Deep brain stimulation for treatment-resistant depression: current status and future perspectives

1. Introduction

Major depressive disorder (MDD) is one of the leading causes of disability worldwide. Conventional treatment consists of psychotherapy and pharmacotherapy. Severity, improvement, and remission of depression are evaluated using clinician-administered questionnaires, most commonly with the Hamilton Depression Rating Scale (HDRS) or the Montgomery Asberg Depression Rating Scale (MADRS). A 40-50% reduction in the HDRS or MADRS is generally considered a ‘response,’ and the patient presenting this reduction, a ‘responder.’ Even after several guideline-concordant trials of therapy, up to 30% of patients remain symptomatic and are therefore classified with treatment-resistant depression (TRD).

Three decades of neuroimaging research have informed the current model of depression neurocircuitry, although significant gaps in knowledge remain. Neural substrates have been identified at the level of individual brain structures, white matter tracts, circuits of interconnected regions, and complex networks. Crucial hubs or nodal points within the neurocircuitry responsible for depression have been identified, which are ideal targets for neuromodulation interventions.

One such neuromodulation treatments is deep brain stimulation (DBS), a neurosurgical procedure involving the insertion of electrodes into deep neural targets. These electrodes are connected subcutaneously to a pulse generator, placed below the clavicle, to deliver low-voltage continuous stimulation. Beginning with a case series in 2005 [1], DBS has increasingly been employed in the treatment of TRD, almost exclusively within the context of clinical trials. At least six neural targets have been used in DBS for TRD trials, highlighting the complexity, heterogeneity, and controversy present in the literature. This editorial will focus on the three DBS targets which are most widely studied. Due to their close anatomical proximity, the nucleus accumbens and the ventral anterior limb of the internal capsule will be referred to as the ventral capsule/ventral striatum (VC/VS).

2. Early experience of DBS for MDD: 2005-2015

Between 2005 and 2015 there were several open-label trials using DBS for MDD. Most of these reported promising clinical results with response rates of 45-75% at 12 months following surgery. The first target, the subcallosal cingulate region (SCC), was chosen based on neuroimaging studies demonstrating local overactivity in depressed patients or healthy volunteers with induced sadness [2]. Another popular target, the VC/VS, was adopted from the obsessive-compulsive disorder (OCD) literature, where DBS induced not only decreases in anxiety but also improvements in mood [3]. Long-term results from the SCC literature demonstrated that improvements were stable, and continued far beyond the 1 year follow up timepoint, with >50% of patients continuing to meet responder status at 8 years [4]. This is substantially better than the natural history of TRD, where <20% are responders at 2 years after conventional treatment [5].

3. Failure of large industry-sponsored trials

Following these promising early results, two large industry-sponsored randomized controlled trials (RCTs) were conducted, targeting the VC/VS [6] and SCC [7]. Based on the impressive and rapid response rates from earlier open-label trials, these studies adopted a conventional design, randomizing patients to stimulation-ON or stimulation-OFF (sham-control) after implantation. The primary outcome was a comparison in depression scores between groups, assessed at 4 months in one study, and 6 months in the other. After interim analyses demonstrating no significant difference between sham and treatment groups, both RCTs were terminated early. The failure of these large conventional RCTs sparked significant introspection within the field, as well as calls for new approaches.

Perhaps, the largest concern levied against the industry-sponsored RCTs was that the primary outcome endpoints were assessed too early to distinguish the DBS effect from that of placebo [8]. In both RCTs, substantial placebo effects were observed in the sham-stimulation groups, with 14% and 20% meeting response criteria. This may be due to the fact that interventions of greater cost or invasiveness amplify placebo effects.

4. Expert commentary

Despite the results of the studies described above, novel approaches to trial design and DBS targeting have been proposed and deserve to be commented. Bergfeld et al. conducted an RCT using a novel, patient-specific design [9]. Rather than randomizing patients to sham or active VC/VS stimulation immediately following surgery, all patients underwent a period of stimulation until they reached relatively stable clinical symptoms; this lasted approximately 1 year. Following this, patients entered a blinded crossover phase, where they were randomized to a period of ON or OFF stimulation. The study’s primary outcome was a comparison between depression scores in the ON and OFF groups, which showed significantly better depression scores during active rather than sham stimulation. The overall response rate during the 12 months open-label phase was 10/25 (40%), and two-year follow up data were recently published, with 8/25 (32%) meeting responder status [10]. This was the first example that DBS for TRD may require an adaptive trial design, with a prolonged optimization period to establish optimal stimulation settings and extinguish placebo responses. It is important to note, however, that this RCT essentially tested the effect of withdrawal of stimulation, rather than truly testing the effect of stimulation.

The field has also benefited from the use of diffusion tensor imaging (DTI). By mapping out white matter pathways at the individual level, DTI allows for personalized DBS target optimization. The superolateral medial forebrain bundle (slMFB) was the first tractographically defined DBS target for TRD [11]. The slMFB is a white matter tract thought to deliver excitatory neurotransmission to the ventral striatum, as well as the prefrontal cortex. It was first identified serendipitously, through the observation that patients treated with Parkinson’s Disease developed hypomania when the stimulation extended more medially than planned, into the slMFB [12]. Subsequently, long-term data have been reported in over 30 patients from 2 centers, with response rates of at least 70% [11]. The slMFB target relies on the mapping of axonal pathways at the patient-level. Open-label clinical results from slMFB stimulation continue to show high responder rates [11].

Riva-Posse et al. (2014) retrospectively evaluated electrode locations in a series of SCC DBS patients using DTI. They observed that compared with non-responders, responders shared a similar white matter tract blueprint – the convergence of four tract pathways – the forceps minor, cingulum, uncinate fasciculus, and frontal-striatal fibers [13]. The same group went on to prospectively test the hypothesis that SCC DBS is optimal when the active electrode is at the convergence of these four white matter tracts. This targeting strategy led to a marked improvement in clinical results – 82% response rate at 12 months – albeit in the context of open-label treatment [14].

As an added layer of complexity, depression is a heterogeneous condition consisting of multiple subtypes, many of which have neurobiological differences [15]. It may be that only certain depressive phenotypes are responsive to DBS or that different DBS targets address some symptoms better than others. As noted by Widge et al. (2018) [8], the three most common DBS targets evoke different intraoperative responses – motivational drive (slMFB) [11], anxiety relief (VC/VS) [3], or a sense of lightness (SCC) – suggesting different mechanisms of action. Patients with prominent symptoms of anhedonia or anxiety may respond preferentially to stimulation of the slMFB or VC/VS, respectively, although this has yet to be demonstrated. This may eventually lead to a scenario akin to DBS for movement disorders, whereby target selection of the thalamus, subthalamic nucleus, or globus pallidus internus, is partially driven by symptom or adverse effect profile. Furthermore, the success of DBS for MDD is currently measured with symptom-based questionnaire scores, such as the HAMD and MADRS, which may not always be the most reflective of patient priorities. Other measures, such as quality of life metrics, may prove to be more meaningful markers of change. The DBS field will likely evolve along with the search for more nuanced diagnostic criteria and holistic outcome measures in the field of psychiatry. Furthermore, as outlined by the National Institutes of Health’s ‘BRAIN’ initiative, neuro-ethics considerations should increasingly play a role in the design of both neuromodulation trials, and neuromodulation devices.

5. Conclusion

DBS for TRD is a complex and dynamic field. Despite two failed industry-sponsored trials, an RCT, using a patient-specific adaptive design, demonstrated a significant therapeutic effect over placebo. Patient-specific DTI targeting has allowed for novel, sophisticated, circuit-based DBS interventions. DBS trials have yet to account for heterogeneity within MDD. The time is ripe for a novel RCT, which will utilize adaptive trial design, DTI-based targeting, and consideration of depression subtypes.

Key issues

Deep brain stimulation has been investigated for the treatment of depression. This editorial discusses the main results of clinical trials and future perspectives in the field.

Declaration of interest

CH was part of an advisory board for Medtronic unrelated to this work. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Additional information

Funding

This paper was not funded.