Effects of acupuncture on vascular dementia (VD) animal models: a systematic review and meta-analysis

The following electronic databases were searched from database inception to December 2016: Embase, PubMed and Ovid Medline (R), and the following keywords were used: acupuncture, electroacupuncture, acupoint, vascular dementia, multi-infarct dementia and multiinfarct dementia. The search language was limited to English, and we also tried to collect records from other sources. The detailed search strategies are shown in Additional file 2.

One reviewer (ZYZ) generated the search strategy, searched the databases, and made a list of all the records. Two evaluators (HHD, QC) independently evaluated the articles based on the inclusion and exclusion criteria. Disagreements were solved together through discussions (ZYZ, HHD, QC). Two reviewers (HHD, QC) independently extracted the data.

The following data were extracted: publication year, the last name of the first author, model of vascular dementia, weight range of the included animals, number of animals included, method of treatment with timing and duration in the acupuncture and control groups, the assessment of trials, and the results of each article (positive or negative). The final outcomes were extracted if several outcomes were presented. When the outcome data were only shown graphically, we attempted to contact the authors to obtain the detailed data; if we received no response to our request, we used GetData software to measure the data. Differences were solved together through discussions (HHD, QC, ZYZ).

We evaluated methodological quality against an ten-item checklist [10]: (1) peer-reviewed journal; (2) temperature control; (3) animals were randomly allocated; (4) blind established model; (5) blinded outcome assessment; (6) anesthetics used without marked intrinsic neuroprotective properties; (7) animal model (diabetic, advanced age or hypertensive); (8) calculation of sample size; (9) statement of compliance with animal welfare regulations; (10) possible conflicts of interest.

The quality of each study was evaluated by a score from zero to ten. Two evaluators (HHD, QC) independently extracted the data and assessed the quality of each study. Disagreements were resolved through discussion (HHD, QC, ZYZ).

For statistical analysis, we used the statistical software package Stata version 12.0. The data of the Morris water maze test, such as escape latency, time in the quadrant in which the former platform was located, and frequency of crossing through the former platform were considered continuous data, and we therefore calculated standard mean differences (SMD) with confidence intervals (CIs) established at 95%. Heterogeneity in the studies was examined using I² statistics. If the I² statistic was higher than 50%, we considered significant heterogeneity to be present, and we used a random effects model. Otherwise, we used a fixed effect model. When significant heterogeneity existed, the subgroup analysis would be conducted based on animal species, acupuncture methods and modeling methods. We used Egger’s test and Begg’s test to assess publication bias.

The 16 studies included involved 363 rats, 174 of which were in an acupuncture group and 189 of which were in a control group. All the studies stated the weight range of the rats, which ranged from 180 to 490 g. A total of 7 of the 16 studies mentioned the age of the animals [11, 14, 16‐18, 23, 25], which ranged from 2 months old to 12 months old. Sixteen of the included studies used different subtests of the Morris water maze test; all the studies met the inclusion criteria by using escape latency as outcome data, while two studies used swimming speed [24, 26], five studies used duration in the quadrant of the former platform position [12, 13, 21, 24, 26], one used swimming distance [19], and three used frequency of crossing former platform [12, 15, 16]. Two species of rats were used in the 16 studies: seven studies used Sprague–Dawley (SD) rats [11, 13, 15, 17‐19, 25], and nine studies used Wistar rats [12, 14, 16, 20‐24, 26]. The basic characteristics of the studies are listed in Table 1 [28].

Various methods were used to establish a vascular dementia (VD) model in the different studies (Table 1) [29]. The 4-vessel occlusion (4-VO) method was used in two studies [11, 13]; 7 studies used bilateral common carotid artery occlusion (2-VO) [15, 17‐20, 24, 26]; 6 studies used embolic occlusion (EO) [12, 14, 16, 21‐24]; and the remaining trial used middle cerebral artery occlusion (MCAO) [25].

A range of acupuncture techniques was employed with regard to the combinations of acupoints, the stimulation method (EA or manual acupuncture) and manipulation. Five studies used acupoints CV6 (Zhongji), CV12 (Zhongwan), CV17 (Danzhong), SP10 (Xiehai), ST36 (Zusanli) [12, 14, 16, 21, 24], which was the most commonly used technique. Three studies chose to apply acupuncture treatment to a single acupoint, ST36 (Zusanli) [22, 23, 26]. Four studies used BL23 (Shenshu), GV14 (Dazhui), and GV20 (Baihui) [17‐19, 25]. Two studies used GV14 (Dazhui) and GV20 (Baihui) [11, 15]. One study used BL17 (Geshu), BL20 (Pishu), BL23 (Shenshu), and GV20 (Baihui) [13]. The final study had two acupuncture groups [20]: the group called the positive control group used GV14 (Dazhui) and GV20 (Baihui), and the other group, the cluster-needling group, used the frontal region, the frontoparietal region, and the parietal region. Nine studies chose manual acupuncture as the method of stimulation [12, 14, 16, 20‐24, 26], while the other seven studies used electroacupuncture [11, 13, 17‐19, 25]. Almost all of the studies that used manual acupuncture adopted 30s as their treatment duration; the exception was the study that had two acupuncture groups [20], the positive control group and the cluster-needling group, which used 60 min and 360 min. Of the 7 EA studies, six of the studies adopted continuous waves [11, 13, 15, 17‐19], of which the frequency ranged from 4 Hz to 150 Hz, and the current density was from 1 to 2 mA. The remaining study adopted sparse wave [25], of which the frequency was 30 Hz. A summary of the acupuncture treatment protocols is shown in Table 1.

Ten studies adopted some interventions in the control groups. The Western medicine nimodipine was used as a control intervention in three studies [11, 13, 15], and the remaining seven studies adopted placebo-acupuncture as a control intervention [12, 14, 16, 22‐24, 26].

The study quality scores ranged from 4 to 8. Four studies scored points in 4 items [13, 16, 21, 23]; four studies scored points in 5 items [11, 12, 14, 17]; six studies scored points in 6 items [15, 18, 19, 24‐26]; one study scored 7 points [22]; and the remaining study scored 8 points [20]. All of the included studies were published in peer-reviewed journals. Eleven of the studies mentioned control of temperature [11‐13, 15, 18‐20, 22, 24‐26], which included room temperature or the temperature of the water in the maze. All of the studies adopted random allocation. Blinded building of the model was adopted in 13 studies [11, 12, 14‐17, 19‐22, 24‐26]. Blinded outcome assessment was adopted in 2 studies [22, 23]. Twelve studies stated the use of anesthetics without marked intrinsic neuroprotective properties [11, 12, 14, 15, 17‐20, 22‐24, 26]. No study adopted a diabetic, hypertensive or aged animal model, and no study reported sample size calculations. Eight studies mentioned compliance with animal welfare regulations [14, 15, 18, 20, 21, 23‐26]. Seven studies stated possible conflicts of interest [13, 16‐18, 20, 22, 25]. The study quality assessment is shown in Table 2.

We performed a publication bias test for the outcome of escape latency using Egger’s test (Pr > |z| = 0.001 < 0.05, continuity corrected) and Begg’s test (Pr > |z| = 0.005 < 0.05, continuity corrected, Fig. 3). The results of these two tests indicated the potential existence of publication bias across all of the included studies [34]. We did not conduct a publication bias test for the other outcome measures, because there were fewer than ten included studies for each measure [35].

Of the sixteen included studies, fifteen described possible mechanisms of acupuncture in ameliorating cognitive function. The main signaling pathways were summarized in three aspects, oxidative stress damage reduction, nerve apoptosis suppression and neurogenesis. See Table 4 [36, 37].

This study indicates that electroacupuncture could increase the duration in the former platform quadrant both in EO models (SMD = 1.56, 95% CI: 1.02 ~ 2.11; p < 0.00001; heterogeneity: I-squared = 0.9%) and 2-VO models (SMD = 4.29, 95% CI: 3.23 ~ 5.35; p < 0.00001; heterogeneity: I-squared = 0%) compared with the control groups. This finding suggests that acupuncture may play a potential role in ameliorating cognitive dysfunction in animal models. A previous study indicates that the addition of acupuncture therapy to routine care may have beneficial effects on improvements in cognitive status but limited efficacy on health-related quality of life in vascular dementia patients [38].

The high heterogeneity between the studies based on escape latency time cannot be completely explained. We conducted subgroup analyses based on the method of stimulation, the modeling method and the animal species used, but no evident cause was found. Meanwhile, the subgroup analysis of the modeling methods that regarded the duration in former platform as an outcome measure obviously reduced the heterogeneity. We found all five of the studies in this analysis adopted the same method of stimulation, the same animal species, and similar acupoint combinations. Four of the five studies used CV6, CV12, CV17, SP10, and ST36, the other study used ST36. These findings indicate that, in addition to the differences in stimulation method, modeling methods and animal species used, acupoint combinations may be another source of heterogeneity among the studies.

Systematic reviews can help promote the methodological quality of preclinical animal studies. Systematic reviews can help promote the methodological quality of preclinical animal studies. Essential methodological details are significant to measure the quality of a body of evidence and to assess the bias risk in animal trials. However, the insufficiency of the methodology is evident in many aspects of the present study. For instance, an adequate target animal model represents an important aspect to improve the quality of the experimental design. All included studies were performed on healthy and young animals. However, vascular dementia [39] generally occurs in aged, diabetic or hypertensive patients; therefore, experiments using young and healthy animals may overestimate the effectiveness of the intervention [10]. Hence, appropriate target animal models (hypertensive, advanced age and diabetic) should be used in future experimental research.

The included studies investigated several signaling pathways to gain a better understanding of the mechanism of improving cognitive function via acupuncture, including modulating the production and degradation of free radicals to reduce brain damage [11], exerting anti-apoptotic effects by up-regulating B-cell lymphoma-2 (Bcl-2) and counter-regulating Bcl-2-associated X protein (Bax) to protect neurons [14], enhancing glucose metabolism in the brain [16], reducing tumor protein P53 and Noxa expression to protect pyramidal cells from apoptosis [17], increasing cerebral blood flow for increased glucose and oxygen supply to neurons [21], exerting neuroprotective effects via an antioxidative pathway mediated by nuclear factor erythroid2-related factor 2 (Nrf2) [24], and increasing complex enzymes to protect neurons from oxidative stress [26]. The present study indicated that acupuncture protects neurons during vascular dementia mainly through enhancing oxygen and glucose metabolism and anti-apoptosis and antioxidant properties. However, the signaling pathways targeted by acupuncture were infrequently and incompletely reported. Therefore, this field should be further explored in future clinical studies.

This systematic meta-analysis has several limitations, the first of which is language bias. We limited the language of our searches to English only, which may cause potential publication bias. The results may have differed if we had included studies reported in Chinese, Korean, Japanese or other languages. Second, the total sample size was still not big enough, although we made a concerted effort to search all of the studies that met the inclusion criteria. Third, the quality of the included studies was unsatisfactory, which had an important influence on the results of the systematic meta-analysis. Fourth, evident availability bias was caused by two included studies from which the data cannot be extracted [35]. We tried to contact the authors by e-mail, but we received no reply. Fifth, we performed Begg’s test and Egger’s test for publication bias assessment, and the results indicated potential publication bias.

In view of the limitations above, we recommend that non-English language literature should be included in future systematic meta-analyses. Furthermore, blind outcome assessments, use of target animal models (hypertensive, advanced age and diabetic), calculation of sample size, statement of compliance with animal welfare regulations, possible conflicts of interest and detailed data publishing should be considered in future animal model studies.