Background
Neuromodulation is a broad term that could technically be considered to cover any medical, surgical, or physiologic therapy designed to alter the function of the nervous system in some manner. In the clinical neurosciences, however, neuromodulation is understood to refer specifically to therapies that involve targeted delivery of drugs or electrical current, frequently through an implanted device, in order to effect a specific change in the function of a target neural structure. [1]
Prior to the development of implantable devices for neurostimulation, surgical procedures designed to alter the function of the nervous system required the destruction of target tissues, with an accompanying irreversible change in function. Neuromodulation, however, provides a paradigm through which therapy can be titrated, or even turned on and off, with changes usually reversing rapidly upon cessation of therapy. While neuromodulation techniques are currently used primarily for the management of chronic pain and movement disorders, there has been considerable interest in their use for medically refractory psychiatric disease. Significant research remains to be done before any of the therapies are likely to become widespread, however. There is also substantial public trepidation about the use of surgical therapies in the treatment of psychiatric disease as a direct consequence of the abuse of psychosurgery in the mid-twentieth century, specifically the heavy human toll exacted by the careless and widespreaduse of the frontal lobotomy. [2] It is therefore incumbent upon the medical community to ensure that research to refine the effectiveness and expand the indications for neuromodulation in psychiatric disease is conducted in multidisciplinary fashion, and with rigid adherence to the highest standards of medical ethics.
History of the Procedure
With the advent of stereotaxis (the use of special targeting and coordinate systems capable of delivering an electrode or other probe precisely into deep brain structures), surgery for psychiatric disease evolved from the open lobotomy into minimally invasive lesioning such as cingulotomy (stereotactic ablation of the anterior cingulate cortex; see the first image below), capsulotomy (surgical ablation of the anterior limb of the internal capsule; see the second image below), subcaudate tractotomy (surgical ablation of the area known as the substantia innominate, a region ventral to head of the caudate; see the third image below), and limbic leucotomy (essentially a combined subcaudate tractotomy and cingulotomy). [3] Nevertheless, because of the mistrust for lesioning procedures that was the legacy of lobotomy, these procedures never came into widespread use and were not subject to strict experimental protocol. The lessons learned from stereotactic lesioning procedures were to play an important role,however, in the development of neuromodulatory therapies. [4]
The two surgical neuromodulatory therapies currently in use for the treatment of psychiatric conditions are deep brain stimulation (DBS) and vagal nerve stimulation (VNS). Because both were developed for non-psychiatric indications, there had been extensive refinement of the techniques and technology prior to their initial use in psychiatric disease. The first report of DBS for a psychiatric indication, published in The Lancet in 1999, described implantation of electrodes into the bilateral anterior limbs of the internal capsule for treatment of refractory Obsessive-Compulsive Disorder (OCD) in a series of four patients. [5] Target selection was based on the usual targeting for stereotactic cingulotomy, with the hope that electrical stimulation could generate the same effects as lesioning but modifiably and reversibly. Larger trials followed, followed by trials using DBS to treat other psychiatric diseases, particularly Major Depressive Disorder (MDD).
The use of VNS to treat MDD was inspired by the observation that VNS seemed to improve mood in patients with epilepsy in a manner that was independent of seizure reduction. [6] Subsequent trials led to FDA approval of VNS for treatment refractory depression in 2005.
Problem
Psychiatric disease is a major source of suffering, disability, and death worldwide. Major Depressive Disorder (MDD) alone accounts for billions of dollars in direct and indirect medical costs. In 2020, the COVID-19 pandemic led to a stark rise in depressive and anxiety disorders globally. The overall number of cases of mental disorders rose dramatically, with an additional 53.2 million and 76.2 million cases of anxiety and MDD, respectively. [7] Unfortunately, many psychiatric conditions have a high rate of resistance to pharmacologic treatment, and may remain refractory even to multimodal therapies utilizing medication, psychotherapy, and lifestyle interventions. [8] Even patients whose disease responds to medication may not achieve remission due to side effects or dose limitations. Past surgical attempts at treatment of psychiatric disease, using open or stereotactic techniques to interrupt white matter tracts, showed some efficacy but were irreversible and led to catastrophic morbidity when misapplied. [9] The challenges facing broader implementation of neuromodulatory techniques can therefore be broken down into three categories: 1) improving anatomic and pathophysiologic models of psychiatric illness to better identify the best substrates for stimulation, 2) optimizing stimulation parameters to maximize therapeutic effect and minimize adverse consequences, and 3) improving the patient selection process to avoid subjecting likely non-responders to the risk and expense of surgery.
Epidemiology
Frequency
Obsessive-compulsive disorder (OCD) is one of the most debilitating and refractory psychiatric disorders. OCD affects 2.2 million adults, or 1.0% of the US population, and almost 50 million people worldwide. [10, 11] Up to 40% of patients with OCD are partial responders or nonresponders. [12] Few patients with OCD experience a complete remission of symptomatology.
Major depressive disorder (MDD) is the leading cause of disability in the United States for patients aged 15–44 years. [13] In any given 1-year period, 9.5% of the population, or about 20.9 million American adults, suffer from a depressive illness. [14] Up to 30% of these patients are refractory to treatment. [15]
Pathophysiology
In general, insight into the pathophysiology of psychiatric disease is much less defined than the pathophysiology of movement disorders. There are many resons for this disparity, among them the complexity and heterogeneity of the underlying diseases and the lack of animal models of depression and OCD, as compared with the 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) primate model of Parkinson disease. Animal models of psychiatric disease are currently being developed but are not nearly as mature as their movement disorder counterparts. The result is that most early models of psychiatric disease were derived from the observation of behavior after brain lesions in humans. This tended to favor oversimplified models reliant on "positive" versus "negative" mood centers.
Increasingly, however, models of psychiatric disease are predicated not on a "center" of mood or behavior but rather an imbalance in communication of multiple neuronal loops. Functional imaging techniques such as fMRI, PET, and MEG have played a significant role in elucidating functional abnormalities in patients suffering from psychiatric illness. For details of the involved circuitry, see the Anatomy section of this article. Different studies, however, have found different patterns of derangement within this circuitry. This may be due to the heterogeneous nature of psychiatric illness itself. It is possible that biologically distinct subtypes of depression may have different patterns of activity on functional imaging. A similar situation exists for the clinical subtypes of OCD.
Most of the currently accepted somatic treatments for MDD and OCD are thought to primarily work through the manipulation of discrete neurotransmitter systems. MDD has been primarily managed clinically through the use of medications that manipulate serotonergic, noradrenergic, and dopaminergic systems, based largely on drug response and monoamine depletion data. OCD has been managed primarily with manipulation of the serotonergic system, based largely on drug response data. These neurotranmitters likely participate in signaling within the circuits listed above, although they also play multiple other roles within the nervous system. This is again analagous to Parkinson's disease, where the neurotransmitter dopamine plays a central role in the pathophysiology of the disease, but where neuromodulation can be used to balance the activity of competing circuits without affecting unrelated circuits which happen to use the same neurotransmitter.
Presentation
Obsessive-compulsive disorder
Obsessive-compulsive disorder (OCD) is defined by the National Institute of Mental Health as an anxiety disorder characterized by recurrent, unwanted thoughts (obsessions) and/or repetitive behaviors (compulsions). Repetitive behaviors such as handwashing, counting, checking, or cleaning are often performed with the hope of preventing obsessive thoughts or making them go away. Performing these so-called "rituals," however, provides only temporary relief, and not performing them markedly increases anxiety.
No definitive agreement exists on what constitutes medically refractory or treatment resistant OCD. The definition most commonly used is an unsatisfactory response to 2 adequate trials of serotonin reuptake inhibitors, [16] although most would suggest a trial of cognitive behavioral therapy (CBT) prior to defining someone as treatment resistant. Determination of "failure" or an “unsatisfactory response” is made when the Yale-Brown Obsessive Compulsive Scale (Y-BOCS) score is reduced by less than 25% or when improvement is greater than 25% but the patient still experiences significant OCD-caused impairment (meaning that the obsessions or compulsions continue to cause impairment in functioning, even in their improved state). Approximately 10% of the OCD population may be candidates for neuromodulation surgery based on these criteria of treatment resistance. [17]
Major depressive disorder
Major depressive disorder (MDD) is defined by the National Institute of Mental health as manifesting a combination of affective, cognitive, and behavioral symptoms that interfere with the ability to work, study, sleep, eat, and enjoy once pleasurable activities. Such a disabling episode of depression may occur only once but more commonly occurs several times in a lifetime.
Treatment-resistant depression also has no agreed upon definition. Attempts have been made to define degrees of treatment refractoriness. [18] Thase and O’Reardon defined treatment refractory depression as treatment nonresponse (ie, persistence of significant depressive symptoms) despite at least 2 treatment trials with drugs from different pharmacologic classes, each used in an adequate dose for an adequate time period. [18, 19]
The FDA went beyond this most commonly used definition when its approved vagus nerve stimulation (VNS) for treatment-resistant depression raised the number of failed adequate trials to 4 but did not define the types or quality of trials needed. These trials may include medications, therapies, and other treatments such as electroconvulsive therapy (ECT). Approximately 10–15% of the major depressive disease (MDD) population may be candidates for neuromodulation surgery based on these criteria of treatment resistance. [20]
Indications
Patient selection is perhaps the most crucial issue facing the renewed interest in neuromodulation for psychiatric disease. Careful patient selection is the key to not recapitulating the mistakes of the lobotomy era. An emphasis is placed on a multidisciplinary approach in which a team led by psychiatrists who are expert in the treatment of refractory obsessive-compulsive disorder (OCD) and major depressive disorder (MDD) carefully reviews all patients prior to surgical intervention.
Recent publications have given the following guidelines for forming such a team: an ethics committee, a patient assessment committee, strict adherence to accepted criteria for treatment-refractory MDD/OCD, limitation of such efforts to tertiary-care academic centers, limitation of patient selection to those patients with decision-making capacity, and the avoidance of procedures whose purpose involves law enforcement, political, or social ends. [21]
Relevant Anatomy
In 1937, James Papez introduced a circuitry that included the hippocampus, the fornix, the mammillary bodies, the mammillothalamic tract, the anterior thalamic, the subgenual cingulate (Brodmann area 25 or Cg25), the parahippocampal gyrus, and the entorhinal cortex. [3]
This circuit, known as the Papez circuit, has been an important heuristic model for psychiatric research and practice (see the image below).
In 1954, Paul McLean described a neural circuit that included cortical and subcortical structures. Known as the limbic system, this has been perhaps the most influential neuroanatomic model of psychiatric phenomena in the 20th and 21st centuries. The limbic system consists of the regions involved in the Papez circuit and adds the amygdala, the hypothalamus, the nucleus accumbens, and the orbitofrontal cortex. [3]
Frontal lobe
Despite the prejudice it cast on the practice of surgery for psychiatric disease, lobotomy emphasized the inherent role the frontal lobe has in the genesis of psychiatric symptoms and behaviors. The evolution of this insight has been the basis of the evolution of lobotomy to stereotactic lesions and now to the use of deep brain stimulation (DBS) for psychiatric disease. The following anatomic areas within the frontal lobe have to be considered:
-
Orbitofrontal cortex (Brodmann areas 10, 11, 12, 47; see the image below)
See the list below:
This area processes tasks related to reward and punishment and extinction behavior in response to aversive stimuli.
This area’s role in psychiatric disease is perseverative cognitions and emotional response.
The anatomic connections in this area include the following:
It receives projections from every sensory modality (unique among any neocortical region).
Its influence over the autonomic nervous system is second only to the Cg25.
This area has extensive reciprocal connections to the dorsolateral prefrontal cortex and cingulate.
Functional imaging data
-
Dorsolateral prefrontal cortex (Brodmann area 9, lateral 10, 46; see the image below)
The dorsolateral prefrontal cortex. Adapted from an image from Professor Mark Dubin, University of Colorado.
See the list below:
This area processes tasks related to working memory, spatial memory and executive function, and mediation of external environment on limbic responses.
This area’s role in psychiatric disease involves the patient’s insight into symptoms, the ability to suppress negative feelings and painful stimuli, and the psychomotor retardation of severe depression.
Anatomic connections include extensive reciprocal connections to OFC and the cingulate.
Functional imaging data include decreased metabolism in negative mood states and untreated depressed patients and increased metabolism with successful treatment. [22]
-
Cingulate (Brodmann areas 24, 32, and 25; see the image below)
The cingulate gyrus (Cg25; in green). Adapted from an image from Professor Mark Dubin, University of Colorado.
See the list below:
This area processes tasks related to attention and influence over visceromotor and vegetative functions.
This area’s role in psychiatric disease is related to disruptions in hedonic tone and motivation.
This area’s anatomic features include the following:
Extensive connections to autonomic circuitry
Extensive reciprocal connections to the dorsolateral prefrontal cortex (DLPFC) and orbitofrontal cortex (OFC)
Functional imaging data include the following:
These frontal lobe regions can be organized into 2 functionally related compartments: the dorsal compartment (DLPFC/ lateral orbitofrontal cortex [LOFC]; see the first image below) and a ventral compartment (medial orbitofrontal cortex [MOFC]/cingulate; see the second image below).


Thalamocortical loop
Evidence shows that neuronal ensemble oscillation and resonance between the thalamus and the cortex is "deeply related to the emergence of brain functions." [26] The thalamocortical (TC) loop is thought to be the basic building block of behaviors that span from motor activity to psychiatric phenomena. Each TC loop consists of a specific region of cerebral cortex and its reciprocal excitatory connections with a specific target within the thalamus. Derangement in these loops can result in neurologic disorders.
In the case of motor disorders such as Parkinson disease, the TC loop in question involves the cortical regions of the motor cortex, the premotor cortex, and the supplementary motor area and the ventral lateral thalamic motor nucleus (VL). With regard to psychiatric disease, the following 2 TC loops are important: an associative loop that consists of the dorsal frontal lobe compartment and its reciprocal projections to the ventral anterior (VA) and the parvocellular dorsomedial thalamic nuclei (DMpc) and a limbic loop that consists of the ventral frontal lobe compartment and its reciprocal projections to the magnocellular portion of the dorsomedial thalamus (DMmc; see the images below).


Cortico-striato-thalamocortical loop
In 1986, Alexander and Delong described a series of 5 loops of information, from cortex to basal ganglia and back to cortex. [27] Each loop activity courses through the basal ganglia in parallel direct and indirect pathways. These heuristic schemes provided the basis for modern movement disorder surgery. In the case of movement disorders, the motor loop is of importance. For psychiatric disease, the dorsolateral, orbitofrontal, and anterior cingulate loops are important. Each loop has a direct and indirect component (see the image below).

One of the features of these basal ganglia loops is that information is segregated according to the anatomic areas of their components. The primary cortical association of the associative loop is the dorsal compartment. Most of the information in the dorsal compartment flows through central striatal regions, such as the head of the caudate and portions of the NA core. The primary cortical association of the limbic loop is the ventral compartment. Most of the information in the ventral compartment flows through ventromedial striatal regions, such as the NA core and the NA shell. Like other cortico-striato-pallido-thalamocortical (CSPTC) loops, information travels through parallel indirect and direct pathways, with the output structures being the globus pallidus pars interna (GPi) and substantia nigra pars reticularis (SNr).
Hypothalamic-pituitary axis
The third anatomic circuitry that must be discussed is the interface of these thalamocortical and basal ganglia loops with the hypothalamic-pituitary axis. Via direct and indirect connections, the associative and limbic loops have access to autonomic machinery via the amygdala, the NA shell, the hypothalamus, and the serotonergic midbrain. The autonomic circuitry is especially important to the so-called vegetative aspects of psychiatric disease, such as wake/sleep cycles, feeding aberrances, and anxiety manifestations (see the image below).

Currently, 7 targets for neuromodulation surgery have been published: Cg25, the anterior internal capsule (AIC), the nucleus accumbens (NA), the ventral striatum (VS), the inferior thalamic peduncle (ITP), the subthalamic nucleus (StN) and the left vagus nerve. Each of these regions can be seen as nodes in the aforementioned circuitry. Putative modulation at these nodes is the basis of the current efforts investigating neuromodulation surgery for refractory psychiatric disease. The highlighted areas of the images below show how neuromodulation at each target may influence the aforementioned circuitry.









-
Cingulotomy. The position of the lesion is shown.
-
Anterior capsulotomy. The position of the lesion is shown.
-
Subcaudate tractotomy. The position of the lesion is shown.
-
The Papez circuit.
-
The orbitofrontal cortex. Adapted from an image from Professor Mark Dubin, University of Colorado.
-
The dorsolateral prefrontal cortex. Adapted from an image from Professor Mark Dubin, University of Colorado.
-
The cingulate gyrus (Cg25; in green). Adapted from an image from Professor Mark Dubin, University of Colorado.
-
The dorsal "compartment" of the frontal lobe. Adapted from an image from Professor Mark Dubin, University of Colorado.
-
The ventral "compartment" of the frontal lobe. Adapted from an image from Professor Mark Dubin, University of Colorado.
-
The limbic thalamocortical loop. MOFC = medial orbitofrontal cortex. DMmc = dorsomedial thalamic nucleus, magnocellular portion.
-
The cortico-striato-thalamocortical loop (CSTC loop) as it relates to neuromodulation for psychiatry. DLPFC = dorsolateral prefrontal cortex. LOFC = lateral orbitofrontal cortex. MOFC = medial orbitofrontal cortex. GPe = globus pallidus pars externa. GPi = globus pallidus pars interna. STN = subthalamic nucleus. SNc = substantia nigra pars compacta. SNr = substantia nigra pars reticularis. VTA = ventral tegmental area.
-
The associative thalamocortical loop. DLPFC = dorsolateral prefrontal cortex. LOFC = lateral orbitofrontal cortex. VApc = ventral anterior thalamic nucleus, parvocellular portion. VAmc = ventral anterior thalamic nucleus, magnocellular portion. DMpc = dorsomedial thalamic nucleus, parvocellular portion.
-
The hypothalamic-pituitary axis as it relates to the aforementioned circuitry. OFC = orbitofrontal cortex.
-
How neuromodulation at the cingulate gyrus (Cg25) interacts at the aforementioned circuitries. The highlighted area represents Cg25.
-
How neuromodulation at cingulate gyrus (Cg25) interacts at the aforementioned circuitries. The highlighted area represents Cg25.
-
How neuromodulation at the anterior internal capsule (AIC) interacts at the aforementioned circuitries. The highlighted area represents AIC.
-
How neuromodulation at the anterior internal capsule (AIC) interacts at the aforementioned circuitries. The highlighted area represents AIC.
-
How neuromodulation at the nucleus accumbens (NA) shell interacts at the aforementioned circuitries. The highlighted area represents the NA shell.
-
How neuromodulation at the nucleus accumbens (NA) shell interacts at the aforementioned circuitries. The highlighted area represents the NA shell.
-
How neuromodulation at the ventral striatum (VS) interacts at the aforementioned circuitries. The highlighted area represents VS.
-
How neuromodulation at the ventral striatum (VS) interacts at the aforementioned circuitries. The highlighted area represents VS.
-
How neuromodulation at the inferior thalamic peduncle (ITP) interacts at the aforementioned circuitries. The highlighted area represents ITP.
-
How neuromodulation at the inferior thalamic peduncle (ITP) interacts at the aforementioned circuitries. The highlighted area represents ITP.
-
Examples of deep brain stimulation (DBS) leads.
-
Example of implantable neurostimulator for deep brain stimulation (DBS) leads (implantable pulse generator, IPG).