2.1. Search Strategy

The following sources were searched from inception to July 2020: Medline, Embase, Chinese National Knowledge Infrastructure (CNKI), Wanfang Data, VIP information database, and Chinese Biomedical Literature Database. In an effort to identify further published and unpublished research studies, we retrieved the reference lists of relevant reports and review articles identified from electronic databases. Following terms combined were used for searching: acupuncture OR acupoint, hypertension OR blood pressure, animal OR pre-clinical study OR mechanism study.

2.2. Inclusion Criteria

Studies were included if all of the following standards were met: sustaining high blood pressure (HBP) animal model, baseline systolic blood pressure (SBP) ≥ 140 mmHg, diastolic blood pressure (DBP) ≥ 90 mmHg [9, 10]; at least one of the following was used as outcome measures: SBP, DBP, and MAP; and SBP/DBP/MAP was compared with those of hypertension animals receiving sham-acupuncture or no treatment.

2.3. Exclusion Criteria

Studies were excluded if any of the following standards was met: the HBP was induced by stress or adrenaline administration and was not lasting; studies were conducted to compare the effects of different acupuncture methods on hypertension, with no treatment or sham-acupuncture treatment as control; studies did not use BP values as outcomes; and duplicate publications.

2.4. Selection of Studies

Two reviewers screened the titles and/or abstracts of searched studies and excluded obviously irrelevant studies, such as clinical studies, reviews, and nonhypertension and nonacupuncture studies. The full texts were obtained for the remaining studies. A flowchart of study selection is presented in Figure 1.

2.5. Data Extraction

Two reviewers independently assessed the eligible studies and extracted data using a predefined template. The following items were extracted: publication information, experimental animal information, type of animal model, acupuncture manipulation methods, outcome measurements, data of mean outcome, standard deviation, and sample of animals in the acupuncture as well as the control groups. If the results were only shown by chart, the data were obtained by accurate measurement from the figures by software Digxy1.0.0.1. The missing information was sought by sending e-mail or telephone call to the corresponding author of the article. Stata 12.0 software was used for data analysis.

2.6. Quality Assessment

The methodological quality of the included studies was assessed according to the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines, the standards of reporting in animal research. 20 items were included in the guideline. Standards of grading were adjusted based on previous report [11, 12]. Items “1” and “11” were marked a lowest score of 0 and a highest score of 1 (0 = inaccurate, not concise, or not reported; 1 = accurate, concise, or reported). The other items (2, 3, 4, 5, 6, 7, 8, 9, 10, 12, 13, 14, 15, 16, 17, 18, 19, and 20) were marked a lowest score of 0 and a highest score of 2 (0 = clearly inaccurate or not reported; 1 = possibly accurate, unclear, or incomplete; 2 = clearly accurate). The total scores of one study ranged from 0 to 38. The category score was sum of scores by each item. The maximum score was the maximum possible score, i.e., when each item was assessed as clearly accurate [11]. We calculated a ratio quality score/maximum score, generating three possible quality intervals of which 0.8–1 was considered “excellent,” 0.5–0.8 was considered “average,” and scores below 0.5 was considered “poor.”

2.7. Bias Assessment

The risk of bias was evaluated with the Systematic Review Centre for Laboratory Animal Experimentation (SYRCLE) tool [13]. This tool contains 10 items about the selection, performance, detection, attrition, reporting bias, etc. Each item was labeled yes, no, or unclear to score selected articles. Studies with 1 item recognized to be high risk of bias were considered to have an entire high risk of bias. Studies with unclear risk of bias for at least one item were considered to be at unclear risk of bias, and studies with low risk of bias in all items were assessed as low risk of bias [14].

The assessment of the methodological quality and risk of bias was performed by two authors separately. Any divergence was resolved through discussion with a third reviewer.

2.8. Statistical Analysis

Meta-analyses were performed using Stata 12.0 software. Outcome measures of SBP, DBP, and MAP were considered as continuous data. The effect size was calculated by combining these data with the random effects model, respectively. Publication bias was assessed with a funnel plot and Egger’s test. The I² statistic was used for detection of heterogeneity. If the I² statistic was higher than 50%, we considered significant heterogeneity was present. To explore the impact of factors potentially influencing the BP outcome, subgroup analyses were conducted for the following factors: acupuncture methods used, age of acupuncture initiation, age of BP measurement, duration of each acupuncture session, total acupuncture sessions, and acupuncture treatment frequency. Sensitivity analyses were performed by deleting one study at a time from the pooled studies.

3.1. Study Characteristics

Totally, 1522 animals were included for comparison. Of the hypertension models in these studies, 63 used SHR, 1 used AngII infusion [44], and 3 used renal artery stenosis [17, 21, 22]. 3 studies simultaneously assessed SBP/DBP/MAP as the outcome measures [42, 48, 52]. 20 studies assessed SBP and DBP as the outcome measures [23, 32, 35, 36, 38, 41, 43, 45, 47, 49, 50, 53, 55, 56, 60, 62, 68, 72, 79, 80]. 3 studies assessed SBP and MAP as the outcome measures [37, 58, 70]. 35 studies assessed only SBP as the outcome measures [9, 17–22, 24–27, 29–31, 33, 34, 39, 40, 46, 51, 54, 57, 59, 61, 63–65, 67, 69, 71, 73, 75–78]. 6 studies assessed only MAP as the outcome measure [15, 16, 28, 44, 66, 74]. 40 studies adopted manual acupuncture [9, 17, 20, 25, 29, 32–41, 43, 45–47, 50, 51, 53, 54, 57–63, 65–68, 70, 71, 75–78], of which 35 adopted manipulation instead of simple needle penetration [9, 17, 20, 25, 29, 32–41, 43, 45–47, 50, 51, 53, 54, 57, 59–61, 63, 65, 66, 68, 71, 76–78]. 20 studies used EA as intervention [15, 16, 18, 21–23, 27, 28, 30, 31, 42, 44, 48, 52, 64, 69, 72–74, 79]. 4 studies combined manipulation and EA as intervention [24, 49, 55, 56]. 2 studies used other auxiliary devices combined with acupuncture as treatment [19, 80]. The needle retaining time ranged from 0.5 minute to 30 minutes. The duration of acupuncture ranged from 1 time to 48 times. Animal age of treatment initiation ranged from 8 weeks to 34 weeks. Study characteristics are shown in Table 1.

3.2. Quality Assessment

Twenty items were evaluated according to the ARRIVE guidelines. According to ratio quality score/maximum score, the items 5/6/10/12/13/14/17/18 were considered to be poor. The items 2/3/4/7/9/11/19/20 were considered to be average. Also, the items 1/8/15/16 were considered to be excellent. Detailed information of identified studies’ quality is shown in Table 2.

3.3. Risk of Bias

Risk of bias was appraised for each included study. 37 studies were thought to have low risk of sequence generation. 23 studies were judged as having low risk of baseline characteristics. No study was considered to have low risk of allocation concealment, random housing, blinding against performance bias, random outcome assessment, and blinding against detection bias. 56 studies were thought to have low risk of incomplete data. All of the 67 studies were thought to have low risk of selective outcome reporting and other bias. Thus, all studies were judged as having high risk of bias. Details of bias assessment information of identified studies are shown in Table 3.

3.4. Effectiveness

3.4.1. SBP

61 studies including 1385 animals conducted the meta-analysis of SBP value. Compared with the hypertension model group, acupuncture showed a mean reduction of SBP for 25.37 mmHg with significant heterogeneity in a pooled analysis of 60 studies (Figure 2) (MD −25.37, 95% CI: −29.18 to −21.56, P < 0.0001; heterogeneity: chi² = 2746.83, df = 59 (P < 0.0001), I² = 97.6%). Compared with the sham-acupuncture group, acupuncture showed a mean reduction of SBP for 21.47 mmHg with statistically significant heterogeneity in a pooled analysis of 19 studies (Figure 3) (MD −21.47, 95% CI: −27.04 to −15.89, P < 0.0001; heterogeneity: chi² = 342.60, df = 18 (P < 0.0001), I² = 94.7%).

3.4.2. DBP

23 studies including 523 animals conducted the meta-analysis of DBP value. Compared with the hypertension model group, acupuncture showed a mean reduction of DBP for 21.26 mmHg with statistically significant heterogeneity in a pooled analysis of 22 studies (Figure 4) (MD −21.26, 95% CI: −26.21 to −16.30, P < 0.0001; heterogeneity: chi² = 425.69, df = 21 (P < 0.0001), I² = 95.1%). Compared with the sham-acupuncture group, acupuncture showed a mean reduction in DBP for 19.11 mmHg with statistically significant heterogeneity in a pooled analysis of 7 studies (Figure 5) (MD −19.11, 95% CI: −27.89 to −10.33, P < 0.0001; heterogeneity: chi² = 58.32, df = 6 (P < 0.0001), I² = 89.7%).

3.4.3. MAP

12 studies including 274 animals conducted the meta-analysis of MAP value. Compared with the hypertension model group, acupuncture showed a mean reduction of MAP for 21.96 mmHg with statistically significant heterogeneity in a pooled analysis of 12 studies (Figure 6) (MD −21.96, 95% CI: −29.31 to −14.62, P < 0.001; heterogeneity: chi² = 284.15, df = 11 (P < 0.0001), I² = 96.1%). Compared with the sham-acupuncture group, acupuncture showed a mean reduction in MAP for 16.72 mmHg with statistically significant heterogeneity in a pooled analysis of 5 studies (Figure 7) (MD −16.72, 95% CI: −23.98 to −9.46, P < 0.0001; heterogeneity: chi² = 51.97, df = 4 (P < 0.0001), I² = 92.3%).

3.5. Subgroup Analysis and Sensitivity Analysis

To investigate potential factors which influenced the BP measures, we stratified the included studies according to factors, as shown in Tables S1–S5. The subgroup analysis of MAP comparing acupuncture and sham-acupuncture was not conducted due to limited number of studies.

For SBP/DBP/MAP, sensitivity analyses showed that the results did not largely change after omitting any one study (Figures S1–S6).

3.6. Assessment of Publication Bias

Funnel plot of all outcome measures between different groups (Figures 8–13) showed symmetry, indicating no significant publication bias. Egger’s test of different comparisons of SBP/DBP/MAP showed no publication bias present. Details of Egger’s test are shown in Table S6.

3.7. Related Mechanisms

The underlying molecular mechanisms of acupuncture involved in the present review for hypertension are summarized as follows: (1) inhibition of inflammatory factors including tumor necrosis factor-α (TNF-α) [21], interleukin-6, and C-reactive protein [40, 46], as well as inflammation-related genes [63], through toll-like receptor 4 [66] in the PVN; (2) reduction of oxidative stress reactions [64, 72–74, 79] by inhibiting nicotinamide adenine dinucleotide phosphate (NADPH) oxidase activity by downregulation of p38 mitogen-activated protein kinases and extracellular signal-regulated protein kinase 1/2 [58]; (3) inhibition of AngII activity in the plasma [23, 32, 62, 67, 75], myocardium [59, 71], thoracic aorta [69], rostral ventrolateral medulla [62], kidney [62], etc. and reversion of the artery remodeling [53] and left ventricle [45, 57] through mitogen-activated protein kinase signaling [24]; (4) inhibition of glucocorticoid system, including corticotropin-releasing hormone, adrenocorticotropic hormone, and cortisol and glucocorticoids receptor [48, 52]; and (5) mediation of the balance between vasorelaxation factors and vasomotor factors and improvement of the endothelial dysfunction [15, 16, 21, 25–27, 34, 39, 49, 57, 59]. This is consistent with the genomic study of acupuncture, which suggested acupuncture may regulate multiple biological processes mainly involving oxidative stress, inflammation, and vascular endothelial function [70].

4.1. Limitations

Limitations are enumerated as follows. First, being limited in English and Chinese databases, other language studies may be missed. The absence of other language studies may bring about selective bias. Second, negative results were hardly published. Thus, the whole body of positive results might bring about exaggerated efficacy. Third, most included studies are defective in terms of randomization, allocation concealment, blinding assessment, and sample size calculation by both assessment scales, which are central for the preclinical study design criteria [81]. Low methodological quality and high risk of bias weakened the robustness of the current preclinical evidence. Fourth, the high level of heterogeneity among different studies implied not only differences of laboratory animals but also the acupuncture method details. Though subgroup analyses according to several factors were conducted to explore possible sources of heterogeneity, the results failed to completely account for the heterogeneity. Thus, the present study should be interpreted cautiously.

4.2. Efficacy Assessment

Although the effectiveness of acupuncture’s antihypertension in most studies was demonstrated by the value of BP, the ultimate goal of hypertension treatment is to prevent target organ damage and decrease mortality. Among the 67 included studies, 15 assessed the target organ damage, including left ventricle, kidney, and blood vessel remodeling. Limited by study period, none of the studies assessed the mortality of hypertension animals. Thus, target organ damage should be paid attention to in future acupuncture antihypertension studies.

Although blood pressure variability (BPV) is not a physiological index daily detected in clinical practice, it is deemed as a prognostic factor independent of BP value for cardiovascular mortality [82, 83]. Evidence shows that cardiac accident of hypertension may be the consequence of elevated variability but not BP alone [84, 85]. In addition, there is a strong correlation between elevated BPV and aggravated target organ damage [86]. Only 1 study assessed the effect of acupuncture on BPV [58]. Analysis of BPV may provide information about autonomic function, and there are also studies which demonstrated that acupuncture has the potential to mediate autonomic function [87]. Thus, exploring the effect of acupuncture on hypertension animal BPV may contribute to finding out the potential efficacy of acupuncture.

4.3. Implications

Preclinical studies usually provide evidence basis for clinical studies. To demonstrate the scientific function pathway of acupuncture, animal studies should be designed, study data should be analysed, and the result should be reported appropriately and transparently. There is a wealth of evidence showing the poor design of animal research [88], which is an obstacle to advance animal research into promising achievement for human disease. The ARRIVE guidelines have been issued since 2010 to unify the criteria of reporting in animal research and were utilized by this review to assess the quality of paper in respect of acupuncture’s antihypertension effects. The ARRIVE [89] is a reporting guideline containing a 20-entry checklist about each part of publishing papers. The SYRCLE tool was applied to assess the internal validity and risk of bias within individual studies [13]. In addition to items of ARRIVE and SYRCLE, acupuncture’s details including acupuncture manipulation, treatment duration, if electroacupuncture is used, the frequency and intensity of pulse method, etc. should be especially noted. Although these assessment tools could be used in acupuncture preclinical study [90], the need for adjusted and unified criteria in reporting of acupuncture preclinical study is imperative. Similar to the Standards for Reporting Interventions in Clinical Trials of Acupuncture (STRICTA) [91], a specialized guideline for acupuncture’s preclinical study should be formulated in order to promote study’s transparency and repeatability.