The application of tDCS for the treatment of psychiatric diseases.

Neuroplasticity represents the dynamic structural and functional reorganization of the central nervous system, including its connectivity, due to environmental and internal demands. It is recognized as a major physiological basis for adaption of cognition and behaviour, and, thus, of utmost importance for normal brain function. Cognitive dysfunctions are major symptoms in psychiatric disorders, which are often associated with pathological alteration of neuroplasticity. Transcranial direct current stimulation (tDCS), a recently developed non-invasive brain stimulation technique, is able to induce and modulate cortical plasticity in humans via the application of relatively weak current through the scalp of the head. It has the potential to alter pathological plasticity and restore dysfunctional cognitions in psychiatric diseases. In the last decades, its efficacy to treat psychiatric disorders has been explored increasingly. This review will give an overview of pathological alterations of plasticity in psychiatric diseases, gather clinical studies involving tDCS to ameliorate symptoms, and discuss future directions of application, with an emphasis on optimizing stimulation effects.


Introduction
The prevalence of psychiatric diseases has increased rapidly in the last decades to almost one third of the global population having experienced a mental disorder at some time during their lifetimes (Steel et al., 2014). This growing incidence also exacerbates the burden for healthcare systems in both social and financial aspects. Therefore, a main focus of neuropsychiatric research is to elucidate pathophysiology underlying psychiatric diseases, and on this foundation to develop effective treatment methods.
During the last years it became increasingly evident that dysfunctional neuroplasticity is involved in various neuropsychiatric disorders. Neuroplasticity encompasses dynamic alterations of neuronal connections in the central nervous system. It serves as the foundation for various brain functions, including cognitive processes such as learning and memory formation. Plasticity of synaptic connections was first proposed by Hebb (1961) and later demonstrated in hippocampal slices as long-term potentiation (LTP) and long-term depression (LTD) of excitability after fast or slow rhythmic stimulation, respectively (Bliss & Lømo, 1973;Dunwiddie & Lynch, 1978). In addition to glutamatergic synapses in the hippocampus, both LTP and LTD can be induced in other synaptic connections including GABAergic and dopaminergic neurons in virtually all areas of the central nervous system in vitro and in vivo, which has been revealed by extensive investigations in the last three decades (for review, see Malenka and Bear (2004)). Furthermore, physiological plasticity has been linked to learning and memory processes, as demonstrated in different modalities such as sensory and motor learning, as well as more complex memory formation (McGann, 2015;Rioult-Pedotti, Friedman, & Donoghue, 2000;Rioult-Pedotti, Friedman, Hess, & Donoghue, 1998;Whitlock, Heynen, Shuler, & Bear, 2006). On the other hand, disruption of long-term plasticity or aberrant plasticity has been shown to compromise learning and memory processing (McNaughton & Morris 1987). Such cognitive deficits are commonly observed in psychiatric diseases.
Neuroplasticity can be induced in humans by noninvasive brain stimulation techniques such as transcranial direct current stimulation (tDCS). Brain stimulation with direct current was first shown in animal models to elicit long-lasting alterations of cortical excitability, possibly via generation of sub-threshold, stimulation polarity-dependent alteration of neuronal membrane potentials, which modifies spontaneous firing rates (Bindman, Lippold, & Redfearn, 1964). For an overview, see . During the last decade, non-invasive transcranial direct current stimulation of the human brain has been introduced following the development of methods that allow probing its neurophysiological effects (e.g. transcranial magnetic stimulation, TMS, and functional magnetic resonance imaging, fMRI). It has become established as a technique that reliably induces and modulates neuroplasticity in the human cerebral cortex, in order to elicit prolonged and, thus, neuroplastic, but yet reversible shifts of cortical excitability (Nitsche & Paulus, 2000, 2001Nitsche et al., 2008). The respective excitability alterations are bi-directional within certain limits, as excitatory and inhibitory effect can be induced by anodal and cathodal tDCS, respectively, and the after-effects can outlast the stimulation for over 1 h or longer when the stimulation is continued in the range of minutes (Nitsche & Paulus, 2000, 2001Nitsche, Nitsche, Klein, et al., 2003). The resulting plasticity is associated with Nmethyl-D-aspartate (NMDA) receptors of glutamatergic synapses and calcium channels, as well as protein synthesis, and thus shares common features with LTP and LTD (Nitsche, Fricke, et al., 2003;. In addition to the neurophysiological effects, as demonstrated primarily by changes of excitability in the human motor system, tDCS can also modulate cognitive functions such as working memory and learning processes in different modalities (Kuo & Nitsche, 2012;Shin, Foerster, & Nitsche, 2015). Therefore, it provides the possibility to counteract cognitive dysfunction observed in psychiatric diseases, and to serve as alternative and adjuvant therapeutic option to conventional pharmaceutic treatment. Extensive investigation in both basic and clinical studies has revealed promising results and facilitated the development of effective stimulation protocols (for review, see Woods et al. (2016)). Here we will review the current state of tDCS application in psychiatric diseases, and provide an outlook based on the up-to-date evidence for the improvement of its efficacy in future studies.

tDCS in psychiatric diseases
In psychiatry, pharmacological intervention remains the first choice of therapeutic option to date. Non-invasive brain stimulation techniques including tDCS have recently been introduced as adjuvant treatment, especially for refractory or treatment-resistant patients. Earlier applications of tDCS in psychiatry were mainly focused on depression, but the field of application has gradually extended to other diseases such as addiction, schizophrenia, anxiety, and obsessive-compulsive disorder (OCD), which will be discussed separately in the following sections.

Depression
Most tDCS studies conducted in psychiatry so far are dedicated to the treatment of depression, possibly due to the underlying pathophysiology, which includes reduced activity of the left dorsolateral prefrontal cortex (DLPFC), which is located at the convexity of the brain, and, thus, provides optimal pre-requisites for stimulation efficacy (Fitzgerald et al., 2006). In addition, right hemispheric hyper-activation is also suggested to be involved in depression, which supports the focus of therapeutic strategies to enhance left DLPFC activity, and/or to decrease right DLPFC activity (Sch€ onfeldt-Lecuona, C ardenas-Morales, Freudenmann, Kammer, & Herwig, 2010). Furthermore, anodal tDCS facilitates LTP-like plasticity, which has been shown to be compromised in depression (Normann, Schmitz, F€ urmaier, D€ oing, & Bach, 2007;Spedding, Neau, & Harsing, 2003). Consistent with the hypothesis that reduced LTP plays a role in depression, an important effect of antidepressant serotonin re-uptake inhibitors might be its enhancing impact on LTP-like plasticity in healthy humans and patients suffering from major depression (Nitsche, Boggio, Fregni, & Pascual-Leone, 2009;Normann et al., 2007).
The application of tDCS for the treatment of depression can be traced back to the 1960s. Bilateral anodal prefrontal stimulation was conducted in those trials combined with an extra-cephalic return electrode. These studies revealed mixed results, and, since physiological effects of such stimulation protocols were not explored, it is unclear what kind of cortical excitability alteration was induced (for details see Lolas (1977) and Nitsche, Boggio, et al. (2009)).
With regard to 'modern' tDCS protocols, excitability-enhancing anodal tDCS of the left DLPFC with the return electrode positioned over the contralateral orbit has turned out to be efficient to ameliorate clinical symptoms in major depression. tDCS improved the pathological attentional bias for emotional cues during working memory tasks in patients with depression (Wolkenstein & Plewnia, 2013). In a double-blinded, sham controlled study, anodal tDCS for five consecutive days in newly-diagnosed patients resulted in a significant improvement of clinical symptoms (Fregni, Boggio, Nitsche, Rigonatti, & Pascual-Leone, 2006). Increasing stimulation intensity to 2 mA for up to 15 sessions resulted in stable clinical effects lasting for up to 1 month after tDCS, as shown in two other double-blinded sham-controlled studies Loo, Alonzo, Martin, Mitchell, Galvez, & Sachdev, 2012). The tDCS efficacy in these studies was similar to the treatment with 20 mg fluoxetine, but evolved earlier than the pharmacological intervention . In other trials, anodal prefrontal tDCS was not superior as compared to placebo stimulation, possibly due to the application of weaker and less frequent stimulation, and the inclusion of more severely affected patients (Bennabi et al., 2015;Loo et al., 2010;Palm et al., 2012). However, it was observed in the latter study that trend-wise more patients in the active tDCS group achieved the response and remission criteria after treatment, as compared to the sham group (Bennabi et al., 2015). Furthermore, additional 'boosting sessions' following two studies with left-DLPFC tDCS (Loo et al., 2010;Martin et al., 2011) revealed beneficial effects, e.g. one study showed remission rates of $80% after 3, and 50% after 6 months, with weekly or second-weekly extra sessions after initial daily tDCS (Martin et al., 2013) (see Table  1).
Bi-frontal tDCS with the anode over the left and the cathode over the right DLPFC, in order to reestablish the balance between right and left DLPFC activation, was demonstrated first to be effective in some open-label studies (Brunoni, Ferrucci, et al., 2013;Brunoni, Ferrucci, Bortolomasi, Vergari, Tadini, Boggio, et al., 2011;Dell'Osso et al., 2012). The clinical improvement lasted for up to 3 months in half of the patients receiving tDCS (Dell'Osso et al., 2014). However, less promising results were obtained in a double-blinded sham-controlled study (Blumberger, Tran, Fitzgerald, Hoy, & Daskalakis, 2012). Factors like patients with higher disease severity and numerous patients under benzodiazepines, which might have reduced the efficacy of tDCS , may have contributed to this negative result (Brunoni, Ferrucci, et al., 2013). It was also demonstrated that moving the right frontal return electrode to an extra-cephalic position resulted in a better initial treatment response in patients resistant to bifrontal stimulation (Martin et al., 2011). In a recently published large-scale double-blinded sham-controlled study, bi-frontal tDCS was applied in 120 patients diagnosed with unipolar depression, and the effects of tDCS or sertraline alone, and combination of both agents was compared (Brunoni, Valiengo, et al., 2013). DC stimulation with 2 mA for 30 min was performed for 10 days, and then repeated every other week once. The results revealed that tDCS alone improved depression ratings significantly, and to a similar extent as antidepressant medication. Interestingly, combination of tDCS and sertraline had larger effects on the respective symptoms than each of the interventions alone. In phase II of the same trial the non-responders of the sham tDCS groups received a 10-day tDCS course and more than half of these responded to active tDCS . Moreover, in the same study another group including all active-tDCS responders were recruited for a follow-up phase-III trial, in which a 24-week tDCS treatment regimen was administered, with a maximum of nine tDCS sessions performed every other week for 3 months and then once monthly for the subsequent 3 months. The results demonstrated an average response duration of 11.7 weeks within the 24-week treatment course . Alternative electrode arrangements in addition to the above-mentioned montages have also been tested. Both frontooccipital and bi-temporal tDCS were reported to be effective in four or 14 patients, respectively, with major depression (Ho et al., 2015;Ho et al., 2014).
With regard to combined treatment approaches, not only tDCS applied during antidepressant medication, but also a combination of tDCS with cognitive therapy has been investigated, based on the rationale that tDCSinduced LTP-like plasticity could further improve cognitive function such as therapy-related learning processes, which are associated with cortical plasticity. Tensession adjunctive bi-frontal tDCS combined with cognitive behavioural therapy in a patient with refractory major depression improved therapeutic endurance, as compared to tDCS alone (D'Urso, Mantovani, Micillo, Priori, & Muscettola, 2013). In a sham-controlled study, in which bifrontal tDCS was combined with cognitive control therapy, real tDCS was, however, not superior to sham control . In contrast, a clinical benefit was reported when the return electrode was placed over the right supra-orbital area, and tDCS was applied with lower current density and shorter treatment sessions (Segrave, Arnold, Hoy, & Fitzgerald, 2014).
With regard to manic symptoms in bipolar disorder, which might be associated with a converse pattern of imbalanced prefrontal activation, i.e. right hypo-and left hyper-activity, preliminary results from a patient showed that anodal tDCS over the right DLPFC induced fast alleviation of acute symptoms (Schestatsky et al., 2013). Stimulation in most studies was applied with current intensities between 1-2 mA and durations between 20-30 min. However, the number of treatment sessions, the interval between sessions, electrode positions, and disease severity vary considerably between studies. So far most studies have shown a potential of tDCS to alleviate depressive symptoms. Stronger stimulation, more stimulation sessions, and tDCS in less severely affected patients might generate larger effects, as implied by the results of the study by Brunoni, Valiengo, et al. (2013). Moreover this study is in favor of a superior efficacy of combined tDCS and pharmacological intervention, which makes sense, given the deficient LTP hypothesis of depression, the positive impact of both agents alone on symptoms, and the strengthening of tDCS-induced LTP-like plasticity by application of serotonin or noradrenaline reuptake-inhibitors (Kuo et al., 2016;Nitsche, Boggio, et al., 2009). tDCS seems to be relatively well tolerated; however, hypomania, and manifest mania were described in relatively rare cases, which might hint to the importance of 'dosed' stimulation. Future studies should be designed to identify optimally suited stimulation protocols with regard to the abovementioned parameters, and, given the positive results of the studies by Brunoni, Valiengo, et al. (2013) and Valiengo et al. (2013), it would make sense to conduct large multi-centre double-blinded sham-controlled trials. Apart from this, it might, furthermore, make sense to explore the potential of tDCS by elucidating the interaction of tDCS with antidepressant medication or medication known to improve the efficacy of anodal tDCS, such as D-cycloserine and reboxetine (Kuo et al., 2016;Nitsche, Jaussi, et al., 2004), combining tDCS with cognitive training of prefrontal functions, comparing the efficacy of different electrode positions directly, testing the impact of remote 'boosting' sessions of tDCS on the maintenance of therapeutic effects, broadening the application of tDCS to specific depression syndromes, such as post-stroke depression, and exploring the cognitive effect of tDCS in more detail. The results of these studies might help to identify optimized stimulation protocols, to learn more about the cognitive impact of tDCS in depression, and suitable patient groups.

Schizophrenia
Schizophrenia is a chronic mental disorder characterized by dysfunction of perception of reality, emotion and cognition. Clinical manifestations include positive (hallucinations, delusions, thought disorders, and bizarre behaviour) and negative (affective flattening, anhedonia, alogia, and attention impairment) symptoms, which are associated with dysregulation of several neuromodulatory transmitters, consequently leading to pathological alterations of cortical activity and plasticity. Deficits of both excitatory and inhibitory neuroplasticity, induced by anodal and cathodal tDCS, respectively, were demonstrated in schizophrenia patients (Hasan, Aborowa, et al., 2012;Hasan, Nitsche, Rein, Schneider-Axmann, Guse, Gruber, et al., 2011). Since tDCSinduced cortical plasticity is dependent on NMDA receptors and is modulated by dopaminergic transmission (Monte-Silva, Liebetanz, Grundey, Paulus, & Nitsche, 2010), this observation can be explained by the imbalance of the glutamatergic and dopaminergic systems in schizophrenia (Goto & Grace, 2007;Javitt, 2010).
For the impact of tDCS on deficient cognitive functions in schizophrenia, one study demonstrated that patients with schizophrenia, as compared to healthy controls, show a more rightward bias in a line bisection task, which was partially corrected by parietal tDCS (Ribolsi et al., 2013). However, in another study, probabilistic associative learning was not improved by simultaneous anodal tDCS over the left DLPFC, with the exception of a sub-set of patients with relatively good baseline performance (Vercammen et al., 2011). These results suggest that tDCS may be able to facilitate cognitive function in schizophrenia, yet more studies are needed to delineate the specific modulatory effects of tDCS on different aspects of cognition, with regard to timing or connectivity between related cortical areas, and to more specific and optimized stimulation protocols.
Regarding the clinical application of tDCS in schizophrenia, inhibiting activity of the left temporoparietal cortex (TPC) to reduce auditory hallucinations (AH), a frequent positive symptom associated with enhanced left-TPC activity, is one potentially relevant target of stimulation. For negative symptoms, which are characterized by dysfunctionally reduced frontal cortex activation, enhancement of this activity by intervention might be a promising approach (Andreasen et al., 1997). Clinical studies using other non-invasive brain stimulation protocols, such as repetitive transcranial magnetic stimulation (rTMS), have demonstrated beneficial effects of respective targeted excitability alterations (for review, see Freitas, Fregni, and Pascual-Leone (2009)).  explored the efficacy of tDCS to ameliorate auditory verbal hallucinations, and to improve negative symptoms, by a bipolar stimulation approach. They applied tDCS in schizophrenic patients within a sham controlled, double-blinded design, in which excitability of the left DLPFC was enhanced by anodal stimulation, and excitability of the left TPC was aimed to be reduced by cathodal tDCS (2 mA for 20 min each session, 2 sessions per day for 5 consecutive days). The authors describe a significant reduction of AHs together with reduced negative symptoms accomplished by active tDCS as compared to sham stimulation, and the effect lasted for up to 3 months after treatment (Brunelin, Mondino, Gassab, et al., 2012). Similar amelioration of AH together with improvement of the deficit in identifying self-generated mental events was also reported in the extended follow-up study (Mondino, Haesebaert, Poulet, Suaud-Chagny, & Brunelin, 2015). Furthermore, the reduction of AH was correlated with a reduction of resting-state functional connectivity (rs-FC) between the left TPC and the left anterior insula (Mondino et al., 2016). A case report with cathodal tDCS over the left TPC describes reduced AH after 10 consecutive daily sessions with 1 mA intensity and 15 min duration of tDCS (Homan et al., 2011). Regional cerebral blood flow measured by arterial spin labelling confirmed a significant reduction of blood flow under the cathode after each session, which might serve as a neurobiological explanation for the effect of tDCS. Persistent AH were also significantly improved in three open-label studies applying the same tDCS montage Brunelin, Hasan, Haesebaert, Nitsche, & Poulet, 2015;Shivakumar et al., 2015). It should be noted that tDCS seems to be more beneficial in non-smoking patients when targeting auditory verbal hallucinations, which could be due to the comorbid plasticity decline observed in long-term smokers present under acute nicotine deprivation (Grundey et al., 2012;Thirugnanasambandam et al., 2011). Furthermore, patients carrying the catechol-O-methyltransferase (COMT) Val158Met polymorphism seem to be more responsive Shivakumar et al., 2015).
Efficacy of tDCS in schizophrenia is further suggested by several case studies Jacks, Kalivas, Mittendorf, Kindt, & Short, 2014;Nawani, Kalmady, et al., 2014;Shenoy et al., 2015) ( Table 2). In single-case studies applying twice daily tDCS for 5 consecutive days, an improvement or even total cessation of AH was shown immediately after the first day of two-session tDCS, and complete cessation after day 5 Rakesh et al., 2013). Moreover, sustained therapeutic effects were maintained for up to 1 year with six add-on boosting sessions after 5-day treatment . Another study reported improvement at the end-point of 10 sessions of tDCS, although 6 days later the symptom rating scores returned to baseline (Praharaj, Behere, & Sharma, 2015). Prolonged treatment courses over years might be more beneficial . On the other hand, a negative result was reported by a recent study implementing 5 days once-daily tDCS sessions targeting the same cortical regions, however with a different electrode arrangement, which might have contributed to the lack of clinical effects (Frohlich, Burrello, Mellin, Cordle, Lustenberger, Gilmore, et al., 2016).
A different stimulation protocol was developed targeting visual hallucinations (VH) and AH simultaneously (Shiozawa, da Silva, Cordeiro, Fregni, & Brunoni, 2013b). Here the cathode was placed over the occipital region with the anode over the left DLPFC for the first 10 sessions in 5 days, and then the occipital electrode was switched to the left TPC for another 5 days. The results revealed improved VH and AH, as well as other negative and positive symptoms (Shiozawa et al. 2013b). On the other hand, an electrode montage with the cathode placed over the right TPC and anode over the left TPC did not result in clinical improvement (Shiozawa, Santos, et al., 2014). Interestingly, tDCS with the anode over the left DLPFC and return electrode over the right supraorbital area also resulted in positive outcomes on both positive and negative symptoms, including catatonia, following 10 treatment sessions (Gomes, et al., 2015;Palm et al., 2013;Shiozawa et al., 2013a), but no effect after only five sessions (Smith et al., 2015). A beneficial effect was also reported when the cathode was moved to an extra-cephalic region, but only for negative symptoms (Kurimori, Shiozawa, Bikson, Aboseria, & Cordeiro, 2015) (Table 2).
Taken together, the results of these pilot studies are promising, but need confirmation by larger multicentre trials. Furthermore, nothing is known about optimally suited tDCS protocols for the treatment of schizophrenia. A number of active mono-centre studies are registered in ClinicalTrials.gov, encompassing the treatment of schizophrenia with tDCS. Some of these studies are dedicated to the improvement of negative symptoms, including cognition, in schizophrenia, and two other studies aim to treat negative   as well as positive symptoms, one of those in childhood-onset schizophrenia. These studies are wellsuited to improve the evidence of an effect of tDCS on clinical symptoms, and to broaden its application range to children. Moreover, some of the studies include measures of physiological effects of tDCS in schizophrenia, which seems to be especially important in this disease because alterations of the glutamatergic and dopaminergic systems have a profound impact on tDCS-induced plasticity (Monte-Silva et al., 2009;Nitsche, M€ uller-Dahlhaus, Paulus, & Ziemann, 2012). Systematic titration of tDCS parameters exploring optimally efficient stimulation protocols is needed for more efficient application in schizophrenia.

Addiction
Substance abuse or dependence remains difficult to treat and relapse rates are high. Addiction is related to abnormal reinforcement of the brain reward circuitry, and prefrontal cortical networks including the DLPFC exert a crucial role in inhibitory control mechanisms (Bechara, 2005;Koob & Volkow, 2010). Indeed, prefrontal tDCS can modify decision-making processes, which may share some common mechanisms with impulsive behaviour in addiction, as shown in healthy subjects (Boggio, Campanhã, et al., 2010;Fecteau, Knoch, et al., 2007;Knoch, Nitsche, Fischbacher, Eisenegger, Pascual-Leone, & Fehr, 2008). Accordingly, decision-making in a risk-taking task similar to the ones applied in the above-mentioned studies was modulated in chronic marijuana users via bilateral DLPFC (R-anodal/L-cathodal) tDCS, and a significant reduction of craving following tDCS was observed in these patients (Boggio, Zaghi, et al., 2010). Similar acute effects of single-session bilateral tDCS were reported for decreasing craving and cigarette consumption in smokers . In a follow-up study, the authors performed 5 consecutive days of bilateral DLPFC stimulation with the same montage, which resulted in not only decreased cigarette consumption but also reduced craving (Boggio, Liguori, Sultani, Rezende, Fecteau, & Fregni, 2009). Similarly, a single tDCS session with the same bilateral DLPFC montage reduced cravings in alcoholdependent patients (Boggio, Sultani, et al., 2008). In a study aiming for a therapeutic effect, it was documented that five weekly sessions of tDCS with the anode positioned over the left DLPFC and the cathode over the right supradeltoid region significantly suppressed craving in alcoholism after tDCS, and in accordance reduced a respective pathological increase of the amplitude of cue-related evoked potentials (da Silva et al., 2013;Nakamura-Palacios, et al., 2012). However, in the latter study the active group also showed a higher tendency to relapse as compared to sham controls (da Silva et al., 2013). In contrast, application of an intensified stimulation protocol with opposite polarity (bilateral DLPFC tDCS, L-cathodal/ R-anodal, 10 twice-daily sessions) showed no effect on craving scores, but a significant decrease in relapse after treatment (Klauss et al., 2014).
The therapeutic potential of tDCS in other substance abuse disorders, such as crack-cocaine addiction, was also explored. Bilateral DLPFC tDCS (R-anodal/L-cathodal) for five sessions resulted in a significant reduction of craving, and also a higher abstinence rate at the 3-month follow-up assessment, as compared to the sham group (Batista, Klauss, Fregni, Nitsche, & Nakamura-Palacios, 2015;Conti & Nakamura-Palacios, 2014). For methamphetamine addicts, single-session tDCS with the anode over the right DLPFC and cathode over the contralateral supra-orbital area showed a state-dependent modulation: tDCS reduced craving during rest, but increased the craving rate when exposed to substance-related cues (Shahbabaie et al., 2014).
In summary, tDCS over the DLPFC shows a potential to reduce substance craving in addiction (Table 3). Bilateral stimulation with both polarities has been shown to be effective, but effects might differ between specific substances, and addiction-related symptoms. In the study with marijuana users, only anodal tDCS over the right DLPFC diminished craving, while L-anodal/R-cathodal tDCS showed no effect (Boggio, Zaghi, et al., 2010). In alcoholism, however, left-DLPFC anodal tDCS (with return electrode over right supradeltoid) reduced craving, but trend-wise increased relapse, while bilateral DLPFC tDCS (R-anodal/L-cathodal) showed opposite effects (da Silva et al., 2013;Klauss et al., 2014). This double dissociation may imply different mechanisms underlying addiction symptoms in alcoholism, and differences between effects of tDCS in different substance disorders. The principal therapeutic effect of tDCS on substance abuse could be related to disruption of reward circuits within and between left and right DLPFC. It may also be associated with modulation of ventromedial PFC (vmPFC) activity via DLPFC tDCS, as suggested in a study showing increased diffusion tensor imaging (DTI) parameters, and, thus, enhanced structural connections between vmPFC and nucleus accumbens following DLPFC tDCS treatment (Nakamura-Palacios et al., 2016). However, further exploration of the pathophysiological mechanisms with optimized stimulation protocols and experimental designs is required for a more effective therapeutic application.

Anxiety disorders
Anxiety disorders, including OCD and general anxiety disorder (GAD), represent another major category of psychiatric diseases. Neuroimaging studies have revealed abnormal patterns of cortical and sub-cortical activation, as well as functional connectivity in OCD patients. Striatal dysfunction, mainly of the caudate nucleus, is thought to result in insufficient thalamic gating, and hyperactivity of orbitofrontal and anterior cingulate cortices, resulting in intrusive thoughts and anxiety. Moreover, the connectivity of the ventral striatum with prefrontal cortices seems to be enhanced in these patients (Del Casale et al., 2011;Sakai et al., 2011). Recently, an inter-hemispheric imbalance with left hyper-and right hypo-activation was suggested by functional imaging in a case report, in which 2 mA 20 min tDCS (cathode -F3/anode -posterior neck) did not alter OCD symptoms, but restored the balance of cortical activity between the two hemispheres, and improved depression and anxiety (Volpato et al., 2013). OCD-associated abnormal hyperactivities in the orbitofrontal-subcortical network, including DLPFC, orbitofrontal cortex (OFC), the anterior cingulate gyrus, the supplementary motor area (SMA), and the basal ganglia, have been shown in neuroimaging studies (Maltby, Tolin, Worhunsky, O'Keefe, & Kiehl, 2005). In a case study, cathodal tDCS over the left OFC with the anode over the right occipital region was applied twice daily for 5 days, and resulted in reduced OCD symptoms (Mondino, Haesebaert, Poulet, Saoud, & Brunelin, 2015). It was also described that 20 sessions of tDCS with the anode positioned over the left pre-SMA/SMA and the cathode over the right supraorbital area significantly improved OCD scores in two patients. This effect was sustained for up to 2 months . In contrast, worsened symptoms were recorded following the same stimulation protocol (but with an extra-cephalic cathode) in another study. However, in this case report, the reversed electrode montage with cathodal tDCS over the right preSMA/SMA for another 10 sessions resulted in reduction of symptom scores for 3 months after treatment (D'Urso et al., 2016). Regarding tDCS effect on general anxiety disorder, tDCS was shown to be effective in a case report with cathodal tDCS over the right DLPFC (anode over the left deltoid, 2 mA 30 min daily for 3 weeks), and this improvement lasted for up to 1 month (Shiozawa, Leiva, et al., 2014). It was also described that anorexia nervosa was ameliorated after 10 sessions of tDCS with the anode over the left DLPFC and the cathode positioned on the right arm (Khedr, Elfetoh, Ali, & Noamany, 2014) (Table 4).
To sum up, the application of tDCS for anxiety disorders might be promising, but results to date are preliminary. One mechanisms of action might be stimulation-induced alteration of dysfunctional cortico-subcortical networks including cortico-striatal, and cortico-thalamic loops (Polan ıa, Paulus, & Nitsche, 2012c), which are involved in the pathophysiology of anxiety disorders.

Outlook: optimizing therapeutic effects of tDCS
One critical aspect of the future application of tDCS in psychiatric diseases is the optimization of stimulation frequency, duration, and strength, as well as electrode position, to achieve optimal clinical effects. In the following section, we will discuss future optimized stimulation protocols based primarily on results of primary motor cortex stimulation, as this area has been studied most thoroughly based on advanced understanding of underlying neurophysiology. In addition, study designs including treatment course and concomitant therapy will also be covered in this section.

Stimulation intensity and duration
For anodal tDCS, stronger and longer-lasting stimulation results in larger effects, as shown by varying stimulation intensity between 0.2-1 mA, and stimulation duration between 1-5 min (Nitsche & Paulus, 2000). On this basis, stimulation duration and stimulation intensity has been extended in many clinical studies, as compared to the initial protocols. For stimulation intensity, tDCS with 2 mA for 10 min resulted in effects similar to stimulation with 1 mA . Extending the stimulation duration to 20 min with 2 mA current strength reversed, however, the effects of cathodal tDCS from excitability diminution to enhancement (Batsikadze, Moliadze, Paulus, Kuo, & Nitsche, 2013). Moreover, prolongation of 1 mA anodal tDCS to 26 min generated excitability diminution (Monte-Silva et al., 2013). These results, obtained in the motor cortex of healthy humans, revealed a non-linear relationship between stimulation parameters and the direction of aftereffects, which constrains simple extension of stimulation duration to obtain stronger and longer-lasting after-effects. It should, however, be taken into account that, in some studies, longer anodal tDCS durations have been performed in neuropsychiatric patients with a positive outcome on clinical symptoms, most probably due to excitability-enhancing effects of stimulation, and the pathophysiological nature of the disease. Brunoni and colleagues reported reduction of depression symptoms by 30 min anodal stimulation, and clinical benefits have also been described in other psychiatric disorders with the same or alike stimulation protocols (Brunoni, Valiengo, et al., 2013;Shiozawa, Leiva, et al., 2014, Shiozawa, Santos, et al., 2014Valiengo et al., 2013). Similar results have been obtained in some studies conducted in neurological diseases (Floel, 2014). Thus, a one-to-one translation of the physiological results obtained in healthy young subjects (Monte-Silva et al., 2013) might be questionable. Specific conditions of the brain state in the target population, such as compromised LTP-like plasticity in depressed subjects, might broaden the range for excitability-enhancing effects of tDCS. However, possible non-linear effects of stimulation on excitability should be considered with regard to the implementation of intensified stimulation protocols. Increasing tDCS intensity is, furthermore, restricted due to the induction of pain sensation at current strengths of $3 mA and a blinding problem at 2 mA with electrode sizes between $20-35 cm 2 , which could be at least partially prevented by the application of topical anaesthetics over the stimulation area (Batsikadze et al., 2013;O'Connell et al., 2012).

Treatment course
Repetition of stimulation can enhance the after-effects of tDCS. The physiological effects of repetitive stimulation on the after-effects of tDCS have been evaluated for relatively short (3 and 20 min), and long (3 and 24 h) inter-tDCS intervals (1 mA, 35 cm 2 electrodes, 9 min cathodal/13 min anodal tDCS) (Monte-Silva et al., 2013;Monte-Silva, Kuo, Liebetanz, Paulus, & Nitsche, 2010). Specifically, the short intervals prolonged the after-effects for at least 24 h after anodal tDCS (Monte-Silva et al., 2013). These results are in accordance with late-phase plasticity induction procedures in animal slice preparations, where a critical time window of $30 min was described (Reymann & Frey, 2007). The treatment course with tDCS in patient studies typically ranged from 5-20 sessions so far. Most studies applied 10 sessions for 2 weeks, with Table 4. tDCS for anxiety disorders. Shown are studies dedicated to treatment of anxiety in chronological order. 5 consecutive days each week (see Tables 1-4). When compared, 10 sessions of stimulation, either once or twice daily, show more benefits than only five sessions, as least in schizophrenia (Gomes et al., 2015;Shiozawa, et al. 2013a;Smith et al., 2015). It has, furthermore, been shown in depression that, after the 2week treatment, a follow-up stimulation protocol with five sessions per week for 6 alternating weeks and then three monthly sessions consolidated and prolonged the clinical effects . In general, study courses with repeated sessions reported so far are restricted to at most 5 days per week, with a 2-day break between each stimulation week. It remains unclear whether continuous long-term treatment with tDCS will generate better therapeutic effects, and results from rTMS studies suggest that even current prolonged tDCS protocols might be too short. The development of home-based tDCS interventions, whose feasibility has been tested in other chronic diseases (Kasschau et al., 2016;O'Neill, Sacco, & Nurmikko, 2015), may enable the investigation and establishment of more advantageous treatment protocols.

Electrode montage
With regard to the focality of tDCS, the conventional bipolar electrode arrangement with large electrodes delivers a relatively non-focal stimulation . More focal effects can be achieved by reducing stimulation electrode size, or increasing the size of the return electrode, thus enabling a functional monopolar stimulation (Nitsche, Doemkes, Karakose, Antal, Liebetanz, Lang, et al., 2007). Moreover, the return electrode can be placed at remote areas distant from the head, although tDCS might be less efficient with this electrode arrangement (Moliadze, Antal, & Paulus, 2010). This does not imply that an extracephalic return electrode position makes stimulation functionally ineffective, as shown by studies in which tDCS for depression (Martin et al., 2011), schizophrenia (Kurimori et al., 2015), and anxiety (Shiozawa, Leiva, et al., 2014) was applied. Statements about the relative clinical efficacy of cephalic vs extracephalic return electrode positions are not possible at present, because no studies have been conducted to compare these protocols directly. Furthermore, different neuronal populations due to different current flow, and electrical field orientation, might be affected by these protocols. A principal problem of an extracephalic return electrode position might be the activation of brainstem structures; however, possible problematic vegetative effects have not been present in a recently conducted study (Vandermeeren, Jamart, & Ossemann, 2010). Another option to focalize the effects of tDCS might be the so-called high definition (HD) tDCS. Here a relatively small central stimulation electrode is surrounded by four return electrodes, which are presumed to be functionally inert.
Modelling suggests that this electrode arrangement results in more focal effects than the conventional electrode arrangement (Bikson, Rahman, & Datta, 2012;Kuo et al., 2013). Moreover, it is effective at physiological and functional levels (Borckardt et al., 2012;Kuo et al., 2013). Physiological validation of increased focality of the effects, however, is missing so far. It waits to be seen if more focal stimulation is more efficient for the treatment of neuropsychiatric diseases. Better-targeted stimulation might result in less side-effects. However, in some diseases relatively large areas would be preferentially targeted for modulation. Therefore, benefits and shortcomings of focal stimulation with regard to clinical application of tDCS should be discussed thoroughly for each project. Whereas the focus of tDCS effects so far was dedicated to regional effects under the stimulation electrodes, it also modulates the activity within and between different cortical networks. Primary motor cortex stimulation has been shown to increase the connectivity of cortico-cortical and cortico-subcortical motor network components, including premotor and parietal areas, as well as thalamic nuclei, and the caudate nucleus, in the resting human brain, as shown by fMRI. An EEG study demonstrated similar effects of tDCS on motor networks in the gamma frequency range. Here tDCS increased respective motor taskrelated activations (Polania, Paulus, & Nitsche, 2011, 2012a, 2012b, 2012c. Beyond the motor cortex, prefrontal tDCS affects resting network connectivity (Keeser et al., 2011), and anodal stimulation of the inferior frontal gyrus, an area critically involved in language production, resulted in increased connectivity of this area with other major hubs of the language network in the resting brain. Interestingly, in this study tDCS improved word retrieval, suggesting a functional relevance of the respective network activation (Meinzer et al., 2012).
Pathophysiological alteration of neural network connectivity associated with psychiatric symptoms has been revealed by neuroimaging techniques (for review, see Narr & Leaver (2015) and Northoff (2016)). For example, abnormal resting-state functional connectivity, which revealed a pathological pattern of connectivity within a fronto-temporal network, has been related to AH in schizophrenia (Hoffman & Hampson, 2011).
Recently it has been shown that tDCS decreased the functional connectivity of the left TPC with the left anterior insula and the right inferior frontal gyrus, and increased connectivity of the left TPC with the left angular gyrus, the left dorsolateral prefrontal cortex, and the precuneus in patients with auditory verbal hallucinations (Mondino, Haesebaert, Poulet, Saoud, et al., 2015;Palm et al., 2013). Moreover, the improvement of clinical symptoms was correlated with respective tDCS-induced changes of functional connectivity within this disease-related network (Mondino et al., 2016). In a study investigating tDCS effect on drug addicts, bilateral DLPFC tDCS increased connectivity between vmPFC and nucleus accumbens, as shown with DTI, and the alteration of DTI parameters was associated with craving reduction (Nakamura-Palacios et al., 2016). These results suggest a modulating effect of tDCS on the symptom-associated cortical networks and connectivity beyond the stimulation sites, which could hint at future applications of tDCS with combined stimulation of disease-relevant connected areas rather than stimulation of a single structure.

Concurrent therapy
Another option to prolong and strengthen the aftereffects of tDCS is the combination of stimulation with pharmacological interventions. The partial NMDA receptor agonist D-cycloserine, amphetamine, serotonin, and reboxetine all have been demonstrated to enhance the efficacy of anodal tDCS (Kuo et al., 2016;Nitsche, Jaussi, et al., 2004, Nitsche, Grundey, et al., 2004, whereas application of L-dopa, as well as dopamine agonists extend the after-effects of cathodal stimulation (Fresnoza et al., 2014;Kuo, Paulus, & Nitsche, 2008;Monte-Silva et al., 2009Nitsche et al., 2006). The latter effects have been shown to be non-linearly dosage-dependent. Combination of pharmacological intervention with stimulation might be especially well-suited for diseases in which the specific drugs are applied for therapeutic reasons, e.g. application of serotonin re-uptake inhibitors with anodal tDCS for treatment of depression (Brunoni, Valiengo, et al., 2013).
For combination of tDCS with cognitive therapy in psychiatric disorders, three studies have explored the effect of combining tDCS and cognitive training in patients with depression, and mixed results have been described D'Urso et al., 2013;Segrave et al., 2014). It is not clear whether the timing of tDCS and cognitive training may contribute to the results in the respective studies. A relevance of timing was observed for the combination of tDCS and behavioural/motor task performance in rehabilitative settings with regard to neurological diseases, e.g. motor rehabilitation after stroke (Floel, 2014). For most rehabilitation protocols, stimulation and rehabilitation therapy were so far conducted simultaneously. In healthy subjects, with regard to a sequential motor learning task, anodal tDCS of the primary motor cortex during, but not before learning improved performance, and premotor cortex stimulation, which did not improve performance during learning, resulted in improved outcome when applied during REM sleep, during which this area is involved in reconsolidation processes Nitsche et al., 2010). For a visuo-motor consolidation task, however, anodal tDCS improved performance when applied not only during, but also before learning (Antal, Begemeier, Nitsche, & Paulus, 2008;Antal et al., 2004). In the latter condition, cathodal tDCS also improved performance. Thus, it might be speculated that anodal tDCS during learning boosts task-related plasticity via the addition of stimulation-induced plasticity, maybe mediated via activity-dependent calcium influx, while anodal stimulation before performance might gate task-related plasticity, and cathodal tDCS before performance might improve it via homeostatic mechanisms (Ziemann & Siebner, 2008). In contrast, a recently conducted study showed superior effects of anodal stimulation, when applied before performance of an implicit visual perceptual learning task (Pirulli, Fertonani, & Miniussi, 2013). However in this study a repetitive tDCS protocol with relatively short stimulation duration was performed, which makes it difficult to speculate about the net impact of this protocol on cortical excitability (Fricke et al., 2011). Although conceptually it makes sense that, for learning, tDCS during performance should be more effective, due to not only NMDA receptor-, but also calcium channel-mediated intracellular calcium increases, the latter induced by tDCS-dependent membrane depolarization, clearly more systematic studies are needed to explore this topic further. For cognitive processes, which might not require the induction of neuroplasticity, e.g. working memory, or attentional processes, similarly most studies have been performed with tDCS during performance, but systematic studies comparing differently timed stimulation protocols are missing (for an overview see Kuo and Nitsche (2012)).

Conclusion
This review gathers clinical trials conducted for the treatment of psychiatric diseases via 'modern' tDCS protocols, i.e. stimulation protocols which have been physiologically validated. In general, the results from most studies are promising, demonstrating effects of stimulation in a variety of psychiatric diseases accompanied by pathological alterations of cortical excitability and activity. In principal, two groups of studies can be discerned: early pilot experiments, which are dedicated primarily to the evaluation of principal efficacy of tDCS to improve symptoms, and later controlled trials, which aim to induce clinically relevant effects. For the latter, a limited number of diseases was explored so far. Relatively clearly, clinically relevant effects seem to be achieved in depression and schizophrenia. Importantly, side-effects so far are rare and mild, if any. Before tDCS can be implemented into clinical practice, however, larger multi-centre studies are needed also for these relatively well-explored diseases. One important aspect to clarify is the definition of optimized stimulation protocols. Here, interesting new approaches do exist, which are, however, based mostly on stimulation results in healthy subjects so far. The transferability of the respective results to patient populations is yet to be investigated.

Disclosure statement
MFK and PSC report no conflicts of interest. MAN is on the Advisory Board of Neuroelectrics.