Acupuncture May Be a Potential Complementary Therapy for Alzheimer's Disease: A Network Meta-Analysis

With Alzheimer's disease (AD) becoming a worldwide problem, traditional Chinese medicine (TCM), especially acupuncture, stands out as a complementary therapy because of its feature—“treatment based on syndrome differentiation”. This systematic review and network meta-analysis (NMA) confirms the complement effect of acupuncture and explores the best combination of therapy for AD based on the total effect and activity of daily living scale (ADL). We searched relevant randomized controlled trials (RCTs) that applied acupuncture for treating AD. 58 studies with 4334 patients were included in accordance with PRISMA guidelines. The results showed that for the total effect, the order of probability for the effect: acupuncture + western medicine > acupuncture + herbal medicine > acupuncture > acupuncture + western medicine + herbal medicine. For the ADL score, the order of probability for the effect: acupuncture + western medicine > acupuncture > acupuncture + western medicine + herbal medicine > acupuncture + herbal medicine. The combination of acupuncture and medicine has a better clinical effect than acupuncture only in a way. Acupuncture + western medicine has an obvious and exact improvement in the curative effect from both total effect and ADL score, but further higher quality studies, which can detail the classification of these interventions, are still needed to verify it.


Introduction
Alzheimer's disease (AD), also known as senile dementia, is a common degenerative disease of the central nervous system in the elderly. AD is the most common type of dementia (accounting for 60% to 80% of all dementia types) [1], mainly manifested as memory impairment, aphasia, apraxia, ignorance, executive dysfunction, as well as personality and behavior changes. With the progression of the disease, patients' abilities in cognition, behavior, and other aspects can gradually decline. Teir living quality can be much lower and they eventually lose their cognition and selfcare abilities. As the aging of the world's population intensifes, the incidence rate of AD has also increased. As of 2017, the prevalence rate of AD in China was 7.5%; the prevalence rate for people over 80 years old was about 30% [2]. From the current situation of clinical treatment, the cure rate of AD is low. Recently, the age of onset has also been getting younger, and the pathogeny of AD has become much more complicated. AD has gradually become a worldwide problem [3].
At present, western medicine is considered the mainstream treatment for AD. Western medicine mainly uses drugs with functions of inhibiting β-amyloid deposition, inhibiting neurofbrillary tangles, increasing cholinergic nerve function, and excitatory neurotransmitters to treat AD [4], such as nimodipine, donepezil hydrochloride, Oracetam, Carbalatin, etc. Because of the complex pathogenesis of AD, drug therapy has its limitations. Drug therapy can only target certain pathogenesis to treat AD, which is defcient in comprehensive treatment. Terefore, various complementary therapies have been developed recently. Traditional Chinese medicine (TCM) therapies, such as herbal medicine and acupuncture, are complementary treatments with huge development potential that have been proven to be efective. Tey all follow the principle of "treatment based on syndrome diferentiation", which means that clinicians can adjust their selection of herbal medicine or acupoint based on the specifc body condition of patients to get a better overall efect. Moreover, previous studies have confrmed that acupuncture has the features of multiple targeting therapy. Moreover, its function of holistic regulation plays an important role in the preventive treatment of AD [5]. Te combination of TCM therapies and western mainstream medicine has been constantly innovated and developed, among which the combination of acupuncture and medicine accounts for a certain proportion, and the clinical efcacy of this combination has also been confrmed. However, due to the variation and diferences in the prescription of TCM and the selection of acupoints, the clinical efcacy of combined interventions is greatly afected by specifc intervention plans. Te various intervention plans with great diferences in clinical randomized controlled trials (RCTs) may afect the comprehensiveness of therapeutic evaluation because of the limitations on sample size. Clinical studies with large sample sizes are needed to provide evidence for comparing the clinical efcacy of various combinations of acupuncture and medicine for treating AD, to help determine the best combination, and to explore whether acupuncture can complement the mainstream drugs for AD, and provide a reference for the clinical treatment of AD.
Network meta-analysis (NMA) can aggregate data from multiple studies and remedy the limitations on the sample size, allowing us to compare and analyze the clinical efcacy of diferent interventions for AD based on the network relationships of multiple trials. Tis study compares and ranks the clinical efcacy of diferent combination interventions for treating AD (acupuncture + western medicine, acupuncture + herbal medicine, acupuncture, acupuncture + western medicine + herbal medicine, western medicine + herbal medicine, and western medicine) based on NMA to provide more intuitive data evidence for the comparison and application of various combinations of acupuncture and medicine in the clinical treatment of AD.

Search Strategy.
Two researchers searched PubMed, Embase, Cochrane Library, CBM, CNKI, WanFang Data, and CQVIP databases until August 21, 2021, independently. Tere were no date limits regarding the publication date of the included studies. In addition, the references of the included studies were traced to obtain other relevant studies to supplement the included studies.

Inclusion and Exclusion Criteria.
Inclusion and exclusion criteria were formulated based on the principle of PICOS (P-population; I-intervention;C-comparison;Ooutcome;S-study design): Te inclusion criteria were as follows: ① Study design: published RCTs. Te language of materials was limited to Chinese or English. ② Population: patients who were diagnosed with AD or met the diagnostic criteria, such as "the Diagnostic and Statistical Manual of Mental Disorders, Revised Fourth Edition" (DSM-IVR) [6], which was published by the American Psychiatric Association. Tere was no limitation in patients' gender, age, nationality, race, occupation, education level, course, and severity of the disease. Te baseline of the same RCT was balanced (P > 0.05). Te participants were allowed to sufer from hypertension, diabetes, hyperlipidemia, and other underlying diseases. ③ Intervention and comparison: Te experimental group was treated with various acupuncture methods alone or combined with herbal or western medicine, such as electroacupuncture combined with donepezil hydrochloride, acupuncture combined with Yizhi Jiannao granule, and so on. Te control group was treated with herbal medicine, western medicine, or a combination of herbal medicine and western medicine. ④ Outcome: Total efect: According to "Criteria of diagnosis and therapeutic efect of internal diseases and syndromes in traditional Chinese medicine" (issued by the National Administration of Traditional Chinese Medicine), the curative efect can be divided into "cured" (all symptoms disappear), "improved" (symptoms are relieved), and "inefective" (aggravation or no change in symptoms). "Cured" and "improved" were regarded as efective. Te total efect � (the number of "cured" and "improved"/the total sample size) * 100%; Activity of Daily Living Scale (ADL). Included studies should address one or both of the outcomes mentioned above.
Te exclusion criteria were as follows: ① Repeated publications. ② Studies in the diagnosis of vascular dementia. ③ Studies without relative data or unavailable for researchers. ④ Participants had a malignant tumor, diseases of the blood or immune system, mental illness, or other obvious complications. ⑤ Te experimental groups or control groups applied other therapies besides acupuncture, herbal medicine, or western medicine, such as doll therapy, hyperbaric oxygen, music-assisted therapy, electric shock therapy, etc. ⑥ Te rate of loss to follow-up or drop-of was more than 50%, the data of outcome were missing or wrong obviously, or the efcacy evaluation was unclear.

Study Selection and Data Extraction.
Four trained researchers were divided into two groups to screen studies and extract data independently, and another two researchers cross-checked the data. Any disagreement was resolved by discussion. Subsequently, the data were extracted into a unifed spreadsheet, and the extraction contents included the following: ① basic information of included studies: title, name, and nationality of the frst author, publication year, source of study, fund status, etc.; ② baseline characteristics of objects: sample size of each group, age, course of the disease, etc.; ③ intervention: acupuncture methods (including acupoint selections, reinforcing and reducing techniques, direction of the needle, retaining time of needle, course of treatment, etc.), drug therapeutic schedule; ④ relative information about bias risk assessment: random method, the situation of drop-of and follow-up, etc.; ⑤ outcome: total efect and ADL.

Risk of Bias.
Two researchers evaluated the quality of included studies independently according to the bias risk assessment tool, namely ROB 2 [7], recommended by Cochrane5.1.0. Subsequently, the results of the assessment were cross-checked and any disagreement was resolved by a discussion.
Te assessment was related to fve major domains: ① the randomization process; ② deviations from the intended interventions; ③ missing outcome data; ④ measurement of the outcome; and ⑤ selection of the reported results. Te answers to questions involved the fve domains were provided as Yes (Y), Probably Yes (PY), Probably No (PN), No (N), or No Information (NI). Te whole process of assessment was based on the Cochrane Handbook.

Statistical Analysis.
Researchers utilized Stata/SE 16.0 software to construct NMA in a frequentist framework. All the statistical data mentioned below were calculated using Stata/SE 16.0. For dichotomous variables (total efect), odds ratio (OR) was adopted as the efective value. For continuous variables (ADL), mean diference (MD) was adopted as the efective value. Te meta-analyses were carried out by calculating the efect values and their 95% credibility interval (CI).
Researchers constructed a network map to depict the comparator arms of various interventions and the relationship between these interventions. Weight the points by the total sample size received for the specifc treatment, and weight the lines by the number of researchers, which compared two interventions connected by the line directly.
Researchers calculated the efect values and their standard error (SE) of each research group and constructed a contribution plot to display the contribution of direct and indirect comparison in NMA.
A heterogeneity test was performed through an I 2 test. Higgins [8] considered that I 2 was between 0% and 100%. Tere was no heterogeneity between studies when I 2 � 0%. Te larger the I 2 , the higher the possibility of heterogeneity. It indicated that there was mild heterogeneity when I 2 � 25%; it indicated that there was moderate heterogeneity when I 2 � 50%. It indicates a high degree of heterogeneity when I 2 � 75%. Te Cochrane manual believed that when I 2 > 50%, the research study was considered to be heterogeneous, and a random efects model should be applied. When I 2 < 50%, the fxed efects model should be applied. If the heterogeneity was high, further subgroup analysis (according to the course of disease and therapy) and meta-regression should be performed to analyze the causes of heterogeneity.
Te inconsistency test of each closed loop in the network map was carried out. Researchers calculated the inconsistency factors (IFs), 95% CI, and the heterogeneity parameter t 2 (t � Standard deviation<SD>) of each loop to analyze whether there was an inconsistency in each closed loop. Te closer the IF gets to 1, the more consistent between Evidence-Based Complementary and Alternative Medicine 3 diferent studies. If the lower limit of 95% CI was 1, it meant that the direct comparison results were consistent with the indirect comparison results. Researchers set "Western medicine" as the original control intervention. We construct an interval prediction graph and an inverted triangle diagram to display the direct and indirect comparison results of diferent interventions. Treatment ranking was related to the area under the curve. Te larger the area, the better the efect of the intervention [9].
A comparison correction funnel plot was applied to analyze whether there was a small sample efect between the studies and to assess the publication bias.
Researchers summarized the selection and usage frequency of acupoint and drugs used in the included studies.   were included in our research, and the process is depicted in Figure 1. Table 2, 58 articles were included in the research, and 4334 AD patients were recruited in the trial, 2190 for experimental groups and 2144 for comparator groups, respectively. Two studies [40,64] collected outcomes, respectively, at diferent stages of treatment. Researchers split these two studies according to the course of treatment into fve independent studies. Ultimately, 60 studies were included in the fnal statistical analysis, with 4542 patients. Te total efect and ADL were the main outcomes. 54 studies reported total efects, and 25 studies reported ADL. 7 interventions were included, A-herbal medicine; B-western medicine; Cacupuncture;D-acupuncture + herbal medicine; Eacupuncture + herbal medicine + western medicine; Facupuncture + western medicine; G-herbal medicine + western medicine.

Risk of Bias and Certainty of Evidence.
Researchers used the bias risk assessment tool, named ROB 2, recommended by Cochrane 5.1.0. A total of 5 aspects of the original study were assessed, including the randomization process, deviation from intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result. Included studies were classifed as high quality, low quality, or unknown risk bias. Te result is depicted in Figure 2 and Table 3.

Total Efect
A total of 59 studies reported the total efect, involving 129 arms and 4,414 patients. Figure 3 depicts the comparative relationship between diferent interventions. Te dots represent the total number of samples in all studies using this intervention. Te lines represent the amount of research evidence that directly compared the two interventions connected. An indirect comparative analysis was carried out based on a network structure for two unconnected interventions. Te studies involved included six kinds of interventions: herbal medicine, western medicine, acupuncture, acupuncture + herbal medicine, acupuncture + western medicine, and acupuncture + western medicine + herbal medicine. fve closed loops have been formed in the network structure ("herbal medicine, acupuncture, acupuncture + herbal medicine", "herbal medicine, western medicine, acupuncture", "western medicine, acupuncture, acupuncture + herbal medicine", "western medicine, acupuncture, acupuncture + western medicine", "herbal medicine, western medicine, acupuncture + herbal medicine"), to provide direct and indirect comparative evidence for NMA. Figure 3 displays the contribution of each direct comparison result to the comprehensive comparison results of NMA, based on the total efect. "Direct comparisons in the network" refers to the direct comparison evidence included in studies. "Mixed estimates" represent comparisons that combine direct and indirect comparison evidence. "Indirect estimates" represent comparisons that are only based on indirect comparison evidence. For example, 25.9 means that the contribution rate of the direct comparison between intervention A (Herbal medicine) and intervention D (Acupuncture + Herbal medicine) for comparing the efcacy of intervention A (Chinese medicine) and intervention B (Western medicine) is 25.9%.

Testing for Heterogeneity and Inconsistency.
According to the results of the heterogeneity test, I 2 � 16.4% < 25%, P < 0.05, regarded as low heterogeneity. NMA was carried out under the fxed efects model; applied the inconsistency model was used for NMA in advance, P � 0.0946 > 0.05. According to the inconsistency test for the closed loop, P > 0.05 for each closed loop (Table 4), which indicates no inconsistency among the groups. Te consistency model was selected for NMA. Figure 4 displays the results of direct and indirect comparisons; _y_A, _y_C, _y_D, _y_E, and _y_F represent comparison results between interventions A, C, D, E, F, and intervention B, respectively. Labels, like_y_C-_y_A, _y_D-_y_A, etc., represent the comparison results between the two interventions mentioned. Te results indicate that the curative efects of acupuncture, acupuncture + herbal medicine, acupuncture + western medicine, and acupuncture + herbal medicine + western medicine are better than that of western medicine. Te curative efects of acupuncture, acupuncture + herbal medicine, and acupuncture + western medicine are better than those of herbal medicine. Te curative efect of acupuncture + western medicine is better than acupuncture + herbal medicine + western medicine. Te diferences in the remaining comparisons were not statistically signifcant.

Network Meta-Analysis.
Te results mentioned above can also be obtained from the inverted triangle diagram (Table 5). Te 95% CI must not contain 1; otherwise, the diferences in comparisons are not statistically signifcant. If the OR value is greater than 1, it means that the interventions sorted by the column have better efcacy than the interventions sorted by the line.
Te surface under the cumulative ranking curve (SUCRA) (Figure 4) shows that the combination of acupuncture and western medicine is the most efective intervention for treatment. Te order of probability for the efect: acupuncture + western medicine > acupuncture + herbal medicine > acupuncture > acupuncture + western medicine + herbal medicine > herbal medicine > western medicine. the order of the curative efect of the intervention combined with acupuncture: acupuncture + western medicine > acupuncture + herbal medicine > acupuncture > acupuncture + western medicine + herbal medicine.

Small Sample Efect and Bias.
Te comparisoncorrection funnel plot ( Figure 5) displays that the dots are slightly asymmetrically distributed on both sides of the vertical line of the X � 0. Five studies, including "western medicine" vs "acupuncture + herbal medicine + western medicine", "western medicine" vs "acupuncture + herbal medicine", "herbal medicine" vs "acupuncture + herbal medicine", "western medicine" vs "acupuncture + western medicine" are from the line of 95% CI in Figure 5, which shows that asymmetry may be caused by heterogeneity.

Heterogeneity Test and Subgroup Analysis.
According to the results of the heterogeneity test, I 2 � 94.4% > 75%, P < 0.05, regarded as high heterogeneity.

Identification of studies via databases
Records removed before screening: Duplicate records removed (n = 2264) Records marked as ineligible by automation tools (n = 0) Records removed for other reasons ( n = 0) Records identified from: Records screened (n = 4074) Reports sought for retrieval (n = 157) Reports assessed for eligibility (n = 157) Evidence-Based Complementary and Alternative Medicine Evidence-Based Complementary and Alternative Medicine  A-herbal medicine; B-western medicine; C-acupuncture. D-acupuncture + herbal medicine; E-acupuncture + herbal medicine + western medicine; F-acupuncture + western medicine; G-herbal medicine + western medicine.

Evidence-Based Complementary and Alternative Medicine
Researchers performed a subgroup analysis of the included materials according to the course of AD (Studies were divided into 4 subgroups: less than 1 year, 1-3 years, 3-5 years, and 5-10 years). I 2 of groups "less than 1 year" and "5-10 years" decreased to 80.6% and 80.3%, respectively, and the I 2 of the remaining groups did not change signifcantly. Researchers performed a meta-regression based on the course of AD and interventions, but the heterogeneity remained unchanged. Moreover, there was no reason for heterogeneity was found. Since the number of studies in the group "less than 1 year" is too small (2 studies in total) and the groups "1-3 years" and "3-5 years" have high heterogeneity (>90%), only the "5-10 years" group was involved in NMA. NMA was carried out under the random efects model.

Network Structure.
A total of 11 studies were involved in the "5-10 years" group, involving 23 arms and 842 patients. Figure 3 depicts the comparative relationship between diferent interventions. Te dots represent the total number of samples in all studies using this treatment. Te lines represent the amount of research evidence that directly compared the two treatments connected. An indirect comparative analysis was carried out based on the network structure of two unconnected interventions. Te studies involved included six kinds of interventions: herbal medicine + western medicine, western medicine, acupuncture, acupuncture + herbal medicine, acupuncture + western medicine, and acupuncture + western medicine + herbal medicine. One closed loop has been formed in the network structure ("herbal medicine + western medicine-acupuncture-acupuncture + herbal medicine + western medicine") to provide direct and indirect comparative evidence for NMA. Figure 3 displays the contribution of each direct comparison result to the comprehensive comparison results of NMA, based on the ADL score. For example, 44.1 means that the contribution rate of the direct comparison between herbal medicine + western medicine and acupuncture for comparing the efcacy of these two interventions is 44.1%.

Testing for Inconsistency
. Applied the inconsistency model for NMA in advance, P � 0.4132 > 0.05. According to the inconsistency test for closed loop, P � 0.279 > 0.05 (Table 6), which indicated no inconsistency among the groups. A consistency model was selected for NMA. Figure 4 displays the results of direct and indirect comparisons. Te results indicate that the curative efects of acupuncture + western medicine and acupuncture + herbal medicine + western medicine are better than those of Western medicine. Te diferences in the remaining comparisons were not statistically signifcant. Te results mentioned above can also be obtained from the inverted triangle diagram (Table 7).

Network Meta-Analysis.
Te SUCRA (Figure 4) shows that Western medicine is the most efective intervention for treatment. Te order of probability for the efect: acupuncture + western medicine > acupuncture > acupuncture + herbal medicine + western medicine > acupuncture + herbal medicine > herbal medicine + western medicine > western medicine. Te order of the curative efect of the intervention combined with acupuncture: acupuncture + western medicine > acupuncture > acupuncture + western medicine + herbal medicine > acupuncture + herbal medicine.

Small Sample Efect and Bias.
Te comparisoncorrection funnel plot ( Figure 5) displays that most of the dots are symmetrically distributed on both sides of the vertical line of X � 0, indicating a low possibility of both bias and the small sample efect.

Usage of Acupoints and Drugs.
Most studies selected the Governor vessel, three foot-yang meridians, extra acupoints, and three foot-yin meridians, with few acupoints selected for three hand-yin meridians and three hand-yang meridians relatively. Compared with other parts of the body, the acupoints on the head, face, and neck, including Governor vessel acupoints, extra acupoints, other acupuncture treatment methods (including the four-shen acupuncture, temporal three-needle, the three-zhi Acupuncture, etc.), acupoints of twelve regular meridians and conception vessel acupoints, were chosen mostly among the 57 kinds of literature included. Te number of selected lower limb acupoints is the second, including only acupoints of twelve regular meridians ( Figure 6). Te top ten ranked frequencies of chosen acupoints are DU20, SP6 (confuent acupoint of three foot-yin meridians), ST36 (He-sea point of foot-yangming meridian), KI3 (Shustream acupoints of foot-shaoyin meridian), EX-HN1, GB39 (marrow convergence), BL23 (kidney back-shu point), PC6 (connecting point of hand-jueyin meridian), DU24, and DU14 (confuent acupoint of governor vessel, three foot-yang, and hand-yang meridians), most for located acupoints and several for nourishing kidney yin (Figure 7). Herbal medicine of the studies included was mainly for tonifying the spleen and kidney by activating blood circulation to dissipate stasis, while donepezil was mostly for western medicine.

Adverse Events.
Eleven included studies reported the presence of adverse events (Table 8). Due to the limited number of included studies that reported adverse events, it was not analyzed using NMA.

Discussion
AD is a common degenerative disease of the central nervous system in the elderly, whose pathogeny is complex and Evidence-Based Complementary and Alternative Medicine difcult to be explained. Tere are many interventions for AD used in clinical settings, such as drug therapy, acupuncture, music therapy, exercise therapy, memory therapy, and so on. Acupuncture has the functions of restoring consciousness and resuscitation, promoting blood circulation, replenishing qi and regulating blood, and replenishing the spleen and kidney [68]. In addition, acupoints can be selected fexibly according to the specifc body condition of the patient, to improve the patient's symptoms and overall physical condition in a targeted manner. Also, because of its small side efects and good tolerance [69], acupuncture is widely applied for treating AD. Te combined application of acupuncture and medicine (herbal medicine or western medicine) has gradually increased recently, and its efcacy has also been confrmed by clinical research studies.
Researchers searched for relative studies and utilized NMA to evaluate the curative efect of acupuncture and the combined treatment of acupuncture and medicine based on the total efect and ADL score. For the total efect, the curative efects of acupuncture, acupuncture + herbal medicine, acupuncture + western medicine, and acupuncture + herbal medicine + western medicine are all better than those of western medicine. Te curative efects of acupuncture, acupuncture + herbal medicine, and acupuncture + western medicine are better than those of Herbal medicine. Te curative efect of acupuncture + western medicine is better than acupuncture + herbal medicine + western medicine. Te diferences in the remaining comparisons were not statistically signifcant. For the ADL score, the curative efects of acupuncture + western medicine and acupuncture + herbal medicine + western medicine are better than those of western medicine. Te diferences in the remaining comparisons were not statistically signifcant. Te SUCRA shows that the top two interventions that have the best efcacy for total efect are acupuncture + western and acupuncture + herbal medicine (acupuncture + western > acupuncture + herbal medicine). Te top two interventions that have the best efcacy for the ADL score are acupuncture + western medicine and acupuncture (acupuncture + western medicine > acupuncture). Results show that acupuncture combined with medicine has a better clinical efect than acupuncture for treating AD. Acupuncture + herbal medicine is more efective for improving the total efect, but there are certain disadvantages in improving the ADL score. Te combination of acupuncture and western medicine has an impressive efect on both the total efect and the ADL score.
Acupuncture + herbal medicine and acupuncture + western medicine both have impressive efects on improving the total efect, but they both have worse efects than only applying acupuncture when adding a variety of medicine (applying acupuncture + herbal medicine + western medicine). Acupuncture + herbal medicine has an impressive efect on improving the total efect, but it has a worse efect on improving ADL scores. Researchers speculated that the reason for this contradiction in the sorting of curative efects may be related to the signaling pathways that various treatments afect the body. Te most commonly used herbal medicines for treating AD, such as Salvia miltiorrhiza, Ligusticum chuanxiong, Noto ginseng, turmeric, Herba epimedium, and so on, can treat AD by inhibiting the formation and deposition of amyloid β-protein (Aβ), inhibiting the hyperphosphorylation of the protein tau, antagonizing oxidative stress damage and neuronal apoptosis, or playing an anti-infammatory efect, etc [70][71][72][73][74]. In particular, herbal medicine for removing blood stasis is closely related to the body's autophagy, which can enhance autophagy and regulate the content of Aβ and protein tau [75]. Te most commonly used Western medicine mentioned in the included studies, such as nimodipine, donepezil, and so on, mostly focus on improving symptoms and have a therapeutic efect on AD by inhibiting cholinesterase, regulating the concentration of calcium ions in the brain, and protecting the structure of neurons [9,[76][77][78]. Acupuncture can promote autophagy at diferent levels to treat AD by stimulating specifc acupoints. Te regulating function of acupuncture on autophagy is bidirectional, which can not only promote but also inhibit autophagy. Acupuncture can also adjust the body's oxidative defense system and reduce the toxic efects of excessive free radicals on the nervous system [79,80]. Te mechanisms of acupuncture, herbal medicine, and western medicine for treating AD have their specifc parts and similar parts. Tere is saturation in signal transduction and various physiological processes. Once the signal stimulation of the same pathway reaches saturation, it may have no obvious enhancement of the efect, even produce a degenerative efect. Terefore, the combined application of acupuncture and herbal or western medicine may produce diferent comprehensive efects due to  Zhang [50] Li and Li [33] Chen et al. [19] Geng [10] Liu [36] Xia [38] Tao and Li [47] Ji et al. [61] Li and Li [33] Jia et al. [15] Wang et al. [20] Assessment  Te results of the usage of acupoints show that the selection of acupoints for treating AD is diverse and complex and distributes in various parts of the body, but all of them have a therapeutic efect on AD indeed. It refects the treatment principles of combining the main symptoms and concurrent syndromes, treating based on syndrome diferentiation, and selecting acupoints based on syndromes. [81] Although compared with western medicine, acupuncture has poor function targeting treating AD, the principle of acupoint selection based on syndrome diferentiation and the multidirectional efect of acupuncture makes it not only have the efect of treating AD but also regulates the whole body condition. Tis may be the reason why the combination of acupuncture and medicine is better than western medicine alone.
A total of 11 included studies mentioned adverse events after treatment. Adverse events mentioned the most frequently were reactions of the digestive system (nausea, vomiting, abdominal distension, diarrhea, loss of appetite) and the central nervous system (dizziness, insomnia). Although the occurrence of adverse events is afected by the patient's age, gender, and other factors [82], interventions must have a certain relationship with the adverse events. We analyzed the types of interventions used in studies with adverse events. We found that the proportion of Western medicines was the highest (81.8%). Among the included studies, donepezil was the western medicine used the most frequently. Studies have shown that adverse events to the digestive system and central nervous system are the most common adverse events of donepezil [83], which is consistent with the adverse events reported in the included studies to some extent. Te mechanism of donepezil's  Figure 3: Network diagram comparing treatment outcomes of AD for total efect (a) and ADL (b). Te diameter of each dot represents the proportional total weight of all trials in the network that investigated that intervention. Te thickness of each line connecting 2 interventions is proportional to the number of trials that investigated that pair of interventions. Contribution plots for treatments of AD for total efect(c) and ADL (d). A-herbal medicine. B-western medicine. C-acupuncture. D-acupuncture + herbal medicine. E-acupuncture + herbal medicine + western medicine. F-acupuncture + western medicine. G-herbal medicine + western medicine. Te size of each circle is proportional to the weight attached to each direct or indirect summary efect. Te numbers re-express the weights as percentages.  A-Herbal medicine. B-Western medicine. C-Acupuncture. D-Acupuncture + Herbal medicine. E-Acupuncture + Herbal medicine + Western medicine. F-Acupuncture + Western medicine. G-Herbal medicine + Western medicine. Y axis represents cumulative probability and X axis represents rank. Comparing the cumulative probability of the same control ranking, the higher ranking (6⟶1) with a higher cumulative probability means a better curative efect.    Terefore, we believe that although the occurrence of adverse events is closely related to the use of western medicine, it is also afected by acupuncture and herbal medicine.
In the included studies, acupoints on the head were mostly used for treating AD, whereas acupoints correlated with gastrointestinal function were rarely used. Based on the brain-gut axis theory, some active peptides and neurotransmitters exist in both the brain and the gastrointestinal tract. Te gastrointestinal function is closely related to the brain function and can interact with each other. [84] Moreover, studies have shown that the dysregulation of intestinal fora may also lead to AD. [85,86] Due to the high frequency of gastrointestinal reactions in adverse reactions, researchers believe that RN12, BL21, and other acupoints correlated with gastrointestinal function can be appropriately selected in clinical practice to supplement the therapeutic efect of acupoints on the head and to prevent and alleviate adverse events. Te original material on adverse events is not adequate enough, so the conclusion about adverse events should be considered comprehensively and carefully used.

Conclusions
In conclusion, the combination of acupuncture and medicine has a better clinical efect than acupuncture in a way. Acupuncture + western medicine has an obvious and exact improvement in the curative efect from both the total efect and ADL score. Terefore, the researchers believe that the development of the combination therapy of acupuncture and medicine is advantageous and reasonable for treating AD and that acupuncture does have a complementary efect on drug therapy. It has prompted clinicians to practice combination therapy of acupuncture and medicine and use the principle of selecting acupoints based on syndrome diferentiation fexibly to improve the therapeutic efect of AD. Acupuncture should be used appropriately to prevent and alleviate the adverse reactions that may occur during the treatment of AD. Researchers can study and compare the clinical efcacy of diferent combinations of acupuncture and medicine on patients with diferent syndromes to determine the combinations that can clearly reduce or improve clinical efcacy and to refne the selection of specifc acupoints or methods of acupuncture and prescriptions, and to provide more accurate guidance for the clinical treatment of AD. Network meta-analysis PICOS: P-Population; I-Intervention; C-Comparison; O-Outcome; S-Study design PRISMA: Preferred reporting items for systematic reviews and meta-analyses RCTs:

Abbreviations
Randomized controlled trials SD: Standard deviation SE: Standard error SUCRA: Te surface under the cumulative ranking curve TCM: Traditional Chinese medicine