The following databases were electronically searched from PubMed, Web of Science, EMBASE, Chinese National Knowledge Infrastructure (CNKI), VIP information database and Wanfang data Information Site from inception to April 2016. The search terms were as follows: “Shexiang Baoxin OR She Xiang Bao Xin” AND “myocardial infarction OR myocardial ischemia” AND “vascular endothelial growth factor OR VEGF OR microvessel count OR MVC.” All searches were limited to studies on animals. We also manually searched published abstracts of scientific meetings and asked senior authors of identified publications for references of related studies.

To prevent bias, pre-specified inclusion criteria were as follows: (1) experimental CHD model was induced by temporary or permanent myocardial infarction (MI); (2) treatment group received the SBP intervention merely; (3) a control group received vehicle or no treatment; (4) the outcome measures were myocardial infarction area and/or VEGF, and/or MVC; (5) regardless of language, blinding, or publication status.

Prespecified exclusion criteria met any one of the following conditions: (1) treatment group without SBP or combined use of any other agents; (2) non-CHD animal model; (3) no control group; (4) duplicate publications; (5) reviews, case reports, abstracts, letters, comments, study protocol, editorials, and clinical guidelines.

Two authors independently screened the abstracts, and the included manuscripts were approved by corresponding author. The information extracted from the complete manuscripts of the qualified studies was as follows: (1) the first author's name and publication year, model of CHD (transient or permanent, ligation or drug); (2) the features of animals such as animal number, species, sex, weight, age, and any comorbidity; (3) the information of treatment used in experimental group including the types of SBP, dose, method of administration, and duration of treatment; (4) outcome measures, especially the primary outcomes containing myocardial infarction area and/or VEGF, and/or MVC. If outcomes were performed at different time points, only the final test was included. If the experimental group of animals received various doses of the drug therapy, only the data of highest dose of the drug was included. If published data were incomplete, we contacted authors to obtain further information. For each comparison, we extracted data of mean value and standard deviation from each experimental and control group of every study.

We evaluated the methodological quality of the included studies by using the Collaborative Approach to Meta-Analysis and Review of Animal Data from Experimental Studies (CAMARADES) with 10-item quality checklist (MacLeod et al., 2004). One point was awarded for each of (1) publication in a peer-reviewed journal; (2) statement of temperature control; (3) random allocation to groups; (4) allocation concealment; (5) blinded assessment of outcome; (6) use of anesthetic without significant internal protection of blood vessel; (7) appropriate animal model (aged, healthy, diabetic, or hypertensive); (8) sample size calculation; (9) compliance with animal welfare regulations; (10) statement of potential conflict of interests. Two authors independently assessed study quality and any disagreements were solved through discussion or consultation with corresponding author.

All values of myocardial infarction area, VEGF and MVC were considered as continuous data, and then an estimate of the combined effect sizes utilizing standard mean difference (SMD) with the random effects model was given. In the present meta-analysis, the results using the random effects model were presented because heterogeneity between multistudies has to be taken into account. I² statistic was used to assess heterogeneity. If significant heterogeneity with I²-values over 50% was identified, then additional subgroup and/or sensitivity analyses were performed. Probability values 0.05 were considered significant. All analyses were performed with Revman version 5.1 provided by the Cochrane Collaboration.

We identified 755 potentially relevant articles from six databases. After removal of duplicates, 639 records remained. After going through the titles and the abstracts, we excluded 459 papers with at least one of following reasons: (1) case report or review; (2) not an animal research; and (3) other diseases. Reading the full text of the 180 articles remaining which reported the efficacy of SBP in animal models of myocardial infarction OR myocardial ischemia, 20 studies were excluded because the outcome measure was not myocardial infarction area or VEGF or MVC; 76 studies were excluded without SBP; 21 were excluded because of inappropriate outcome indicators; 38 studies were removed due to the deficiency of useful data. Ultimately, 25 eligible studies were identified (Figure 1; Wang et al., 2002, 2004, 2006, 2007, 2014; Han et al., 2006, 2007; Huang and Huang, 2006; Ling et al., 2007; Chen and Yuan, 2008; Li et al., 2008; Shen and Fan, 2008; Tian and Wang, 2008; Wang and Fan, 2008; Xie and Chen, 2008; Yang et al., 2010; Zhang Q. Y. et al., 2011; Zhang S. J. et al., 2011; Guo et al., 2013; Huang et al., 2013, 2014; Luan et al., 2013, 2014; Zang et al., 2014; Yuan et al., 2015).

A total of 439 subjects were included in the 25 studies, of whom 228 were in the experimental group and 211 were in the control group. Twenty-one studies were published in Chinese between 2002 and 2015, and seven studies have not been published as Mphil or PhD thesis. Nine studies (Wang et al., 2004, 2014; Wang and Fan, 2008; Xie and Chen, 2008; Zhang Q. Y. et al., 2011; Guo et al., 2013; Huang et al., 2013, 2014; Zang et al., 2014) used male/female Sprague-Dawley rat models; 14 studies (Wang et al., 2004, 2006, 2007; Han et al., 2006, 2007; Huang and Huang, 2006; Chen and Yuan, 2008; Li et al., 2008; Tian and Wang, 2008; Yang et al., 2010; Zhang S. J. et al., 2011; Luan et al., 2013, 2014; Yuan et al., 2015) used male Wistar rats; one study (Shen and Fan, 2008) used male New-Zealand rabbits; one study (Ling et al., 2007) used female Japan-Sino hybridization white rabbits. All the studies used permanent CHD models which were divided into ligation (Wang et al., 2002, 2004, 2006, 2007, 2014; Han et al., 2006, 2007; Huang and Huang, 2006; Ling et al., 2007; Chen and Yuan, 2008; Li et al., 2008; Tian and Wang, 2008; Wang and Fan, 2008; Xie and Chen, 2008; Yang et al., 2010; Zhang Q. Y. et al., 2011; Zhang S. J. et al., 2011; Guo et al., 2013; Huang et al., 2013, 2014; Luan et al., 2013, 2014; Zang et al., 2014; Yuan et al., 2015) and embolism with the obstructive glue (Shen and Fan, 2008) of the left descending coronary artery. The dosage, duration and initial administration time in the studies aren't all the same. The dosage administrated is various as follows: 0.03 g/kg in 11 studies (Wang et al., 2002, 2006, 2007, 2014; Han et al., 2006, 2007; Yang et al., 2010; Zhang Q. Y. et al., 2011; Zhang S. J. et al., 2011; Guo et al., 2013; Yuan et al., 2015); 0.08 g/kg in four studies (Huang and Huang, 2006; Xie and Chen, 2008; Huang et al., 2013, 2014); 0.1 g/kg in two studies (Wang and Fan, 2008; Zang et al., 2014); 0.05 g/kg in two studies (Wang et al., 2004; Ling et al., 2007); 0.0142 g/kg in two studies (Luan et al., 2013, 2014); the remaining four studies different from each other. The durations of administration time are diverse, including 6 weeks in 12 studies (Wang et al., 2002, 2006, 2007; Han et al., 2006, 2007; Shen and Fan, 2008; Xie and Chen, 2008; Yang et al., 2010; Zhang S. J. et al., 2011; Luan et al., 2013, 2014; Huang et al., 2014), 2 weeks in five studies (Huang and Huang, 2006; Ling et al., 2007; Chen and Yuan, 2008; Guo et al., 2013; Wang et al., 2014), 4 weeks in three studies (Tian and Wang, 2008; Huang et al., 2013; Yuan et al., 2015), 8 weeks in two studies (Wang et al., 2004; Wang and Fan, 2008), 1 week in two studies (Li et al., 2008; Zhang Q. Y. et al., 2011) and 3 weeks in one study (Zang et al., 2014). The initial time to be given SBP is as follows: 1 day after myocardial infarction in 23 studies (Wang et al., 2002, 2004, 2006, 2007, 2014; Han et al., 2006, 2007; Huang and Huang, 2006; Ling et al., 2007; Chen and Yuan, 2008; Li et al., 2008; Tian and Wang, 2008; Wang and Fan, 2008; Yang et al., 2010; Zhang Q. Y. et al., 2011; Zhang S. J. et al., 2011; Guo et al., 2013; Huang et al., 2013, 2014; Luan et al., 2013, 2014; Zang et al., 2014; Yuan et al., 2015) and 1 week after myocardial infarction in two studies (Shen and Fan, 2008; Xie and Chen, 2008).

Eight studies (Wang et al., 2002, 2004; Li et al., 2008; Tian and Wang, 2008; Wang and Fan, 2008; Xie and Chen, 2008; Huang et al., 2013, 2014) reported myocardial infarction area, 12 studies (Wang et al., 2002, 2006, 2014; Huang and Huang, 2006; Ling et al., 2007; Tian and Wang, 2008; Xie and Chen, 2008; Yang et al., 2010; Zhang Q. Y. et al., 2011; Guo et al., 2013; Huang et al., 2014; Zang et al., 2014) VEGF protein, five studies (Chen and Yuan, 2008; Li et al., 2008; Shen and Fan, 2008; Zhang S. J. et al., 2011; Luan et al., 2013) VEGF mRNA, 12 studies (Han et al., 2006, 2007; Wang et al., 2006, 2007; Chen and Yuan, 2008; Yang et al., 2010; Zhang Q. Y. et al., 2011; Zhang S. J. et al., 2011; Huang et al., 2013; Luan et al., 2014; Zang et al., 2014; Yuan et al., 2015) MVC. The characteristics of the 25 included studies were summarized in detail in Table 1.

Eighteen studies were publications in a peer reviewed journal. Seventeen studies (Han et al., 2006; Huang and Huang, 2006; Ling et al., 2007; Chen and Yuan, 2008; Li et al., 2008; Shen and Fan, 2008; Tian and Wang, 2008; Xie and Chen, 2008; Yang et al., 2010; Zhang Q. Y. et al., 2011; Guo et al., 2013; Huang et al., 2013, 2014; Luan et al., 2013, 2014; Zang et al., 2014; Yuan et al., 2015) reported the control of temperature. All studies described random allocation to groups, of which two studies (Huang and Huang, 2006; Tian and Wang, 2008) used random number table method. None of the studies used either allocation concealment or blinded assessment of outcome. Chloral hydrate was used as anesthetic in eight studies (Ling et al., 2007; Chen and Yuan, 2008; Tian and Wang, 2008; Yang et al., 2010; Guo et al., 2013; Huang et al., 2013; Wang et al., 2014; Zang et al., 2014); pentobarbital was used in 10 studies (Han et al., 2006, 2007; Wang et al., 2006, 2007; Li et al., 2008; Wang and Fan, 2008; Xie and Chen, 2008; Zhang S. J. et al., 2011; Huang et al., 2014; Yuan et al., 2015); sumianxin (compound preparation, including Xylidinothiazole, Ethylenediamine tetraacetic acid, Dihydroetorphine Hydrochloride, and Haloperidol) was used in one study (Shen and Fan, 2008); ether was used in two studies (Huang and Huang, 2006; Luan et al., 2013, 2014); ketamine hydrochloride was used in two studies (Wang et al., 2002, 2004); ethyl carbamate was used in one study (Wang et al., 2004). No studies described a sample size calculation, and none of studies reported compliance with animal welfare regulations or mentioned a statement of potential conflict of interests. One study (Yuan et al., 2015) chose the healthy rats with the standard II lead ECG normal, one (Guo et al., 2013) chose the healthy adult rats, one (Huang et al., 2014) chose the healthy rats, two (Shen and Fan, 2008; Xie and Chen, 2008) chose the rats high-fat, and the others chose the appropriate animal models but not described the characteristics. The quality score of studies ranges from 2 to 5, and the median was 3.6. The methodological quality of each study was summarized in Table 2.

Twenty-five studies with 439 animals were included in the present study. The findings preliminarily demonstrated that SBP can reduce myocardial infarction area through an increase of VEGF and MVC in CHD, suggesting that SBP exerted potential cardioprotection largely through promoting angiogenesis.

The strengths of the study are that systematic reviews of pre-clinical animal data could inform the planning to focus on the method of randomization to treatment group, blinded assessment of outcome, allocation concealment, and explicit sample size calculation (MacLeod et al., 2004), and to affirm the optimum dosage, duration and initial administration time identified for further “basic” research. Moreover, they could preclude unnecessary study replication, and contribute to both “reduction” and “refinement” in animal experimentation, improve the likelihood of success of future clinical trials (Murphy and Murphy, 2010). In the present study, we provided a novel study method for synthesizing animal studies that can independently evaluate the current preclinical evidence and possible mechanism of SBP for CHD.

Some limitations of the study are listed as follows: previous meta-analyses have suggested that animal studies, less rigorously designed, may overestimate treatment effects (Rooke et al., 2011). In the present study, all the studies failed to mention the blinded assessment of behavioral outcome, which made the overall quality of the containing studies moderate. It may lead to performance bias and detection bias (Kahan et al., 2014). Thus, the results should be interpreted with caution. Secondly, possible heterogeneity problems are as follows: different conditions include different animal species, dosage, duration, initial administration time, or administration route and various assessment methods of treatment effects, including the myocardial infarction area, VEGF, and MVC. Third, the patients suffered from CHD often with comorbidity. However, in the present study animal models used were healthy that were inconsistent with humans. Fourth, our search strategy included only Chinese and English databases leading to selective bias to some certain degrees (Guyatt et al., 2011). Finally, the primary aim of the studies reviewed was not to assess the angiogenesis effects of CHD. Some studies did not include an impartial measurement of myocardial infarction area, VEGF and/or MVC which may lead to bias in their outcome, or did not state the method of measurement.

A forest plot is a graphical representation of the individual results of each study included in a meta-analysis together with the combined meta-analysis result according to the Cochrane Collaboration's definition (2014). The forest plot is able to demonstrate the degree to which data from multiple studies observing the same effect overlap with one another, allowing readers to see the heterogeneity among the results of the studies. Thus, the forest plot provides a quick visual representation of overall effect estimates and the heterogeneity (Callcut and Branson, 2009; Israel and Richter, 2011). The sensitivity analysis is based on Cochrane Reviews which is used to guide systematic reviews of intervention (Verhagen and Ferreira, 2014). A sensitivity analysis is a repeat of the primary analysis or meta-analysis, substituting alternative decisions or ranges of values for decisions that were arbitrary or unclear. In the present study, first of all, timing of initiation of treatment contributed to the heterogeneity according to myocardial infarction area outcome measure. Xie's study (Xie and Chen, 2008) administrated SBP after 1 week the models established other than after 1 day in the remaining seven studies (Wang et al., 2002, 2004; Li et al., 2008; Tian and Wang, 2008; Wang and Fan, 2008; Huang et al., 2013, 2014). Second, based on the VEGF outcome measure, a test method of the VEGF expression is undefinite in the study by Wang et al. (2006) while the remaining four studies (Wang et al., 2002; Huang and Huang, 2006; Tian and Wang, 2008; Yang et al., 2010) are clearly provided it; the duration of the administration by the study of Zhang Q. Y. et al. (2011) is shorter than that of the other five studies (Zhang Q. Y. et al., 2011; Guo et al., 2013; Huang et al., 2014; Wang et al., 2014; Zang et al., 2014); Shen et al. (Shen and Fan, 2008) used the embolic model with the obstructive glue while the other four studies (Chen and Yuan, 2008; Li et al., 2008; Zhang S. J. et al., 2011; Luan et al., 2013) used the ligate. Third, the duration of the administration by two studies (Chen and Yuan, 2008; Zhang Q. Y. et al., 2011) is shorter than the other 10 studies (Han et al., 2006, 2007; Wang et al., 2006, 2007; Yang et al., 2010; Zhang S. J. et al., 2011; Huang et al., 2013; Luan et al., 2014; Zang et al., 2014; Yuan et al., 2015) according to MVC outcome measure.

Previous studies (Murphy and Murphy, 2010; Wei et al., 2013) suggested that the quality of the research design is an important factor affecting the outcome. A lower-quality study trends toward better outcomes, leading to the global estimated effect overstated (García-Bonilla et al., 2012). In the present study, the quality need be promoted by means of incorporating the ARRIVE guidelines (Kilkenny et al., 2010). In particular, we should focus on the method of randomization to treatment group, blinded assessment of outcome, allocation concealment and explicit sample size calculation (MacLeod et al., 2004). Second, according to the effect size, this study indicated that a comparison among varieties of the administration duration showed the shorter administration duration using SBP was less effectiveness in MVC improvement. The young animals distinguish from the pathology of CHD, a slowly developing and chronic disease, and with frequently co-morbidities such as diabetes, hypertension, atherosclerosis, or advanced age (Gianaros and Sheu, 2009; Mohamed Omer et al., 2016). Therefore, selecting suitable timing of initiation of treatment, the duration of the administration, the optimal animal models and a standardized test of VEGF expression are required in the future.

Ischemia and hypoxia of the myocardium induced by acute or chronic CHD reduces myocardial infarction area and promotes the angiogenesis by up-regulating VEGF (Ramakrishnan et al., 2014; Möbius-Winkler et al., 2016). Additionally, ischemic myocardium initiates a severe inflammatory response indirectly to promote the angiogenesis (Braunwald, 2013). The blood flow is interrupted so that the myocardium lacks the supply of the oxygen and the nutrition. Hypoxia induces inflammation, wherein inflammation causes hypoxia (Braunwald, 2013). Hypoxia inducible factor (HIF) is an essential factor induced in hypoxia condition which transactivates or transcriptionally regulates many hypoxia responsive genes such as VEGF (Leung et al., 1989). Then VEGF sites on endothelial cells as a stimulatory factor for proliferation, sprouting, migration, and luminal formation (Des Guetz et al., 2006; Konopka et al., 2013) and inhibits the calcium sensitive receptor to the apoptosis on myocardial ischemia from reperfusion injury (Hoffmann et al., 2013). Nevertheless, VEGF improves new blood vessels and enhances collateral development termed angiogenesis to guarantee the microcirculation establishment for the requirement of the ischemic myocardium (Banai et al., 1994; Tammali et al., 2011). In this study, SBP could reduce myocardial infarction area, up-regulate VEGF in the edge of myocardial infarction (Zhu et al., 2010), and increase blood vessel density, which promotes blood circulation to protect ischemic myocardium (Xiang et al., 2012). In addition, animal experiments have contributed to our understanding of angiogenesis for CHD by the signal way of VEGF or other growth factors (Hackam and Redelmeier, 2006; Hackam, 2007). Further investigation is needed to determine the specific signal way of angiogenesis. And the underlying molecular mechanisms also require further study.

Multiple active compounds in essence combined to enhance the effectiveness of TCM therapy (Zhou and Wang, 2014). SBP, known as the polypill, can be adapted for secondary prevention of cardiovascular disease based on existing evidence confirming that SBP can reduce the ischemia myocardium to keep the myocardial function and promote the angiogenesis to increase the blood flow (Lafeber et al., 2013; Working Group on the Summit on Combination Therapy for CVD et al., 2014). SBP also plays an important role in anti-thrombosis and anti-artherosclerosis (Zhou and Wang, 2014). SBP for the second prevention of CHD need to be further confirmed by RCTs.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewers OC and LX and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review.