Deep Brain Stimulation for Post-Traumatic Stress Disorder: A Review of the Experimental and Clinical Literature

Post-traumatic stress disorder (PTSD) emerged after the Vietnam War as a clinical diagnosis in the 1980 Diagnostic and Statistical Manual of Mental Disorders (DSM-III) [1]. It is related to persons who were exposed to life-threatening events, such as combat, rape, or confinement to a concentration camp and who showed symptoms such as tormenting flashbacks and hypervigilance. In the DSM-IV, the diagnosis was broadened and was defined as an anxiety disorder that can develop after exposure to traumatic events, such as threat of death, serious injury, or a serious threat to an individual’s bodily integrity. This could lead to individuals experiencing reoccurring intense fear, nightmares, horror, and helplessness [2].

It is estimated that PTSD has a prevalence of about 7% of the population in the USA and 5% in other high-income countries, making it the fourth most common psychiatric disorder, and a major global health problem [3-5]. PTSD has the highest comorbidity rates for developing other psychiatric disorders such as depression and has a 15 times higher risk for suicide attempt compared to other psychiatric conditions [6]. The estimated lifetime prevalence is between 5 and 8% for men and 10–14% for women [7]. In combat veterans, this is almost doubled [8, 9]. PTSD results in substantial economic costs through losses in productivity at work, and healthcare costs are estimated to be in the billions of dollars [10].

The symptoms of PTSD are clustered into 3 main categories: (i) the traumatic event is persistently and intrusively re-experienced, triggering flashbacks, and/or hallucinations, with intense psychological and physiological reactivity; (ii) individuals persistently avoid stimuli associated with the trauma, displaying anhedonia; and (iii) individuals display enduring symptoms of increased arousal such as insomnia, irritability, hypervigilance, and an exaggerated startle response [2].

Current treatment for PTSD consists of pharmacological (such as selective serotonin reuptake inhibitors) and/or psychological interventions (cognitive behavioral therapy, exposure therapy, and eye movement desensitization and reprocessing) [11]. Although research demonstrates clinical efficacy for these treatments [12], 30% of individuals are still debilitated by this condition 10 years after initial diagnosis and are considered treatment refractory [13, 14], and may suffer from low life satisfaction, marital problems, psychiatric disorders such as depression and higher suicide risk [15-17]. Johnson et al. [18] found a sobering 17% mortality rate in 51 PTSD combat veterans over 6 years. Combat-related PTSD compared to civilian PTSD is associated with more severe symptoms [19-21] and is more likely to be treatment refractory [14, 22, 23].

Evidence from brain imaging studies on people with PTSD indicate abnormalities in specific brain regions that may either contribute to, or result from, PTSD [24]. These abnormalities concern the hippocampus (HPC), amygdala, medial prefrontal cortex (mPFC), anterior cingulate cortex (ACC), and ventral striatum (VS) [11].

Structural neuroimaging studies of PTSD have focused on hippocampal volume, due to findings from animal research demonstrating the destructive nature of stress on the HPC [25]. Human studies frequently report reduced hippocampal volume in individuals with PTSD [26-28] and abnormal hippocampi are argued to mediate specific PTSD symptoms [24]. It is suggested that under extreme stress the HPC is unable to encode reliably, leading to dysfunction in memory retrieval, specifically in relation to safe spaces and context [29].

The prefrontal cortex (PFC) has been found to be abnormal in structure and function in patients with PTSD, especially the mPFC and ACC. Abnormalities in these regions are associated with deficits in emotional regulation in PTSD. Patients with PTSD have been demonstrated to show diminished hemodynamic responses in both ventromedial PFC (vmPFC) and ACC [30, 31], and this hypoactivation correlates with PTSD symptom severity [32]. Furthermore, a reduction of PTSD symptoms following successful treatment is associated with increased vmPFC activation [33].

The most consistent finding in human functional imaging research is hyperactivity of the amygdala. Hyperactive amygdalae are associated with heightened fear and hyperarousal in individuals with PTSD. Specifically, the basolateral nucleus of the amygdala (BLa) is thought to be critical in acquisition and expression of fear. Several researchers [34-36] used single-photon emission computed tomography in veterans with PTSD and found that the amygdala had increased activation to combat sounds compared to controls. A meta-analysis of functional neuroimaging studies established the focus of this hyperactivity within the BLa [37]. Research in combat veterans with traumatic brain injury found that lesions to the amygdala and vmPFC were protective against the development of PTSD relative to other brain injury locations [38]. Furthermore, functional MRI studies have shown that PTSD patients who have improved with CBT showed a reduction in amygdala hyperactivity [39].

The PFC can inhibit amygdala activation and decrease its responsivity, and the hyporesponsiveness of the vmPFC in PTSD is argued to give rise to amygdala overactivation [40]. The BLa receives afferents from the mPFC and the HPC; the input from the mPFC is suggested to mediate fear extinction and its underactivity in PTSD is suggested to lead to a failure in fear extinction [41]. The input from the HPC is thought to relate to contextual information regarding events, allowing emotional responses to develop. During stressful events, it is suggested that neutral contextual information is not encoded reliably [42]. Overall, the lack of regulation of the amygdala by the PFC may allow the amygdala to inappropriately generalize an emotional response of fear, across multiple contexts [41].

Additionally, autonomic responses from the hypothalamic-pituitary-adrenal axis are regulated by the amygdala and HPC, controlling the release of corticotrophin-releasing hormones [43]. HPC degradation found in PTSD may result in a more pronounced stress response leading to greater cortisol release and therefore more HPC damage [43]. Thus, the amygdala is responsible for the generation of exaggerated fear responses in PTSD, the hypothalamus produces the autonomic responses, and the vmPFC provides inhibitory regulation of the amygdala. Since the vmPFC is found to be underactive in PTSD, the amygdala is insufficiently inhibited which drives exaggerated fear response, resulting in PTSD symptomology [43] (Fig. 1).

Deep brain stimulation (DBS) is an established treatment in Parkinson’s disease, tremor, and dystonia, and a promising – albeit still investigatory – procedure in psychiatric disorders, especially in obsessive-compulsive disorder (OCD), Gilles de la Tourette’s syndrome and depression [44]. From these promising results, the potential role of DBS in treating refractory PTSD has been proposed [41, 45]. It is in this context that the USA Government’s Defense Advanced Research Projects Agency (DARPA) granted in 2003 seventy-million USD to support development of therapeutic brain stimulation technologies [46].

DBS as a treatment for PTSD is in its infancy [47], and comprehensive surveys of the rationale for this potential treatment, its applicability and efficacy are lacking. The present review aims to address this by examining animal model research and published case reports from ongoing clinical trials. Overall, the review aims to provide a comprehensive overview of the research area and discuss current debates within the field.

A broad search was performed using the following databases: PubMed, Ovid MEDLINE, Cochrane Library, and PsycINFO. A combination of words was used: “Deep Brain Stimulation,” “DBS,” “Functional Neurosurgery,” “Neuromodulation” and “Post Traumatic Stress Disorder,” “PTSD,” “Post-traumatic stress disorder,” “treatment resistant,” “treatment resistant post-traumatic stress disorder,” “treatment resistant post-traumatic stress disorder,” “anxiety disorder.” Bibliographies of relevant journal articles were then searched to retrieve additional articles not obtained in the initial search. References were excluded if they were not in the English language and did not specifically refer to DBS and psychiatric disorders.

Following search methodology, articles were combined, and duplicates removed. This resulted in 87 articles that met the search criteria; 57 articles were excluded since they did not meet the inclusion criteria (Fig. 2). The 30 included publications were grouped into 2 areas: animal models (n = 24) and human clinical reports (n = 6).

Table 1 details DBS for PTSD in animal models according to brain region stimulated: PFC including mPFC and vmPFC, HPC, VS, and basolateral amygdala (BLa). These articles are summarized below.

Fear conditioning and extinction paradigms have been utilized to explore the role of PFC structures. Vidal-Gonzales et al. [48] found that subdivisions of the mPFC into prelimbic and infralimbic (IL) regions gave opposing results to DBS. Stimulation applied to prelimbic during extinction increased freezing behavior during extinction recall, whereas stimulation to the IL decreased freezing behavior, indicating an anxiolytic effect – a result also found by others [49, 50]. Research by Milad and Quirk [51] and Milad et al. [52] only observed these effects of IL stimulation when it was administered 100–400 ms after tone presentation. Others applied high-frequency stimulation to IL for 10 min after fear conditioning and extinction, and when rats were reexposed to the fear-conditioned context, freezing behavior was reduced [53-56]. DBS for PFC may thus depend on the PFC region and stimulation parameters.

High-frequency stimulation to the HPC after fear extinction resulted in reduced freezing behavior during recall sessions [56, 57]. Research by Deschaux et al. [58] and Garcia et al. [59] using similar methodology but applying DBS at low frequencies (2 Hz) for 25 min to dorsal Cornu Ammonis 1 and Cornu Ammonis 2 regions found that this increased freezing behavior during retention training. However, Cleren et al. [60], delivered low frequency stimulation for 25 min to ventral Cornu Ammonis 1 instead, at either 6 or 14-h after fear conditioning and found freezing behavior was reduced during extinction. Hence, hippocampal studies suggest that the improvements in anxiety-like behavior may depend on the target regions and stimulation protocols.

Rodriguez-Romaguera et al. [61, 62] and Do-Monte et al. [63] showed that rats with electrodes implanted in the VS dorsal to the anterior commissure (AC) demonstrated significantly less freezing behavior than sham controls. The opposite effect was observed in those with electrodes implanted in the VS ventral to the anterior commissure. Whittle et al. [64] used a fear-conditioning paradigm in mice to measure the effect of DBS to the VS core. DBS was administered during fear conditioning, during extinction training and extinction retrieval. DBS during conditioning and extinction was found to not affect freezing behavior compared to sham controls. However, DBS during extinction retrieval significantly reduced freezing behavior.

Research into BLa DBS has mainly used the defensive burying paradigm. This is an innate behavior observed in rats, whereby objects that are threatening, dangerous, or associated with unpleasant experiences are buried. Treatments reducing an animal’s tendency to defensively bury objects are seen as anxiolytic. Saldivar-Gonzalez et al. [65] used this paradigm in which rats were to bury a shocking rod. DBS was delivered in a single session prior to behavioral testing, at 3 different intensities. Animals were reported to show reduced burying at 150 and 300 μA.

Langevin et al. [66] and Stidd et al. [67] used defensive burying paradigms, exposing rats to a ball (the conspicuous object), and giving them a series of inescapable shocks. DBS to the right BLa or sham treatment was administered for 4 h a day over 7 days, prior to defensive burying. Rats treated with right BLa DBS, spent significantly less time burying the ball than sham controls. Additionally, Sui et al. [68] used same protocol of right BLa DBS following fear conditioning and found significant decrease of rats’ freezing behavior in response to tone. Hashtjini et al. [69] used a fear-conditioning paradigm and found that DBS to right BLa significantly decreased rats’ freezing. In a follow-up study [70], a combination of DBS to the right BLa with saffron treatment reduced freezing behavior but not with DBS alone. Recently, Dengler et al. [71] reported that bilateral DBS to the BLa decreases avoidance behavior when using a predator scent avoidance paradigm – which is argued to more closely replicate human PTSD as this is a life-threatening stressor for the rat.

Table 2 details DBS for PTSD in human clinical research, comprising 6 publications on 3 individual patients, summarized below. One pilot trial of DBS of the BLa [72] has resulted so far in 2 published cases with 4 years follow-up for one of the 2 operated patients [73-77]. The authors’ rationale for targeting the BLa was based on one hand on the reported rat model research, and on the other hand on human imaging studies using fMRI, positron emission tomography and single-photon emission computed tomography showing that when patients with PTSD are exposed to imagery or audio related to their trauma, the BLa is hyperactive compared to healthy controls [37].

Langevin et al. [73, 74] and Koek et al. [75, 76] described a combat veteran who despite 20 years of pharmacological and psychological treatment remained severely symptomatic. His baseline clinician-administered PTSD scale (CAPS) score was 119, classifying his PTSD as extremely severe. After enrollment, the patient underwent 2-fluorodeoxyglucose positron emission tomography scans, one during resting conditions and 1 during an activated condition consisting of the patient recalling the traumatic event. Images obtained during recall demonstrated higher amygdala metabolism, compared to at rest. Subsequently, bilateral DBS electrodes were implanted in the BLa. Targeting was performed on stereotactic 3T-MRI with gadolinium and with the help of the Schaltenbrand and Wahren Atlas, and the authors used microrecording and macrostimulation through the contacts of the 3,387 Medtronic electrode. The inferior limit of the BLa was located 16 mm lateral and 4 mm posterior to the AC, and 18 mm inferior to the AC-PC plane (although these coordinates may show variations between individuals). The authors used a rather steep trajectory at an angle of 0–10° from the midline and an anterior angle of 70–80° from the AC-PC plane, making sure to avoid the ambient cistern and the lenticulostriate vessels. This approach allowed for a DBS lead placement that spanned the central nucleus of the amygdala, the BLa, and the head of the HPC [74]. Substantial clinical improvement was reported 8 months postoperatively, with a reduction in CAPS score to 74 (37.8% reduction) and at 15 months a reduction to 62 (48% reduction) [73, 75]. Stimulation parameters were 1.4 V, 60 μs, 160 Hz on the right side and 0.7 V, 60 μs, and 160 Hz on the left side. In a 2-year update Koek et al. [76] reported “over a year of essentially complete suppression of previously nightly severe combat nightmares,” but hospitalization at 17 months for suicidality. At 4 years, the patient had maintained a 40% reduction in CAPS scores [76]. Recently, Koek et al. [76, 77] presented the results of a second operated patient, a 40-year-old Iraq combat veteran, who showed >30% amelioration in CAPS scores at 7 months post-surgery.

In 2020, Hamani et al. [78] published the results of DBS in the mPFC and the uncinate fasciculus in a 46-year-old woman with a 17-year history of PTSD due to domestic abuse. For surgical targeting, the authors relied on a 3T-MRI scan using a T1-magnetization prepared rapid acquisition gradient-echo sequence and diffusion tensor imaging sequences. They centered the DBS lead on the subgenual cingulum with manual adjustments allowing to maximize contact with the uncinate fasciculus. They used Boston Scientific directional Vercise electrodes so that one of the 2 middle directional electrodes was placed near the uncinate fasciculus. They state that this allowed stimulation of the uncinate fasciculus as well as the cingulate bundle, forceps minor, and frontostriatal projections. Their rationale for using these brain targets were based on experiments on rat models of PTSD by Reznicov et al. [49]. At 6-month follow-up, the patient, who scored 56 on the CAPS before surgery, showed 100% improvement as well as substantial improvement in depression, global assessment of functioning, and quality of life [78].

Several issues were revealed by this review of the literature on experimental and clinical DBS for PTSD. First, when compared to virtually all other human DBS applications – whether in movement disorders or in neuropsychiatric illness – PTSD is the only potential clinical indication of DBS that shows an extensive animal research background prior to pilot human applications. Indeed, our review revealed 24 published reports on DBS in hundreds of rat models of PTSD, compared to 3 published PTSD patients who received DBS.

In the rat models of PTSD, DBS has been trialed on various brain areas considered to be implicated in the circuitry underpinning fear and behavioral reactions to fear: amygdala, PFC including the anterodorsal cingulum, VS, and HPC. In the 3 published clinical cases of PTSD, the DBS targets included the BLa in 2 patients and the mPFC in one, both yielding encouraging results.

In translating animal models research to humans, several criteria must be met. The behavior observed must be analogous to that in humans, models should test predictions regarding the disorder’s mechanism and etiology, and must share comparable neural mechanisms [10]. Neural mechanisms in animals seem to agree with neuroimaging research in humans on the neurocircuitry involved. While structure and connectivity of the amygdala in rodents correlates well with human anatomy [79], comparisons between prefrontal and hippocampal structures in rodents and humans are controversial due to topological differences [80, 81].

Animal models typically study short-term fear/anxiety, rather than long-term symptomatology of PTSD [29]. To accurately reflect PTSD, animals should develop long-lasting fear/anxiety responses when reexposed to cues [82, 83]. Contrary to patients with treatment refractory PTSD who have undergone many treatments, most animal models are administered DBS as an initial treatment. Furthermore, delivering short time intervals of DBS in animals does not adequately emulate the continuous (for years) application in humans, failing to explore chronic consequences of DBS [29]. As such, models used to mimic PTSD states in rodents fail to meet the above criteria and may need interpreting with caution [84].

Treatment refractory PTSD is a heterogenous and multifaceted complex disorder, with a multitude of cognitive and physiological symptomology including impact on individuals’ disease presentation by variable social and environmental factors [29]. Animal models cannot replicate this complexity.

For over 30 years DBS has been used to treat >208,000 patients worldwide, the majority suffering from Parkinson’s disease, tremor, and dystonia [47]. Although modern DBS in neuropsychiatry was inaugurated in 1999 with DBS for OCD [85], and DBS for Gilles de la Tourette syndrome [86], then in 2005 with DBS for depression [87], so far DBS in these conditions has not made a dent, nor has it reached the level of routine surgical treatment as has been the case since long time for DBS in movement disorders [88]. However, the mere probing of DBS as a treatment of severe refractory neuropsychiatric conditions has extended the potential applications of DBS to new “indications,” one of which is thus PTSD. It is in this perspective that one can understand the interest of DARPA in late 2013 to finance research in DBS with 70 million USD over 5 years [46]. However, a consultation of registered clinical trial on https://clinicaltrials.gov/, (accessed September 19, 2021), using search words “deep brain stimulation” rendered a total of 563 trials, of which 50 are for DBS in OCD and 152 for DBS in depression. For DBS and PTSD, the site lists had only 3 trials, of which one is “Active not recruiting,” one “Withdrawn,” and one “Recruiting.” The latter trial indicates a “study start date” in 2014, aims at recruiting 6 patients, and indicates an “Estimated study completion date” by December 31, 2025. The brief summary provided for this trial states that PTSD “affects 30% of American veterans returning from Iraq and Afghanistan” and that “combat PTSD has a tendency to be resistant to current treatments.” Hence, the authors propose to trial “DBS of the basolateral nucleus of the amygdala.” As shown in Table 2, the authors of this trial have presented the 4 years results of the first patient and the 7 months results of the second patient.

The most recently registered trial (https://clinicaltrials.gov/ct2/show/NCT04152993) deals with responsive DBS for PTSD and aims to enroll 6 participant veterans with severe PTSD. It proposes to target the BLa, and based on the fact that in PTSD, local field potentials recordings from the BLa reveal a pattern of signals that correspond to an exaggerated state of fear, the trial aims at using these local field potentials signals as a biomarker in order to program the DBS device to detect them and deliver the stimulation when needed, that is, on demand, instead of using continuous stimulation.

Although neurocircuitry of OCD is well studied in animal and in man [89], and data from pilot patients arguably demonstrate a potential for DBS to improve PTSD in combat veterans, over the last 7 years only 2 patients were recruited and published. Experience in these 2 patients has shown that DBS should be carefully monitored because it may yield different and conflicting clinical responses dependent on the strength of stimulation and exact location of the electrode within the BLa [90]. Although Langevin et al. and Koek et al. found significant improvements in their first patient following BLa DBS, a CAPS score of 62 is still deemed as rather severe PTSD [91]. Additionally, as in DBS for other neuropsychiatric disorders, there is a frequent need for battery changes and battery depletion or accidental disconnections may result in rebound of symptoms and even suicidal crisis [92, 93]. PTSD in humans is further complicated in this respect as differences in trauma type can differentially affect responsiveness, as well as sleep disturbances, hyperarousal, and dissociation [11], and symptoms involved are not only limited to extinction deficits and anxiety, but also disturbances in cognition, difficulties with affect regulation, self-perception, and interpersonal relationships.

Importantly, there may be ethical issues in relation to obtaining full informed consent from individuals who are severely ill [94-96]. Within this, PTSD is related to changes in cognition and in affect, which can further exacerbate concerns around informed consent. Potential ethical concerns may also arise from clinical trial funded by the department of veteran affairs and the DARPA [97, 98]. In an article published in Nature in 2015 [98], Sara Reardon voiced concerns in relation to DARPA’s involvement in some neurological research projects, stating that DARPA’s program managers can fund projects without waiting for peer review, and researchers are often termed as “performers” with projects axed if milestones are not on time. Be it as it may, ethical issues surrounding potential indications for DBS are not confined to use of DBS for PTSD, which after all is an illness and a highly relevant disorder in need of treatment. There may be other more debatable issues related to DBS “indications” such as reports showing support for the idea of using DBS to enhance memory in healthy people [99], or to treat “antisocial behavior,” and improve “morality” [100].

PTSD is a heterogenous disorder involving multiple brain circuitries. There is a solid wealth of data from DBS applied to various nodes in the brain circuitry of rodent models of PTSD, of which one is investigated in a clinical trial of DBS targeting the BLa. While the first 2 combat veterans who received BLa DBS for PTSD in a registered clinical trial have shown promising results, it seems that there are difficulties in recruitment of patients for this trial. This may be probably due to cautious interpretation of results for such a complex condition and perhaps to reluctance of combat veterans with severe PTSD to undergo such trial. Though DBS is shown to potentially help treat fear extinction and anxiety, it is as yet unclear whether other aspects of PTSD will improve, and whether patients are able to reintegrate in their social role and in the community. Despite initial enthusiasm and funding, and despite the obvious need to address and treat the not so uncommon neuropsychiatric consequences of combat on veteran soldiers, it is uncertain as to what extent DBS will become a common, accepted, and efficient treatment of PTSD in the future. In this, DBS for PTSD shares even more the fate of all DBS in psychiatry, including DBS for OCD and DBS for depression which is still not considered “established” despite all reports and case series and clinical trials that have been published in the last 2 decades [101].

J.F. declares no conflicts of interest. M.H. received speaker honoraria from Boston Scientific.

The authors declare no funding sources.

J.F. contributed to study conduct; data collection; analysis and interpretation; first manuscript drafting, revision, and approval. M.H. contributed to study design; study conduct; data collection; analysis and interpretation; manuscript drafting, revision, and approval.