糖心vlog

Key Points

Question听 Which noninvasive brain stimulation (NIBS) is associated with the best efficacy and acceptability in reducing negative symptoms in schizophrenia?

Findings听 In this systematic review and network meta-analysis, the excitatory NIBS strategies (ie, high-definition transcranial random noise stimulation, intermittent theta-burst stimulation, anodal transcranial direct current stimulation, high-frequency repetitive transcranial magnetic stimulation, and extreme high-frequency repetitive transcranial magnetic stimulation) over the left dorsolateral prefrontal cortex with/without other inhibitory stimulation in the contralateral brain regions were associated with significantly larger reductions in negative symptoms than sham control. Acceptability did not significantly differ between groups.

Meaning听 Excitatory NIBS over the left dorsolateral prefrontal cortex was associated with significantly large improvements in negative symptoms.

Importance听 Negative symptoms have a detrimental impact on functional outcomes and quality of life in people with schizophrenia, and few therapeutic options are considered effective for this symptomatic dimension. Studies have suggested that noninvasive brain stimulation (NIBS) interventions may be effective in treating negative symptoms. However, the comparative efficacy of different NIBS protocols for relieving negative symptoms remains unclear.

Objective听 To compare the efficacy and acceptability of different NIBS interventions for treating negative symptoms.

Data Sources听 The ClinicalKey, Cochrane CENTRAL, Embase, ProQuest, PubMed, ScienceDirect, ClinicalTrials.gov, and Web of Science electronic databases were systematically searched from inception through December 7, 2021.

Study Selection听 A frequentist model network meta-analysis was conducted to assess the pooled findings of trials that evaluated the efficacy of repetitive transcranial magnetic stimulation (rTMS), theta-burst stimulation, transcranial random noise stimulation, transcutaneous vagus nerve stimulation, and transcranial direct current stimulation on negative symptoms in schizophrenia. Randomized clinical trials (RCTs) examining NIBS interventions for participants with schizophrenia were included.

Data Extraction and Synthesis听 The Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline was followed. Data were independently extracted by multiple observers. The pair-wise meta-analytic procedures were conducted using a random-effects model.

Main Outcomes and Measures听 The coprimary outcomes were changes in the severity of negative symptoms and acceptability (ie, dropout rates owing to any reason). Secondary outcomes were changes in positive and depressive symptoms.

Results听 Forty-eight RCTs involving 2211 participants (mean [range] age, 38.7 [24.0-57.0] years; mean [range] proportion of female patients, 30.6% [0%-70.0%]) were included. Compared with sham control interventions, excitatory NIBS strategies (standardized mean difference [SMD]: high-definition transcranial random noise stimulation,鈥夆垝2.19 [95% CI,鈥夆垝3.36 to 鈭�1.02]; intermittent theta-burst stimulation, 鈭�1.32 [95% CI,鈥夆垝1.88 to 鈭�0.76]; anodal transcranial direct current stimulation, 鈭�1.28 [95% CI,鈥夆垝2.55 to 鈭�0.02]; high-frequency rTMS, 鈭�0.43 [95% CI,鈥夆垝0.68 to 鈭�0.18]; extreme high-frequency rTMS, 鈭�0.45 [95% CI,鈥夆垝0.79 to 鈭�0.12]) over the left dorsolateral prefrontal cortex with or without other inhibitory stimulation protocols in the contralateral regions of the brain were associated with significantly larger reductions in negative symptoms. Acceptability did not significantly differ between the groups.

Conclusions and Relevance听 In this network meta-analysis, excitatory NIBS protocols over the left dorsolateral prefrontal cortex were associated with significantly large improvements in the severity of negative symptoms. Because relatively few studies were available for inclusion, additional well-designed, large-scale RCTs are warranted.

In the 2019 Global Burden of Disease Study, the number of schizophrenia cases was reported to have increased from 14.2 million in 1990 to 23.6 million in 2019. Furthermore, schizophrenia has contributed to 12.2% disability-adjusted life-years globally.¹ The main symptomatic dimensions of schizophrenia included positive, cognitive, and negative symptoms.² Although antipsychotic drugs are relatively effective in mitigating positive symptoms of schizophrenia, their efficacy is limited for the treatment of the negative symptoms of the illness.² Evidence indicates that the negative symptoms play a key detrimental role in overall disability and quality of life of people with schizophrenia.²

Negative symptoms are believed to have a greater negative impact on patients鈥� quality of life and social function than positive symptoms.³ Regardless of the potentially beneficial effect on negative symptoms by oral antipsychotics, those regimens were frequently associated with poor acceptability and safety profile (ie, high all-cause discontinuation and overall adverse events).⁴ Noninvasive brain stimulation (NIBS), such as transcranial direct current stimulation (tDCS) or repetitive transcranial magnetic stimulation (rTMS), exerts various effects on the brain. Different stimulation protocols appear to have distinct effects on cortical activity. For example, high-frequency rTMS (hf-rTMS), extreme hf-rTMS, and intermittent theta-burst stimulation (iTBS) increase cortical excitability, whereas low-frequency rTMS and continuous theta-burst stimulation decrease it.^5,6 Studies have suggested that NIBS helps manage neuropsychiatric diseases^7-11 or modulate cognitive function.¹² Although several meta-analyses have addressed the benefits of NIBS on positive symptoms of schizophrenia,^13,14 no conclusive evidence of their efficacy on negative symptoms has been obtained.

According to previous functional magnetic resonance imaging studies, patients with schizophrenia present abnormal functional connectivity between dorsolateral prefrontal cortex (DLPFC) and midbrain region,¹⁵ which is referred to as the mesocortical pathway. A dysfunctional mesocortical pathway is known to be related to the presence of negative symptoms.¹⁶ Therefore, treatment-induced changes in connectivity in these regions could theoretically improve negative symptoms in schizophrenia. Randomized clinical trials (RCTs) examining the efficacy of NIBS for treatment of negative symptoms have provided inconsistent findings. A 2020 RCT demonstrated that application of anodal tDCS (a-tDCS) over left DLPFC (F3 region) resulted in greater improvements in negative symptoms than sham control.¹⁷ Similar findings were observed in an RCT focusing on hf-rTMS (10 Hz) and demonstrated a reduction in negative symptoms in schizophrenia.¹⁸ However, another RCT of low-frequency (1 Hz) deep rTMS failed to demonstrate any beneficial effects.¹⁹ Several pairwise meta-analyses of different NIBS for relieving negative symptoms in schizophrenia have revealed that only tDCS^20,21 elicited a substantial decrease in these symptoms. However, other meta-analyses have challenged these findings.^22-24 Although some studies have produced valuable findings, several factors must still be considered. First, the RCTs have used a wide variety of study designs. Second, conclusive evidence for the efficacy of various NIBS modalities, such as low-frequency rTMS, hf-rTMS, tDCS, high-definition tDCS (hd-tDCS), transcranial random noise stimulation (tRNS), transcutaneous vagus nerve stimulation, and iTBS, are lacking. Finally, direct comparisons between interventions are required to evaluate their relative efficacy. Because of these major gaps in the literature, further research is necessary.

A network meta-analysis of RCTs may help to evaluate the comparative efficacy and acceptability of different NIBS to manage negative symptoms of schizophrenia. Furthermore, it may clarify the relative merits of multiple interventions that the aforementioned pairwise meta-analyses were unable to address.^25,26 To our knowledge, no network meta-analysis has been conducted on this topic. Therefore, we conducted this systematic review and network meta-analysis to compare the efficacy and acceptability of various NIBS protocols in negative symptoms management of schizophrenia.

This detailed description of this systematic review and network meta-analysis is available in the eMethods in the Supplement. We followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses () reporting guideline (eTable 1 in the Supplement)²⁷ and the flowcharts in accordance with the procedures of other network meta-analyses (eTable 2 in the Supplement).^{7,8,10,12,28-31} After registration in PROSPERO (), 2 independent authors (P.-T.T. and B.-S.Z.) searched and screened ClinicalKey, Cochrane CENTRAL, Embase, ProQuest, PubMed, ScienceDirect, Web of Science, and ClinicalTrials.gov to find RCTs comparing the effects of different NIBS methods on the severity of negative symptoms among participants with schizophrenia. Two independent authors (P.-T.T. and C.-M.-H.) evaluated risk of bias according to Cochrane risk of bias tool³² and extracted data of coprimary outcomes: (1) changes of negative symptoms after NIBS and (2) acceptability (ie, dropout). The dropout rate was defined when patients left the study for any reason before the study鈥檚 completion. We summarized the measurement with the standardized mean differences (SMDs) with 95% CIs for continuous outcome and odds ratios and 95% CIs for categorical data. The inclusion criteria included (1) RCTs, (2) application of NIBS interventions, (3) participants with schizophrenia, and (4) studies comparing the efficacy of different NIBS strategies to manage participants鈥� negative symptoms. The diagnosis of schizophrenia could be based on a well-established criterion or by physicians鈥� clinical diagnosis. To include more comprehensive studies, we did not set limitations to patients with predominant negative symptoms. The nomenclature (ie, node definition) of the NIBS was based on our previous 6 network meta-analyses on NIBS approaches for other conditions based on 10 to 20 electroencephalography mapping.^7-10,12,33 The network meta-analysis used mvmeta command in Stata version 16.0 (StataCorp).³⁴ All pairwise meta-analyses and network meta-analysis procedures were conducted using random-effects and frequentist models, respectively. This study used a mixed comparison with generalized linear mixed models to analyze direct and indirect comparisons in the network meta-analysis.³⁵ This study calculated the surface under the cumulative ranking curve (SUCRA) to rank the probabilities of the effects of all treatments on the target outcomes.³⁶ Heterogeneity among the included studies was evaluated using the tau value, whereas inconsistencies were evaluated by using the loop-specific approach, node-splitting, and design-by-treatment model.³⁷ We followed Cochrane Handbook for GRADE (Grading of Recommendations, Assessment, Development and Evaluations) ratings in the BMJ³⁸ and 1 important network meta-analysis in the Lancet³⁹ for quality assessment. We used funnel plots and Egger regression to evaluate potential publication bias. Finally, we assessed the reported efficacy of the different sham interventions (ie, changes in negative symptom severity) or waiting list controls to justify our assumption of transitivity. Furthermore, we conducted subgroup analyses focusing on RCTs with definite diagnostic criteria for schizophrenia.

A total of 101 articles were considered for a full-text review (Figure 1). Fifty-three articles were excluded for various reasons (Figure 1 and eTable 3 in the Supplement). Finally, 48 RCTs were included (eTable 4 in the Supplement).^{15,17,19,40-83}

A total of 2211 participants were included. The mean (range) age was 38.7 (24.0-57.0) years, and the mean (range) proportion of female participants was 30.6% (0%-70.0%). The mean (range) treatment duration was 2.8 (1-12) weeks. The mean (range) overall study duration (ie, treatment鈥�+鈥塮ollow-up duration) was 9 (1-32) weeks. The baseline characteristics of the included participants are listed in eTable 4 in the Supplement. All the included RCTs allowed concurrent treatment with antipsychotics during the study period.

Among the 48 RCTs, 37 recruited patients with schizophrenia and 11 recruited patients with schizophrenia or schizoaffective disorder were analyzed. Only 1 RCT⁸⁴ did not clearly mention the diagnostic criteria, whereas the others applied codes from the DSM-IV, DSM-IV-TR, and DSM-5, Chinese Classification of Mental Disorders, International Classification of Diseases, 9th Revision, and Clinical Modification, International Classification of Diseases, 10th Revision, Clinical Modification as the criteria.

The following NIBS were associated with significantly greater improvements in negative symptom severity than sham control: hd 2-mA anode tRNS at AF3 plus a cathode at AF4, F2, F6, and FC4 (hd-tRNS-AF3AF4F2F6FC4; SMD鈥�=鈥夆垝2.19 [95% CI,鈥夆垝3.36 to 鈭�1.02]; iTBS at left DLPFC: iTBS-F3: SMD鈥�=鈥夆垝1.32 [95% CI,鈥夆垝1.88 to 鈭�0.76]), 2-mA anode tDCS at F3Fp1 plus a cathode at F4Fp2 (a-tDCS-F3Fp1 + cathodal [c]-tDCS-F4Fp2; SMD鈥�=鈥夆垝1.29 [95% CI,鈥夆垝2.27 to 鈭�0.31]), 2-mA anode tDCS at F3 plus a cathode at Fp2 (a-tDCS-F3 + c-tDCS-Fp2; SMD鈥�=鈥夆垝1.28 [95% CI,鈥夆垝2.55 to 鈭�0.02]), 2-mA anode tDCS at F3 plus a cathode at left TPJ (a-tDCS-F3 + c-tDCS-TP3; SMD鈥�=鈥夆垝0.85 [95% CI,鈥夆垝1.56 to 鈭�0.13]), 20-Hz rTMS at left DLPFC (extreme hf-rTMS-F3; SMD鈥�=鈥夆垝0.45 [95% CI,鈥夆垝0.79 to 鈭�0.12]), and 10-Hz rTMS at left DLPFC (hf-rTMS-F3; SMD鈥�=鈥夆垝0.43 [95% CI,鈥夆垝0.68 to 鈭�0.18]; eTable 10A in the Supplement and Figure 2A). According to SUCRA, hd-tRNS-AF3AF4F2F6FC4 was ranked the highest probability of being the best, followed by iTBS-F3 and a-tDCS-F3Fp1 + c-tDCS-F4Fp2 (SUCRA value: 2.4, 12.4, and 15.7, respectively) (eTable 5A in the Supplement). eFigure 6A in the Supplement shows the heat map diagram representation of the improvement of negative symptoms.

The test of transitivity revealed that there was significant alleviation of negative symptom severity in both the rTMS鈥搕heta-burst stimulation and tDCS sham therapy groups (SMD鈥�=鈥�0.468 [95% CI鈥�=鈥�0.313-0.622]; P鈥�<鈥�.001 and SMD鈥�=鈥�0.251 [95% CI,鈥�0.022-0.480]; P鈥�=鈥�.03, respectively). Furthermore, no significant between-group differences were detected between the rTMS鈥搕heta-burst stimulation and tDCS sham therapy groups (rTMS sham control: SMD鈥�=鈥�0.47 [95% CI, 0.31-0.62] vs tDCS sham control: SMD鈥�=鈥�0.25 [95% CI, 0.02-0.48]; P鈥�=鈥�.12; eFigure 1 in the Supplement).

The subgroup analysis focusing on RCTs with definite diagnostic criteria revealed similar findings: hd-tRNS-AF3AF4F2F6FC4 (SMD鈥�=鈥夆垝2.19 [95% CI,鈥夆垝3.36 to 鈭�1.02), iTBS-F3 (SMD鈥�=鈥夆垝1.32 [95% CI,鈥夆垝1.88 to 鈭�0.76]), a-tDCS-F3Fp1 + c-tDCS-F4Fp2 (SMD鈥�=鈥夆垝1.29 [95% CI,鈥夆垝2.27 to 鈭�0.31]), a-tDCS-F3 + c-tDCS-Fp2 (SMD鈥�=鈥夆垝1.28 [95% CI,鈥夆垝2.55 to 鈭�0.02]), a-tDCS-F3 + c-tDCS-TP3 (SMD鈥�=鈥夆垝0.85 [95% CI,鈥夆垝1.56 to 鈭�0.13]), extreme hf-rTMS-F3 (SMD鈥�=鈥夆垝0.45 [95% CI,鈥夆垝0.79 to 鈭�0.12]), and hf-rTMS-F3 (SMD鈥�=鈥夆垝0.43 [95% CI,鈥夆垝0.68 to 鈭�0.18]) were associated with significantly greater alleviation of negative symptom severity than sham control (eTable 6A, eFigure 2A, and eFigure 3A in the Supplement). According to the SUCRA, hd-tRNS-AF3AF4F2F6FC4 was ranked the highest probability of being the best, followed by iTBS-F3 and a-tDCS-F3Fp1 + c-tDCS-F4Fp2 (SUCRA鈥�=鈥�3.0, 12.4, and 16.2, respectively) (eTable 5B in the Supplement).

None of the investigated NIBS modalities were associated with significantly different acceptability rates relative to sham control groups (eTable 5C and eTable 10B in the Supplement, Figure 3B, and Figure 2B). eFigure 6B in the Supplement represent the heat map diagram representation of the tolerability in aspect of dropout rate.

None of the investigated NIBS approaches was associated with significantly different changes in positive symptom severity compared with sham control groups, except for iTBS-F3. iTBS-F3 (SMD鈥�=鈥�1.06 [95% CI,鈥�0.32-1.80]) was associated with significantly worse positive symptom severity than sham control (eTable 6B, eFigure 2B, and eFigure 3B in the Supplement). According to the SUCRA, 6-Hz priming rTMS at PT3 plus 1-Hz rTMS at PT3 (priming [pr] TMS-PT3 + low-frequency [lf]鈥搑TMS-PT3) (SUCRA value, 22.8) was associated with the most improvement but did not achieve statistical significance (SMD鈥�=鈥夆垝0.75 [95% CI,鈥夆垝2.38 to 0.88]; eTable 5D in the Supplement).

Only a-tDCS-F3Fp1 + c-tDCS-F4Fp2 (SMD鈥�=鈥夆垝0.79 [95% CI,鈥夆垝1.43 to 鈭�0.15]) was associated with significant alleviation of depressive symptom severity compared with sham control. By contrast, the use of 10-Hz rTMS at the left prefrontal cortex and right prefrontal cortex (hf-rTMS-F3F4; SMD鈥�=鈥�0.73 [95% CI,鈥�0.10-1.36]) was associated with significantly worse depressive symptoms (eTable 6C, eFigure 2C, and eFigure 3C in the Supplement). According to the SUCRA, a-tDCS-F3 + c-tDCS-Fp2 was associated with the most alleviation but did not achieve statistical significance (SMD鈥�=鈥夆垝0.93 [95% CI,鈥夆垝1.94 to 0.07]), followed by a-tDCS-F3Fp1 + c-tDCS-F4Fp2 (SUCRA value: 10.1 and 10.1, respectively) (eTable 5E in the Supplement).

Of the included studies, 74.4% (250 of 336), 25.0% (84 of 336), and 0.6% (2 of 336 items) had overall low, unclear, and high risks of bias, respectively. The unclear reporting of allocation concealment contributed to bias risk (eFigure 4A and B in the Supplement). Funnel plots of publication bias across the included studies (eFigure 5A-H in the Supplement) revealed a general symmetry. No significant publication bias was detected among the articles included in network meta-analysis by using the Egger test. The network meta-analysis did not demonstrate inconsistency in terms of either local inconsistency (assessed using the loop-specific approach and node splitting) or global inconsistency (determined using the design-by-treatment method; eTable 7 in the Supplement). The heterogeneity test revealed some significant heterogeneity noted in some treatment comparison (eTable 8A and B in the Supplement). The GRADE ratings revealed that the quality of evidence in the network meta-analysis ranged from low to medium (eTable 9A and B in the Supplement).

To our knowledge, this is the first network meta-analysis to directly investigate the potential benefits of NIBS for treating negative symptom severity in patients with schizophrenia. This study demonstrated that, compared with sham controls, excitatory stimulations (ie, tRNS, iTBS, a-tDCS, hf-rTMS, and extreme hf-rTMS) over left DLPFC (F3 region) with/without other inhibitory stimulation protocols in the contralateral brain regions were associated with the highest probability of being the best reductions in negative symptom severity. Of these, a-tDCS-F3Fp1 + c-tDCS-F4Fp2 was also associated with significant improvements in depressive symptom severity compared with sham controls. None of the investigated NIBS were associated with significantly different dropout rates compared with sham controls.

The key finding of this network meta-analysis was that the excitatory NIBS applied over left DLPFC (F3 region) were associated with significant alleviation of negative symptoms compared with sham controls. A functional magnetic resonance imaging study demonstrated abnormal functional connectivity of the DLPFC to the ventral tegmental area,¹⁵ which is the origin of mesocorticolimbic dopamine projections. Previous studies have indicated significant association between this abnormal functional connectivity and abnormal working memory performance⁸⁵ and anticipated reward (wanting) function.⁸⁶ Among these deficits, anticipated reward is considered to be linked to the negative symptoms of schizophrenia.⁸⁷ Therefore, the restoration of this abnormal connectivity has become a rationale for negative symptom management. A recent seed-based functional magnetic resonance imaging in treatment-resistant schizophrenia demonstrated that iTBS over left DLPFC could both decrease patients鈥� negative symptom severity and increase functional connectivity between left DLPFC and brain regions with dopamine neuron cell bodies.⁴⁰ Therefore, excitatory stimulation over left DLPFC would be a hypothetically reasonable strategy. This hypothesis is supported by the main finding of the present study, in which the most excitatory NIBS (tRNS, iTBS, a-tDCS, hf-rTMS, and extreme hf-rTMS) over F3 region significantly alleviated negative symptom severity in schizophrenia.

Of the excitatory NIBS, tRNS (hd-tRNS-AF3AF4F2F6FC4) was associated with the greatest alleviation of negative symptoms. This is a newly developed NIBS technique in which the current is delivered at randomly alternating intensities and frequencies and is believed to induce random neural activity, resulting in neural noise effects.⁴¹ The theory of neural noise effects⁸⁸ explains the contradictory mechanism of transcranial electrical stimulations.^89,90 The hypothesis of neural noise proposes that transcranial electrical stimulations, either through depolarization or hyperpolarization in different cortices, would improve cognitive performance.¹² The application of tRNS has been revealed to be effective in different cortical dysfunction diseases, such as tinnitus with unknown origin.¹⁰ In a head-to-head trial, tRNS was revealed to be superior to tDCS at modulating neural activity after a single stimulation session.⁹¹ This superiority was also identified in the present network meta-analysis, in which tRNS stimulation was ranked superior to tDCS strategies in relieving negative symptom severity. However, only 1 RCT in this study, with a relatively short duration (a total of 5 weeks with a study duration of 1 week), reported the effect of tRNS on schizophrenia and schizoaffective disorder.⁴¹ Therefore, clinicians should pay particular attention when applying these findings in their clinical practice.

In addition to the aforementioned neural noise hypothesis, bimodal tDCS (a-tDCS-F3Fp1 + c-tDCS-F4Fp2) demonstrated favorable results. Specifically, a-tDCS-F3Fp1 + c-tDCS-F4Fp2 was beneficial to both negative and depressive symptoms in schizophrenia. This result was derived from 1 pioneer RCT,⁴² in which the authors applied a bianodal-bicathodal tDCS mode to successfully reduce patients鈥� negative and depressive symptoms. This finding might be explained by the 2-fold accumulated electrical dosage, synergistic effect, and deeper brain stimulation.⁴² However, because this result is derived from a pioneer RCT, the underlying mechanism beyond the bimodal tDCS remains unclear.

This study has some limitations. First, this network meta-analysis may have been underpowered because of the heterogeneity of participants (eg, comorbidities, concomitant antipsychotics, baseline negative symptom severity, instruments used in each study, timing of NIBS intervention, and follow-up duration). All included studies were relatively small trials, presenting small samples and heterogeneous technical procedures. Second, although most RCTs included a sham control in their study design, the blindness of the RCTs may not have been complete because of the limitations of the commercial instruments used. Third, significant alleviation of negative symptom severity was demonstrated in both the rTMS-TBS and tDCS sham therapy groups (P鈥�<鈥�.001 and P鈥�=鈥�.03, respectively). Furthermore, no significant between-group difference was detected between these 2 groups (P鈥�=鈥�.12), which might indicate fair transitivity in this network meta-analysis. The significant placebo effect in the rTMS-TBS and tDCS sham therapy groups might be caused by (1) direct placebo effect of sham therapy and (2) potential therapeutic effect of concurrent antipsychotics on negative symptoms. However, we cannot completely exclude the potential therapeutic effect of the concurrent antipsychotics because no RCTs excluded antipsychotics from their study design. Fourth, this network meta-analysis only investigated dropout rate as an indicator of safety. Fifth, the study was designed to assess changes in negative symptom severity as a primary outcome. Therefore, we might have missed RCTs that mainly reported changes in positive or depressive symptoms and other outcomes in schizophrenia. Sixth, a new hypothesis has been proposed that potential beneficial effects of different NIBS would result from their neuroprotection, antiapoptosis, neurogenesis, angiogenesis, or neuroplasticity effects.⁹² However, we could not further investigate whether the negative symptoms were alleviated because of the neurogenesis effect of NIBS because of limited information. Seventh, our study was limited by the number of RCTs available for inclusion and the absence of long-term studies. In addition, because of the limited numbers of RCTs available, the overall network structures were relatively weak (ie, star-shaped network). Therefore, the interpretation of the SUCRA results should be more conservative. This weak network structure would also limit the application of inconsistency evaluation. To be specific, the inconsistency model would be less suitable for those network with star-shaped network structure. Besides, the heterogeneity test revealed some significant heterogeneity noted in some treatment comparisons. Therefore, clinicians should pay special attention when applying the results of our study in clinical practice.

This network meta-analysis revealed that excitatory stimulation over left DLPFC with/without other inhibitory stimulation protocols in contralateral brain regions is associated with alleviation of negative symptom severity. None of the investigated NIBS were associated with significantly different dropout rates compared with sham controls. In addition, this study noted significant alleviation of negative symptom severity in both the rTMS-TBS and tDCS sham therapy groups. However, because the NIBS had not been approved to have indication of negative symptoms management in schizophrenia, the results of the current network meta-analysis should be interpreted and applied in a more preserved way. Our findings might serve as a starting point for future large-scale RCTs with longer follow-up periods and sham control to investigate the association between NIBS and negative symptoms in schizophrenia.

Accepted for Publication: April 24, 2022.

Published Online: June 22, 2022. doi:10.1001/jamapsychiatry.2022.1513

Corresponding Authors: Tiao-Lai Huang, MD, Department of Psychiatry, Kaohsiung Chang Gung Memorial Hospital and Chang Gung University College of Medicine, Kaohsiung, Taiwan, No. 123, Dapi Road, Niaosong District, Kaohsiung City, 833 Taiwan (a540520@adm.cgmh.org.tw); Cheng-Ta Li, MD, PhD, Division of Community and Rehabilitation Psychiatry, Department of Psychiatry, Taipei Veterans General Hospital, Functional Neuroimaging and Brain Stimulation Lab, Taipei Veterans General Hospital, Taiwan, No. 201, Section 2, Shipai Road, Beitou District, Taipei City 11267, Taiwan (on5083@msn.com).

Author Contributions: Drs Tseng and Zeng had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Tseng, Zeng, and Hung contributed equally as first authors.

Concept and design: Tseng, Zeng, Hung, Stubbs, Brunoni, Su, T. Chen, D. Li, Lin, C. Wu, Shiue, C. Li.

Acquisition, analysis, or interpretation of data: Zeng, Hung, Liang, Stubbs, Carvalho, Brunoni, Tu, Y. Wu, Hsu, Y. Chen, Suen, Satogami, Takahashi, Yang, Huang, C. Li.

Drafting of the manuscript: Tseng, Zeng, Hung, Stubbs, Brunoni.

Critical revision of the manuscript for important intellectual content: Liang, Stubbs, Carvalho, Su, Tu, Y. Wu, T. Chen, D. Li, Lin, Hsu, Y. Chen, Suen, Satogami, Takahashi, C. Wu, Yang, Shiue, Huang, C. Li.

Statistical analysis: Tseng, Hung, Liang, Stubbs, Carvalho, Brunoni, Tu, Y. Wu, Lin, Y. Chen, C. Wu, Huang.

Administrative, technical, or material support: Tseng, Zeng, Hung, Stubbs, T. Chen, Hsu, Y. Chen, Suen, C. Wu, Shiue, Huang, C. Li.

Supervision: Stubbs, Brunoni, Su, Tu, T. Chen, Y. Chen, Suen, Satogami, Takahashi, Yang, Shiue, Huang, C. Li.

Conflict of Interest Disclosures: Dr Stubbs is supported by a clinical lectureship jointly funded by Health Education England and the National Institute for Health Research (NIHR); is part funded by the NIHR Biomedical Research Centre at South London and Maudsley NHS Foundation Trust; and is supported by the Maudsley Charity, King鈥檚 College London, and the NIHR South London Collaboration for Leadership in Applied Health Research and Care funding. Dr Brunoni reported grants from S茫o Paulo Research Foundation, Brazilian National Council of Scientific Development, and University of Sao Paulo Medical School; support from the Newton Advanced Fellowship; and in-kind support from Flow Neuroscience and MagVenture during the conduct of the study. Dr Su is supported by grants from An Nan Hospital, China Medical University, Tainan, Taiwan and China Medical University, Taichung, Taiwan. Dr Tu was supported by a grant from the Ministry of Science and Technology in Taiwan. Dr Lin is supported by grants from the Ministry of Science and Technology in Taiwan and Kaohsiung Chang Gung Memorial Hospital in Taiwan. Dr Hsu is supported by grants from the Ministry of Science and Technology in Taiwan. Dr Takahashi reported speaker鈥檚 honoraria from Teijin Pharma Ltd during the conduct of the study and from Dainippon Sumitomo Pharma, Eisai, Meiji Seika, Mochida Pharmaceutical, Ono Pharmaceutical, Otsuka Pharmaceutical, and Takeda Pharmaceutical outside the submitted work. No other disclosures were reported.

Disclaimer: This article presents independent research. The views expressed in this publication are those of the authors and not necessarily those of the acknowledged institutions.

Additional Contributions: This manuscript was edited by Wallace Academic Editing.