Effectiveness and Safety of Compound Chinese Medicine plus Routine Western Medicine in In-Stent Restenosis: A Meta-Analysis and Systematic Review

Objective To examine the effects and safety of oral compound Chinese medicine (CCM) plus routine western medicine (RWM) in in-stent restenosis (ISR). Methods Various electronic databases (CBM, CNKI, VIP, Wanfang, PubMed, EMBASE, and Cochrane Library) were searched until April 2017. The quality of the included studies was evaluated, and meta-analyses were performed using RevMan5.3 and STATA 12.0 software. Moreover, funnel plot and Egger's publication bias plots were analysed to identify publication bias and adverse reactions were reported. A sensitive analysis was carried out according to the quality score. Results In all, 40 RCTs involving 4536 patients were selected for this review. The pooled estimates of three studies showed that the benefit to the number of ISRs (NoR) was more substantial for CCM plus RWM than for RWM alone (RR 0.24, 95% CI 0.10 to 0.57, P = 0.001; I2 = 0%, P = 0.81). The rate of ISR was significantly lower for CCM plus RWM than for the same RWM alone (RR 0.44, 95% CI 0.37 to 0.53, P < 0.00001; I2 = 0%, P = 0.95). CCM plus RWM benefitted the rate of ISR when a CM placebo plus RWM was used as the control intervention (RR 0.34, 95% CI 0.20 to 0.57, P < 0.0001; I2 = 0%, P = 0.95). The difference of adverse reactions was not significant. For secondary outcomes, the CCM plus RWM group did not reduce the rates of revascularization and cardiac death, but it did reduce the rate of recurrent angina over the results observed in the RWM alone group. In addition, funnel plot and Egger's publication bias plot indicated that there was publication bias. The association between the use of CCM plus RWM and RWM alone remained significant after the sensitivity analysis excluding studies with low quality score (quality score ⩽ 4) with a pooled RR of 0.41 (95% CI, 0.34–0.50). Conclusion Oral CCM plus RWM clearly benefitted patients with percutaneous coronary intervention (PCI) because it prevented and treated ISR better than was observed for either RWM alone or a CM placebo plus RWM.


Introduction
Currently, percutaneous coronary intervention (PCI) is widely used around the world to treat coronary artery disease (CAD) and has significantly reduced mortality in patients with acute coronary syndromes [1]. However, the incidence rate of in-stent restenosis (ISR) is still approximately 10%, even accounting for the introduction of drug-eluting stents (DES) [2], and reaches as high as 40%-50% in patients with multivessel involvement [3]. ISR therefore inflicts a heavy burden on both the lives of patients with CAD and the economy and remains a challenge to the implantation of PCI and CAD prognoses.
In traditional Chinese medicine (TCM), the typical symptoms of ISR, such as chest pain and chest tightness, are referred to as "Xiongbi" [4]. TCM presents mature theories and is supported by abundant clinical experience for treating "Xiongbi". In recent years, the efficacy and safety of TCM in 2 Evidence-Based Complementary and Alternative Medicine ISR has been widely studied with remarkable results. Studies have shown that TCM not only clearly alleviates the typical symptoms of ISR but also improves its long-term prognosis, suggesting that TCM may lead to promising applications for the treatment of ISR.
Previous reviews focused on this issue were published in 2008 [5], 2012 [6], and 2014 [7]. The first two of these studies were systematic reviews (SRs), and the latter was a metaanalysis. They explored the effects of many types of PCI and included studies with small sample sizes (fewer than 30 cases) and that explored a single TCM herb. In addition, a presearch showed that, in the 3 years since 2014, more studies have supported the efficacy and safety of oral compound Chinese medicines (CCMs) in ISR. In this SR, we applied more rigorous inclusion and exclusion criteria and aimed to explore whether oral CCM plus routine western medicine (RWM) is effective and safe for treating ISR after stent implantation. We use CCM because it has wider applications in clinical practice and conforms better to TCM theory.

Methods
This SR was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Statement [8] and the Cochrane Handbook [9] Systematic review and is registered at PROSPERO with the registration number CRD42017075368 (https://www.crd.york.ac.uk/ PROSPERO).

Search Strategy.
A comprehensive search of 7 medical databases, including the PubMed, EMBASE, Cochrane Library, China National Knowledge Infrastructure (CNKI), Chinese Scientific Journals (VIP), Wanfang, and Chinese Biomedical (CBM) databases, was conducted through April 2017 without a language restriction. We performed the search using individually or combined Mesh terms for all fields relating to the patients (coronary heart disease and coronary artery disease) and interventions (traditional Chinese medicine, Chinese herbal medicine, compound Chinese medicine, Chinese prepared medicine, Chinese herbal formula, and Chinese herbs) of interest. All titles/subjects related to the outcome (in-stent restenosis) were searched. When searching Chinese databases, the above terms were searched in Chinese.

Inclusion Criteria.
Only randomized controlled trials (RCTs) were included. We focused on trials with participants diagnosed with major angiographic criteria-documented [10] CAD who were eligible for stent implantation (with either bare metal stents or drug-eluting stents) regardless of their gender, age, disease course, comorbidity, and ethnic origin. The baseline characteristics of each study were consistent. The interventional treatment was any oral CCM plus RWM administered for at least 1 month regardless of dosage. The control group was treated with the same RWM or with a CM placebo plus the same RWM. "CCM" included Chinese prepared medicines or Chinese herbal formulas. The compositions of the CCM were detailed. Our primary outcomes were restenosis occurring within at least 6 months of followup. Coronary in-stent restenosis is classically defined as the angiographic detection of a recurrent stenosis with a diameter greater than 50% at the stent segment or the 5 mm segments adjacent to it (in-segment restenosis). The drop-out rate for coronary angiography (CA) review was required to be lower than 20%. Each included study was required to report the primary outcome. Secondary outcomes could be major adverse cardiac events (MACE), including recurrent angina, myocardial infarction, revascularization, and cardiac death. Only secondary outcomes that improved in equal to or more than 50% of cases were considered responses. In addition, adverse reactions were considered as a primary safety outcome. When there were several follow-up points, only the last one was considered.

Exclusion Criteria.
Studies with a small sample study (less than 30 cases), duplicate reports, and pilot studies were excluded. Any trial that failed to satisfy the inclusion criteria or for which required data were unavailable was excluded. Li) independently extracted the data using a standardized extraction form. Disagreements were resolved by census or consultation with a third reviewer (Wenhao Jia). The following items were extracted: study name, year of publication, sample size, details of the trial design (i.e., randomization, allocation concealment, and blinding), eligibility criteria, general characteristics of patients, details of intervention and control therapies, details related to outcomes, details related to drop-outs, and other information that may help detect bias.

Quality Assessment.
Two reviewers (Xiaoyun Cui and Kun Zhou) independently assessed the methodological quality of the included studies using the Cochrane Collaboration's tool [11], which is a domain-based evaluation tool used to generate a "risk of bias" table for each study. Any disagreement was resolved by consensus or consultation with a third reviewer (Yanchao Huang). The domains used for assessment were sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective outcome reporting (reporting bias), and other potential sources of bias (e.g., early termination, contamination, and conflict of interest). Because it is difficult to blind participants in studies of Chinese herbal medicines, performance bias was likely present in all trials. With regard to the objective outcomes adopted in this SR, such as ISR and MACE, performance bias and detection bias might not be as important, and we therefore summarized these trials as low-risk. In addition, with regard to selective outcomes, this domain was ranked as "low-risk" unless the outcomes were critical to our issue.
2.6. Data Analysis. Data synthesis and analysis were performed with RevMan software 5.3 and STATA 12.0 software (StataCorp LP, College Station, TX) [9]. Dichotomous data Evidence-Based Complementary and Alternative Medicine 3 12 records identified through searching 3 English databases 2,186 records identifi ed through searching 4 Chinese databases 1,231 records screened out by reading their titles or abstracts 74 excluded: -Non-RCT or quasi-RCT (n=9) -follow up < 6 months (n=4) -drop-outs of CA review > 20% or failing to report the prespecified primary outcome (n=42) -failing to report the composition of CCM (n=3) -control group combining with CM injection or another CCM (n=2) -review (n=3) -treatment duration < 1 month or unclear (n=2) -unavailable data (n=3) -patients without stent implantation (n=3) -small sample study (n=2) -unclear baseline characteristics(n=1)  were measured as relative risk (RR), while continuous data were measured as the mean difference (MD), both with corresponding 95% confidence intervals (CIs). Heterogeneity across trials was measured with the Cochran test and is presented as 2 statistics. Only trials with 2 lower than 85% were used for the meta-analysis, and the characteristics of the included trials were similar. A fixed effect model was used if 2 was lower than 25%. Otherwise, a random effect model was applied under the assumption that any heterogeneity was readily explainable. Furthermore, funnel plot and Egger's test were performed to detect heterogeneity and publication bias, respectively, if sufficient sources were available. A 2-tailed value less than 0.05 was considered significant. A sensitivity analysis was conducted to assess the stability of the results.

Study Identification.
All eligible studies were screened and identified (Figure 1). A total of 2198 records were retrieved. Of these, full-text evaluations were conducted on 114 studies. In all, 74 of these 114 studies were excluded for the following reasons: non-RCT or quasi-RCT ( = 9); followup < 6 months ( = 4); drop-outs from CA review > 20% or failure to report the prespecified primary outcome ( = 42); failure to report the composition of the CCM ( = 3); control group consisted of a CM injection or another CCM ( = 2); review ( = 3); treatment duration < 1 month or unclear ( = 2); unavailable data ( = 3); patients without stent implantation ( = 3); small study sample size ( = 2); and unclear baseline characteristics ( = 1). Finally, a total of 40 RCTs with a total of 4536 patients were included in this SR .

Characteristics of Included RCTs.
All studies were conducted in China from 2010 to 2017. The average age of the patients ranged from 52.93 to 69.27 years old. Most trials had more males than females. The diagnostic criteria for CAD were mainly based on CA criteria or the Nomenclature and Diagnostic Criteria for Ischemic Heart Disease (World Health Organization, WHO) [52]. All patients successfully underwent stent implantation. The distributions of baseline characteristics were basically the same in each group. Four trials [25,26,36,49] were double-blinded, and, in three trials, the control group used a CM placebo plus RWM [19,25,26]. The remaining trials were designed to compare a CCM plus RWM group to a group treated with same RWM alone. Of the 40 studies, 20 used a decoction, 1 used an oral liquid, 2 used granules, 4 used pills, 12 used capsules, and 1 used tablets. RWM mainly consisted of aspirin, Clopidogrel, angiotensinconverting inhibitor (ACEI)/angiotensin II receptor blocker (ARB), beta-blockers, and statins. The dosage and types of RWM were prescribed according to the recommendations of Chinese Society of Cardiology Guideline [10]. Overall, 33 kinds of CCM were used, and the treatment courses ranged from 3 to 12 months with a follow-up time of 6 to 18 months. For outcomes, ISR was assessed by computed tomography angiography (CTA) or CA in all trials; adverse reactions caused by CCMs were reported in 11 studies but were significant in only 2 studies [26,38]; MACE were reported in 12 studies and included recurrent angina ( = 12), myocardial infarction ( = 3), revascularization ( = 2), and cardiac death ( = 5) (Tables 1 and 2).

Methodological Quality of Included RCTs.
Of the 40 included studies, 13 were randomized by random number         tables or SPSS software. The other 27 studies used the phrase "randomly allocating" but did not describe the method of randomization. Only 1 trial reported allocation concealment as "by sealed, opaque envelopes". Participants or outcome assessors were blinded in only 4 trials. All studies reported the primary outcome and clearly described the data collection methods, and we therefore believed that they were free from selective reporting bias. According to the criteria we prespecified, as shown in the Quality Assessment section, we judged the performance bias and detection bias of these studies to be low-risk. In conclusion, the overall methodological quality of the 40 trials was rated as low-risk. The details of this analysis are shown in Figure 2. The mean score for the RCTs included in this analysis was 5.15 (Table 3).

In-Stent Restenosis.
Three studies provided data for the number of ISR (NoR), and the pooled estimates showed that, in patients with CAD, NoR was lower for CCM plus RWM than for RWM alone (RR 0.24, 95% CI 0.10 to 0.57, = 0.001; 2 = 0%, = 0.81) after 6 months of follow-up. In all, 34 studies provided data on the number of cases of ISR that had a follow-up time of 6 to 18 months. The rate of ISR was obviously lower in the CCM plus RWM group than in the group treated with the same RWM alone (RR 0.44, 95% CI 0.37 to 0.53, < 0.00001; 2 = 0%, = 0.95). A metaanalysis of an additional 3 studies also showed that CCM plus RWM exerted a beneficial effect on the rate of ISR when CM placebo plus RWM was used as the control intervention with a 6-to 12-month follow-up (RR 0.34, 95% CI 0.20 to 0.57, < 0.0001; 2 = 0%, = 0.95). In addition, we evaluated these 34 studies to detect publication bias. The funnel plot was asymmetrical, indicating publication bias. Because funnel plot is used as a qualitative method to detect the publication bias of an article, we also used Egger's method to detect publication bias, and the results again implied the presence of publication bias ( = 0.00 < 0.05). The details of these analyses are shown in   Table 4.
Difference in the dosage forms of drugs (e.g., decoction, granules, pills, capsules, and tablets) might impact results, and we therefore carried out a subgroup analysis. A pooled estimate of cases of ISR showed that there was a significant difference between the decoction plus RWM and RWM alone groups in 18 studies (RR 0.28, 95% CI = 0.20 to 0.38, < 0.00001; 2 = 0%, = 0.97). No additional benefit was      exerted in ISR in the granules plus RWM group than in the RWM alone group (RR 0.39, 95% CI 0.13 to 1.21, = 0.10; 2 = 0%, = 0.46). A meta-analysis of another 3 studies also showed that there was no significant difference between the pills plus RWM group and the RWM alone group (RR 0.59, 95% CI 0.34 to 1.01, = 0.60; 2 = 0%, = 0.89). There were significantly fewer cases of ISR in the capsules plus RWM group than in the group treated with the same RWM alone (RR 0.49, 95% CI 0.36 to 0.67, < 0.00001; 2 = 0%, = 0.97). The details of these analyses are shown in Table 6 and Figure 7. The association between the use of CCM plus RWM and RWM alone remained significant after the sensitivity analysis excluding studies with low quality score (quality score ⩽ 4) with a pooled RR of 0.41 (95% CI, 0.34-0.50). (Figure 8).

Adverse Reactions.
Only 2 studies reported the details of adverse reactions related to CCM, including hepatic dysfunction, gingival bleeding, diarrhoea, stomach discomfort,  flushing, dizziness, and headache. More adverse reactions were reported in patients treated with a Huoxue yiqi decoction and a Shenshao decoction, but most of these reactions were not severe and disappeared without special treatment.
There was no significant difference in adverse reactions between the CCM plus RWM and RWM alone groups (Table 7 and Figure 9).  Evidence-Based Complementary and Alternative Medicine       (Table 8) Revascularization. Revascularization was reported in 3 studies, in which 308 patients were treated with 3 different CCMs, including Anxin granules, Shexiang baoxin pills, and a Huoxue yiqi tongmai decoction. After 6-18 months of follow-up, the rate of revascularization was not significantly different between the CCM plus RWM and the RWM alone groups (RR 0.33, 95% CI 0.07 to 1.62, = 0.17; 2 = 0%, = 0.89) ( Figure 10).   Angiographic Measurements. Follow-up angiography performed in patients with ISR at 6-12 months after the index PCI in 7 studies that evaluated 4 Chinese Herbal Medicines (CHMs) in 1917 patients. Baseline information showed that the mean minimal luminal diameter (MLD) before and immediately after the index PCI and the gain in luminal diameter following stent placement were comparable between groups. MLD is defined as the smallest diameter in the treated lesion area [53] and was reported in 5 studies (Table 9). MLD was significantly better in the CCM plus RWM group than  The primary angiographic endpoint was late lumen loss (LLL) after 6 to 12 months. LLL was evaluated by determining the difference between the minimum lumen diameter after the procedure and at follow-up using quantitative coronary angiography [54]. LLL was measured in 4 studies of 3 CHMs after 6 and 12 months of follow-up (Table 10). There was a significant difference between the CHM plus RWM group and the control group treated with the same RWM alone when the CHM being tested was Tongxinluo capsules (MD LU HW2014 = −0.36 mm, 95% CI −0.42 to −0.30 mm), a Xinmai futong decoction (MD ZHANG PF2014 = −0.3 mm, 95% CI −0.41 to −0.19 mm) and a Shenshao oral lotion (MD ZHANG Q2013 = −0.05 mm, 95% CI −0.09 to −0.01 mm). Finally, LLL was lower in the RWM plus Tongxinluo capsule group than in the group treated with the same RWM plus placebo (MD YANGN TL2016 = −0.24 mm, 95% CI −0.28 to −0.20 mm).

Myocardial
In this SR, 40 studies involving 4536 CAD patients who underwent stent implantation were identified. All studies were RCTs. Approximately 75% of the studies were performed in the last five years. A domain-based evaluation showed that the mean score of the included RCTs was 5.15. Of the included studies, three were designed to compare a CCM plus RWM versus a CM placebo plus the same RWM, and 37 studies were designed to compare a CCM plus RWM versus the same RWM alone.
In this SR, all studies assessed ISR using CTA or CA. A three subgroup meta-analysis of 33 CCM plus RWM groups showed that they produced an absolute decrease in the ISR rate. Given the low risk of bias (demonstrated by analyses of methodological quality and publication bias), the following robust conclusion can be drawn: CCM plus RWM reduces ISR. Furthermore, there was no significant difference in adverse reactions between the CCM plus RWM and RWM alone groups in 2 of the studies, potentially indicating 24 Evidence-Based Complementary and Alternative Medicine that CCM plus RWM is a safe treatment. However, further observations are needed. An Egger's linear regression test was performed to statistically assess funnel plot asymmetry and publication bias. Substantially more studies in the literature have positive than negative results, and this produced socalled publication bias.
With regard to secondary outcomes, we concluded that CCM plus RWM did not exert a benefit against revascularization, MI and cardiac mortality during a 6-18-month follow-up period after stent implantation. This finding may be attributed to the limited number of patients and the short follow-up times reported in the studies. However, 11 studies explored CCMs plus RWMs, and recurrent angina was significantly reduced by this treatment during the 6-18month follow-up period. This may indicate that treatment with a CCM plus RWM exerts a better effect on symptom improvement. Some studies have concluded that a major determinant of the restenosis rate is the final MLD after all interventions [55]. MLD can be evaluated as a continuous end point from 6 to 9 months after intervention and is precisely associated with stable long-term results in the treated coronary segment [56]. In addition, LLL is monotonically correlated with the probability of restenosis and, when evaluated, is a more efficient estimate of restenosis progress in this era of lower binary restenosis rates [54]. Hence, MLD and LLL are two major determinants of the restenosis rate. In this SR, 7 studies that evaluated 4 CHMs reported MLD and LLL. Although pooled results were not available for either MLD or LLL because there was high heterogeneity ( 2 = 97% or 2 = 99%), in most of the studies, CCM plus RWM exerted significant beneficial effects.
The pathogenesis of restenosis is not yet fully understood. However, it is generally accepted that when a stent causes injury to a coronary artery, it triggers a series of inflammatory reactions that result in the migration and proliferation of vascular smooth muscle cells (SMCs) within the vessel lumen and neointimal hyperplasia. In TCM, CAD is thought to be a result of "Qi deficiency and Blood Stasis". The severity of Blood Stasis syndrome is significantly correlated with the complexity of coronary lesions and the degree of stenosis and is an important factor affecting the occurrence of restenosis after PCI [57]. Many studies have demonstrated that Astragalus [58], salvia [59], tetramethylpyrazine [60], and Chuangxingol and paeoniflorin [61], which are main ingredients in TCM, play effective anti-inflammatory roles and inhibit SMC proliferation and migration. This may be the mechanism by which the above-mentioned CCMs prevent and treat restenosis.
Ideally, RCTs should adhere to known research design standards. In our examination of these studies, we did not have access to enough details related to these characteristics, although the baselines were comparable between comparison groups in the included studies. Details about randomization methodology were also lacking. In the 40 studies we reviewed, 11 reported that they implemented randomization using a random number table or a computer random number generator, such as SPSS software, and only 1 mentioned using sealed, opaque envelope concealment. Participants or outcome assessors were blinded in only 4 of the trials. In addition, studies that involve therapeutic trials should also report the rates of adverse reactions regardless of whether or not they occur. Reporting of adverse reactions is very important for evaluating the safety of interventional measures even though there is no guarantee that the adverse reactions were related to the interventional measure. Subgroup analyses showed that different types of Western medicines and drug forms affected outcomes. Furthermore, most of the 40 studies failed to mention the type of stent that was deployed (i.e., whether it was a bare metal stent or a drugeluting stent). Therefore, the difference in the curative effects observed between the CCM plus RWM and control groups is uncertain, and there is bias in our results. In our review, only 11 of the 40 included trials reported adverse reactions, rendering it difficult to systematically evaluate the safety of the evaluated CCMs in ISR. Future research on these topics will help to clarify the effectiveness and safety of CCMs in ISR.

Conclusions
Based on the results of this SR, we conclude combining RWM with CCMs may provide moderate efficacy in preventing ISR following PCI with stent placement. This is despite the fact that our investigation revealed the potential presence of bias in the identified studies. CCMs used to supplement Qi and activate the blood are suggested for preventing and treating restenosis. In addition, additional experimental studies should explore the mechanisms by which the main ingredients in CCM act to supplement Qi or activating the blood to prevent and treat restenosis. Future rigorously designed RCTs that explore CCM plus RWM therapies aimed at preventing post-PCI ISR should adhere to established design standards to overcome the limitations presented in this review. In particular, they should ensure adequate concealment of allocation and blinding of primary outcomes assessors.
Yan Li, Xiaoyun Cui, and Kun Zhou established the Data Management Committee and were separately responsible for study selection, data extraction, and risk of bias assessment procedures. Wenhao Jia, Yanchao Huang, and Wenbai Qu took charge of coordinating each process of this study, settling disagreements, and ensuring data safety. Qian Lin and Lu Liu contributed to the design of the statistical analyses. Lu Liu and Yang Wu were responsible for the overall data synthesis and analysis. Jing Liu contributed to polishing the language of the text. Qian Lin was responsible for the overall quality of this SR. All authors have read this manuscript and approved the publication of this SR.