Introduction

Stroke is a major global health issue, and it is a leading cause for mortality and adult disability [1–4]. In China, up to 1.6 million people were dead for stroke every year [5], which is a huge burden for society and patients’ family. IS accounts for about 70 % of total stroke patients [6]; recurrent stroke and stroke deterioration often come in the acute phase of IS occurrence. Without proper treatment, the probability of a recurrent stroke after the first stroke is about 3–10 % in the first month and about 5–14 % in the first year [7, 8]. Antithrombotic therapy can effectively prevent recurrence stroke or stroke deterioration, which significantly improves the prognosis of stroke patients.

Aspirin is a mainstay for antithrombotic therapy, two large RCTs of aspirin in acute IS reported that aspirin reduced the odds of early recurrent stroke by about 12 % [odds ratio (OR) 0.88, 95 % CI 0.79–0.97] and the odds of death or dependency at the end of follow-up by about 5 % (OR 0.95, 0.91–0.99) [9]. In the medication guideline of myocardial infraction (MI) published in June 2016 [10], dual-antiplatelet regimen is strongly recommended for MI patients or patients after percutaneous coronary intervention (PCI). Different types of antiplatelet drugs have distinct antithrombotic mechanisms; combination of these may strengthen antithrombotic effect and further improve the treatment of IS. Stroke has a similar formation mechanism to MI, it is reasonable and proper to hypothesize that DAPT would bring more benefits than mono-antiplatelet therapy in treating IS.

There are many trials [11–20] and meta-analyses [21–23] conducted to investigate the efficacy and safety of dual-antiplatelet therapy. FASTER [16] is a randomized controlled trial comparing the effect and safety of dual-antiplatelet therapy in treating stroke or TIA; their outcomes showed that DAPT is not superior to monotherapy in treating IS or TIA. A few meta-analyses [21, 22] were conducted in recent years, but they did not focus on short-term regimen and they only included the combination of aspirin and clopidogrel. Both analyses indicated that short-term dual therapy is better than monotherapy, but they also suspected that beneficial outcomes may be related to timely treatment initiation time. A meta-analysis [23] also showed that the acute phase is critical for treatment of stroke patients. Our meta-analysis aims to further justify the efficacy and safety of comprehensive DAPT in treating stroke patients and test if the beneficial outcomes associated with initiation treatment time.

This meta-analysis complied with preferred reporting items for systematic review and meta-analysis (PRISMA) [24] and was registered in PROSPERO website (CRD42016033680).

Methods

Data source and searching

We made a systematic search in databases of Pubmed, Embase and Cochrane library through July 2016, using the following medical subject heading (MeSH) and free-text terms: stroke, cerebral infraction, cerebrovascular disease, TIA, aspirin, clopidogrel, cilostazol, dipyridamole, tirofiban, ticlopidine, triflusal, terutrobran, aspirin and clopidogrel, aspirin and dipyridamole, aspirin and cilostazol, aspirin and ticagrelor, aspirin and ticlopidine, aspirin and triflusal, aspirin and terutobran, dual-antiplatelet, mono-antiplatelet. No other search restrictions were applied. To find out newly developed clinical trials, we searched the clinicaltrials.gov. Finally, we also searched references of former meta-analyses and trials for additional trials which were not identified in databases.

Study selection

Two researchers (Y. L. and Zx. F.) independently selected eligible studies which are included in our meta-analysis. If there exists a disagreement, they would resolve it by consulting another researcher (Jx. F.). Inclusion criteria were listed as following: (1) randomized controlled trials; (2) DAPT versus monotherapy in any doses were assessed; (3) treatment duration is no more than 1 year; (4) patients involved were with a clinical diagnosis of IS or TIA; (5) sufficient data for outcomes were provided. Papers were excluded if they are (1) non-RCTs; (2) papers only with abstract; (3) anticoagulant drugs, like warfarin, were also tested in the trial; (4) case report; (5) retrospective studies. If several papers have published about one trial, the paper which contains more detailed information needed was included in our meta-analysis.

Data extraction

We abstracted the information in included studies from three aspects: the baseline characteristics of included trials, the baseline characteristics of participants, and the basic outcomes. Two researchers (W. W. and Mj. Z.) independently abstracted the needed information, if there exists a disagreement, they will reach a consensus by discussing with other researchers (G. C., Y. L. and Jx. F.). We collected the following information in each trial: trial name, country, sample size, blinding, treatment group and dosage, intention to treatment (ITT) analysis, DAPT treatment duration, follow-up, mean age, gender percentage, hypertension and diabetes mellitus (DM) percentage, patients disease, treatment initiation time, and loss to follow-up. We also abstracted following trials’ outcomes: recurrent stroke, the net clinical outcome, IS recurrence, TIA, composite outcome of major vascular events, bleeding episodes, major bleeding, and intracranial hemorrhage. The information on major bleeding and intracranial hemorrhage was collected according to the definition in each study. We counted Jadad score of trials to make a primary assessment of each trial. Composite outcome of major vascular events include myocardial infraction, stroke and vascular death. If included studies did not report the composite outcome of major vascular events while gave the separated information on myocardial infraction, stroke and vascular death, we will add these figures together to calculate this outcome.

Based on the Cochrane collaboration tool for assessing risk, we abstracted following information to further assess the quality of included studies: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting and other forms of bias. Authors of these included studies were not contacted for additional information.

Quality assessment

With Cochrane collaboration tools, we assessed bias of included studies in following seven aspects: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting and other forms of bias. For the limitation of sample size, we assessed the publication bias by visual funnel plot without conducting the Egger’s or Begg’s test. Finally, we made recommendation ranks of studied outcomes by Grading of Recommendations Assessment, Development, and Evaluation (GRADE) systems. Two authors (Y. L. and Zx. F.) independently carried out the risk of bias assessment and quality assessment of evidence, and any differences were resolved by discussion.

Statistical analysis

The primary efficacy outcomes are recurrent stroke and the net clinical outcomes, and the main safety indexes are major bleeding and intracranial hemorrhage. We made subgroup analysis according to races, treatment initiation time and different combinations of DAPT.

RR of all outcomes was calculated with 95 % CI. Two-tailed P < 0.05 was considered statistically significant. Heterogeneity was assessed by Cochrane’s Chi-square test, P < 0.10 and I 2 > 50 % was considered significant heterogeneity. Pooled analyses were conducted using a fixed-effects model, whereas a random-effects model was applied if there was heterogeneity (P < 0.10 and I 2 > 50 %). All the analyses were conducted by Revman5.2 software (Nordic Cochrane Centre, Cochrane Collaboration, 2013) and sensitivity analysis was conducted by Stata 11.0 (Stata Corp, College Station, TX, USA). With STATA software, we tested the robustness of primary outcome by omitting the included studies once a time.

Results

Search results

We identified 3581 potentially eligible records and included 10 studies that met our inclusion criteria. The selected procedure is shown in Fig. 1. There are totally 8969 patients involved in our meta-analysis: 4481 patients were randomized to DAPT (experimental group) and 4488 patients were randomized to monotherapy (control group). The characteristics of the included trials are described in Tables 1 and 2. All trials were published between 2005 and 2014. Five [11–13, 16, 20] of ten trials are double-blind design, four trials [14, 17–19] were open-label design but they were blindly assessed. All trials but two [18, 19] are ITT analysis. DAPT treatment duration ranges from 7 days to 6 months, and the onset-to-treatment interval ranges from 24 h to 3 months. Among ten trials, seven trials [11–17] were allocated to receive aspirin and clopidogrel; the other three trials [18–20] were allocated to aspirin and dipyridamole. The bias assessment of all ten trials is detailed in Fig. 2.

Fig. 1
figure 1

Flowchart illustrating the process used to select the studies included in this meta-analysis

Full size image
Table 1 Baseline characteristics of included studies
Full size table
Table 2 Baseline characteristics of participants
Full size table
Fig. 2
figure 2

Risk of bias graph and summary for included studies

Full size image

Risk of bias in included studies, and quality of evidence

Eight studies [11, 13–18, 20] explicitly described the random sequence generation, mainly by a computer random number generator, a random number table, choosing marked ball, or an interactive voice response system. Eight studies [11, 13–18, 20] all used unpredicted methods to generate random sequence which stated a low risk of allocation concealment process, and two studies [12, 19] which lack random sequence generation method were regarded as having an unclear risk of bias in this domain. There are five trials [11–13, 16, 20] having double-blind and blind assessment design, the trials’ risk of these two domains was low. Four open-label trials [14, 17–19] have adopted blind assessment design and the open-label design has little influence on final outcomes, so the risk of performance bias for these four trials were low. Only one study [19] described neither the number of withdrawal nor loss to follow-up and the reason for these aspects; therefore, this study [19] was regarded as having a high risk of bias in the domain of incomplete outcome data. All except one [20] study was considered to have a high risk of bias in selective reporting because it is a subgroup of PRoFESS trial. Because of premature termination, FASTER [16] was regarded as having a high risk in the part of other bias, and other trials were regarded as having an unclear risk in this domain.

The evidence classification results, summarized from the GRADE evidence profile assessed by the GRADEpro software, are shown in Table 3. The quality of evidence was high for the composite outcome of major vascular events, moderate for the net clinical outcome, low for the outcome of recurrent stroke, IS recurrence, bleeding episodes, and very low for TIA, major bleeding and intracranial hemorrhage.

Table 3 Summary of GRADE evidence profile
Full size table

Clinical results

All outcomes were reported in total analysis and subgroup analysis. We made subgroup analysis of both efficacy and safety outcomes in accordance with predefined groups.

Recurrent stroke is the primary outcome of our meta-analysis. Totally 10 trials [11–20] reported the outcome of recurrent stroke (Fig. 3). Pooled evidence indicated that comparing with monotherapy, DAPT reduced the risk of recurrent stroke by 35 % (RR 0.65, 95 % CI 0.56–0.76).There was no statistical heterogeneity between included studies (P = 0.67, I 2 = 0 %). To detect the influence of initiation treatment time, we conducted a subgroup analysis according to acute phase and non-acute phase. The stratified analysis of acute phase combination therapy included 7 RCTs [13, 15–20] and revealed that compared with monotherapy, DAPT significantly decreased the risk of stroke recurrence (RR 0.66, 95 % CI 0.57–0.77) (Table 3). Three RCTs [11, 12, 14] were included in the stratified analysis of non-acute phase combination therapy; Table 3 shows significant reduction in stroke recurrence with DAPT group (RR 0.19; 95 % CI 0.03–1.07, P = 0.06).

Fig. 3
figure 3

Individual and summary risk ratios (RRs) with 95 % CIs of recurrent stroke

Full size image

Information regarding the net clinical outcome is reported in 6 trials [11, 13, 14, 16, 17, 20]. The pooled evidence showed that compared with monotherapy, DAPT decreased the risk of the net clinical outcome by 33 % (RR 0.67, 95 % CI 0.58–0.79) (Fig. 4). There did not exist a significant statistical heterogeneity between included studies (P = 0.36, I 2 = 8 %). Data of IS recurrence were available in eight trials [11–14, 16, 17, 19, 20]. As shown in Fig. 5, pooled evidence showed DAPT reduced the risk of IS recurrence by 35 %. No statistical heterogeneity was found between the eight trials.

Fig. 4
figure 4

Individual and summary risk ratios (RRs) with 95 % CIs of the net clinical outcome

Full size image
Fig. 5
figure 5

Individual and summary risk ratios (RRs) with 95 % CIs of ischemic stroke recurrence

Full size image

The analysis of composite outcome of major vascular events included five trials [11, 13, 16, 18, 20]. Pooled evidence indicated that DAPT decreased the risk of composite outcome of major vascular events by 30 % (RR 0.70, 95 % CI 0.60–0.81) (Fig. 6). There is no statistical heterogeneity between the five trials (P= 0.94, I 2 = 0 %).

Fig. 6
figure 6

Individual and summary risk ratios (RRs) with 95 % CIs of composite outcome of major vascular events

Full size image

Five trials [11, 13, 14, 18, 19] reported data on TIA (Fig. 7). Based on the overall pooled evidence, compared with monotherapy, DAPT has a nonsignificant reduction in TIA (RR 0.81, 95 % CI 0.56–1.17) and heterogeneity (P = 0.80, I 2 = 0 %).

Fig. 7
figure 7

Individual and summary risk ratios (RRs) with 95 % CIs of TIA

Full size image

The safety endpoints were major bleeding, intracranial hemorrhage and bleeding episodes (Figs. 8, 9, 10). Major bleeding is a primary index for safety; totally eight trials [11, 13–18, 20] reported values on major bleeding. As shown in Fig. 8, there is no significant difference in the risk of major bleeding between DAPT and monotherapy (RR 1.44, 95 % CI 0.72–2.88). Eight trials [11, 13–17, 19, 20] have reported information regarding intracranial hemorrhage. The pooled evidence indicated that compared with monotherapy, DAPT has led to a nonsignificant increase in this domain (RR 1.29, 95 % CI 0.56–2.93). The bleeding episodes were available on seven trials [11, 13–17, 20]; DAPT can lead to a significant increase in bleeding episodes compared with monotherapy (RR 1.66, 95 % CI 1.19–2.30). No statistical heterogeneity was found in all three safety outcomes.

Fig. 8
figure 8

Individual and summary risk ratios (RRs) with 95 % CIs of major bleeding

Full size image
Fig. 9
figure 9

Individual and summary risk ratios (RRs) with 95 % CIs of intracranial hemorrhage

Full size image
Fig. 10
figure 10

Individual and summary risk ratios (RRs) with 95 % CIs of bleeding episodes

Full size image

Through sensitive analysis conducted by STATA software, we found similar overall results for primary outcome after excluding each individual study shown in Fig. 11.

Fig. 11
figure 11

Sensitivity analysis of primary outcome

Full size image

Discussion

The meta-analysis of 8969 participants provided evidence about the efficacy and safety of short-term DAPT in treating IS or TIA. A sensitivity analysis on three main outcomes generated similar results, which indicated that results of the present meta-analysis were generalizable. In the subgroup analysis of clopidogrel plus aspirin or dipyridamole plus aspirin, significant reductions were found in the risks of stroke recurrence, the net clinical outcome, composite outcome of vascular events and IS recurrence (Table 4). We also performed an acute phase (<72 h) and non-acute phase subgroup analysis of eight outcomes. The data indicated that either in acute IS or non-acute phase of IS, DAPT brought more beneficial outcomes without increasing the risk for major bleeding and intracranial hemorrhage. In addition, subgroup analysis based on different races was also carried out. Both arms have similar effect outcomes, whereas non-Asian have more bleeding risk compared with Asian which may be due to different physiology character and stroke type.

Table 4 Subgroup analysis of eight outcomes on baseline characteristics (treatment initiation time, races, different DAPT combination)
Full size table

Although many clinical trials such as MATCH [25] or SPS3 [26] and meta-analyses [27, 28] have shown that treatment with DAPT offers no better clinical outcomes while having more bleeding danger, many participants in these trials were given long-term treatment of DAPT which may increase risks of bleeding. Three network meta-analyses [29–31] were published recently which compared different antiplatelet therapy for treating IS. All three analyses concluded that cilostazol has advantages over DAPT. Whereas the conclusions lack credibility for they are indirect comparisons of different antiplatelet therapy; large head-to-head RCTs are needed to further confirm their conclusions.

To our knowledge, there are two recently published meta-analyses that mentioned the comparison of short-term DAPT and monotherapy. They also recommended short-term DAPT for IS or TIA which is consistent with our results. Compared with these two analyses, first we focused on the short-term application of DAPT. Second, we included more dual-antiplatelet combinations and made subgroup analysis according to different combinations. Third, quality of the included studies in the meta-analyses by Zhang et al. [22] and Ge et al. [21] was only assessed with Jadad scoring system, which is explicitly discouraged in Cochrane reviews because it has a strong emphasis on reporting rather than conducting and does not cover one of the most important potential biases in randomized trials (allocation concealment). Fourth, we used Cochrane collaboration’s tool for assessing the risk of bias, which is recommended in Cochrane reviews. Moreover, our meta-analysis is the first study of this subject to assess the quality of evidence with the GRADE system, which was used to ensure the reliability of our results.

There are two ongoing multicenter clinical trials [32, 33] which are relevant to our analysis. The double-blind POINT [32] (Platelet-Oriented Inhibition in New TIA and Minor Ischemic Stroke, ClinicalTrials.gov Identifier: NCT00991029) trial aims to assess the safety and efficacy of clopidogrel (75 mg od) plus aspirin (50–325 mg od) versus aspirin alone (50–325 mg od) for reducing risk of major ischemic vascular events at 90 days, in TIA or minor stroke patients within 12 h of symptom onset. It was initiated in October 2009 and aims to enroll 5840 patients until December 2018. Upon finished, the POINT trial will be included in our updated meta-analysis. In addition, the open-label, blinded end point TARDIS [33] (triple antiplatelets for reducing dependency after ischaemic stroke) trial (ISRCTN47823388) is investigating the efficacy and safety of more intensive antiplatelet therapy (combined aspirin, clopidogrel, and dipyridamole) in treating IS and TIA. Both these trials will bring fresh air to the establishment of future guidelines for treatment of acute IS and TIA.

We note several limitations in our study. First, our meta-analysis included studies that varied in relation to the study population, stroke severity, comparator, antiplatelet medications, onset-to-treatment interval, and treatment duration. All of these factors could be potential confounders for accurate inclusions. Second, CHANCE is a large trial which account for about 50 % of participants in the meta-analysis; therefore, its results drove much of the findings. Third, only published data were included, which may lead to a reporting bias by overestimating the effect of dual therapy. In addition, the majority of participants included in trials are from Asia; therefore, the application of our inclusions may have some limitations in whites. All these aspects reinforce the need to perform more large, well-designed trials involving the effects of short-term DAPT in the secondary prevention of IS to obtain more reliable conclusions.

Conclusions

In conclusion, the current study shows that short-term (≤1 year) DAPT offers protection effects against stroke recurrence and major vascular events without increasing the risk of hemorrhagic stroke and major bleeding events in patients with prior stroke or TIA. Either in acute phase or non-acute phase, DAPT will provide more beneficial outcomes for patients with IS or TIA. For both the combination of aspirin plus clopidogrel and aspirin plus dipyridamole, short-term DAPT is superior to monotherapy in treating IS or TIA. Results of ongoing large trials will provide more conclusive evidence on the use of DAPT for stroke patients of other ethnic descents than Asian region and for IS patients.