Therapeutic brain stimulation intends to induce durable treatment effects by exploiting brain capacity for plasticity. Neuroplasticity is the process where alterations in brain structure promote functional changes, and represents the process of normal brain functions during learning, adaptation to change, and recovery from brain injury. It is thought that early changes with brain stimulation involve alteration in synaptic strength, with longer exposures trigger longer-lasting anatomical changes such as neuronal sprouting and alterations of dendritic spines. Brain stimulation is best viewed as an intervention that targets specific brain circuits, rather than brain transmitter chemicals (neurotransmitters) [1,2]. Two processes that occur during brain stimulation-induced neuroplasticity [1]:

Brain stimulation therapy is not new to the 20th century. In the Roman era, electric torpedo fish were placed on the scalp to treat headache or epilepsy, and gouty arthritis was treated by placing painful extremities in pools with torpedo fish. In the mid-1700s, advances in electrophysiology inspired the use of transcranial electrical stimulation with direct currents to treat mental disorders [3,4]. Transcranial electrical stimulation machines for private use became widely available, and study of transcranial electrical stimulation intensified in the 1800s. Transcranial electrical stimulation was claimed to generate euphoria and improve mental performance by some patients and physicians, who advised that currents to the head not exceed 10 mA from risks of burning and shock. Common side effects were headaches, dizziness and nausea; Benjamin Franklin suffered retrograde amnesia after accidentally administering an electric shock to his head [3].

Erratic results and the advent of ECT led to waning interest in direct current brain stimulation [4]. ECT was introduced in 1938 to replace drug-induced convulsive therapy of severe psychosis; epilepsy was mistakenly believed antagonistic to schizophrenia. Depression was later known as a more suitable indication [5]. Psychosurgery using stereotactic lesioning in specific deep brain structures was introduced in 1947 to avoid the side effects of the widely used frontal lobotomy. Deep brain stimulation followed in the early 1950s as treatment for psychiatric illness, Parkinson disease, and pain [5,6]. Primitive forms of magnetic stimulation were first investigated in the 1890s, first shown to stimulate isolated nerves in 1959, and the first modern device was introduced in 1976, the precursor for the first TMS technique in 1985 [3].

In contrast to other non-invasive brain stimulation modes, tDCS does not induce neuron action potentials but instead modulates neuron membrane excitability [15]. Parameters that influence patient response and functional outcomes with tDCS are current polarity, delivered dose, and electrode positions.

ECT is reserved for use in complex and acutely severe clinical presentations of MDD; in these patients, ECT is considered unrivaled for rapid induction of antidepressant effects [7]. ECT is recommended as first-line treatment in the following patients: extremely severe melancholic depression; those who refuse to eat or drink; have a very high suicide risk; very high levels of distress; psychotic depression; refractory MDD; catatonia; or previous positive ECT response [53].

Ultra-brief pulse width unilateral ECT allows effective treatment with markedly fewer cognitive side effects. Ultra-brief pulse efficacy appears comparable to standard brief pulse unilateral ECT but may require more treatments to achieve remission. Ultra-brief pulse bifrontal ECT is effective in cognitive sparing but ultra-brief pulse bitemporal is ineffective and not recommended [53,56,57,58].

Initial ECT response can be maintained with medication or ECT. Post-ECT antidepressant use reduces relapse rates by roughly 50% [60]. Strongest evidence supports the post-ECT relapse reduction efficacy of nortriptyline plus lithium or venlafaxine plus lithium, and both show comparable efficacy [63,64].

Seizure induction is the most serious adverse effects with rTMS, but fewer than 25 cases have been reported worldwide to date. The estimated incidence of spontaneous seizures is 0.01% to 0.1% with rTMS, 0.1% to 0.6% with antidepressant drugs, and 0.07% to 0.09% in the general population. HF-rTMS is contraindicated in patients with seizure history. Safety of LF-rTMS has been demonstrated in patients with epilepsy but is not established in patients with depression and seizures. Seizure history is usually considered an absolute contraindication [8].

Compared with rTMS, tDCS is more recently introduced, with fewer published efficacy and safety studies in psychiatric disorders. Antagonist effects on tDCS treatment response from two drug agents were first identified in a 2016 review of tDCS in psychiatric disorders [16]. Nicotine inhibits schizophrenic patient response to tDCS. This effect is confirmed experimentally, and evident empirically in tDCS studies outcomes of schizophrenia; higher rates of current smokers repeatedly showed direct association with lower tDCS response rates [16]. In patients with MDD, co-use of benzodiazepines blunts therapeutic response to tDCS [16].

A meta-analysis compared the efficacy of ECT and rTMS in MDD. ECT was superior to HF-rTMS in response (64.4% vs 48.7%) and remission (52.9% vs 33.6%), and discontinuation was similar (8.3% vs 9.4%). ECT was superior in psychotic depression, but HF-rTMS and ECT were comparable in non-psychotic depression. ECT had a slight advantage over rTMS in overall improvement in HAM-D scores. Data on medium or long-term efficacy was insufficient. The same results were found with ECT versus LF-rTMS. ECT led to greater impairment in cognitive domains such as visual memory and verbal fluency. ECT and rTMS seemed comparably effective in patients with MDD without psychosis, but ECT was more effective with psychotic depression [92].

The number of sessions, and anode/cathode polarity are thought to influence tDCS efficacy [16]. Most MDD have trials consist of placement of the anode over the left dorsolateral PFC and cathode over a noncortical region, or left dorsolateral PFC anodal stimulation combined with right dorsolateral PFC cathodal stimulation (bilateral tDCS). Minimum stimulation with 2 mA for ≥30 minutes per day over two weeks is required for an antidepressant effect.

A review of six randomized sham-controlled trials found active tDCS significantly superior to sham in response (34% vs 19%), remission (23.1% vs 12.7%), and improvement in depression. Treatment-resistant depression and higher tDCS "doses" were negative and positive predictors of tDCS efficacy, respectively. The authors concluded the tDCS treatment effect size was comparable to those of rTMS and antidepressant drug treatment in primary care [104].

A randomized sham-controlled trial evaluated 30-session (over six weeks) LF-rTMS in 25 patients with GAD. Patients undergoing active rTMS (versus sham) experienced higher response rates (71% vs 25%) and significant reductions in anxiety, worry, and depressive symptoms. At three months post-rTMS, 43% of active rTMS patients showed remission, versus 8% with sham. Response rates with active rTMS were maintained during follow-up, with some additional gains in remission rates. At post-treatment, right dorsolateral PFC activation was increased for active rTMS only, and changes in neuroactivation significantly correlated with changes in worry symptoms. These findings provide preliminary evidence that rTMS may improve GAD symptoms by modifying neural activity in the stimulation site [114].

A secondary analysis study was performed to determine if dorsolateral PFC neuromodulation improved emotion regulation in patients with GAD [114]. Statistically significant improvements in self-reported emotion regulation difficulties were found at post-treatment and three-month follow-up with active rTMS only. Improvements primarily involved the domains of goal-directed behaviors and impulse control, and were significantly associated with global clinician ratings of improvement. These preliminary results support rTMS as a treatment for GAD and suggest improved emotion regulation as a possible mechanism of change [34].

Bilateral LF-rTMS was evaluated in 13 patients with comorbid GAD and MDD, who received 24 to 36 rTMS sessions over five to six weeks. Following the last treatment, 11 of 13 (84.6%) patients achieved remission in anxiety symptoms (<5 on the GAD-7), and 10 of 13 patients (76.9%) achieved remission in depressive symptoms (<8 on the HAM-D). In this small pilot study, most patients with comorbid GAD/MDD achieved significant improvement in anxiety and depressive symptoms after bilateral rTMS [115].

rTMS and tDCS have been increasingly researched in OCD. In clinical trials, LF-rTMS of the supplementary motor area, orbital frontal cortex, or right dorsolateral PFC shows the most promising efficacy, while older studies targeting the prefrontal dorsal cortex were not as successful. Larger-scale investigations of tDCS have yet to be published in OCD [1].

Patients with OCD were randomized to rTMS frequencies of 1 Hz, 10 Hz, or sham of the right dorsolateral PFC for 10 sessions. Patients were assessed after the last session and three months later. Compared with 10 Hz or sham, 1 Hz led to significantly greater improvements in obsessive-compulsive and anxiety symptoms, greater clinical benefit, and significantly larger percentage change in global improvement. One Hz LF-rTMS of the right dorsolateral PFC is a promising treatment approach in OCD [121].

The efficacy of rTMS in OCD was evaluated by reviewing 15 randomized sham-controlled trials. rTMS was significantly superior to sham for OCD symptom reduction. The risk of publication bias was low, and between-study heterogeneity was low. Meta-regression showed no particular influence of any variable on the results. In all, rTMS was superior to sham for amelioration of OCD symptoms [122].

In two trials, patients with treatment-resistant OCD received rTMS or sham for two and four weeks. Compared with sham, active rTMS led to significantly greater improvements in symptom severity; and cognitive performance in auditory perception, visual perception, short-term memory, and processing speed [123,124].

Studies of rTMS treatment in schizophrenic symptoms such as auditory hallucinations have found contradictory results. A review of 10 randomized sham-controlled trials using LF-rTMS reported a positive effect size favoring rTMS over sham, with the left temporoparietal cortex appearing an effective target [127]. Studies published after 2013 were not available for this review. A study of HF-rTMS to the left temporoparietal cortex over 2 days found no difference between active rTMS or sham [128]. The overall sham-controlled outcomes of rTMS to the left temporoparietal cortex are mixed; even when rTMS was effective following 10 sessions, treatment effect began dissipating one month later [129].

Auditory verbal hallucinations are commonly observed persistent symptoms in schizophrenia, even when patients are stabilized by antipsychotic medication. The primary indication for tDCS in psychotic disorders is reduction of auditory verbal hallucinations. Eliminating or reducing these debilitating residual symptoms requires inhibition of neuronal activity of the left temporoparietal junction (TPJ) that mediates auditory verbal hallucinations [17].

Most tDCS treatment studies in schizophrenia have used anode placement over the l-dorsolateral PFC and the cathode over the left TPJ. After 10 twice-daily sessions, this placement has been consistently shown in sham-controlled randomized controlled trials to ameliorate symptoms of the illness, with robust reductions in auditory verbal hallucinations at acute and three-month follow-up, improvements in other schizophrenic symptoms, and substantive decreases in treatment-resistant auditory verbal hallucination frequency. Further support for this tDCS protocol comes from uncontrolled trials of schizophrenic patients with persistent auditory verbal hallucinations, with significant reductions in psychotic and auditory verbal hallucination symptoms reported. Several case reports described refractory schizophrenic patients who achieved responses ranging from significant reductions in psychotic and auditory verbal hallucinations, to full remission. Greater therapeutic response in nonsmokers was noted [16].

The dorsolateral PFC plays a key role in self-regulatory control mechanisms and is a common brain stimulation target in disorders of fronto-limbic dysregulation. rTMS of the dorsolateral PFC has demonstrated potential in reducing addictive behaviors and craving for nicotine, alcohol, and cocaine [46].

Placement of right-anodal plus left-cathodal tDCS over the bilateral dorsolateral PFC has reduced a variety of substance cravings in patients with substance use disorders, and bilateral stimulation with both polarities may be equally effective [142]. tDCS can enhance patient ability to inhibit potent responses, a clinical important finding because impaired inhibitory control contributes to relapse in patients with substance use disorder [17].

The FDA approved VNS in 2005 as adjunctive therapy in refractory MDD with four or more failed adequate antidepressant trials. Some studies below also enrolled patients with treatment-resistant bipolar depression, and VNS outcomes were similar for both [165].

Worth noting in the VNS studies is the severity of MDD in the patient population, distinguished by chronicity (average duration of illness: >25 years, and current episode: 7 years), treatment-resistance (average of seven drug failures, and ECT failure in >50%), high rates of lifetime hospital admissions and suicide attempts [165]. However, VNS was found ineffective in patients with eating disorders or schizophrenia [10,166].

Short-term outcomes with deep brain stimulation show 30% to 60% response rates and 20% to 40% remission rates at three or six months, and medial forebrain bundle deep brain stimulation in seven patients reported response and remission rates of 85.7% and 57.1% [172,173].

As with VNS, two randomized sham-controlled trials of deep brain stimulation were halted early from lack of benefit, later seen as premature with recognition that efficacy increases with longer stimulation [8]. Outcomes of four studies showed depression severity was significantly reduced by 12 months. At 3, 6, and 12 months, response rates were 36.6%, 53.9%, and 39.9%, with remission rates 16.7%, 24.1%, and 26.3% [174].

Subcallosal cingulate deep brain stimulation studies of greater than one year duration reported response rates of 36% and 92% at one and two years, and a 58% remission rate at two years. A longer trial reported 62.5%, 46.2%, and 75% response rates at one, two, and three years, with 20% and 40% remission rates at two and three years. Long-term subcallosal cingulate and medial forebrain bundle deep brain stimulation led to improved health-related quality of life [173,175,176].

Transient psychosis and hypomania have emerged when changing parameters during nucleus accumbens stimulation, resolved by switching parameters. No hypomania episodes were reported with subcallosal cingulate deep brain stimulation, including patients with bipolar disorder. Oculomotor adverse events of blurred vision and strabismus occurred in all patients receiving higher amplitude medial forebrain bundle deep brain stimulation. Reports of suicidality and completed suicide were deemed not device-related, but risk of suicidality may be increased by a history of pre-deep brain stimulation suicidality or major life stressors [8].

Deep brain stimulation efficacy in severe, refractory OCD was reviewed in 25 trials. Outcome was change in Y-BOCS scores; response was ≥35% improvement in Y-BOCS (Table 3) [41]. Reported adverse effects included serious events during surgery (i.e., two seizures, three intracerebral hemorrhages), device-related adverse events, and stimulation-related events. Device-related events included breaks in stimulation leads and battery failure resulting in acute psychiatric symptoms until replacement [41]. Stimulation-related adverse events were acute increases in anxiety and hypomania, induced by changing stimulation parameters or battery depletion. All cases resolved with parameter adjustment. Cognitive problems resolved with time or parameter adjustment. No personality changes during deep brain stimulation were observed.

Neural circuits involved in OCD and depression include the VC/VS and anterior cingulate gyrus. VC/VS deep brain stimulation combined with anterior cingulotomy was evaluated in patients with refractory OCD and MDD. Mean baseline Y-BOCS scores were 34.7, decreasing to 23.0 at three months, and stable two years post-surgery at 19.0. These outcomes were comparable, but not superior, to published outcomes for VC/VS deep brain stimulation alone, and deep brain stimulation may be sufficient to control refractory OCD [179].

In the early 1990s, epidural motor cortex stimulation (eMCS) showed repeated efficacy in pharmaco-resistant neuropathic pain. These methods were adopted by TMS researchers in chronic pain using stimulation to the primary motor cortex (M1) and precentral gyrus in the hemisphere contralateral to pain [27]. Stimulation parameters were refined over time to maximize efficacy. An early 10-day protocol of 5 Hz rTMS of M1 in patients with diverse chronic neuropathic pain reported minimal benefit, the result of using low frequency (5 Hz) stimulation and small number of pulses (500) per session. A subsequent five-day protocol of 20 Hz HF-rTMS of M1 led to durable pain reduction in patients with post-stroke pain, trigeminal neuropathic pain, and phantom limb pain [19].

When used in pain treatment, a figure-8 coil delivering biphasic pulses should be placed over the precentral gyrus (M1) contralateral to the painful side with a posteroanterior orientation [13,27]. HF-rTMS (10–20 Hz) is used to activate projecting axons and local interneurons and applied below the motor activation threshold to avoid inducing muscle contractions. Focal neuropathic pain can be relieved by HF-rTMS (but not LF-rTMS) to the contralateral M1 area. Repeated rTMS sessions can induce cumulative pain reductions for at least several weeks following 10 consecutive sessions, but optimal timing for long-term efficacy and safety remain studied.

Higher stimulation intensity or duration to enhance efficacy has been examined, but prolonged M1 stimulation at higher intensities may reverse the intended effects on neural excitability [185].

In the United States, more than 300,000 spinal surgeries are performed annually, mainly in the lumbar spine, and as many as 100,000 result in failure where the patient experiences new-onset pain in addition to unresolved pain from the original problem. The level of pain is widely variable and may occur with neurologic deficits. Contributors to pain and the clinical features of failed back surgery syndrome include recurrent disk herniation, epidural abscess, scar tissue formation around the nerve root, facet joint syndrome, and muscle spasm. Patients with persistent radicular pain, usually from chronic nerve injury, greatly benefit from treatment that addresses the neuropathic pain [193].

Neuropathic pain syndromes tend to show greatest benefit from rTMS to the M1, but some nonneuropathic chronic pain syndromes, such as CRPS-I or fibromyalgia, may have a neuropathic component. Focal lesions with defined onset, such as pain from shingles or trauma, have advantages of known localization and time of onset, but early cases often improve spontaneously which complicates treatment outcomes in research; thus, established cases with pain duration of at least one year are preferable [13].

The treatment efficacy of rTMS and tDCS of fibromyalgia was evaluated by reviewing 16 randomized sham-controlled trials. Significant improvements in fibromyalgia domains were found [204]. Overall treatment effect sizes were large for pain, sleep disturbance, fatigue, and tender points. Effect sizes were medium for depression and general health/function. rTMS showed a significantly greater effect size than tDCS. Primary motor cortex (M1) stimulation produced a subtle but greater effect size in pain reduction versus dorsolateral PFC, and dorsolateral PFC stimulation may be better for depression improvement. Clinically meaningful improvements in anxiety or cognition, or dose-response effect in fibromyalgia pain reduction, were not found for either modality.

Deep brain stimulation is more invasive than motor cortex stimulation because electrodes are implanted through the skull, dura, and brain to stimulate deep targets. Morbidity associated with motor cortex stimulation and deep brain stimulation have greatly improved over time since their introduction [13]. For intractable pain, patients selected for motor cortex stimulation/deep brain stimulation should have evidence of exhaustive but failed conservative management, screening for secondary gain, and receive a psychological evaluation. Trial stimulation is mandatory with deep brain stimulation, but is symptom dependent with motor cortex stimulation. Motor cortex stimulation/deep brain stimulation in chronic intractable pain are both proven effective in specific pain indications (>40% pain reduction for ≥12 months) [216].

The American Headache Society practice recommendation for unilateral hypothalamic deep brain stimulation was downgraded to Negative B, based on a study showing active deep brain stimulation no different from sham in pain reduction and serious adverse effects during deep brain stimulation treatment [201].