 33–48% participants displayed an antidepressant response and 25% achieved remission with up to 2 years of open-label SCG-DBS.75, 76 Nevertheless, the lack of therapeutic response in TRD patients after SCG-DBS may be due to the wrong placement of electrodes or misleading points of stimulation.45 Preliminary evidence for site-specific clinical responses following SCG-DBS was provided in the small open-label trial conducted by Accolla et al.77 that enrolled five participants with TRD; noticeable antidepressant responses following stimulation of the posterior gyrus rectus region in one patient with TRD was observed, although none of the participants in whom DBS was applied to the originally planned Brodmann area 25 were considered responders.

Preclinical studies

The infralimbic cortex, which is part of the rodent ventromedial PFC (vmPFC) is thought to represent the rodent homologous of the human SCG (Brodmann area 25).78 Preclinical studies indicate that DBS applied to the vmPFC may lead to antidepressant-like effects across several preclinical models including the forced swimming test (FST), sucrose preference test and novelty suppressed feeding (Table 2).79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90 Electrical stimulation of this neuroanatomical target has also been shown to increase hedonic and motivational states in animal models of depression,91, 92 and to reverse depressive-like phenotypes induced by chronic unpredictable stress (CUS), chronic social defeat stress, olfactory bulbectomy and in a putative therapy-refractory depressive-like rat line.82, 87, 88, 89, 93, 94, 95

Hamani et al.79 characterized the optimal brain stimulation settings for DBS applied to the vmPFC in rats. DBS targeted to the infralimbic cortex also resulted in antidepressant-like effects, may be through a reversal of synaptic metaplasticity and increments in mitosis.85, 96, 97, 98 In another study, DBS applied to prelimbic mPFC led to different functional brain alterations.99 Moreover, chronic DBS applied to the vmPFC may reverse stress-induced behavioral deficits in the sucrose preference test, FST, novelty suppressed feeding and elevated plus maze models, and may also increase brain-derived neurotrophic factor levels, blood vessel size, synaptic density and astrocyte size in the hippocampus.82, 100, 101, 102 Altogether, these data suggest that stimulation parameters need to be precisely set for achieving meaningful responses after prelimbic or infralimbic PFC-DBS.

Hamani and colleagues80, 82, 103 provided preclinical data to support a putative role for the serotonergic system as a mediator of the antidepressant-like effects of vmPFC-DBS. Moreover, vmPFC stimulation was found to produce antidepressant, anxiolytic and hedonic effects through the modulation of dorsal raphe nucleus (DRN) circuitry in the CUS animal model of depression.86, 87, 94, 104 Furthermore, adjuvant treatment with monoamine oxidase inhibitors was reported to potentiate behavioral effects of vmPFC-DBS in the FST.81 However, co-administration of buspirone, pindolol or risperidone did not significantly alter antidepressant-like effects of DBS.105 This further suggests that vmPFC-DBS might involve the modulation of prefrontal projections to the DRN, which is a brain region involved in serotonin (5-HT) synthesis and release. Apart from the DRN, neurostimulation of the vmPFC may also remotely affect activity of the ventral tegmental area (VTA) and locus coeruleus.106, 107

Possible MOA

Depression has been associated with increased activity of the SCG and dysregulated corticolimbic networks.30, 96, 108, 109, 110, 111 Furthermore, SCG neurons are preferentially more responsive to negative (unpleasant) emotions.112 SCG-DBS may ameliorate depressive symptoms or more specifically anhedonia in patients with TRD.113 Antidepressant response to SCG-DBS in participants with TRD has been associated with frontal asymmetry, higher frontal theta cordance (θ) and local field potential broad α-band activity.114, 115, 116 Suppression of gamma oscillations and increased θ-gamma coupling by active SCG-DBS stimulation may also enhance gamma-aminobutyric acid neurotransmission.117 Therefore, it seems that SCG-DBS stimulation may have a key role in normalizing spectral rhythms in brain networks related to depression neurobiology.

DBS applied to the NAc

Clinical studies

Anhedonia was one of the first manifestations to improve during NAc stimulation in participants with TRD, which also reported a heightened perception of pleasurable activities.34, 36, 113, 118 The ideal parameters of electrical stimulation that could provide adequate antidepressant responses are summarized in Table 1. Schlaepfer et al.34 reported short-term outcomes in three patients with TRD who underwent NAc-DBS. Twelve months of chronic, open-label NAc-DBS led to a decrease in the metabolism of the SCG, amygdala and prefrontal regions with 45 and 9% response and remission rates, respectively, in 10 participants with TRD.118 The long-term open-label trial conducted by Bewernick et al.36 reported a sustained antidepressant effect for NAc-DBS in 11 participants with TRD (45.5% response rate at 48-month follow-up). Yet, only five participants completed this 4-year trial.36 Moreover, there was no evidence of cognitive deterioration in agreement with data from the same research group.36, 119 An open-label trial with six participants reported 50% responders, and did not show signs of cognitive deterioration.120 Altogether, these data suggest that NAc-DBS could be efficacious for the management of TRD.

Preclinical studies

Application of DBS to the NAc-shell triggered impulsive behaviors accompanied by significant increases in dopamine and 5-HT levels in the NAc. However, DBS applied to the NAc core led to antidepressant-like effects without significantly altering levels of 5-HT and dopamine,121, 122 although 2 consecutive days of bilateral stimulation elicited a rapid increase in dopamine and 5-HT release in the orbital PFC.123 HFS or low frequency stimulation of the NAc-DBS also produced distinct region-specific and frequency band-specific changes in local field potential oscillations,124 therefore, suggesting that different stimulation parameters may engage distinct brain areas, which could then influence antidepressant responses to NAc-DBS. Furthermore, a recent study reported that DBS applied to the lateral NAc-shell reduced motivation for sucrose, whereas stimulation of the medial NAc–shell selectively increased the intake of chow.125 These findings suggest that subdivisions of the NAc–shell may influence motivational eating behavior, and may point to dissociable effects of NAc-DBS in alleviating anhedonia in depression. Yet, the field awaits further investigations. Recently, Lim et al.87 observed reduced anxiety-like behaviors and increase in motivation for chow intake in the CUS depression model after HFS of the NAc–core as compared to that in the NAc–shell.

Accumulating evidence indicates that NAc-DBS may decrease depressive-like behavior in CUS-induced animal model of depression.86, 87, 93, 126, 127 Similar results were documented in the high anxiety-related behavior mouse model and in a chronic adrenocorticotropic hormone model of TRD after NAc stimulation.91, 128 Hamani et al.126 observed comparable antidepressant-like effects after stimulation either the vmPFC or the NAc in the FST. Nevertheless, only NAc-DBS influenced different subcortical relay centers in the brain reward circuitry. In contrast, in another study vmPFC-DBS outperformed NAc-DBS.127 Antidepressant-like effects were significantly higher after interrupted stimulation of the NAc compared with intermittent stimulation,128, 129 which was associated with decreased levels of tyrosine hydroxylase, dopamine and norepinephrine in the PFC. Although acute DBS-NAc did not significantly alter hippocampal neurogenesis,130 DBS-NAc–core lowered CUS-induced increase in c-Fos expression in the magnocellular part of the medial vestibular nucleus compared with CUS sham.131

Possible MOA

Consistent with imaging studies in humans, a significant increase in blood oxygenation level-dependent signal in the insula, thalamus and parahippocampal cortex and a decrease in the SCG and PFC during stimulation of the NAc functional magnetic resonance imaging was reported in a pig model.132 Moreover, modulation of the NAc may normalize disease-related hypermetabolism in the SCG and in prefrontal regions including the orbitofrontal cortex, with possible procognitive effects,118 which are similar metabolic decreases observed in patients undergoing SCG-DBS.47 Thus, it has been hypothesized that effects on the SCG could also mediate antidepressant effects of NAc-DBS.118 In addition, as reviewed in the section above, site-specific effects on monoamine neurotransmission have been implicated as a putative antidepressant mechanism of NAc-DBS.

DBS applied to the VC/VS or vALIC

Clinical studies

Application of DBS to the VC/VS (also referred as vALIC in some studies) significantly decreased anxiety and depressive symptoms in participants with obsessive–compulsive disorder, thus providing a rationale for testing its efficacy in samples with TRD.65, 133, 134 Obsessive–compulsive disorder patients who underwent DBS of the VC/VS showed a reduction of cerebral blood flow in the SCG, which appears to be metabolically hyperactive in patients with MDD.30 An open-label pilot trial conducted by Malone et al.32, 135 assessed the efficacy of VC/VS-DBS in 17 patients with TRD. Response rates of 53 and 71% at 12 month and last (14–67 months) follow-up visits, respectively, and a 40% remission rate at last follow-up (6–51 months) were observed. A case report described smoking cessation in a single responder after VC/VS-DBS.136 However, double-blind, randomized, sham-controlled trials of VC/VS-DBS for MDD have thus far provided inconsistent findings.137, 138 In a 16-week sham-controlled randomized trial followed by an open-label continuation phase, Dougherty et al.137 did not observe significant differences in treatment response rates in the active DBS group. The same research group subsequently reported that vALIC-DBS did not influence cognitive function compared with sham.139 In addition, adverse events were more severe for vALIC-DBS compared with the sham group (Table 3). Nevertheless, 25 participants underwent 52-week open-label vALIC-DBS (optimization phase), and 10 participants out of 25 with TRD were classified as responders (40%).140, 141 Sixteen participants were subsequently randomized to active-sham or sham-active groups in a cross-over design, and participants scored significantly lower during active rather than during sham DBS.140 Therefore, the antidepressant efficacy of DBS primarily applied to the VC/VS (or vALIC, a brain structure slightly anterior and ventral to the VC/VS) remains to be established.

Table 3 Summary of most common observed adverse effects after DBS of various brain targets
Full size table

Preclinical studies

As ALIC is not well developed in rodents, Hamani et al.126 had chosen white matter fibers of the frontal region for electrical stimulation as this neuroanatomical structure resemble the ALIC in human. Application of DBS in white matter fiber influenced the large brain regions of the cortical and subcortical structures, without producing a significant antidepressant-like effect in FST.126

Possible MOA

Neuroimaging studies conducted in participants with obsessive–compulsive disorder who underwent DBS in this target demonstrated modulation of different nodes of the cortico–striatal–thalamic–cortical circuitry, including the orbitofrontal cortex, basal ganglia, along with a reduction in metabolic hyperactivity of the SCG, observed particularly in participants with co-occurring MDD.142, 143 In addition, the VS encompasses structures like the bed nucleus of the stria terminalis (BNST) and the NAc, which are regions putatively involved in the regulation of stress and reward-motivational pathways in individuals with depression.144, 145 Nevertheless, the antidepressant efficacy as well as possible MOA of VC/VS-DBS remains unclear.

DBS of the MFB

Clinical studies

Three different academic institutions have assessed the efficacy of MFB-DBS in samples with TRD (Figure 1 and Table 1). However, evidence for putative antidepressant effects of MFB-DBS remains relatively unexplored as only data from 11 participants with TRD were provided from two uncontrolled studies.33, 35, 39 Despite these limitations, findings suggest that MFB-DBS could confer rapid and long-lasting antidepressant effects. Short-term bilateral stimulation of the superolateral-MFB showed a rapid reduction in the severity of depressive symptoms in six out of seven participants within 2 days of stimulation, and four out of seven patients met criteria for treatment response after 1-week stimulation,33 whereas at the last observation (after 12–33 weeks) six participants (85.7%) were treatment responders. Fenoy et al.39 also reported in their interim analysis a robust and rapid antidepressant response in an open-label trial of bilateral MFB-DBS, in which three out of four participants with TRD were responders after 1 week of DBS initiation, and two of four participants displayed >80% decrease in MADRS scores after 26 weeks of stimulation. Recently, Bewernick et al.35 provided long-term data for their open-label trial.33 At the time of analysis, six out of eight participants (75%) were treatment responders at 12-month follow-up; these antidepressant effects remained stable for up to 4 years. Furthermore, no evidence of cognitive impairments were noted even after several months of stimulation in this target.35, 39 The most frequently reported adverse effect associated with MFB-DBS were oculomotor disturbances (Table 3), which in most cases diminished over time. In a recent case study, a 58-year-old patient perceived marked mood improvement effect after 1 week of MFB-DBS.146 However, this patient experienced oculomotor side effects that failed to remit over time. The patient was re-operated after 2 years, and responded to DBS applied to the BNST without troublesome oculomotor side effects. These data provide further evidence that in selected clinical situations this adverse effect could be a reason for discontinuing MFB-DBS.39 Taken together, these preliminary data suggest that MFB-DBS could be efficacious for the management of TRD. Thus, the design of a randomized, sham-controlled trial is warranted to confirm those promising findings.

Preclinical studies

Similarly to the NAc, the MFB has a critical role in the regulation of motivation, and thus may contribute to anhedonia.147 In preclinical studies, the MFB-HFS generated antidepressant-like effects was associated long-lasting neural adaptation in target regions of the mesolimbic–mesocortical circuitry (Table 2).92, 148, 149 When Flinders sensitive line rats received bilateral HFS to the MFB, antidepressant-like effects in the FST were observed.92 Similar antidepressant-like effects were reported following bilateral HFS in rats accompanied by an increase in expression of the immediate early gene, zif268, in the piriform cortex, prelimbic cortex, NAc shell, anterior regions of the caudate/putamen and the VTA.150 However, no significant changes in the release of either dopamine or serotonin at the level of the NAc were observed. Although MFB-DBS was reported to mitigate depressive-like behaviors and increase pleasurable or rewarding experiences after mild VTA lesion in rats, it was unable to reverse a despair phenotype after severe VTA lesions in rats.151 Recently, rapid antidepressant-like effects and increased expression of dopamine D2 receptors in the PFC have been reported following acute stimulation of MFB in rats.152

Possible MOA

The MFB is a central component of the mesolimbic–mesocortical dopamine reward system,153, 154, 155 and it is interconnected with several other DBS targets.24, 34, 156, 157 MFB-DBS may activate the mesocorticolimbic system by increasing neuronal activity within these regions through the modulation of dopaminergic and glutamatergic neurotransmission.33, 39, 158, 159 Recently, optogenetic activation of VTA dopaminergic neurons led to increased functional magnetic resonance imaging blood oxygenation level-dependent signals in the NAc concomitantly with an increase in motivational behavior in rats.160 It is thought that modulation of the MFB via DBS may recruit descending glutamatergic fibers from the mPFC to the VTA, and may thus indirectly modulate dopaminergic firing at the VTA.161 MFB-DBS may also modulate upstream cortical regions.33 Yet, the precise MOA underlying putative antidepressant effects of MFB-DBS remain to be elucidated.

DBS of the LHb

Clinical studies

Hyperactivity of neurons at the LHb has been suggested to have a pathophysiological role in MDD.83, 162 Bilateral LHb-DBS may decrease activation within this neuroanatomical region.163 Sartorius et al.163 observed a sustained remission of depressive symptoms after 4 months of DBS in a patient with TRD. A marked re-emergence of depressive symptoms elapsed after the erroneous cessation of stimulation, further pointing to the potential therapeutic usefulness of LHb-DBS for TRD. As evidence from single case report is available, more studies would decide the relevance of this target for TRD.

Preclinical studies

LHb-DBS has been shown to reverse depressive-like behaviors in the CUS as well as in the chronic adrenocorticotropic hormone and learned helplessness models of TRD.83, 84, 87, 164, 165 DBS applied at the LHb gradually increased peripheral and brain levels of norepinephrine, dopamine and 5-HT, which peaked after 28 days of treatment.84 In a comparative study of vmPFC-DBS versus LHb-DBS, Lim et al.87 showed that, although vmPFC-DBS produced a fourfold increase in hippocampal 5-HT release, LHb stimulation produced an ~55–70% increase in striatal 5-HT release. Moreover, HFS of the LHb was reported to counteract depressive-like behavior in the CUS animal model.87

Possible MOA

Electrical stimulation of the LHb was found to significantly inhibit the firing of dopaminergic neurons in the substantia nigra pars compacta and the VTA.166, 167 In a genetic animal model of TRD, a significant alteration of regional cerebral blood volume was observed within the LHb.168 Furthermore, the antidepressant effects of LHb-DBS may involve monoamine pathways as there are strong interactions and direct efferents from the LHb to the DRN, locus coeruleus and the VTA.169

DBS of the ITP

The ITP encompasses a bundle of fibers connecting the dorsomedial thalamus to the orbitofrontal cortex,170 which is dysregulated in participants with MDD.171 Moreover, surgical lesions to the ITP may disrupt the inhibitory action of the thalamo–orbitofrontal system, and may also promote antidepressant-like effects in preclinical models.172 Two case reports have described the results of ITP-DBS in a single patient with TRD.173, 174, 175 A significant decrease in Hamilton depression rating scale scores (from 42 to 6) without side effects at high-frequency DBS settings of 3.5 V and 450 μs of pulse width was verified.173 Recently, antidepressant efficacy of ITP and ALIC/BNST stimulation were compared in blinded crossover study.176 Although no superiority of either targets were noticed, 6/7 patients preferred stimulation in ALIC/BNST.176 Clearly, this brain target deserves further exploration in clinical trials.

DBS of the STN

Acute treatment with STN-DBS significantly improved depression-like behaviors in preclinical models (Table 2).177, 178 Similarly, post hoc clinical evidence suggested possible antidepressant effects following STN-DBS in samples with Parkinson’s disease (PD).179, 180, 181, 182, 183 In contrast, worsening of depressive-like behaviors in rats, and an increased risk of transient cognitive and psychiatric complications have been reported after STN-DBS for PD.184, 185, 186 However, depressive symptoms were less frequent after pallidal-DBS compared with STN-DBS in PD patients.187 In a case report, an increase in current frequency of STN-DBS from 60 to 185 Hz led to a significant improvement of depressive symptoms with complete resolution of his suicidal thoughts in a PD patient.188 In sum, divergent findings in both the clinical and preclinical literature have diminished enthusiasm in the field for further testing this target in clinical trials involving participants with TRD.

Conclusion

Evidence indicates that DBS can be delivered to discrete brain targets and may directly modulate brain activity in a limited brain structure and also activity of interconnected (that is, more distant) brain networks. Furthermore, available evidence suggests that it could be a relatively safe and well-tolerated non-pharmacological therapeutic option for TRD.24, 39, 47, 189, 190 Six prominent DBS targets have been tested for the management of TRD. Interestingly, in most circumstances those targets have been tested in clinical trials prior to the conduction of preclinical studies. However, the underlying mechanisms of DBS-related antidepressant effects remain elusive, and may vary as a function of the primary brain target as well as with stimulation parameters. In addition, clinical trials to date have methodological shortcomings such as a lack of proper randomization, the lack of a control (that is, sham-stimulated) group, small sample sizes, heterogeneity of participants across trials (for example, different definitions of TRD) as well as a lack of consensus algorithms for DBS delivery.40 Herein, the underlying mechanisms for putative antidepressant effects of DBS identified in clinical trials and preclinical studies are reviewed. We sought to determine which brain target(s) most consistently elicited antidepressant responses. Out of six DBS targets, only VC/VS and the SCG have been investigated in multicenter, randomized, sham-controlled trials. However, these trials failed to confirm the efficacy of those targets in participants with TRD.57, 75, 137 On the other hand, open-label clinical studies verified high response rates (ranging from 60 to 78%) in participants with TRD after the application of DBS to several targets.191 Furthermore, discrepant findings across these open-label trials may be ascribed to interpersonal anatomical differences (size and shape) of DBS targets, which could be rectified by mapping the internal neuroanatomical structure of individual patients.35, 39, 45, 77, 192, 193 Taken together, detailed characterization of the anatomical, physiological and neurochemical substrates underlying the effects of DBS may delineate suitable brain targets for the management of TRD.

Clinical implications and future directions

The use of DBS in the management of TRD remains at investigational stages. How research in this treatment modality for depression should continue and improve from this point represents a meaningful challenge for neuroclinicians.38, 40, 194 As most neuroanatomical centers regulating affect and motivation are interconnected,3, 195, 196, 197 DBS could influence distant brain networks regardless of the initial target.198 Tractography and other imaging techniques could provide valuable resources to guide surgical procedures, and also for post-surgery monitoring.39, 40, 45, 47, 72, 118 The identification of individual clinical characteristics as well as biomarkers of treatment response within the emerging framework of precision psychiatry also provides a relevant yet relatively unexplored research direction.114, 115, 165, 193, 199 It should be noted that the effectiveness of unilateral versus bilateral DBS of brain targets warrant confirmation in randomized, double-blind, sham-controlled trials.39, 79, 80, 152, 200 Importantly, it is yet debatable whether non-responders to one target might benefit from DBS delivered to an alternative target. Taken together, the selection of patients, target brain regions, parameters of stimulation and identification of early biomarkers are necessary steps to be taken to provide more consistent evidence for this promising treatment modality for intractable depression. It should be noted that RCTs of DBS applied to the SCG, VC/VS or vALIC have thus far provided inconsistent results. Therefore, adequately powered and well-designed double-blind RCTs are warranted to provide a more accurate assessment of the efficacy and safety for DBS delivered at different brain targets as a therapeutic option for TRD.