Acupuncture For Major Depressive Disorder: A Systematic Review and Meta-Analysis of High-Quality Randomized Control Trials

Background: The effects of acupuncture for major depressive disorder (MDD) uncertain. This review aims to determine the effects of Acupuncture for MDD . Methods: Eight Database were searched to identify randomized control trials (RCTs) on Acupuncture for MDD. All RCTs with adult participants undergoing acupuncture treatment for MDD were included. The primary outcome measure was the 24-item Hamilton rating scale for depression (HAMD-24). We used random-effects meta-analysis to synthesis the results with mean difference or odds ratio. Furthermore, the potential heterogeneity was tested through meta-regression/subgroup analyses/sensitive analysis. The quality of evidence for each outcome was assessed by the Grading of Recommendations Assessment, Development and Evaluation approach. Results: Forty-three studies were included: 9 acupuncture versus sham-acupuncture (n=920), 26 acupuncture versus antidepressants (n=2169), 9 acupuncture plus antidepressants versus antidepressants (n=667). Of the 43 high-quality articles, 24 and 8 were determined to have a low and moderate risk of bias, respectively. The pooled results for HAMD-24 and SDS revealed the clinical benets of Acupuncture or Acupuncture plus antidepressants compared to sham-acupuncture or antidepressants, with high quality evidence. Furthermore, high quality of evidence showed that acupuncture led to fewer adverse effects compared to antidepressants. Conclusions: Acupuncture or acupuncture plus antidepressants were signicantly associated with reduced HAMD-24 scores, with high-quality evidence. More rigorous trials are needed to identify the optimal frequency of Acupuncture for MDD and integrate such evidence into clinical care to reduce antidepressant use.


Background
Major depressive disorder (MDD) is a common psychological condition with an estimated lifetime prevalence of 16%, affecting more than 320 million people across the globe 1 . As of April 2017, depression has been recognized by the World Health Organization (WHO) as the leading cause of health-related disability, accounting for approximately 4.4% of all disabilities and premature deaths worldwide 2 .
Antidepressants such as selective serotonin reuptake inhibitors or selective serotonin-norepinephrine reuptake inhibitors are commonly used to treat MDD.
Most evidence-based guidelines recommend antidepressants to be the rst-line therapy because of their favorable outcomes and superior characteristics, including broad-spectrum effectiveness, safety, tolerability, simplicity of use, and low cost 3 . Nevertheless, pharmacological interventions also have undesirable side effects, including central nervous system and gastrointestinal disorders, weight gain, sexual dysfunction, and adverse emotional effects 4 .
In addition, long-term use may also cause drug tolerance, withdrawal symptoms when discontinued and increased suicidal ideation in certain patient populations 5 .
As an essential component of traditional Chinese medicine, acupuncture therapy has been practiced in China for thousands of years in disease prevention and treatment, functional improvement, longevity enhancement, and regulating emotions. As early as 1979, the WHO held a symposium on Acupuncture and created a list of 43 diseases suitable for Acupuncture 6 . In 2002, the WHO recommended Acupuncture as a treatment for depression (including depressive neurosis and depression following stroke) 7 . However, signi cant heterogeneity has been reported in ndings of several randomized control trials (RCTs) [8][9][10][11][12][13][14] .
Clinical trials have demonstrated that the effect of acupuncture therapy was partly induced via the autonomic nervous system [15][16][17] . In this regard, manual acupuncture leads to a wide range of central nervous system responses involving the amygdala, hippocampus, hypothalamus, cerebellum and other limbic structures, documented by functional magnetic resonance imaging and EEG 17 . Moreover, animal studies have shown that acupuncture therapy exerts its effects via multi-receptor and multi-pathway regulation related to amino acid metabolism and in ammatory pathways, especially the Toll-like receptor signaling pathway, tumor necrosis factor signaling pathway and the nuclear factor kappa-light-chain enhancer of activated B cells (NF-κB) signaling pathway 18 . Besides, acupuncture therapy has been reported to in uence the neurotransmitter levels of serotonin and noradrenaline and the adenylate cyclase cyclic adenosine monophosphate-protein kinase A cascade within the central nervous system via mechanisms similar to antidepressant medication 19 .
Although several systematic reviews and meta-analyses suggested that acupuncture therapy effectively treated MDD 20-26 , some review of these studies found some quality issues, and the original studies did not yield consistent results [20][21][22] . Therefore, we conducted this meta-analysis, including only high-quality RCTs, to determine the effectiveness and safety of acupuncture in treating MDD.

Study registering and reporting
This protocol was registered on INPLASY International Prospective Register of Systematic Reviews (INPLASY2021100073). It was conducted in accordance with the Measure Tool to Assess Systematic Reviews (AMSTAR 2) 23 , and the nal report was presented following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statements 2425 .

Selection criteria Study design
We included all RCTs with eligible intervention(s) and outcome(s) for MDD published in Chinese and English. For the crossover trials with the randomized control design, the rst stage consisted of data collection for this meta-analysis. Furthermore, we only included RCTs with a Jadad score ≥4 since the previously published systematic review could not obtain accurate conclusions due to the low quality of RCTs included.

Interventions/Comparators
The experimental group consisted of patients that received Acupuncture or Acupuncture plus antidepressants. The acupoint numbers, retaining time and frequency, treatment sessions were not limited.
The control group included non-acupuncture techniques, such as placebo control or other active therapies.

Outcome measures
The primary outcome was the Hamilton Depression Rating Scale for Depression (HAMD17/24). The secondary outcomes included changes in the Zung Self-rating Depression Scale (SDS) and the incidence of adverse effects.
The data at each time point was extracted from the original trials to be analyzed in this study.

Selection of studies
The process of data screening and study selection is displayed in Figure 1. First, studies were imported into Endnote X9 to remove duplicates, including duplications from different publications and multilingual publications and reports on different aspects of the same research. Two investigators (XGX and HBQ) independently assessed the titles and abstracts of the articles to select eligible studies based on the inclusion criteria. Subsequently, two other inves tigators (HLY and XQW) performed full-text assessments to exclude articles according to the exclusion criteria. Points of disagreement were resolved through consensus or by consulting a fth investigator (LFR).

Risk of bias and quality assessment
The risk of bias of eligible trials was measured by the Risk of Bias (ROB) Tool in Cochrane Handbook (5.1.0) 30 by two independent researchers (HFY and LYL). Each criterion was graded as "low" risk of bias, "high" risk of bias and "unclear" risk of bias. The methodology quality was assessed by two independent investigators (XQW and HBQ) with the Jadad scale. Any dissent occured in the assessment procedures was judged by a third investigator (SMS).

Data extraction
First, a standard table for electronic data extraction was established during a general meeting. After crosschecking for duplicates, the data were extracted by two authors (ZZ and TH) independently. The collected data included: basic information of studies (Study ID, the published date, author information, title, publication); characteristic of trials (study design, sample size, grouping method, blinding, assessment of outcomes, objectives of the study, etc.); the participants (age, gender, ethnicity, country, diagnosis, duration); method of intervention/control (number of treatment, frequency, duration of a session, additional treatment, etc.); outcome measurements (primary outcome, secondary outcome, assessment timeline, length of follow-up, etc.); results (mean, SD, adverse event, etc.). In addition, the extracted data was saved in an Excel format, and a third researcher (XQW) crosschecked the data input to ensure consistency and validity.

Uni cation of data and dealing with missing data
Before statistical analysis, the unit and time of extracted data were uni ed to ensure the accuracy of the data analysis. For studies with incomplete data, our researchers invested time and effort to contact the rst author or corresponding author to collect the incomplete data as far as possible. If the data was not obtained, a systematic review will be conducted for these data instead of statistical synthesis.

Data synthesis
Meta-analysis of RCTs with available data was performed by calculating the effect size and 95% CI using the random-effects model. Heterogeneity among trials was identi ed by the χ 2 test and reported as I 2 . Statistical analyses were performed with RevMan 5.2 and Stata 15.0. Two-sided P-value < 0.05 was considered statistically signi cant.
Studies were grouped according to the type of intervention (acupuncture, Acupuncture plus antidepressants) and the controls (sham-acupuncture, antidepressants, and sham-acupuncture plus antidepressants). For studies with more than 1 control group, such as Acupuncture versus sham acupuncture versus antidepressants, the results were split into pairwise comparisons by the different comparators.
Given the reported strong correlation between HAMD-17 and HAMD-24, the HAMD-17 scores were converted to the corresponding HAMD-24 scores.
Meta-regression and sensitivity analyses were conducted to explore potential sources of heterogeneity. Meta-regression was used to explore whether the age of patients, baseline HAMD-24, course of MDD, and acupuncture session affect the effectiveness of Acupuncture. Sensitivity analysis was used to identify studies that signi cantly affected the overall effect.
Publication bias was assessed by funnel plots for asymmetry when at least ten trials were included.
Evidence quality evaluation: The quality of evidence for the outcomes was assessed with the Grading of Recommendations Assessment, Development, and Evaluations (GRADE) system 31 and rated as high, moderate, low, and very low. A summary of ndings is presented in Table 1.

Description of included studies
The initial electronic search yielded 9943 unique records. 296 RCTs on Acupuncture for MDD were selected. Screening and full-text article analysis identi ed 43 high-quality RCTs with Jadad score ≥4, including 4037 patients (Figure 1, appendix 2) comparing acupuncture/acupuncture plus antidepressants versus sham-acupuncture/antidepressants. Appendix 3 shows the Jadad score in included trials (Jadad≥4), while Appendix 4 presents the Jadad score in excluded trials (Jadad<4). The sample size in the included trials ranged from 20 to 176, with a mean age of 41.46 (range 30-52.33) and included 1466 men. At the baseline, the mean HAMD-24 score was 32.27 (range 18.2-58.14). Eligible RCTs included one study from India 8 and forty-two from China 9-13 32-68 . Qualitative synthesis was conducted in two studies 10 43 due to lack of raw data, and quantitative synthesis in forty-one studies 8 Table 1). Analysis of follow-up outcomes showed that treatment with acupuncture plus antidepressants was more effective than with antidepressants alone, with moderate-quality evidence (Table 1 and Appendix 6).
During the meta-regression analysis of acupuncture versus sham-acupuncture, we found that the disease course of MDD could signi cantly reduce the heterogeneity (adjusted R2 = 51.87%, I2resid = 60.27%, P=0.039, Figure 6). Meta-regression for age (adjusted

SDS
Pooled analysis of four studies showed that Acupuncture (MD: -8.54 [95% CI: -13.01 to -4.06]; I 2 =81%) was more effective than sham acupuncture, with moderate quality of evidence (Table 1 and Appendix 11). Furthermore, high-quality evidence from 10 RCTs showed that Acupuncture and antidepressants were of equal effectiveness for MDD but was potentially in uenced by publication bias (Table 1, Appendix 12 and Appendix 13). Moreover, pooled results from 4 RCTs provided low-quality evidence that Acupuncture plus antidepressants were not as bene cial as antidepressants for MDD treatment (Table 1 and Appendix 14).
No studies with signi cant differences were found in the sensitivity analyses of Acupuncture versus sham acupuncture (Appendix 16).

Adverse Events
Pooled data from 9 studies provided high-quality evidence that acupuncture (OR: 0.27 [95% CI: 0.12 to 0.61]; I 2 =34%) led to fewer adverse events compared to antidepressants (Table 1 and Figure 9) and low-quality evidence that acupuncture was not statistically different from sham-acupuncture (Table 1 and Appendix 17). Furthermore, low-quality evidence showed that acupuncture plus antidepressants were not statistically different from antidepressants (Table 1 and Appendix 18).

Discussion
The systematic review included 43 high-quality RCTs involving 4037 MDD patients, while the meta-analysis included 41 RCTs with 3387 patients. The high quality evidence demonstrated an association between acupuncture with or without antidepressants and signi cant reduction in HAMD-24 scores.
Importantly, relatively few adverse events were associated with acupuncture, consistent with previous studies and reviews 8-11 45 . Our study provided the latest combined evidence for acupuncture alone or in combination with antidepressants in the treatment of MDD and identi ed research gaps that remain to be addressed.
Unlike previous systematic reviews, only high-quality RCTs were included in this study, ensuring that high-quality results were produced  . Furthermore, the interventions in this study were restricted to acupuncture or acupuncture plus antidepressants, which greatly reduced the impact of diversiform acupuncture therapies [71][72][73][74] . Accordingly, the present meta-analysis found that acupuncture and acupuncture plus antidepressants were associated with signi cantly reduced HAMD-24 scores than sham acupuncture or antidepressants, which was not documented in previous reviews 226970 , which could be accounted for by our strict inclusion of high-quality RCTs 8-1332-68 . In addition, more stringent inclusion criteria were used to ensure the quality of the source of the included randomized controlled trials.
Positive results from sham acupuncture RCTs suggested that acupuncture in combination with antidepressants had more bene cial effects than sham acupuncture plus antidepressants. Importantly, sham acupuncture helped to exclude the placebo effect of acupuncture 8 . Insu cient blinding in included studies also increased the risk of bias of this meta-analysis. However, in recent years, pragmatic unblinded trials have been recommended to obtain clinically relevant results since they emphasized practical applicability in the real world and extrapolation (increased external validity) rather than treatment effects 75 . This design of our study was well suited to complex and exible interventions despite the inability to blind acupuncturists [75][76][77] .
Interestingly, it has also been suggested that placebo could play a role in the effects of acupuncture 78-81 , and pragmatic nonblinded trials could provide more useful evidence for clinical guidelines of acupuncture 82 . However, a gap still exists between the ndings of RCTs of Acupuncture and our observations during clinical practice.
Clinically, acupuncture is often used to treat pain and plays an adjunctive treatment role in depression. The intended meaning is not clear. Did the author mean "The evidence from our study substantiated that acupuncture could be used as a major treatment approach for depression; however, the results of the heterogeneity analysis suggested that the results of data synthesis were not robust. Indeed, heterogeneity is an issue that cannot be avoided in metaanalysis 83 , and in the present study, greater heterogeneity was found after data pooling. Importantly, a meta-regression analysis found an association between baseline HAMD scores and acupuncture treatment effectiveness, consistent with some previous studies 1 70 .
Therefore, we do not believe that Acupuncture is suitable as the sole treatment for depression. However, the latest clinical evidence suggests that acupuncture has good clinical effectiveness in improving HAMD scores and improving 5-HT and GABA levels 32 41 48 68 . However, these studies had small sample sizes, emphasizing the need for high-quality studies with large samples to corroborate these ndings.

Limitations
There were several limitations to this study. Signi cant heterogeneity was found in the results of this study, which lowered the level of evidence, while the subgroup analysis did not reduce heterogeneity. There are many similar treatment modalities for Acupuncture, and this study was unable to determine whether Acupuncture was the best treatment modality. The number of acupuncture treatments varied considerably in the included studies, but no association with effectiveness was found after meta-regression analysis. Accordingly, it was not possible to clarify the optimal number of acupuncture sessions for MDD. Funnel plots were not feasible for most outcomes due to the limited number of trials included while assessing each outcome in the meta-analysis, explaining why publication bias could not be fully assessed. Only one of the studies included in this review was not done in China; accordingly, the fact that Chinese people tend to believe in the therapeutic effects of acupuncture may be another source of bias.

Conclusions
The ndings of this systematic review and meta-analysis provided high-quality evidence that acupuncture or acupuncture in combination with antidepressants signi cantly reduced HAMD-24 scores.

Patient and Public Involvement
No patient was involved.

Funding
This study has been supported by National Natural Science Foundation (No. 81590950, No.81590951), China. Funding sources had no role in the design of this study and will have no role during its execution, analysis, interpretation of data, or decision to present results.

Availability of data and materials
The ndings of this meta-analysis and systematic review will be published in peer-reviewed publications or conference presentations, and all of the data will be reported.

Competing interests
None.

Consent for publication
Not applicable.
Ethics approval and consent to participate Ethics approval and patient consent were not required since this is secondary research without patient involvement.  Figure 1 Flowchart of study selection Figure 2 Forest plot of HAMD-24 for Acupuncture versus sham-acupuncture Forest plot of HAMD-24 for Acupuncture versus antidepressants Funnel plot of HAMD-24 for Acupuncture versus antidepressants Forest plot of HAMD-24 for Acupuncture + antidepressants versus antidepressants Meta-regression with baseline disease course as a single continuous covariate