**1. Introduction**

Major depressive disorder (MDD) is a common mood disorder that a ffects one's feelings, thoughts, and behavior. According to the DSM-5, for a diagnosis of MDD, five of the following symptoms need to be present for at least two weeks: depressed mood, reduced interest or pleasure, weight loss or reduced appetite, insomnia or hypersomnia, psychomotor agitation or retardation, fatigue or loss of energy, worthlessness or excessive guilt, impaired concentration or indecisiveness, and recurrent thoughts of death or suicidal ideation or attempts. Either 'depressed mood' or 'loss of interest or pleasure' is essential for a diagnosis [1]. The total number of people su ffering from MDD worldwide was estimated to be 322 million in 2015 and its prevalence increased by 18.4% between 2005 and 2015 [2]. Therefore, effective treatment of MDD merits intense consideration.

Whereas psychotherapy and antidepressant medication are e ffective in the majority of patients, approximately one third of patients do not respond to these therapies. In the Sequenced Treatment Alternatives to Relieve Depression (STAR\*D) trial, the cumulative remission rate of MDD patients after four successive treatments was 67% [3]. In line with this, a meta-analysis of 92 studies of the e ffectiveness of psychotherapy showed that 62% of patients no longer met the criteria of depression after treatment [4]. Failure to respond to a treatment algorithm of several steps is commonly referred to as treatment resistance, although there is still discussion about the exact definition of treatment refractoriness [5]. Treatment-resistant depression (TRD) is associated with more (comorbid) mental health disorders, a higher number of hospitalizations, and more suicide attempts, leading to higher treatment costs compared to non-TRD [6]. In addition, patients with TRD show a higher demand of healthcare resources and costs of health care compared to non-TRD patients [7]. Various alternative treatment options for TRD are currently being investigated, including vagal nerve stimulation [8], repetitive transcranial magnetic stimulation (rTMS) [9], and deep brain stimulation (DBS) [10].

The aim of this narrative review is to provide an overview of recent studies of the e ffectiveness of DBS in TRD with a special focus on the relationship between the targeted brain structures and clinical outcomes. Based on these findings, we discuss the importance of distinguishing between di fferent clinical phenotypes of depression that would allow for more personalized symptom-based treatment approaches, which may be a key factor in improving treatment outcomes.

### **2. Recent Insights on the Pathophysiology of Depression**

It is hypothesized that in depression, there is an imbalance in the limbic cortico-striatal-thalamo-cortical (CSTC) mood circuits [11,12], ye<sup>t</sup> many aspects of circuitopathy in MDD remain largely unknown. Based on di fferent models [11,12], three main components of the CSTC mood circuits have been proposed (Figure 1). First, the ventral component is essential for recognizing emotions and initiating an adequate emotional and behavioral response. In this circuit, the amygdala, ventral striatum, ventral part of the anterior cingulate cortex, orbitofrontal cortex, ventrolateral prefrontal cortex, and downstream structures such as the hypothalamus and locus coeruleus are involved. Second, the dorsal component that regulates the emotional responses and requires cognitive processing. Here, the dorsolateral and dorsomedial prefrontal cortex, the dorsal part of the anterior cingulate cortex, and the hippocampus are involved. Third, a modulating region is present, although no consensus has been made about its precise anatomical organization and function. Some have suggested that this component consists of the thalamus and the rostral part of the anterior cingulate cortex [11–13]. As implied by Mayberg et al., the model of depression indicates that depression is associated with a decreased activity in dorsal limbic and neocortical regions and a relative increase in ventral paralimbic regions. Treatment of depression therefore requires the inhibition of the overactive ventral regions, resulting in the disinhibition of the underactive dorsal regions. To mediate this process, proper functioning of the rostral cingulate cortex is required [12]. These mood circuits overlap with the circuitry involved in compulsive traits; DBS of the ventral capsule/ventral striatum (VC/VS) in treatment resistant obsessive-compulsive disorder (OCD) patients has led to improvements in mood which prompted studying the application of DBS in TRD patients [14,15].

**Figure 1.** Schematic representation of emotional processing and its neurobiological base. Figure from Moonen et al. (2017) [16] with permission.

### *Expanding the Cortico-Striatal-Thalamo-Cortical Mood Circuits*

One region that is not included in the CSTC mood circuits and ye<sup>t</sup> has been a region of interest for DBS targeting in depression for over a decade is the subgenual cingulate gyrus/cortex (SCG/SCC) [10]. This region has shown hyperactivity in untreated depressed patients [17], is part of the ventral component, and has projections to the amygdala, hippocampus, superior and medial temporal gyri, ventral striatum, mid- and posterior cingulate cortex, thalamus, hypothalamus, periaqueductal gray, and lateral habenula [18,19]. Furthermore, in recent years, it has become known that several other brain areas all belonging to the ventral component play a role in the pathophysiology of depression. Among these are the thalamic peduncles (THp) that interconnects with the prefrontal cortex including the orbitofrontal cortex (OFC) [20], the medial forebrain bundle (MFB) that projects to the frontal cortex, the NAcc and ventral striatum [21], and the ventral part of the anterior limb of the internal capsule (vALIC) which forms a homeostatic system with the MFB and the bed nucleus of the stria terminalis (BNST) [22] (Figure 2).

**Figure 2.** Cortico-striatal-thalamic-cortico mood circuits divided in a dorsal, ventral, and modulating compartment based on Alexander et al. [11], Mayberg et al. [12], and Moonen et al. [16] expanded with regions researched with deep brain stimulation (DBS) for treatment-resistant depression (TRD). DLPFC; dorsolateral prefrontal cortex, DMPFC; dorsomedial prefrontal cortex, ACC; anterior cingulate cortex, THp; thalamic peduncles, OFC; orbitofrontal cortex, VLPFC; ventrolateral prefrontal cortex, MFB; medial forebrain bundle, vALIC, ventral part of the anterior limb of the internal capsule, BNST; bed nucleus of the stria terminalis, SCG; subgenual cingulate gyrus, HPA axis; hypothalamic pituitary adrenal axis.

### **3. Deep Brain Stimulation for Treatment-Resistant Depression**

DBS is an invasive neuromodulation technique that is effective in managing clinical symptoms of neurological and psychiatric disorders, such as Parkinson's disease (PD) [23,24] and OCD [25]. At stimulation settings commonly used in clinical practice, DBS decreases the spontaneous firing of neuronal populations and activates axonal projections near the electrode [26]. This modulates pathological activity and replaces it with regular patterns of discharge with intervals of burst activity [27, 28]. More recent theories sugges<sup>t</sup> that DBS destabilizes abnormal synchronous oscillatory activity within the basal ganglia circuitry improving hyperkinetic symptomology [24]. However, the exact mechanism(s) by which DBS normalizes electrical activity in the basal ganglia and exerts beneficial effects on PD symptoms remain unknown. In DBS for TRD, target selection has mostly been based on either neuroimaging studies or clinical observations of mood improvement following DBS in OCD [10,15,29]. For these reasons, the underlying mechanisms of action are poorly studied. DBS studies for TRD (Table 1) and the outcomes for selected brain targets (Table 2) are described below.


*J. Clin.*  **2020**,

*Med.*

 *9*, 2729


*J. Clin. Med.* **2020**, *9*, 2729


### *J. Clin. Med.* **2020**, *9*, 2729


*J.* 


"-": not investigated; (d) ACC; (dorsal) anterior cingulate cortex, AG; amygdala, CES-D score; center for epidemiologic studies depression score, (d) CN; (dorsal) caudate nucleus, DMN; default mode network, DTI; deterministic diffusion tensor imaging, GP; globus pallidus, Hip; hippocampus, Hth; hypothalamus, MFB; medial forebrain bundle, OFC; Orbitofrontal cortex, (m/dl/dm/vm/vl) PFC (medial/dorsolateral/dorsomedial/ventromedial/ventrolateral) Prefrontal cortex, PGR; posterior gyrus rectus, PTN; pedunculopontine tegmental nucleus, Pt: putamen, RN; raphe nuclei, SN; substantia nigra, Th; thalamus, TRN; thalamic reticular nucleus, VP; ventral pallidum, VS; ventral striatum, VTA; ventral tegmental area.

### *3.1. Subgenual Cingulate Gyrus*/*Cortex*

The first clinical trial of DBS of the SCG for TRD was performed in 2005 and included six patients with MDD [10]. The severity of depression was measured using the Hamilton Depression Rating Scale (HDRS) and the Montgomory Asberg Depression Rating Scale (MADRS). The HDRS has been the gold standard for the assessment of depression for years [51]. A clinical response is commonly defined as a decrease in the HDRS score of more than 50% compared to baseline, and clinical remission is defined as a decrease in the HDRS score to eight or less. After one month, two out of six patients met the criteria for response. At the end of the sixth month, a response was seen in four out of six patients, with three of the patients reaching remission or near remission. Preliminary observations with positron emission tomography (PET) showed a metabolic hyperactive SCG (Brodmann area 25, Cg 25) during depressive states. It was speculated that DBS would reduce this hyperactivity [17] (Table 2). The improvement in depression scores after DBS was thought to be due to e ffectively disrupting focal pathological activity in limbic-cortical circuits. After 3 months of stimulation of the subgenual cingulate region (CG25) in patients su ffering from TRD, local cerebral blood flow (CBF) was decreased in CG25 and the adjacent orbitofrontal cortex (Brodmann area 11). Moreover, after three and six months of stimulation, CBF was decreased in the hypothalamus, anterior insula, and medial frontal cortex of long-term responders, while CBF increased in the dorsolateral prefrontal cortex (dlPFC), dorsal anterior, posterior cingulate, and premotor and parietal regions (Table 2) [10]. In the di fferent open-label trials, response rates varied from 20 to 57% after 1 month, 33.3 to 87.5% after 6 months, and 29 to 62.5% after 12 months (Table 1) [10,33,49,52–61]. In a long term follow-up, Kennedy et al. (2011) reported response rates at 1, 2, and 3 years after DBS implantation in the SCC of TRD patients of 62.5%, 46,2%, and 75%, respectively [52] (Table 1). In a case series of DBS of the SCG in five TRD patients, a decrease in the score of the depression rating scale was only found in one of the five TRD patients. This patient turned out to be stimulated in the posterior gyrus rectus (PGR) based on single subject tractography results rather than the initially targeted CG25 [62]. A recent exploratory meta-analysis of four observational studies investigating DBS for TRD (Holtzheimer et al. 2012, Lozano et al. 2012, Puigdemont et al. 2012, and Kennedy et al. 2011) reported relatively large response and remission rates following DBS treatment: the twelve-month response and remission rates were 39.9% (95% CI = 28.4% to 52.8%) and 26.3% (95% CI = 13% to 45.9%). The included studies reported a significant decrease in depression scores between 3 and 6 months (Hedges' g = −0.27, *p* = 0.003), while no additional decrease was found between 6 and 12 months, suggesting that maximal antidepressant e ffects occur mostly within the first 6 months of treatment [63]. However, adverse events can occur, including worsening of depression, suicidal ideation, and seizures (Table 1). A study consisting of a double-blind active vs. sham stimulation phase of four weeks, followed by an open-label stimulation for up to 24 months, reported no significant di fferences between the active and sham stimulation of the SCG and no reduction in HDRS scores in the first four weeks. In the open-label phase, response rates were 37.5%, 43% and 23% after 6, 12 and 28 months, respectively. Remission rates were 12.5% and 14.2% at 6 and 12 months, respectively, and 33.3% at 24 and 28 months [64].

A randomized controlled trial (RCT) investigating DBS of the subcallosal cingulate, known as the BROADEN trial, was aborted prematurely. The study lasted six months, during which all patients should have received SCC implantation surgery. After six months, blinding would have been uncovered and both groups would have been o ffered open-label DBS for another six months. At the end of the first six months, responses of the treatment group and control group were predicted to be 40% and 18.5%, respectively. In this trial, the response rate was defined as more than or equal to a 40% decrease in MADRS scores from baseline. However, after six months, only 20% of patients (*n* = 12) in the treatment group showed a response versus 17% of patients (*n* = 5) in the control group. At that time, a futility analysis predicted the probability of a successful study outcome to be 17% or less leading to the funding for DBS electrodes for this study to be discontinued. The actual study was never published, but results were published and mentioned in Morishita et al. (2014) [65,66]. It has been postulated that the patients enrolled in the BROADEN trial had extreme and chronic depression with a mean duration of the current depressive episode of 12 years, nearly twice that of previous open-label studies. Therefore, these patients could have required a longer treatment period before significant results emerge. Long-term outcomes of SCG DBS in TRD patients for up to 8 years show that most patients have a sustained antidepressant response [60]. However, these results need to be interpreted carefully as the patient group consisted of both MDD and bipolar type-II disorder patients. Further comparison between high- and low frequency DBS in the SCG in TRD showed no significant di fference in e ffectiveness between the two groups and a 44.44% response rate at 13 months of stimulation [67].
