Comparison of Efficacy of Acupuncture-Related Therapy in the Treatment of Rheumatoid Arthritis: A Network Meta-Analysis of Randomized Controlled Trials

Background The refractory, repetitive, and disabling characteristic of rheumatoid arthritis (RA) has seriously influenced the patients’ quality of life, and makes it a major public health problem. As a classic complementary and alternative therapy, acupuncture is usually applied for RA combined with disease-modifying anti-rheumatic drugs (DMARDs). However, there are various types of acupuncture, and the curative effects are different in different acupuncture therapies. In this study, we evaluated the clinical efficacy of different acupuncture therapies combined with DMARDs in the treatment of RA. Methods The randomized controlled trials (RCTs) of acupuncture combined with DMARDs in the treatment of RA were searched in both English and Chinese database of PubMed, Cochrane Library, EMBASE, Web of Science, CNKI, VIP database, Wanfang, and SinoMED, up to October 2021. Literature screening, data extraction, and evaluation of the risk of bias were carried out independently by two researchers, and the data were analyzed by Stata14.2 and GeMTC 0.14.3 software. Results A total of 32 RCTs were included, including 2,115 RA patients. The results of network meta-analysis were as follows: in terms of improving DAS28 score, Electro-acupuncture + DMARDs has the best efficacy. In terms of improving VAS score, Fire Needle + DMARDs showed the best efficacy. In terms of improving morning stiffness time, acupuncture-related therapies combined with DMARDs were not better than DMARDs alone in improving morning stiffness time in RA patients. In terms of reducing CRP and ESR, Fire Needle + DMARDs showed the best efficacy. In terms of reducing RF, Moxibustion + DMARDs has the best efficacy. Conclusions The comprehensive comparison of the outcome indicators in 8 different treatments indicates that electro-acupuncture combined with DMARDs is the best combined therapy in improving DAS28 score, while in terms of improving pain and serological markers, fire needle combined with DMARDs and moxibustion combined with DMARDs were the best combined therapies. However, it is impossible to find out which is better between fire needle and moxibustion due to the limited studies. Clinically, appropriate treatment should be selected according to the actual situation. Systematic Review Registration https://www.crd.york.ac.uk/prospero/#recordDetails, CRD42021278233.


Exclusion Criteria
a. Patients with other rheumatic immune diseases; b. Without clear diagnosis; c. Without outcome indicators; d. With more than two TCM therapies, such as cupping, traditional Chinese herbs, or a combination of two or more acupuncture therapies, such as acupuncture combined with moxibustion and electro-acupuncture combined with moxibustion; e. Repetitively published studies; f. Without complete data in the study even after contacting the authors.

Literature Search Strategy
RCTs of acupuncture combined with DMARDs in the treatment of RA in PubMed, EMBASE, Web of Science, Cochrane Library, China Knowledge Network (CNKI), WanFang, VIP Database, and SinoMed were searched. The search terms were acupuncture, electro-acupuncture, warm needle, fire needle, blood-letting therapy, moxibustion, acupoint catgut embedding, acupoint injection, rheumatoid arthritis, and RA in both Chinese and English. The PubMed database search strategy was shown in Table 1.

Literature Screening and Data Extraction
Exclusion of duplicate literature was performed in EndNote X9 software, and then preliminary screening was performed by reading the title and abstract. After that, the full text was further screened to exclude the literature that did not meet the inclusion criteria. For data extraction, two researchers (RW and YF) separately conducted data extraction based on the inclusion and exclusion criteria. If there was any disagreement, the third researcher (LZ) would make a final decision. The data extraction content included title, author, publication year and month, sample size, diagnostic criteria, interventions of treatment group and control group, dosage, course of treatment, and outcome indicators, among others.

Evaluation of the Risk of Bias
The quality evaluation was performed by two separate researchers (RW and YF) using RCT Bias Risk Assessment Tool of the Cochrane System Review Manual Version 5.1.0, and the third researcher (LZ) would assist in judging the divergence between the two researchers. Evaluation items included random sequence generation, allocation concealment, blinding of patients and investigators, blinding of outcome evaluators, incomplete result data, selective reporting, and other biases.

Statistical Analysis
A directly compared meta-analysis was performed using Stata14.2 software. For continuous variables, mean difference (WMD) or standard mean difference (SMD) was used for analysis. c 2 test was used to analyze the heterogeneity among the included study results, and I 2 was used to quantitatively judge the heterogeneity. If p ≥ 0.10, I 2 < 50%, there was no significant heterogeneity between studies, and meta-analysis was performed using a fixed effect model. If p < 0.10, I 2 ≥ 50%, the heterogeneity between studies was considered significant, and a random effect model was used for meta-analysis. Stata14.2 software was used to make an evidence network diagram to show the comparative relationship in interventions for each outcome indicator. The small sample effects or publication bias was detected by comparison-correction funnel plots. At the same time, network meta-analysis was conducted by GeMTC 0.14.3 software based on the Markov Chain Monte Carlo (MCMC) consistent model under the Bayesian framework. Four chains were used for simulation, with the number of iterations set as 50,000 times (the first 20,000 times for annealing, and the last 30,000 times for sampling), and were estimated and inferred under the assumption that MCMC reached a stable state of convergence evaluated by Potential Scale Reduction Factor (PSRF). The stability and consistency of the results were evaluated using the MCMC inconsistent fitting model.

Results of the Risk of Bias
For random sequence generation, random number tables were used in 12 studies (22, 26-28, 31, 32, 35, 36, 45, 49, 50, 52), random numbers were generated by computer in 6 studies (21,23,29,37,42,48), and the remaining 14 studies (24, 25, 30, 33,    (32) used central allocation, one study (29) used lottery, and the remaining 24 studies did not report allocation concealment. For blinding of investigators and participants, due to the limitation of interventions, double-blind was not applied in all studies. For blinding of the outcome assessors, the outcome assessors were blinded in four studies (21,23,35,42), and the remaining 28 studies did not report blinding of the outcome evaluators. For incomplete reporting, selective reporting, and other biases, all the 32 studies (21-52) reported complete data, without selective reporting and others bias. The results of the risk of bias evaluation are shown in Figure 2.

DSA28 Scores
The results of meta-analysis showed that the DSA28 scores of the Moxibustion + DMARDs group and Acupuncture + DMARDs group were lower than that of the DMARDs group (p < 0.05). The DSA28 scores of the Warm Needle + DMARDs group and Acupoint catgut embedding + DMARDs group had no difference compared with that of the DMARDs group (p > 0.05).
Descriptive analysis results showed that the DSA28 scores of the Electro-acupuncture + DMARDs group and Auricular Needle + DMARDs group were lower than that of the DMARDs group (p < 0.05). The DSA28 scores of the Electroacupuncture + DMARDs group were lower than that of the Acupuncture + DMARDs group (p < 0.05). See supplementary materials (Table S2).

VAS Scores
The results of meta-analysis showed that the VAS scores of the Moxibustion + DMARDs group, Acupuncture + DMARDs group, and Electro-acupuncture + DMARDs group were lower than that of the DMARDs group (p < 0.05). Descriptive analysis results showed that the VAS scores of the Warm Needle + DMARDs group and the Fire Needle + DMARDs group were lower than that of the DMARDs group (p < 0.05). See supplementary materials (Table S2).

Morning Stiffness Time
The results of meta-analysis showed that morning stiffness time in the Acupuncture + DMARDs group and Warm Needle + DMARDs group was lower than that of the DMARDs group (p < 0.05). The morning stiffness time in the Moxibustion + DMARDs group was not different from that of the DMARDs group (p > 0.05). Descriptive analysis results showed that the morning stiffness time of the Electro-acupuncture + DMARDs group and Auricular needle + DMARDs group was lower than that of the DMARDs group (p < 0.05). The morning stiffness time of the Electro-acupuncture + DMARDs group was lower than that of the Acupuncture + DMARDs group (p < 0.05). See supplementary materials (Table S2).

CRP
The results of meta-analysis showed that the CRP of the Moxibustion + DMARDs group, Acupuncture + DMARDs group, Electro-acupuncture + DMARDs group, and Warm Needle + DMARDs group were lower than that of the DMARDs group. The CRP of the Fire Needle + DMARDs group was lower than that of the Acupuncture + DMARDs group (p < 0.05).
Descriptive analysis results showed that the CRP of the Auricular needle + DMARDs group was lower than that of the DMARDs group. The CRP of the Electro-acupuncture + DMARDs group was lower than that of the Acupuncture + DMARDs group (p < 0.05). See supplementary materials (Table S2).

ESR
The results of meta-analysis showed that the ESR of the Moxibustion + DMARDs group, Acupuncture + DMARDs group, Electro-acupuncture + DMARDs group, and Warm Needle + DMARDs group was lower than that of the DMARDs group. The ESR of the Fire Needle + DMARDs group was lower than that of the Acupuncture + DMARDs group (p < 0.05). Descriptive analysis results showed that the ESR of the Auricular needle + DMARDs group was lower than that of the DMARDs group. The ESR of the Electro-acupuncture + DMARDs group was lower than that of the Acupuncture + DMARDs group (p < 0.05). See supplementary materials (Table S2).

RF
The results of meta-analysis showed that the RF of the Moxibustion + DMARDs group, Acupuncture + DMARDs group, Electro-acupuncture + DMARDs group, and Warm Needle + DMARDs group was lower than that of the DMARDs group (p < 0.05). There was no difference in RF in the Fire needle + DMARDs group as compared to the Acupuncture + DMARDs group (p > 0.05). See supplementary materials (Table S2).

Subgroup Analysis
To further explore the effect of different treatment duration on the results, we conducted subgroup analysis. Due to the small

Heterogeneity Analysis
In the directly compared meta-analysis, some results were heterogeneous. Through the analysis of the original data, it was found that there may be methodological heterogeneity due to less description of the blind method and allocation concealment in the included studies. At the same time, the clinical heterogeneity may be caused by factors such as the inclusion population, acupoints, and operation methods. However, due to the lack of specific description of these details in the original study and the small number of studies in some results, it was impossible to further explore the source of heterogeneity by subgroup analysis.
Although we have carried out subgroup analysis on some results, all heterogeneity has not been eliminated. However, we did a sensitivity analysis and found that the results were stable after we excluded either study. Therefore, we can ignore this heterogeneity and adopt a random effect model for meta-analysis.    namely, DMARDs and Acupuncture + DMARDs and Electroacupuncture + DMARDs. The thicker the line between the two interventions was, the greater the number of studies between the two measures was. The larger the node was, the larger the research sample size was. Except for DMARDs, Acupuncture + DMARDs, and Electro-acupuncture + DMARDs, there was no closed loop among the other interventions, which indicated that there was no direct comparison between those interventions, as shown in Figures 3-8.

Results of DSA28 Scores
Fourteen studies reported DAS28 (21-24, 27, 31, 32, 34, 36-38, 42, 46, 52). The convergence evaluation results showed that the PSRF value was close to 1, and the inconsistent fit model result was similar to the consistent fit model (p = 0.407 > 0.05), indicating fine stability and consistency of results, so the MCMC fitting consistency model was used for analysis. The results showed that Moxibustion + DMARDs was better than DMARDs, and Electro-acupuncture + DMARDs was better than Moxibustion + DMARDs, DMARDs, and Acupoint catgut embedding + DMARDs. There was no statistically significant difference between the other therapies, as seen in Table 4. The probability ranking results of improving DSA28 score were as follows: Electro-acupuncture + DMARDs > Auricular needle + DMARDs > Acupuncture + DMARDs > Moxibustion + DMARDs > Warm needle + DMARDs > DMARDs > Acupoint catgut embedding + DMARDs, as shown in Table 5.

Results of VAS Scores
Twenty-three studies reported VAS (21-23, 26, 28-30, 32, 33, 35-37, 40-42, 44-50, 52). The results of convergence evaluation showed that the PSRF value was close to 1, and the inconsistent fit model result was similar to the consistent fit model (p = 0.592 > 0.05), indicating fine stability and consistency of results, so the MCMC fitting consistency model was used for analysis. The results showed that Moxibustion + DMARDs was better than DMARDs, Acupuncture + DMARDs was better than DMARDs, Electroacupuncture + DMARDs was better than DMARDs, and Fire needle + DMARDs was better than DMARDs, Moxibustion + DMARDs, and Acupuncture + DMARDs. There was no statistically significant difference between the other therapies, as seen in Table 6.

Results of Morning Stiffness Time
Fourteen studies reported morning stiffness time (21-24, 26-28, 31, 37, 42, 43, 46, 49, 52). The results of convergence evaluation showed that the PSRF value was close to 1, and the inconsistent fit model result was similar to the consistent fit model (p = 0.843 > 0.05), indicating fine stability and consistency of results, so the MCMC fitting consistency model was used for analysis. The results showed that there was no statistically significant difference between the therapies ( Table 8), indicating that the combined therapies were not better than DMARDs in improving morning stiffness time. The probability ranking results are shown in Table 9.
The results of convergence evaluation showed that the PSRF value was close to 1, and the inconsistent fit model result was similar to the consistent fit model (p = 0.052 > 0.05), indicating fine stability and consistency of results, so the MCMC fitting consistency model was used for analysis. The results showed that Moxibustion + DMARDs was better than DMARDs, Acupuncture + DMARDs was better than DMARDs, Electro-acupuncture + DMARDs was better than DMARDs, Warm needle + DMARDs was better than DMARDs, and Fire needle + DMARDs was better than DMARDs, Moxibustion + DMARDs, Acupuncture + DMARDs, Electroacupuncture + DMARDs, Warm needle + DMARDs, and Auricular needle + DMARDs. There was no statistically significant difference between the other different therapies, as seen in Table 10. The probability ranking results of reducing CRP were as follows: Fire needle + DMARDs > Electroacupuncture + DMARDs > Acupuncture + DMARDs > Moxibustion + DMARDs > Warm needle + DMARDs > Auricular Needle + DMARDs > DMARDs, as shown in Table 11. MCMC fitting consistency model was used for analysis. The results showed that Moxibustion + DMARDs was better than DMARDs, Acupuncture + DMARDs was better than DMARDs, Electro-acupuncture + DMARDs was better than DMARDs, Warm needle + DMARDs was better than DMARDs, Auricular Needle + DMARDs was better than DMARDs, and Fire needle + DMARDs was better than DMARDs, Acupuncture + DMARDs, Electro-acupuncture + DMARDs, and Warm needle + DMARDs. There was no statistically significant difference between the other treatments, as seen in Table 12.

Results of ESR
The probability ranking results of reducing ESR were as follows:

Results of RF
Twenty-three studies reported RF (21-23, 26, 28-30, 32, 33, 35-37, 40-42, 44-50, 52). Those were all indirect comparisons to form a closed loop, and consistency test was not performed. The convergence evaluation results showed that the PSRF value was close to 1, indicating stable results. Therefore, the MCMC fitting consistency model was used for analysis. The results showed that Moxibustion + DMARDs was better than DMARDs, Acupuncture + DMARDs, and Warm needle + DMARDs. There was no statistically significant difference between the other different therapies, as seen in Table 14. The probability ranking results of improving ESR were as follows: Moxibustion + DMARDs > Fire needle + DMARDs > Warm needle + DMARDs > Acupuncture + DMARDs > DMARDs > Electro-acupuncture + DMARDs, as shown in Table 15.

Small Sample Effect Estimation
Comparison-correction funnel plot of the main outcome indicator, DAS28 score, was drawn by Stata 14.2 software for evaluation, as shown in Figure 9. The results showed that the funnel plot was not completely symmetrical, suggesting that there might be a certain publication bias or small sample effect in the research network.

Adverse Reactions
Ten studies reported adverse reactions (23,28,30,32,34,35,39,43,50,51), as shown in Table 16. On the whole, the number of adverse reactions of different acupuncture therapies combined with DMARDs was lower than that of DMARDs, and there are no serious adverse reactions reported.

DISCUSSION
At present, the pathogenesis of RA is not fully understood, and it might be related to autoimmunity, infection, and heredity (54).
Regardless of the length of the disease, early effective treatment of RA could reduce the disability, control disease activity, and prevent and delay the patient's condition (55,56). Studies have shown that the use of non-steroidal anti-inflammatory drugs,  1.93, 14.77) G The above data represent the confidence interval. The bold font indicates that there was a statistically significant difference between the two treatments. A, DMARDs, B, Moxibustion + DMARDs, C, Acupuncture + DMARDs, D, Electro-acupuncture + DMARDs, E, Warm needle + DMARDs, F, Auricular Needle + DMARDs, G, Fire needle + DMARDs. The above data represent the confidence interval. The bold font indicates that there was a statistically significant difference between the two treatments. A, DMARDs, B, Moxibustion + DMARDs, C, Acupuncture + DMARDs, D, Electro-acupuncture + DMARDs, E, Warm needle + DMARDs, F, Fire needle + DMARDs.  The above data represent the confidence interval. The bold font indicates that there was a statistically significant difference between the two treatments. A, DMARDs, B, Moxibustion + DMARDs, C, Acupuncture + DMARDs, D, Electro-acupuncture + DMARDs, E, Warm needle + DMARDs, F, Auricular Needle + DMARDs.      DMARDs, and steroid would cause serious side effects, and drug resistance in some patients, which may seriously reduce the therapeutic effect (57). Acupuncture, as a reliable and safe alternative therapy, plays an important role in the treatment of RA (58). The efficacy and safety of acupuncture combined with DMARDs in the treatment of RA have been clinically verified, but the selection of the optimal combination has become a current research priority.
In this study, we evaluated the effects of acupuncture-related therapies combined with DMARDs on DAS28, VAS, morning stiffness time, CRP, ESR, and RF in patients with RA. DAS28 could continuously measure RA disease activity with information of swollen joints, tender joints, acute phase response, and general health, and it has been widely used to evaluate the remission of RA patients (59). Morning stiffness and pain are the main symptoms that accompany the progression of RA, which can reflect the severity of RA. Generally speaking, the longer the morning stiffness and the more severe the pain was, the worse the condition was (60). Serological disease markers (CRP, ERS, and RF) are important indicators for judging the active stage of RA, which can reflect the degree of inflammation and tissue damage in patients (61). In particular, RF is an important indicator for diagnosing RA and judging its prognosis (62). The results of our study showed that in terms of improving DAS28 scores, electro-acupuncture combined with DMARDs had the best effect according to the probability ranking results, and Electro-acupuncture + DMARDs was superior to Moxibustion + DMARDs, DMARDs, and Acupoint catgut embedding + DMARDs according to the NMA results. In terms of improving VAS score, fire needle combined with DMARDs had the best effect according to the probability ranking results, and the NMA results showed that Fire needle + DMARDs was better than Moxibustion + DMARDs, Acupuncture + DMARDs, and DMARDs. In terms of improving morning stiffness time, there was no statistically significant difference between all the therapies, which meant that acupuncture-related therapies combined with DMARDs were not better than the use of DMARDs alone in improving morning stiffness in RA patients. In terms of reducing CRP, fire needle combined with DMARDs had the best effect according to the probability ranking results, while the NMA results showed that Fire needle + DMARDs was better than DMARDs, Moxibustion + DMARDs, Acupuncture + DMARDs, Electroacupuncture + DMARDs, Warm needle +DMARDs, and Auricular needle + DMARDs. In terms of reducing ESR, the probability ranking results showed that fire needle combined with DMARDs had the best effect, and the NMA results showed that Fire needle + DMARDs was better than DMARDs,  Acupuncture + DMARDs, Electro-acupuncture + DMARDs, and Warm needle + DMARDs. In terms of reducing RF, the probability ranking results showed that Moxibustion + DMARDs had the best effect, followed by Fire needle + DMARDs, while the NMA results showed that Moxibustion + DMARDs was superior to DMARDs, Acupuncture + DMARDs, and Warm needle + DMARDs, but there was no significant difference between Moxibustion + DMARDs and Fire needle + DMARDs. The above results indicated that though the curative effects of different indicators were different, Electroacupuncture + DMARDs, Fire needle + DMARDs, and Moxibustion + DMARDs ranked first among the multiple indicators. It can be seen that Electro-acupuncture + DMARDs, Fire needle + DMARDs, and Moxibustion + DMARDs have outstanding efficacy in the treatment of RA.
Considering that the quality of the included studies is moderate, it is necessary to make a reasonable selection based on the characteristics of the patient's condition in clinical practice. According to the results of the study, the top rankings are as follows: Fire needle combined with DMARDs, Electroacupuncture combined with DMARD, and Moxibustion combined with DMARD. Modern studies have found that acupuncture can effectively relieve the pain and improve the quality of life in RA patients. The curative effect is related to antiinflammation, antioxidant, immune system, endorphins, and serotonin (13,63). Fire needle, electro-acupuncture and moxibustion are further improved and developed on the basis of acupuncture theory, which could enhance the curative effect. Fire needle could effectively inhibit inflammation of RA by downregulating Anti-cyclic citrullinated peptide antibody (ACPA) and tumor necrosis factor-a (TNF-a) (64). Electroacupuncture can reduce the levels of TNF-a and vascular endothelial growth factor (VEGF) in peripheral blood and synovium of joints, improve the internal environment, and relieve joint symptoms of RA patients (65). Moxibustion could downregulate the levels of interleukin-1b (IL-1b), TNF-a, matrix metalloproteinase 1 (MMP-1), matrix metalloproteinase 3 (MMP-3), and hypoxia-inducible factor-1a (HIF-1a)/VEGF, and inhibit angiogenesis to show a potential protective effect on bones (66). For the commonly used acupoints, modern studies have shown that stimulation of ST36 with electroacupuncture could activate the anti-inflammatory pathway of vagus nerve-adrenal gland in mice to exert anti-inflammatory effects (67), while stimulating Ashi points can inhibit the expression of phosphorylated c-Jun N-terminal kinase in dorsal root ganglion of mice and thus play an analgesic role (68).
There are many limitations in this study. First, many studies included did not specifically report random methods, allocation concealment, and blinding, which influenced the testing power of the research results. Second, the sample size of the included studies was small, which might limit the accuracy of the results. Third, the type and dosage of DMARDs, the point selection of acupuncture-related therapies, and the course of treatment were different in the included studies, which might increase clinical heterogeneity. Fourth, there was certain publication bias and small sample effects in the studies, which might influence the reliability of the results. Fifth, there is a lack of other acupuncture therapies, such as bloodletting therapy and acupoint injection, because of the limited amount of original studies, which made it impossible to compare the efficacy of all acupuncture-related therapies. Sixth, as for DAS28 score, it is not clear whether DMARDs combined with fire acupuncture is superior to Electroacupuncture combined with DMARDs because the treatment of DMARDs combined with fire acupuncture was not included in the primary outcome indicators.
In conclusion, after a comprehensive comparison of the outcome indicators of 8 different therapies, electro-acupuncture combined with DMARDs is the best therapy to improve the DAS28 score. In terms of improving pain and serological markers, fire needle combined with DMARDs and moxibustion combined with DMARDs are the best, but it is impossible to tell which is better. In clinical practice, the appropriate treatment method should be selected according to the actual situation. Due to the current limited literature reports and the poor quality of some of them, more multi-center, large-sample, prospective RCT studies are needed to verify the conclusions.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding authors.