key: cord-0318049-fxu5r2qq
authors: Jafari, E.; Alizadehgoradel, J.; Pourmohseni Koluri, F.; Nikoozadehkordmirza, E.; Refahi, M.; Taherifard, M.; Nejati, V.; Hallajian, A.-H.; Ghanavati, E.; Vicario, C. M.; Nitsche, M. A. A.; Salehinejad, M. A.
title: Intensified stimulation targeting lateral and medial prefrontal cortices for the treatment of social anxiety disorder: A randomized, double-blind, parallel-group, dose-comparison study
date: 2021-06-11
journal: nan
DOI: 10.1101/2021.06.08.21258427
sha: 2d9ec6e8afa4f486f592ef9172ea85dc4a972f98
doc_id: 318049
cord_uid: fxu5r2qq

Background: Social Anxiety Disorder (SAD) is the most common anxiety disorder while remains largely untreated. Disturbed amygdala-frontal network functions are central to the pathophysiology of SAD, marked by hypoactivity of the lateral prefrontal cortex (PFC), and hypersensitivity of the medial PFC and amygdala. The objective of this study was to determine whether modulation of dorsolateral and medial PFC activity with a novel intensified stimulation protocol reduces SAD core symptoms, improves treatment-related variables, and reduces attention bias to threatening stimuli. Methods: In this randomized, sham-controlled, double-blind trial, we assessed the efficacy of an intensified stimulation protocol (20 min, twice-daily sessions with 20 min intervals, 5 consecutive days) in two intensities (1 vs 2 mA) compared to sham stimulations. 45 patients with SAD were randomized in three tDCS arms. SAD symptoms, treatment-related variables (worries, depressive state, emotion regulation, quality of life), and attention bias to threatening stimuli (dot-probe paradigm) were assessed before and right after the intervention. SAD symptoms were also assessed at 2-month follow-up. Results: Both 1 mA and 2mA protocols significantly reduced fear/avoidance symptoms, worries and improved, emotion regulation and quality of life after the intervention compared to the sham group. Improving effect of the 2 mA protocol on avoidance symptoms, worries and depressive state was significantly larger than the 1 mA group. Only the 2 mA protocol reduced attention bias to threat-related stimuli, the avoidance symptom at follow-up, and depressive states, as compared to the sham group. Conclusions: Modulation of lateral-medial PFC activity with intensified stimulation can improve cognitive control, motivation and emotion networks in SAD and thereby results in therapeutic effects. These effects can be larger with 2 mA vs 1 mA intensities, though a linear relationship between intensity and efficacy should not be concluded.

either exaggerated activation such as the medial prefrontal cortex (PFC), anterior cingulate cortex, orbitofrontal cortex [18, 21] or altered activity (e.g. decreased or increased) during emotional and threat processing including dorsolateral PFC (DLPFC) [21] [22] [23] . Recently important insights were delivered by nuclear neuroimaging as well. These findings support functional models of SAD by showing abnormalities of neuronal activity in several key limbic and paralimbic regions, including the medial frontal cortex, and also neurochemical pathologies of SAD (i.e., dopaminergic and serotonergic systems) [24] .

At the network level, disturbed and aberrant connectivity between the amygdala and frontal cortex [25, 26] , and the parietal-occipital network [19] are assumed to be involved in core symptoms of SAD such as emotional reactivity, poor cognitive control over negative emotions and negative self-referential processing [18, 19, 25] . These amygdala-frontal cortex network alterations can be explained as disturbances between the emotion network (involving the amygdala), the motivation network (involving medial and orbitofrontal PFC), and the cognitive control network (involving the DLPFC) [18] . A hyper-responsive emotional network interacts with a sensitive motivation system that rewards and punishes non-social and social stimuli respectively and both of these networks are associated with a diminished cognitive control and emotion regulation network [18] . This is in line with a hot-cold cognition perspective of anxiety disorder where a deficient cold regulatory process (DLPFC, anterior cingulate cortex) and an overresponsive system for emotional and reward/punishment processing (medial PFC, subcortical regions) are involved [20] . Similarly, there are both functional and structural alterations in the parietal-occipital network, and its connections to the anterior network. Here a hyperactivation of medial parietal and occipital regions (posterior cingulate, precuneus, cuneus) and reduced connectivity between parietal and limbic and executive network regions are well-documented [19] that are closely related to negative self-image . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ;  https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint and biased self-referential processing in patients with SAD. Modulating cortical activity by noninvasive brain stimulation may be useful in order to directly target and alter functionality of the brain networks involved in SAD [10, 27] . Neuromodulation studies that tackle these treatmentrelevant variables might therefore be useful for developing innovative treatments for SAD.

Transcranial direct current stimulation (tDCS) is emerged promising for the treatment of anxiety disorders [10] [11] [12] and fear response [13] . tDCS is a non-invasive, painless, and welltolerated brain stimulation technique that applies a weak direct current (typically 0.5 mA-2 mA) through surface electrodes on the scalp. It can induce acute and neuroplastic alterations of cortical excitability via subthreshold neuronal depolarization and induction of LTP-like plasticity (anodal stimulation), or hyperpolarization and LTD-like plasticity (cathodal stimulation) [28, 29] . tDCS is increasingly used for studying physiological and neurocognitive functions in the healthy brain, and for clinical applications (for a detailed review see [30, 31] ). Cognitive, emotional and motivational functions of the lateral and medial PFC are increasingly being studied with tDCS [32] [33] [34] [35] [36] . The efficacy of tDCS, especially for therapeutic purposes, depends on different factors including stimulation parameters (e.g. target region, intensity, duration, repetition rate, repetition interval) [37] [38] [39] some of which such as target region, repetition rate and interval, are not wellstudied in randomized clinical trials in SAD. Very few studies so far, have investigated efficacy of tDCS in SAD. A recent tDCS study investigated the effects of single-session anodal DLPFC stimulation on attentional bias to threat in SAD and found a significant decrease of attentional bias during anodal tDCS [40] . No other tDCS study in SAD is available at present. Two case studies, that applied low-frequency excitability-diminishing repetitive transcranial magnetic stimulation (r-TMS) over the right ventromedial PFC (VMPFC) of three patients with SAD resulted however in a significant reduction of anxiety levels after 12 sessions of rTMS [41, 42] .

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint

In order to evaluate the potential of tDCS for the treatment of SAD, trials with longer courses of stimulation, inclusion of primary and secondary clinical scales and cognitive measures, and monitoring of long-term outcomes are required. Accordingly, in this registered, randomized, sham-control clinical trial we aimed to (1) investigate the effects of intensified tDCS (repeated stimulation with 20 min interval) over the left DLPFC and medial PFC on primary and secondary clinical variables in patients with SAD, (2) compare the effect of different stimulation intensities (1 vs 2 mA) on treatment efficacy in comparison to the sham condition, (3) examine the effects of these interventions on attentional bias to threat in patients with SAD, and (4) to see whether the expected improvement in attentional bias to threat is correlated with clinical improvement and symptom reduction. The rationale behind this stimulation protocol is related to functional relevance of target regions and stimulation parameters. With regard to the former, our protocol was designed to upregulate the executive control network via anodal DLPFC stimulation and downregulate exaggerated emotional reactivity by stimulation of the medial PFC via its connection to subcortical limbic structures [10, 18] . With regard to stimulation parameters, we were interested to examine the efficacy not only of different stimulation intensities but also of repetitive stimulation with a short (20 min) interval, which induces late-phase plasticity at the physiological level [43] . This is the first tDCS study in SAD with a randomized parallel-group design that explores the effects of a novel intensified tDCS intervention at two different stimulation intensities on symptoms reduction and attentional bias.

Fifty-six individuals with SAD (18-50 years, mean age=32.36±6.99) were initially recruited from those referred to six Neuropsychiatric Clinics in Ardabil, Iran. Forty-five patients . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint who met the inclusion criteria were randomized into three study arms. All patients were diagnosed with SAD according to the DSM-5 criteria [3] by experienced, licensed psychiatrists. Sample size was a-priori calculated (f=0.35 equivalent to a medium partial eta squared of 0.10, α=0.05, power=0.95, N=30, for a mixed-model ANOVA with 3 measurements, group (1 mA, 2 mA, sham) as the between-subject factor and time (pre, post, follow-up) as the within-subject factor. We added 15 more participants to compensate for potential dropouts. Four patients from all groups could not complete the whole treatment, and thus final analysis of post-intervention measurements was conducted on 41 participants (1mA tDCS N=14, 2mA tDCS N=13, sham tDCS N=14). The followup measurement was conducted in 36 patients (12 subjects per group) due to dropout of five more participants (Fig 1) . The inclusion criteria were: (1) diagnosis of SAD via the structured clinical interview according to the DSM-5, (2) 18-50 years old, (3) non-smoker, (4) no previous history of neurological diseases, brain surgery, epilepsy, seizures, brain damage, head injury or metal brain implants, and (5) absence of other psychiatric disorders except for SAD, as confirmed by a structured clinical interview conducted by a professional licensed psychiatrist. Overall, 19 patients had comorbidities with other anxiety disorders, including generalized anxiety disorder, posttraumatic stress disorder, and panic disorder (Table 1) . Patients who were taking anxiolytic (BZD; n=23) and/or anti-depressant (SSRI; n=12) medication had a stable medication regime from one month before the start of the study up to the follow-up. All participants were native speakers and had normal or corrected-to-normal vision. The study was a registered clinical trial (trial ID: . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint [ Table 1 here]

The LSAS is one of the most commonly used clinical scales for the assessment of social phobia [44] . It is a 24-item scale assessing the range of social interaction and performance situations that individuals with SAD may fear and/or avoid, and a valid and treatment-sensitive measure of social phobia [44] . The clinician asks patients to rate fear and avoidance during the past week on a 0-3 Likert-type scale with 0 and 3 indicative of "never" (0%) and "usually" (67%-100% of the time) occurring. The overall total score is calculated by summing up fear and avoidance . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint scores. The LSAS has a Cronbach's alpha of 0.96 for the overall score and 0.92 for total fear and total avoidance scores, respectively [44] . Cronbach's alpha in our sample was 0.91 for the overall score, 0.90 for the "fear" subscale, and 0.92 for the "avoidance" subscale. Patients' symptoms were evaluated before (pre-intervention), immediately after (post-intervention), and two months after the end of the intervention (follow-up).

Additionally, we measured worry with the Penn State Worry Questionnaire (PSWQ) [45] , depressive states with the Beck Depression Inventory-II (BDI-II), emotion regulation with the Difficulties in Emotion Regulation Scale (DERS) [46] , and quality of life with the WHOQUL questionnaire [47] . These measures were used to assess the clinical efficacy of the intervention on additional aspects impaired in patients with SAD [7, 48, 49] . Patients completed all measures before and after the intervention. A detailed description of these measures is presented in the supplementary materials.

In addition to clinical measures, attention bias to threat stimuli was measured using the dotprobe paradigm [50] . Attention bias is considered a robust cognitive construct for measuring treatment efficacy in emotional disorders [51] . Studies using this paradigm have shown that social anxiety is associated with an attentional bias towards disorder-related stimuli such as threatening faces [52] . Two stimuli are presented to the participants that appear randomly on either side of the screen for a pre-determined time before an asterisk is presented at the location of one former stimulus. Participants are instructed to indicate the location of the asterisk as quickly as possible via the response box. In this computerized task, 80 pictures of facial emotional expressions (40 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. [55, 56] . The two experimenters who applied tDCS were blind to study hypotheses but not to the tDCS conditions and did not participate in any other parts of the study.

A side-effect survey was done after each tDCS session (Fig. 2a) . A 3D model of the current flow . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint in the head was created to determine induced electrical fields in the brain for the above-mentioned tDCS protocol (anodal F3 -cathodal medial PFC) using ROAST [57] , an open-source pipeline for transcranial electrical stimulation (tES) modeling. Details of the modeling procedure are summarized in Fig 2C and supplementary information.

Prior to the experiment, participants completed a brief questionnaire to evaluate their suitability for brain stimulation. All groups of participants received 10 sessions of stimulation (2 sessions daily, 1 week in total) with 20 min intervals between daily sessions. Participants were asked to sit relaxed in a waiting room during stimulation intervals. Clinical measures (e.g. social anxiety symptoms, worry, depressive states, emotion regulation and quality of life) and attention bias task performance were evaluated immediately before the first tDCS session (pre-intervention), right after the end of the last tDCS session (post-intervention), and 2 months following the last stimulation session (follow-up) (Fig 2A) . tDCS sessions were conducted between 2:00-5:00 PM in all patients and across all sessions. The 2-month follow-up measurement was conducted during the COVID-19 pandemic and participants were not allowed to physically attend the assessment session. We did, however, assess the LSAS, as the primary clinical measure, via an online interview. Patients were instructed about the task before the beginning of the experiment. Except for the follow-up session which took place remotely, each measurement session took around 2 hours. To guarantee double-blinding, examining outcome measures, data analysis and group assignment were performed by independent researchers who were not involved in delivering stimulation sessions.

[ Figure 2 here] . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint

Our study had a randomized, double-blind, parallel-group design to prevent blinding failure and carry-over effects. Participants were blind to study hypotheses and stimulation conditions. The experimenter who conducted the outcome measures was blinded to the tDCS conditions. To guarantee blinding of this investigator, tDCS was applied by other investigators [56] . Data analyses were conducted with the statistical package SPSS, version 27.0 (IBM, SPSS, Inc., Chicago, IL).

The normality and homogeneity of data variance were confirmed by Shapiro-Wilk and Levin tests, respectively. Mixed model ANOVAs were conducted for the dependent variables (LSAS, PSWQ, BDI-II, WHOQUL scores; attention bias task RT) with "group" (active 1mA, active 2mA, sham) as the between-subject and time (pre-intervention, post-intervention, follow-up only for LSAS) as the within-subject factors. Mauchly's test was used to evaluate the sphericity of the data before performing the respective ANOVAs (p<0.05). In case of violation, degrees of freedom were corrected using Greenhouse-Geisser estimates of sphericity. Post hoc analyses were calculated using Bonferroni-corrected Student´s t-tests and included the pairwise comparisons of interest within intervention groups (pre-intervention vs post-intervention, pre-intervention vs follow-up for LSAS only) and between intervention groups (pre-intervention: 1 mA vs 2 mA; postintervention: 1 mA vs 2 mA). Associations between clinical symptoms and behavioral performance were explored via the Pearson correlation (p<0.05).

Patients tolerated the stimulation well and no adverse effects were reported during and after stimulation. No significant difference was found between the group ratings of tDCS side effects . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint (Table S1 ). The mean and standard deviation of the dependent variables before and after the intervention is presented in Table 2 . No significant between-group differences were observed for baseline measurements.

[ Table 2 here]

The results of a 2 (domain: fear, avoidance) ×3 (time: pre, post, follow-up) ×3 (group) mixed ANOVA revealed a significant interaction of group×time (F2.29,66=10.7, p<0.001, ηp2=0.37) on LSAS scores. The interaction of group×time×domain, group×domain, and time×domain was not significant (Table 3) indicating that fear and avoidance subscales were not differently affected by the intervention. Bonferroni-corrected post hoc analyses (p=0.005) showed a significant decrease of both fear and avoidance scores at post-intervention and 2-month followup measurements in both, 1 mA (post-intervention: tfear=7. 18 (Fig 3a,b) . Baseline . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint between-group comparisons (active groups vs sham) showed no significant differences preintervention, indicating that reduced fear and avoidance symptoms were specific to tDCS effects.

No significant difference was found between intervention groups (1 vs 2 mA) for the fear scores in the post-intervention (p=0.524) and follow-up (p=0.083) measurement. The reduction of the avoidance symptom was, however, significantly larger in the 2 mA vs 1 mA condition right after (t=2.47, df=35, p=0.045) and 2 months (t=3.45, df=35, p=0.002) following intervention (Fig. 3a,b) .

Individual scores of LSAS, as the primary measure, are displayed in scatter plots for each group in the supplementary materials (Fig. S1 ).

For worries, the results of the 2×3 mixed model ANOVA showed a significant group×time interaction (F2,38=12.83, p<0.001, ηp2=0.40) on PSWQ scores. The main effects of group and time were significant as well (Table 3) . Bonferroni-corrected post hoc t-tests (p=0.016) showed a significant decrease of worries between pre-and post-intervention measurements in the 1 mA (t40=5.56, p<0.001) and 2 mA (t40=7.85, p<0.001) groups, but not the sham group (t=0.54, p>0.999). Between-group comparisons of PSWQ scores with adjusted p value of 0.008 showed no significant difference in the pre-intervention measurements, but significant enhancement of scores in the post-intervention measurements (1 mA vs sham: t40=4.52, p<0.001; 2 mA vs sham: t40=7.14, p<0.001; 1 mA vs 2 mA: t40=2.75, p<0.025) (Fig. 3c) . This indicates that decreased worry was specific for tDCS effects with a significantly larger effect for the 2 mA group. In the BDI-II, we observed an interaction of group×time (F2,38=4.05, p=0.025, ηp2=0.17) . The main effects of group and time were significant as well (Table 3) . Post hoc t-tests (adjusted p=0.016) showed a significant decrease of depressive state between pre-and post-intervention measurements in the 1 mA (t40=3.57, p=0.001) and 2 mA (t40=4.86, p<0.001) groups only. However, when compared to . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint the sham group (adjusted p=0.008), a significant decrease of depressive states was observed only in the 2 mA (t40=5.36, p<0.001) but not the 1mA group (t40=2.44, p=0.056), and the magnitude of change was significantly larger in the 2 than in the 1 mA group (t40=3.01, p=0.010) (Fig. 3d) . No significant differences in depressive state were found between groups in the pre-intervention measurement.

For emotion regulation abilities, a significant interaction of group×time (F2,38=4.16, p=0.023, ηp2=0.18) was observed on the DERS score. The main effects of group and time were also significant ( Table 3) (adjusted p=0.008). There were no significant between-group differences in the pre-intervention . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint measurements, but significant differences were found in the post-intervention measurements for all domains of quality of life (Fig. 3f). [ Table 3 and Fig 3 here] . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint

The results of the 2×2 ANOVA showed no significant group×time interaction or main effects of time and group on the mean bias score (Table 3) . However, when we entered stimuli (threat vs neutral) in a 2×2×3 mixed ANOVA, the results showed a significant group×time interaction (F2,38=12.01, p<0.001, ηp2=0.38 ) as well as significant main effects of time, group, and stimuli on the RT of task performance. Bonferroni-corrected post hoc t-tests (adjusted p=0.016) revealed a significant pre vs post-intervention RT reduction of the threat-related stimuli only in the 2 mA group (t40=3.12, pA=0.007) but not 1 mA (t40=1.88, pA=0.190) or sham (t40=1.39, pA=0.504) groups (Fig. 3g) . No significant difference in the post-intervention RT was observed between the 1 mA vs 2 mA groups for both threat (p=0.595) and neutral stimuli (p=0.413). The same pattern of response was found for the neutral stimuli except that here both interventions significantly reduced RT after intervention and compared to the sham group (Fig. 3h) . These results show that the intervention had an overall improving effect of patients' RT, regardless of the stimuli valence, but did not improve the bias score in patients.

Lastly, we calculated Pearson's correlations to see if SAD symptoms in the postintervention and follow-up measurements correlate with attentional bias task performance. No significant correlation was found between the post-intervention bias scores and fear, avoidance and LSAS total score. We however found significant positive correlations between reduced RT for both threat-related and neutral stimuli and alleviated SAD symptoms, including fear symptom . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint

In this randomized, sham-controlled, parallel-group clinical trial, we investigated the impact of a novel, intensified tDCS protocol (stimulation twice per day with 20 min interval), with intensity dosage comparison (1 mA vs 2 mA) on primary symptoms, clinical measures, and attentional bias in patients with SAD. Both active stimulation conditions, compared to the sham group, significantly reduced fear and avoidance symptoms after the intervention. Fear symptom improvement continued for up to 2 months in both active groups, while avoidance was significantly reduced for up to 2 months only in the 2 mA stimulation group. In treatment-related measures, except for depressive state, which was significantly reduced only in the 2 mA group, other measures (worries, emotion regulation, quality of life) were significantly improved in both active stimulation conditions versus the sham group after the intervention. Moreover, depressive symptoms and worries reduction were significantly larger in the 2 mA vs the 1 mA group. Finally, we found that only the 2 mA protocol reduced RT of both threat-related and neutral stimuli as compared to the sham group although attention bias score was not significantly different across groups.

Reduction of SAD symptoms and clinical improvement following active stimulation can be first and foremost explained via the assumed changes in the lateral-medial prefrontal network, which is involved in the pathophysiology of SAD. Our findings support the assumed pathophysiological mechanism that impaired executive control network functionality (involving DLPFC), along with disturbed motivational and emotional networks (involving VMPFC, amygdala) are central for the development and maintenance of SAD symptoms [18, 20] . The stimulation protocol in our study was designed based on evidence showing a hypoactivation of the lateral PFC [58, 59] along with exaggerated activation of the medial PFC in trait anxiety, SAD [14, . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint 18, 21] and anxiety disorders [10, 13] although functional abnormalities have not been consistent across studies. Based on these studies, we speculate that anodal stimulation-generated excitability enhancement of the DLPFC increased functional connectivity of this region with SAD-relevant networks hereby increased control of the threatening stimuli. On the other hand, cathodal stimulation of the medial PFC was expected to decrease excitability of this region and modulate its connectivity with emotion networks and hereby alters emotional experience of social threats.

Reduced fear induced by, and avoidance of social stimuli, lower worries and depressive states, and improved emotion regulation may be explained via concurrent upregulation of executive control and downregulation of threat-sensitivity (Fig. 4) . Indeed, such cognitive control-emotion regulation association and its improvement via prefrontal tDCS has been shown in other neuropsychiatric disorders associated by emotional disturbances [60] [61] [62] .

Moreover, we found overall reduced attention to both threat-related and neutral stimuli which were specific to the 2 mA group. Although we expected a specific reduced bias to threatrelated stimuli only, this can be interpreted as an overall enhancement of attentional resources, regardless of the valence of stimuli as a result of DLPFC activation. In accordance, in healthy individuals anodal left DLPFC stimulation (2 mA) reduced vigilance to threatening stimuli in a similar dot-probe task [27] . In another tDCS-fMRI study in individuals with high trait anxiety, active DLPFC stimulation increased activity of cortical regions associated with attentional control and was associated with behavioural improvement in a threat-related attentional task [63] . Our findings did not show a threat-specific reduction of attention bias, but an overall performance enhancement (i.e., reduced RT) for all stimuli which can be attributed to the facilitation of attentional resources as a result of prefrontal tDCS [64, 65] .

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint negative emotions and overreactions to threatening stimuli. It is of note, this is a simple proposed mechanism for the effects of the applied intervention (anodal left DLPFC, cathodal medial PFC). While this model considers mostly the cognitive control network (related to DLPFC) and motivational network related to medial PFC, ventral ACC, and subcortical regions, other involved networks in SAD including the default mode network (DMN) and dorsal attention network (DAN) are not depicted in this model. Note: PFC = prefrontal cortex; DLPFC = dorsolateral prefrontal cortex; ACC = anterior cingulate cortex; vACC = ventral ACC; VMPFC = ventromedial prefrontal cortex; OFC = orbitofrontal cortex.

In addition to the cortical regions (DLPFC, medial PFC) which are assumed to be directly modulated by our intervention, there are subcortical regions responsible for assigning salience to threatening stimuli, which are connected to the deficient cognitive control and motivation networks (i.e., lateral-medial PFC). Lateral and medial PFC regions are connected with subcortical areas like the insula and amygdala [66, 67] . Previous studies have shown that activation of the DLPFC with anodal tDCS modulates reactivity of the insula [62] and amygdala [63] to threat, regions that are hyperresponsive to threat and thus involved in the pathophysiology of SAD [21] . Interestingly, it has been shown that effective amygdala-prefrontal connectivity predicts successful emotion regulation [68] , the variable which significantly improved in both active tDCS groups and is linked to SAD core symptoms (fear, avoidance). Accordingly, we assume that our stimulation protocol affected also the emotional network, which is hyper-responsive to social situations in SAD, and linked with cognitive control and motivation networks [18] . It is important to consider here that involved networks in SAD, which also include the parietal-occipital networks, cannot be reduced to prefrontal-cingular network and thus other areas might be promising candidates for brain stimulation interventions in SAD.

A main objective of this study was to compare the efficacy of an intensified stimulation protocol (twice per day with 20 min interval, as compared to the more conventional once per day protocols), delivered with different intensities (1 vs 2 mA). This protocol has not been explored before in SAD and other anxiety disorders to the best of our knowledge. The rationale behind the . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint protocol comes from a study showing that twice stimulation with 20 min interval leads to longer aftereffects on cortical excitability compared to non-repeated stimulation or stimulation with long intervals, and resembles features of late-phase LTP, [43, 69] as well as a recent study demonstrating that 20 min of anodal stimulation-induced larger aftereffects for both 1 and 2 mA intensities, as compared to other durations (15 and 30 min) [70] . We chose specifically 1 and 2 mA stimulation intensities because results of previous studies show that the effects of stimulation on psychological processes are not in all cases larger with higher intensity [71, 72] . Our findings showed that overall, both interventions in this intensified protocol significantly reduced SAD symptoms, worries, depressive states and emotion regulation after intervention. However, the 2 mA stimulation condition led to significantly larger effects as compared to the 1 mA condition for some measures including depressive states and worries after intervention, and the avoidance symptom at both postintervention and 2-month follow-up measurements (Fig.3) . Moreover, attentional bias to the threat-related stimuli was significantly reduced only after 2 mA stimulation, while 1 mA protocol induced only a trend-wise change.

These results have at least three clinical implications. First, intensified stimulation (twice per day with a 20 min interval), has significant acute clinical efficacy for SAD symptoms and treatment-related variables. This is in line with physiological studies that have shown that repeated tDCS sessions induce larger increases of excitability, [73] as well as with the results of clinical studies in which larger efficacy with repeated sessions was reported [38] . Here, stimulation duration (e.g. 20 min) and the interval between stimulations (e.g. 20 min) are also important as shown in previous studies which were considered in our protocol [43, 69, 70] . Second, 2 mA stimulation is associated with higher clinical efficacy in SAD, and probably in other anxiety disorders. This is a novel finding with important clinical implications. Although some other studies . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint reported no outcome differences between 2 mA vs 1 mA tDCS in healthy individuals [70, 73] , our study showed a relatively larger clinical efficacy induced by 2 mA tDCS with this specific intensified protocol. Partially heterogeneous effects between studies might be based on different factors, including different neurotransmitter availability and electrode distance to the brain in respective cortical regions under study (e.g. motor vs prefrontal). Finally, it should be noted that because we could not secure follow-up measures on most outcome parameters, the long-term clinical efficacy of this intervention cannot be evaluated with certainty by this preliminary, but promising data.

It is also important to interpret the observed effects, which are remarkably large, with some precautions due to specific limitations. First, follow-up measures could not be secured for all measures (due to pandemic restrictions) except for the LSAS scores. Therefore, magnitude and duration of the treatment efficacy should be the topic of future replication studies. Small sample size is another detrimental factor in interpreting the results. Finally, neurophysiological and brain functional measures (e.g., fMRI, EEG, TMS-EEG) were not obtained in the present study, but would be valuable to provide a comprehensive picture of treatment efficacy, and clarify mechanisms. This is important, because all of the measures used in this study, except for the dotprobe task, were based on subjective self-reports scales. Nevertheless, a potential placebo effect is unlikely as the effects were not observed in the sham group.

Taken together, our findings suggest that the intensified prefrontal tDCS condition introduced for the treatment of SAD in the present study is promising. Both, primary SAD symptoms (fear, avoidance) and secondary treatment-related variables (worries, depressive state, emotion regulation, quality of life) improved after the intervention (compared to baseline performance). Selective attention to both threat-related and neutral stimuli was significantly . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint reduced after the 2 mA intervention as well. 2 mA tDCS, as compared to 1 mA tDCS, appears to induce larger effects on some clinical and behavioral performance measures. The observed effect sizes, however, are unusually large compared to previous tDCS works. Considering the relatively small sample size and variability in expression of anxiety, our results need to be replicated in larger clinical trials addressing the following limitations. First, our follow-up measures were compromised due to the current pandemic and we could only remotely measure SAD symptoms.

Long-term improvement of other measures including attentional bias to threat was not investigated. Furthermore, the intrinsically limited focality of tDCS can result in a relatively diffuse stimulation, and thus additional cortical and subcortical areas might also have been affected. Neuroimaging methods will help to identify the regions directly affected by tDCS more accurately in future studies. Future studies need to examine the magnitude and duration of effects in studies with longer follow-up assessments and added physiological measures.

MAN is a member of the Scientific Advisory Board of Neuroelectrics and NeuroDevice. All other authors declare no competing interests. 

The data that support the findings of this study are openly available at 10.17605/OSF.IO/2TKB3

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 11, 2021. ; https://doi.org/10.1101/2021.06.08.21258427 doi: medRxiv preprint

Social Anxiety Disorder

Social anxiety disorder

American Psychiatric Association. Diagnostic and statistical manual of mental disorders

American Psychiatric Pub

Having the heart to be evaluated": The differential effects of fears of positive and negative evaluation on emotional and cardiovascular responses to social threat

The Bivalent Fear of Evaluation Model of Social Anxiety: Further Integrating Findings on Fears of Positive and Negative Evaluation

The etiology of social anxiety disorder: An evidence-based model

Social anxiety disorder and quality of life: How fears of negative and positive evaluation relate to specific domains of life satisfaction

The Role of Emotion and Emotion Regulation in Social Anxiety Disorder

Social fears and social phobia in the USA: results from the National Comorbidity Survey Replication

A systematic review on the therapeutic effectiveness of non-invasive brain stimulation for the treatment of anxiety disorders

Transcranial Direct Current Stimulation (tDCS) in Anxiety Disorders

Anodal transcranial direct current stimulation over the ventromedial prefrontal cortex enhances fear extinction in healthy humans: A single blind sham-controlled study

A Systematic Review on the Effect of Transcranial Direct Current and Magnetic Stimulation on Fear Memory and Extinction

Neuroimaging in social anxiety disorder: A systematic review of the literature

The evidence-based pharmacotherapy of social anxiety disorder

Comprehensive Cognitive Behavioral Therapy, and Placeboin Generalized Social Phobia

Self-stigma and treatment effectiveness in patients with anxiety disorders-A mediation analysis

Social anxiety disorder: a critical overview of neurocognitive research

Neuroimaging in social anxiety disorder-A metaanalytic review resulting in a new neurofunctional model

Hot and cold executive functions in the brain: A prefrontal-cingular network

Neuroimaging Predictors and Mechanisms of Treatment Response in Social Anxiety Disorder: an Overview of the Amygdala

Patients with anxiety disorders rely on bilateral dlPFC activation during verbal working memory

Hyperfunction of left lateral prefrontal cortex and automatic thoughts in social anxiety disorder: A near-infrared spectroscopy study

Nuclear Neuroimaging in Social Anxiety Disorder: A Review

Reduced resting-state functional connectivity between amygdala and orbitofrontal cortex in social anxiety disorder

Aberrant amygdala-frontal cortex connectivity during perception of fearful faces and at rest in generalized social anxiety disorder

Frontal Cortex Stimulation Reduces Vigilance to Threat: Implications for the Treatment of Depression and Anxiety

Excitability changes induced in the human motor cortex by weak transcranial direct current stimulation

Shaping the Effects of Transcranial Direct Current Stimulation of the Human Motor Cortex

Studying and modifying brain function with non-invasive brain stimulation

Transcranial Direct Current Stimulation in Psychiatric Disorders: A Comprehensive Review

Interaction of the Left Dorsolateral Prefrontal Cortex (l-DLPFC) and Right Orbitofrontal Cortex (OFC) in Hot and Cold Executive Functions: Evidence from Transcranial Direct Current Stimulation (tDCS)

Differential role of prefrontal, temporal and parietal cortices in verbal and figural fluency: Implications for the supramodal contribution of executive functions

Neurocognitive correlates of self-esteem: From selfrelated attentional bias to involvement of the ventromedial prefrontal cortex

The role of dorsolateral and ventromedial prefrontal cortex in the processing of emotional dimensions

Meta-analysis of the effects of transcranial direct current stimulation on inhibitory control

Complexity of cathodal tDCS: Relevance of stimulation repetition, interval, and intensity

Transcranial Direct Current Stimulation in ADHD: A Systematic Review of Efficacy, Safety, and Protocolinduced Electrical Field Modeling Results

A Systematic Review and Meta-Analysis of the Effects of Transcranial Direct Current Stimulation (tDCS) Over the Dorsolateral Prefrontal Cortex in Healthy and Neuropsychiatric Samples: Influence of Stimulation Parameters

Impact of transcranial direct current stimulation on attentional bias for threat: a proof-of-concept study among individuals with social anxiety disorder

Repetitive transcranial magnetic stimulation (rTMS) to treat social anxiety disorder: case reports and a review of the literature. Clinical practice and epidemiology in mental health

rTMS to treat social anxiety disorder: a case report

Induction of Late LTP-Like Plasticity in the Human Motor Cortex by Repeated Non-Invasive Brain Stimulation

Psychometric properties of the Liebowitz Social Anxiety Scale

Development and validation of the penn state worry questionnaire

Multidimensional assessment of emotion regulation and dysregulation: Development, factor structure, and initial validation of the difficulties in emotion regulation scale

The World Health Organization's WHOQOL-BREF quality of life assessment: Psychometric properties and results of the international field trial. A Report from the WHOQOL Group

Normative values for the Beck Anxiety Inventory, Fear Questionnaire, Penn State Worry Questionnaire, and Social Phobia and Anxiety Inventory

Emotion Regulation Difficulties in Social Anxiety Disorder and Their Specific Contributions to Anxious Responding

Attentional bias in emotional disorders

Cognitive training for modifying interpretation and attention bias in depression: Relevance to mood improvement and implications for cognitive intervention in depression

What does the facial dot-probe task tell us about attentional processes in social anxiety? A systematic review

Warsaw set of emotional facial expression pictures: a validation study of facial display photographs

User's guide. Psychology Software Incorporated

Evaluation of Sham Transcranial Direct Current Stimulation for Randomized, Placebo-Controlled Clinical Trials

Transcranial DC stimulation (tDCS): A tool for double-blind sham-controlled clinical studies in brain stimulation

Realistic volumetric-approach to simulate transcranial electric stimulation-ROAST-a fully automated open-source pipeline

Trait anxiety and impoverished prefrontal control of attention

Prefrontal mediation of emotion regulation in social anxiety disorder during laughter perception

Repeated transcranial direct current stimulation of dorsolateral-prefrontal cortex improves executive functions, cognitive reappraisal emotion regulation, and control over emotional processing in borderline personality disorder: A randomized, sham-controlled, parallel-group study

Cognitive control dysfunction in emotion dysregulation and psychopathology of major depression (MD): Evidence from transcranial brain stimulation of the dorsolateral prefrontal cortex (DLPFC)

Repeated stimulation of the dorsolateral-prefrontal cortex improves executive dysfunctions and craving in drug addiction: A randomized, double-blind, parallel-group study

Effect of Prefrontal Cortex Stimulation on Regulation of Amygdala Response to Threat in Individuals With Trait Anxiety: A Randomized Clinical Trial

The Causal Role of the

Dorsolateral Prefrontal Cortex in the Modification of Attentional Bias: Evidence from Transcranial Direct Current Stimulation

Stimulation of Dorsolateral Prefrontal Cortex Enhances Adaptive Cognitive Control: A High-Definition Transcranial Direct Current Stimulation Study

Abnormal functional connectivity of the amygdala-based network in resting-state FMRI in adolescents with generalized anxiety disorder

A validated network of effective amygdala connectivity

Effective amygdala-prefrontal connectivity predicts individual differences in successful emotion regulation

Induction of long-term potentiation-like plasticity in the primary motor cortex with repeated anodal transcranial direct current stimulation -Better effects with intensified protocols?

Expanding the parameter space of anodal transcranial direct current stimulation of the primary motor cortex

Enhancing cognitive control training with transcranial direct current stimulation: a systematic parameter study

Is Motor Learning Mediated by tDCS Intensity?

The Effect of Transcranial Direct Current Stimulation (tDCS) Electrode Size and Current Intensity on Motor Cortical Excitability: Evidence From Single and Repeated Sessions

Note: tDCS = transcranial Direct Current Stimulation

SD = Standard Deviation

State Worry Questionnaire; BDI-II = Beck Depression Inventory-II

* = Values marked by (*) are given in seconds. ** = p values refer to baseline measurement comparisons