Low-Intensity Transcranial Current Stimulation in Psychiatry

The spatial positioning of stimulation electrodes on the scalp can generate the misleading perception that only the brain underneath the electrodes, and no other area, is stimulated. This notion is mostly incorrect, since the human head exhibits heterogeneous electrical properties. For example, when the current is delivered through scalp electrodes, a large fraction of it is shunted away through the skin and does not penetrate the skull. Current may also travel through the orbits, foramen magnum, or cranial nerve foramina, as low-resistance interstitial fluid creates electrical shunts at these sites. Several tCS devices deliver stimulation through one or more electrodes placed on the ears, face, or elsewhere below the head and neck. It is possible that nonspecific cranial nerve stimulation plays an important part in the effects of tCS. Once the electrical field reaches the brain, tCS has a certain strength and direction; both are relevant for modulating the activity of individual neurons or networks of neurons. Similar to antennae, neurons must be positioned so they are aligned with the direction of an oncoming electrical field if the field is to influence them. When this happens, a series of events leads to a change in the voltage across the neuron’s membrane (7); stronger electrical fields (i.e., those with greater amplitude) have greater effects on the neuronal membrane. Spatial targeting using computer simulations of the electrical field distribution, as a function of electrode number, size, and location, has been proposed (8) but lacks validation as an approach to guide clinical tCS. Moreover, given the distributed and complex deficits in neuronal networks associated with psychiatric disorders, identifying the correct target area(s) for therapeutic stimulation in a specific disorder or symptom remains an important challenge for the field.

The electrical fields used in tDCS are generally considered a subthreshold perturbation, meaning that tDCS, by itself, is not thought to cause neuronal depolarization (Figure 1). However, the net effect of tDCS does not occur in isolation. Communication between individual neurons and neuronal networks is nonlinear and complex, with a large number of inputs influencing the activity of any individual neuron. Therefore, even a small change in the membrane voltage may impact neuronal firing.

Variation in the direction of current flow also impacts neuronal firing (Figure 2). As described above, when current travels in one direction, the effect is to depolarize or enhance the chance of firing. However, current traveling in the opposite direction causes hyperpolarization of the membrane, making the neuron less likely to fire relative to its resting state. Unfortunately, this neurophysiological principle is associated with the unproven model wherein “anodal tDCS” excites brain activity in the region under that electrode and “cathodal tDCS” inhibits brain activity in the region beneath that electrode. While application of this simplistic, and likely incorrect, model (9) has been used to support montages implemented in clinical trials (reviewed in Section 2), further research is needed to characterize the relationship between cellular physiology and clinical outcomes.

The potential therapeutic benefit of tCS arises because the neurophysiological effect of current applied during a single session is durable, to some extent, over time after the stimulation ceases. This phenomenon was demonstrated by a series of experiments wherein motor cortex neurons were stimulated with tDCS, and their excitability was measured after stimulation stopped (10). It is important to recognize that much of what we know about tDCS comes from studies of the motor cortex, and it remains unclear if the same principles apply to other brain regions, such as the prefrontal cortex. Furthermore, since neuronal organization may differ across the brain, it is possible that the same stimulation can result in varied effects when applied to different regions. Nevertheless, a number of experiments (e.g., 11–13) have now demonstrated enduring functional effects of tDCS on (nonmotor) cortical activity, persisting in the hour after stimulation ceases.

Efficacy studies for depression represent the largest group of available data for randomized controlled trials of tCS. tDCS is the dominant modality (Table 1), typically with the anode placed over the left dorsolateral prefrontal cortex (DLPFC). Some studies restricted enrollment to (unipolar) major depressive disorder, and others included participants with unipolar or bipolar major depressive episodes. Some of these studies allowed participants to remain on stable regimens of psychotropic medications while others required medication-free participants.

Initial studies of tDCS generated mixed results regarding potential efficacy. Fregni et al. (25) performed the first clinical trial of tDCS for major depressive disorder (N=10) and found efficacy of active over sham treatment (p<0.05). This was followed by a larger study (N=40) by Boggio et al. (23) that also showed superiority of active stimulation. Subsequently, Loo and colleagues (26) found no difference between active and sham tDCS (N=40) (p>0.1). However, when they conducted a larger study (27) (N=64) with more treatment sessions, they found a significant advantage of active tDCS (p<0.05) but no difference in response rates; one bipolar patient receiving active tDCS became hypomanic. Palm et al. (28) (N=22) and Blumberger et al. (22) (N=24) also found no difference between active and sham tDCS. Bennabi et al. (21) (N=24) tested tDCS plus escitalopram (10–20 mg/day) and found no difference between active and sham tDCS.

In the largest study (N=120) of tDCS to date, Brunoni et al. (4) gave twelve 30-minute sessions of 2-mA tDCS (10 consecutive workday sessions followed by a single session delivered every other week) and/or a low dose of sertraline (50 mg/day) in a 2×2 factorial design; two of the groups (each N=30) were randomized to active tDCS. This approach enabled comparisons of active versus sham tDCS, placebo pill versus sertraline, and a drug-stimulation combination. The investigators observed greater reduction of depression in the group receiving combined sertraline plus active tDCS than in the groups receiving sertraline monotherapy (p=0.002), tDCS monotherapy (p=0.03), and both inactive treatments (placebo plus sham tDCS, p<0.001). Treatment with tDCS monotherapy was superior to placebo plus sham tDCS (p=0.01) but comparable to sertraline monotherapy (p=0.35). In comparisons of response rates, tDCS monotherapy (43.3%, p<0.001) and tDCS plus sertraline (63.3%, p=0.03) did better than placebo plus sham stimulation (16.7%). Remission followed a similar pattern, with worse outcomes for placebo plus sham stimulation (13.3%) compared with tDCS monotherapy (40.0%, p=0.02) and active tDCS plus sertraline (46.7%, p=0.007). Sertraline monotherapy did not statistically separate from placebo plus sham stimulation on any outcome measure. Seven episodes of treatment-emergent mania or hypomania were observed, with the majority (N=5, 17%) in the group receiving combined active tDCS plus sertraline.

Several research groups have evaluated the combined effect of tDCS plus psychotherapy for depression, an approach informed by data indicating that tDCS can facilitate neuronal firing in the context of appropriate environmental cues (10). Segrave et al. (5) (N=27) reported improved depressive symptoms when tDCS was combined with cognitive control therapy, although both Brunoni et al. (24) (N=37) and Vanderhasselt et al. (29) (N=33) found no difference between active and sham stimulation combined with therapy. Some have identified the timing of stimulation relative to therapy as a possible limitation of these studies, theorizing that “online” stimulation, occurring concurrent with therapy, might be superior to “offline” stimulation that precedes the session (35).

The anxiolytic/antidepressant effects of other types of tCS have also been investigated. Over a dozen CES devices received Food and Drug Administration (FDA) clearance for treatment of “insomnia, depression, or anxiety” on the basis of technical features that were considered substantially equivalent to older CES devices already on the market before Congress introduced the Medical Device Regulation Act in 1976. While an older literature (36, 37) suggested clinical efficacy of CES, that body of evidence comprises trials that would not be considered rigorous by modern standards of clinical trial design. A 1995 meta-analysis of CES therapy raised questions regarding data reporting bias and adequacy of blinding (37). While the use of proprietary waveforms by most CES devices has created an obstacle for independent evaluation of efficacy and potential mechanisms of action, Barclay et al. (30) (N=115) conducted an investigation of CES efficacy using the Alpha-Stim device in patients with a primary anxiety disorder and some (unspecified) degree of comorbid depressive symptoms, and they reported significantly improved depression (p<0.001) and anxiety (p<0.001) after treatment. However, subsequent studies by Lyon et al. (31) (N=163) and Mischoulon et al. (32) (N=30) found no advantage of CES over sham stimulation in depressive symptoms (all p>0.1). One recent pilot study of bipolar II depression by McClure et al. (33) (N=16) indicated that 2 weeks of CES could reduce depressive symptoms (p<0.003).

Cranial nerve stimulation is another tCS approach under investigation. Shiozawa et al. (34) (N=40) reported the first randomized controlled trial evaluating the efficacy of eTNS and observed that active stimulation significantly reduced depressive symptoms (p<0.01). Rong et al. (38) (N=160) conducted a pseudo-randomized controlled trial of tVNS for major depressive disorder. While active tVNS was associated with greater reduction in depressive symptoms (p<0.001), no differences in response or remission were observed at endpoint.

To date there are four meta-analyses of tDCS for depression. Although earlier reports were negative (14, 38), recent analyses (incorporating larger studies) are positive. Shiozawa et al. (18) (N=259) found a significant advantage of active tDCS over sham (g=0.37, 95% confidence interval [CI] 0.04–0.7). Odds ratios (ORs) for response and remission were 1.63 (95% CI 1.26–2.12) and 2.50 (95% CI 1.26–2.50). Most recently, Brunoni et al. (39) (N=289) found similar results for response (OR=2.44, 95% CI 1.38–4.32) and remission (OR=2.38, 95% CI 1.22–4.64), and they also reported that treatment resistance predicted nonresponse, whereas higher tDCS dose (longer duration and higher current density) predicted response.

Questions remain about potential side effects or synergistic therapeutic effects when tCS is combined with psychotropic medications, since no large studies have investigated the use of tDCS concurrent with adequate doses of antidepressant medication. The currently available data do not support the use of tDCS as a method to accelerate or enhance the short-term effects of psychotherapy. While risk of adverse events appears modest, the incidence of (hypo)manic induction in larger trials is noteworthy and deserves greater study.

Taken together, the available evidence from randomized controlled trials generally supports the use of tDCS to relieve symptoms of depression, with other stimulation modalities yielding mixed results. To date there is no defined regulatory pathway for tDCS devices, and none is approved or cleared for treating psychiatric disorders. On the other hand, despite having an FDA indication for depression, CES devices have not consistently demonstrated clinical efficacy.

tCS has been investigated as a treatment approach for schizophrenia (Table 2), mostly utilizing tDCS. Montages have typically utilized placement of the anode over the left DLPFC, with the cathode over the temporoparietal junction or over the supraorbital area. Brunelin et al (40) (N=30) conducted the first randomized controlled trial and observed that active tDCS reduced auditory hallucinations acutely (p<0.001) and over 3 months (p<0.001) and reduced negative symptoms (p=0.01). This was followed by a study by Smith et al. (45) (N=33) that found active stimulation improved cognition (p=0.008) but had no effect on positive or negative symptoms (all p>0.1), whereas Palm et al. (44) found that tDCS reduced negative symptoms (p=0.016) and Mondino et al. (43) found that tDCS reduced hallucinations (p<0.001). Several studies using tDCS (Fitzgerald et al. [41], N=24, and Frohlich et al. [42], N=26) and tVNS (Hasan et al. [46], (N=20) found no difference between active and sham stimulation.

The currently available data do not support use of tCS for schizophrenia. The evidence base comprises a small number of randomized controlled trials with conflicting results. More work is clearly needed to develop tCS for treatment of patients with schizophrenia.

Dementia and cognitive deficits are other therapeutic areas of investigation (Table 3), inspired by the potential for tDCS to enhance attention, learning, and memory in healthy adults (reviewed in reference 50). While meta-analyses of single-session tCS (51, 52) indicate benefit in patient samples, the results of most clinical randomized controlled trials have been negative (47, 49), although Manenti et al. (48) (N=20) found that tDCS improved cognition in patients with Parkinson’s disease. On the basis of these results, the available data do not support the use of tDCS for patients with dementia or cognitive deficits.

A number of studies have evaluated tCS for substance use disorders (Table 4). Da Silva et al. (57) (N=13) investigated tDCS for alcohol dependence, and they reported significant reductions in depressive symptoms (p<0.001) and craving (p=0.015), although they also reported a statistical trend toward a higher relapse rate (p=0.053). Klauss et al. (58) (N=33) found active tDCS improved alcohol abstinence (p=0.02). Regarding nicotine, two studies, by Boggio et al. (55) (N=27) and Fecteau et al. (56) (N=12), found that tDCS reduced nicotine craving and cigarette consumption (p<0.05). Findings in cocaine use are mixed; Conti et al. (54) (N=13) found no effect of tDCS on cocaine use (p>0.1), whereas Batista et al. (53) (N=36) found that tDCS reduced cocaine craving (p=0.028). There are some proof-of-concept studies of tDCS for other substances, with potentially concerning results. Boggio et al. (59) (N=25) found that tDCS increased risk-taking behaviors in chronic cannabis users (p<0.001), and Shahbabaie et al. (60) (N=22) found that tDCS increased cue-induced methamphetamine craving (p=0.012).

While the available data appear to provide some support for the use of tDCS for some substance use disorders, there have been very few clinical trials, and several suggest potential harms, such as increased relapse (57), greater risk taking (59, 61), and heightened craving (60).

Data describing tCS for therapeutic areas beyond those reviewed here are quite limited. For example, one study (62) (N=60) did not find efficacy of a single tDCS session for attention deficit hyperactivity disorder, and several case series or open-label studies suggested potential efficacy of tCS for working memory in posttraumatic stress disorder (PTSD) (63) and symptoms of comorbid PTSD and major depressive disorder (64). There are also a growing number of studies for nonpsychiatric conditions that may be of interest to psychiatrists, described elsewhere (14, 65).

The classic risk when stimulating the brain is seizure generation, although the energy used in tCS is orders of magnitude lower than in ECT (e.g., 800 mA) or rTMS (66). Therefore, seizures would be very unlikely in the absence of intracranial pathology. Additionally, the interaction between tCS and metal in the head or neck represents a major unknown risk. Most tCS studies excluded participants with head or neck metal, which could divert and adversely focus applied currents. While tCS in patients with head or neck metal may be safe in some cases, it should not routinely be considered outside of specialized research-based settings.

Perhaps the greatest device-related risk is skin burns from excess energy, although these are generally preceded by pain and redness as warning signs (6). Recent studies, with experienced investigators using devices with adequate safety features, have not resulted in skin burns, e.g., the study by Brunoni et al. (4). Self-administration of tCS by untrained individuals may present greater burn risk. A closely related risk is delivering more (or less) current than desired. Direct-to-consumer devices typically do not include instructions for the consumer to calibrate or otherwise assess the function of the device. Of concern, a growing community of DIY enthusiasts is building and using their own devices for noninvasive brain stimulation. For example, a 2015–2016 Internet search we did yielded five DIY device designs that could be constructed for $50–$100 USD and would likely be capable of delivering 1–2 mA. DIY interest is growing; a user support website with 2,700 registered users in 2013 (69) had grown to over 8,700 in 2016. Purported uses include improving mood and anxiety symptoms, enhancing exercise endurance, and gaining an edge in online gaming. Accessible plans for DIY devices did include multiple statements about safety precautions in building and using the device. Such disclaimers may protect DIY proponents from liability (69), but the information is likely insufficient for patients. Furthermore, since the FDA does not regulate direct-to-consumer devices or DIY device construction documents, serious adverse events may be occurring but are not reported: one DIY tCS website included subjective descriptions of migraines, photophobia, vivid dreams, increased anxiety, and possible mania. Such reports represent important safety information that is otherwise not recorded.

Although claims that tCS improves brain function have been made (50–52), stimulation may also impair cognition (70). It may induce a functional trade-off, improving a single cognitive function at the cost of impairing another. For example, one study of healthy individuals found tDCS improved the learning of new associations at the cost of worse performance of old ones (71). Another reported that tDCS increased mathematics performance but reduced executive function (72). These effects may be greater in psychiatric patients, whose cognitive reserve may be reduced as a consequence of illness. Specific electrode configurations may also be associated with adverse cognitive effects. Several studies described learning and working memory impairments when the tDCS cathode was applied over the parietal lobe or cerebellum (73, 74), and another found reduced cognitive performance when the tDCS anode was placed over the DLPFC (i.e., the configuration used by the vast majority of tDCS studies) (70). Worsened working memory has also been reported after use of a commercial tDCS device (75).

The ostensibly benign profile of tCS could lead patients with mental illness to substitute stimulation for evidence-based care. In a large-scale survey of the DIY community, depressive symptoms were cited as a common reason for trying tCS. Less than half (44%) of those using tCS for a medical condition were seeing a physician for that same condition (69). As reviewed in Section 2, only a small handful of studies systematically evaluated the effects of stimulation concurrent with psychopharmacology or psychotherapy. Given that tCS effects are likely state-dependent, the field should expect to find significant, unexpected, and potentially harmful interactions between tCS and other interventions. As described above, Brunoni et al. (4) described an elevated rate of conversion from depression to hypomania in participants receiving tDCS and sertraline. As tCS becomes widely available to consumers, more patients with a bipolar diathesis may try it and switch into a (hypo)manic state. Clinicians might erroneously attribute the change in mood state to pharmacotherapy, thereby removing a potential treatment option. Several of the reviewed substance abuse studies showed an increase in cravings or related symptoms (57, 59–61), suggesting that occult tCS could attenuate the efficacy of substance abuse treatment. Therefore, unreported or unsupervised tCS may pose a significant risk to patients by interfering with evidence-based psychiatric treatments.