Aspirin 160–300 mg daily started within 48 h of onset of acute ischaemic stroke is associated with a small beneficial reduction in recurrent ischaemic stroke (6 fewer per 1000 patients treated) and pulmonary embolism (1.5 fewer per 1000) that outweighs increased risk of bleeding (2 extra symptomatic ICHs and 4 extra major extracranial haemorrhages). The net effect is that, for every 1000 patients treated early with aspirin, 22 have reduced long-term disability, including 11 more achieving full recovery. Only two single antiplatelet regimens have been compared head to head against aspirin alone: cilostazol (a phosphodiesterase inhibitor) performed similarly to aspirin; ticagrelor (a GP IIa/IIIb receptor antagonist) showed tended to reduce ischaemic events but increased minor bleeding and dyspnea. In minor, non-cardioembolic ischaemic stroke or TIA, early dual antiplatelet therapy (DAPT) has shown advantages over early monotherapy. Most well-studied is clopidogrel and aspirin, with similar findings for dipyridamole and aspirin. DAPT reduces all-type (ischaemic and haemorrhagic) recurrent stroke (27 fewer per 1000 treated patients), but minimally increases major extracranial bleedings (3 more per 1000). Confining DAPT to the first 3 w maximizes the benefit to harm ratio. Anticoagulants alone and arterial-dose anticoagulants added to antiplatelet agents offer no net advantages over antiplatelet drugs alone. Venous prophylaxis-dose anticoagulants and aspirin, compared with aspirin alone, reduced recurrent ischaemic stroke more than it tend increased major extracranial haemorrhage.
The purpose of antiplatelet therapy, like that of anticoagulation (Chapter 10), in patients with acute ischaemic stroke or transient ischaemic attack (TIA) is to prevent ischaemic stroke progression, recurrent ischaemic stroke, and other serious vascular events.
Antiplatelet and anticoagulant agents target two complementary segments of the haemostatic system. Platelet adhesion and aggregation particularly tends to occur when blood passes rapidly over irregular surfaces, causing high-speed, dyslaminar flow – the resulting ‘white thrombi’ are platelet-rich. In contrast, coagulation proteins particularly tend to precipitate when blood flow slows or stops altogether, in low-perfusion pressure settings – the resulting ‘red thrombi’ are rich in interlinked fibrin and trapped erythrocytes. Consequently, antiplatelet agents are often more effective for thrombi arising from arteriopathies like atherosclerosis, while anticoagulants are often more effective for thrombi arising from venous disease and the cardiac atria.
Cyclo-oxygenase inhibitors (e.g. acetylsalicylic acid [ASA], triflusal)
Thienopyridine derivatives (e.g. clopidogrel, prasugrel)
Phosphodiesterase inhibitors (e.g. dipyridamole, cilostazol)
Glycoprotein (GP) IIb/IIIa receptor inhibitors (e.g. abciximab, eptifibatide)
P2Y12 inhibitors (e.g. ticagrelor)
In patients with definite or presumed acute cerebral ischaemia, early antiplatelet therapy (started within 72 hours [h] of onset) has been compared with no antithrombotic therapy in 11 randomized trials enrolling 42,852 patients (Abciximab in Ischemic Stroke Investigators, 2000; Siebler et al., 2011; Ciccone et al., 2014; Sandercock et al., 2014).
Antiplatelet agents tested included aspirin, ticlopidine, and two glycoprotein IIb/IIIa receptor inhibitors, abciximab, and tirofibran. However, two aspirin mega-trials, testing aspirin 160–300 mg daily started within 48 hours of onset, contributed 95% of the data (CAST [Chinese Acute Stroke Trial] Collaborative Group, 1997; International Stroke Trial Collaborative group [IST], 1997).
Permitted time windows for trial enrolment included patients randomized within 6 hours (MAST-I, 1995; AbESTT, 2005), 12 hours (Ciufetti et al., 1990), 24 hours (Abciximab in Ischemic Stroke [AIST] Investigators, 2000), 48 hours (CAST, 1997; IST, 1997), and 72 hours (Roden-Jullig et al., 2003), between 3 and 22 hours (SaTIS – Siebler et al., 2011), and, in one study, 3 separate populations within 5 hours, between 5 and 6 hours, and greater than 6 hours but within 3 hours of awakening (AbESTT-II – Adams et al., 2008). Actual median or mean onset to enrolment or treatment times ranged from 3.7 hours (AbESTT II/Primary) and between 3 and 6 hours (MAST-I), through 5.1 hours (AbESTT II/Companion), 9.8 hours (AbESTT II/Wake-up), 12 hours (AIST), 13 hours (SaTIS), 19 hours (IST), and 25 hours (CAST).
Nine of the eleven trials performed computed tomography (CT) scans in all patients prior to entry to exclude individuals with intracranial haemorrhage (ICH) from entry into the trial. Two trials (CAST, 1997; IST, 1997) performed a CT scan in almost all patients; in these trials, clinicians had to have a low threshold of suspicion of ICH prior to randomization. In the Chinese Acute Stroke Trial (CAST), 87% had a CT scan prior to randomization; by discharge, this number had risen to 94%. In the International Stroke Trial (IST), 67% were scanned before randomization and 29% after randomization, so that, overall, 96% of patients were scanned.
The two large aspirin trials, CAST and IST, enrolled patients without requiring a minimum initial neurological deficit to be present. Several of the other trials did require a minimal degree of deficits, including at least 1 point on the Scandinavian Stroke Scale (Roden-Jullig et al., 2003) or at least 4 points on the National Institutes of Health Stroke Scale (NIHSS) (AIST 2000; AbESTT 2005; AbESTT-II 2008; SaTIS 2011).
The scheduled duration of treatment varied from one-time bolus (lower-dose tiers of AIST 2000), to be continued until 12 hours (higher-dose tiers of AIST 2000; AbESTT 2005; AbESTT-II 2008), 48 hours (SaTIS 2011), 5 days (Roden-Jullig et al., 2003), 10 days (MAST-I 1995), up to 2 weeks or discharge if earlier (IST 1997), 3 weeks (Ciufetti et al., 1990), and up to 4 weeks or discharge if earlier (CAST 1997).
Death or Dependency
Data regarding death or dependency at final follow-up were available from 9 trials, including 4 trials of aspirin enrolling 41,291 patients and 5 trials of abciximab enrolling 1275 patients (Figure 9.1). Overall, random allocation to antiplatelet therapy was associated with a decrease in death or dependency at the end of follow-up, relative risk (RR) 0.98, 95% confidence interval (CI): 0.96–0.99; p = 0.01.
Figure 9.1 Dependency or death (modified Rankin Scale [mRS] 3–6 or nearest equivalent) at 1–6 months, early antiplatelet agents vs avoid early antiplatelet agents.
Although formal testing did not show heterogeneity among agents, assessment power was limited by the large discrepancy in sample sizes between the aspirin and the abciximab groups, suggesting that consideration of each agent class separately is warranted. For aspirin, random allocation to active treatment was associated with a decrease in death or dependency at the end of follow-up, 45.0% versus 46.2% (RR 0.97, 95% CI: 0.96–0.99; p = 0.01). In absolute terms, 12 more patients were functionally independent at the end of follow-up for every 1000 patients treated with aspirin. For abciximab, random allocation to active treatment was not associated with an alteration in death or dependency outcomes (RR 0.99, 95% CI: 0.87–1.12; p = 0.84).
Data regarding death at final follow-up (or end of treatment if later data not available) were available from 11 trials, including 4 trials of aspirin enrolling 41,291 patients, 1 trial of a thienopyridine (ticlopidine) enrolling 30 patients, and 6 trials of glycoprotein IIb/IIIa receptor inhibitors (abciximab and tirofibran) enrolling 1531 patients (Figure 9.2). By the end of follow-up (or end of treatment period if later data not available), antiplatelet therapy was associated with a decrease in mortality (RR 0.94, 95% CI: 0.89–0.99; p = 0.01).
Figure 9.2 Death from all causes at 1–6 months, early antiplatelet agents vs avoid early antiplatelet agents.
The preponderance of patients in the aspirin group limited the power of tests for heterogeneity, supporting consideration of each agent class separately. For aspirin, random allocation to active treatment was associated with a decrease in death, 12.1% versus 12.9% (RR 0.94, 95% CI: 0.89–0.98; p = 0.01). In absolute terms, there were 8 more patients alive at the end of follow-up for every 1000 patients treated with aspirin. For abciximab, random allocation to active treatment was not associated with an effect on mortality (RR 0.98, 95% CI: 0.74–1.30; p = 0.91).
The composite of progressive and recurrent ischaemic stroke usefully captures post-onset events that antithrombotics may in part beneficially avert. Early non-haemorrhagic progression of neurological deficits may be due to several mechanisms. Two are potentially modifiable by antithrombotic treatment – further propagation of the initial thrombus or a new cerebral thromboembolic event causing recurrent ischaemic stroke. Several other mechanisms are not modifiable by antithrombotic treatment, including haemodynamic-collateral failure, oedema, seizure, and infection. Recognition of recurrent ischaemic stroke is more straightforward, between 5 days to 6 months after onset of an initial cerebral ischaemic event.
Antiplatelet therapy was associated with a reduction in progressive or recurrent ischaemic stroke (RR 0.80, 95% CI: 0.72–0.89; p < 0.0001) (Figure 9.3). There was evidence of benefit both for aspirin and for glycoprotein IIb/IIIa receptor inhibitors. For aspirin, random allocation to active treatment was associated with a decrease in progressive/recurrent ischaemic stroke, 2.6% versus 3.2% (RR, 0.80, 95% CI: 0.72–0.89; p < 0.0001). In absolute terms, 6 progressive/recurrent ischaemic strokes were avoided for every 1000 patients treated. For abciximab, random allocation to active treatment was also associated with progressive/recurrent ischaemic stroke rates of 9.8% versus 12.0% (RR 0.80, 95% CI: 0.60–1.06; p = 0.12).
Figure 9.3 Death from all causes at 1–6 months: early antiplatelet agents vs avoid early antiplatelet agents.
Antiplatelet therapy was associated with an increase in symptomatic intracranial haemorrhage (SICH) (RR 1.33, 95% CI: 1.09–1.61; p = 0.005) (Figure 9.4). There was evidence of heterogeneity by agent class, with more magnified increase in SICH with glycoprotein IIb/IIIa receptor inhibitors than with aspirin, p(heterogeneity) = 0.006, I2 = 86.6%. For aspirin, random allocation to active treatment was associated with an increase in SICH, 1.0% versus 0.8% (RR 1.22, 95% CI: 1.00–1.49; p = 0.05). In absolute terms, 2 additional symptomatic intracranial haemorrhages occurred for every 1000 patients treated. For glycoprotein IIb/IIIa receptor inhibitors, random allocation to active treatment was associated with an increase in SICH, 3.4% versus 0.8% (RR 4.30, 95% CI: 1.78–10.35; p = 0.001).
Figure 9.4 Symptomatic intracranial haemorrhage, early antiplatelet agents vs avoid early antiplatelet agents.
Antiplatelet therapy was associated with an increase in major extracranial haemorrhage (RR 1.72, 95% CI: 1.39–2.13; p < 0.0001) (Figure 9.5). There was evidence of heightened extracranial haemorrhage rates for both aspirin and for glycoprotein IIb/IIIa receptor inhibitors. For aspirin, random allocation to active treatment was associated with an increase in major extracranial haemorrhage, 1.0% versus 0.6% (RR 1.70, 95% CI: 1.36–2.1; p < 0.00001). In absolute terms, 4 more major extracranial haemorrhages occurred for every 1000 patients treated. For abciximab, random allocation to active treatment was also associated with major extracranial haemorrhage, RR 1.84, 95% CI: 1.02–3.32; p = 0.04.
Figure 9.5 Major extracranial haemorrhage, early antiplatelet agents vs avoid early antiplatelet agents.
For symptomatic pulmonary embolism, events occurred infrequently, so that among agent classes, only aspirin, investigated in large mega-trials, had sample sizes sufficient to be informative (Figure 9.6). Aspirin therapy was associated with a reduction in pulmonary embolism, 0.34% versus 0.48% (RR 0.71, 95% CI: 0.53–0.97; p = 0.03). In absolute terms, 1.5 pulmonary emboli were avoided for every 1000 patients treated with antiplatelet agents. Few trials reported results for deep vein thrombosis or for asymptomatic pulmonary embolism, so the profile of antiplatelet agent effect on venous thrombosis complications after acute cerebral ischaemia is incomplete.
Figure 9.6 Pulmonary embolism, early antiplatelet agents vs avoid early antiplatelet agents.
Two large trials examined the effect of early aspirin versus control on shift in functional outcomes across 4 levels of disability – fully recovered, independent but not fully recovered, dependent, and dead – at 1 month (CAST, 1997) or 6 months (IST, 1997) in a total of 39,866 randomized patients. Combining the trial data without adjustment, aspirin was associated with a favourable shift to less disabled outcomes across the 4 disability-level states, Mann–Whitney p = 0.02 (Figure 9.7). Using algorithmic joint outcome table analysis, aspirin yielded a lower disability level for a net 22 of every 1000 patients treated. For the best outcome, aspirin versus no aspirin was associated with increased attainment of full recovery, 27.8% versus 26.7% (RR 1.04, 95% CI: 1.01–1.07; p = 0.01).
Analyses for differential treatment response have been conducted among nearly 40,000 patients enrolled in 3 trials of aspirin versus no aspirin, evaluating 28 patient subgroups defined by single baseline characteristics (Chen et al., 2000), and 3 patient subgroups defined by models predicting risk of thrombotic, haemorrhagic, and poor final functional outcomes (Thompson et al., 2015). No patient profile was identified that predicted a heightened or reduced benefit of aspirin therapy.
Antiplatelet therapy with aspirin 160–300 mg daily, given orally (or per nasogastric tube or rectum in patients unable to swallow), and started within 48 hours of onset of acute cerebral ischaemia is associated with a modest reduction in recurrent ischaemic stroke and pulmonary embolism, and this benefit is partially offset by a small risk of bleeding. Overall, among every 1000 patients treated, 6 fewer will have progressive/recurrent ischaemic stroke and 1.5 fewer will have pulmonary embolism, 2 more will have SICH, and 4 more will have major extracranial haemorrhage.
Through these contrasting, overtly recognizable effects, and others, aspirin confers an overall net benefit, modest in degree, upon final outcome. For every 1000 patients treated early with aspirin, compared with no aspirin, 22 have reduced disability, including 11 more who are fully recovered and 8 more alive. For reduced disability, the number needed to treat for 1 patient to benefit is 47.
The only other antiplatelet agent class with any substantial data against control in the early setting is the glycoprotein IIb/IIIa receptor inhibitors, with most of the data arising from abciximab trials. Reflecting trial design (earlier start of therapy, greater deficit severity) and more powerful pharmacological effects, both beneficial ischaemic event reductions and adverse bleeding events were more pronounced with glycoprotein IIb/IIIa receptor inhibitors. Among every 1000 patients treated, 23 fewer had progressive/recurrent ischaemic stroke but 26 more had symptomatic ICH and 17 more had major extracranial haemorrhage, with overall no net positive or negative effect on final functional independence or survival.
In acute cerebral ischaemia patients not being treated with intravenous thrombolytics, aspirin 160–300 mg is a beneficial regimen when started within the first 48 hours of onset and continued as a daily dose.
In patients who are unable to swallow safely, aspirin may be given via a nasogastric tube, per rectum as a suppository, or, in countries where available, intravenously (as 100 mg of the lysine salt of ASA).
In patients with aspirin allergy, an alternative antiplatelet agent from a class that has comparable potency and effects to aspirin should be considered. Glycoprotein IIb/IIIa receptor inhibitors are not good alternative choices based on completed studies. Randomized trials of other agents given acutely are lacking, but thienopyridines (e.g. clopidogrel) and phosphodiesterase inhibitors (e.g. dipyridamole) are potential options.
In patients who are being treated with intravenous fibrinolytic therapy, aspirin should be avoided for the first 24 hours after lytic start, given its known potentiation of bleeding risk (see Chapter 6).
The long-term antithrombotic regimen eventually finalized for a patient will depend upon findings from the diagnostic workup regarding ischaemic stroke mechanism (large artery atherosclerotic [LAA], cardioembolic, small vessel disease, etc), indications or contraindications arising in other vascular beds, and patient passage through the initial several days post-stroke, when risk of haemorrhagic conversion is highest. Initial daily aspirin is a net beneficial regimen that can be deployed during the first hours and days after onset, while information is being gathered to guide the eventual selection of the final long-term regimen.
The overall magnitude of clinical benefit of antiplatelet therapies alone in acute ischaemic stroke is modest (Cranston et al., 2017). Reperfusion therapies have far greater benefits for the patients eligible for them, with intravenous tissue plasminogen activator (tPA) improving the outcomes of 10-fold more patients, and endovascular mechanical thrombectomy 20-fold more cerebral ischaemia patients. This limited impact reflects the fact that the preponderance of brain injury and functional disability in acute cerebral ischaemia arises from the initial occlusion and the initially threatened brain parenchyma, which are addressed directly with reperfusion interventions. Delayed clot propagation causing stroke progression and new thromboembolism causing early recurrent ischaemic stroke are less frequent events that account for much less neural injury and activity impairment at the population level. As a result, stand alone antiplatelet therapy has less scope for benefit in acute cerebral ischaemia.
Using antiplatelet agents to enhance reperfusion therapies is a promising strategy meriting further investigation. If a combination of antiplatelet and fibrinolytic agents can be found that achieves more frequent and sustained recanalization than intravenous fibrinolysis alone, without increasing haemorrhagic risk, clinical benefits would be substantial. Continuous ultrasound studies in patients with acute ischaemic stroke have shown that thrombi behave dynamically after first exposure to intravenous fibrinolytics, with frequent early re-occlusion after initial recanalization, potentially avertable by concomitant antiplatelet therapy. While aspirin added to full-dose intravenous tissue plasminogen activator was not beneficial, initial small trials of various dose combinations of glycoprotein IIb/IIIa receptor inhibitors and fibrinolytics have been encouraging (Pancioli et al., 2013; Adeoye et al., 2015). In the Combined Approach to Lysis Utilizing Eptifibatide and rt-PA in Acute Ischemic Stroke-Enhanced Regimen (CLEAR-ER) trial, 101 patients treated with a combination of reduced-dose tissue plasminogen activator and eptifibatide, compared with historical controls from the NINDS rt-PA Study, had directionally favourable rates of disability-free (modified Rankin Scale [mRS] 0–1) outcome at 3 months, 49.5% versus 36.0% (odds ratio [OR] 1.74; 95% CI:, 0.70–4.31; p = 0.23) (Pancioli et al., 2013). Bleeding rates were not increased.
With regard to preventing early recurrent ischaemic stroke, the initial large aspirin trials established aspirin as being of benefit, though modest. Accordingly, aspirin became the standard of care therapy for this target, and aspirin monotherapy, rather than no antithrombotic treatment, served as the control arm in subsequent trials of other antiplatelet agents or combinations of antiplatelet agents seeking to reduce subacute recurrent ischaemic stroke.
In patients with acute ischaemic stroke or transient ischaemic attack early after onset (within 72 h) and of non-cardioembolic origin, different antiplatelet regimens have been compared with each other in 22 randomized trial comparisons enrolling 31,549 patients. The greatest number of studies have been on dual versus mono antiplatelet therapy regimens, evaluated in 17 trial comparisons enrolling 13,982 patients. Head-to-head comparison of one mono antiplatelet regimen versus another mono antiplatelet regimen has been studied in 3 trials enrolling 13,796 patients. In addition, 1 trial enrolling 675 patients has performed a head-to-head comparison of one dual antiplatelet regimen versus another dual antiplatelet regimen, and 1 trial enrolling 3096 patients has compared a triple antiplatelet therapy regimen versus mono and dual antiplatelet regimens.
Three trials have compared different monotherapy antiplatelet regimens in patients with acute non-cardioembolic ischaemic stroke or TIA early after onset (within 72 h), including two trials of the cyclo-pentyl-triazolo-pyrimidine ticagrelor versus aspirin, enrolling 13,349 patients, and one trial of the phosphodiesterase inhibitor agent cilostazol versus aspirin, enrolling 448 patients. The disparity in agent class sample sizes limits power to interrogate for treatment effect heterogeneity, suggesting the different regimens be evaluated separately.
For ticagrelor versus aspirin, the predominant source of data is the large SOCRATES trial, enrolling 13,199 patients (Johnston et al., 2016), with the smaller SETIS trial contributing an additional 150 patients (Torgano et al., 2010). In the large SOCRATES trial, enrolled patients had to be within 24 hours of a minor ischaemic stroke (NIHSS score of 5 or less) or high-risk TIA (ABCD2 score of 4 or higher), and 63% were enrolled between 12 and 24 hours post-onset. Patients received oral ticagrelor alone or aspirin alone for 3 months. In the smaller SETIS trial, patients were enrolled within 6 hours of a moderate-major ischaemic stroke (NIHSS score of 5–25), with mean time to enrolment of 4.4 hours, and patients received intravenous ticagrelor or aspirin for 3 days.
In combined analysis of the two trials, ticagrelor was associated with a reduction in progressive/recurrent ischaemic stroke (5.9% vs 6.7%, RR 0.88, 95% CI: 0.77–1.00; p = 0.05; Figure 9.8). In addition, ticagrelor did not increase SICH (0.2% vs 0.3%, RR 0.62, 95% CI: 0.31–1.24; p = 0.18; Figure 9.9) or major extracranial haemorrhage (0.17% vs 0.13%, RR 1.21, 95% CI: 0.54–2.58; p = 0.69; Figure 9.10). Mortality by 3 months did not show a treatment group difference (RR 1.15, 95% CI: 0.83–1.60; p = 0.39; Figure 9.11). However, in the large SOCRATES trial, for the trial’s primary endpoint composite of any stroke (ischaemic or haemorrhagic), myocardial infarction, or all-cause death, directionally favourable ticagrelor effects did not reach statistical significance (6.7% vs 7.5%, hazard ratio [HR] 0.89, 95% CI: 0.78–1.01; p = 0.07). The SETIS trial did not report data for this composite endpoint. Also, in SOCRATES, ticagrelor versus aspirin was associated with more combined major or minor bleeding events (1.1% vs 0.7%, p = 0.005) and more dyspnoea events (1.4% vs 0.3%, p = 0.0001).
Figure 9.8 Progressive/recurrent ischaemic stroke, head-to-head comparisons of early single antiplatelet agents.
Figure 9.9 Symptomatic intracranial haemorrhage, head-to-head comparisons of early single antiplatelet agents.
Figure 9.10 Major extracranial haemorrhage, head-to-head comparisons of early single antiplatelet agents.
For cilostazol, the single CAIST trial enrolled 458 patients with mild-to-moderate, but not severe, ischaemic stroke (NIHSS score up to 15) within 48 hours of onset (mean actual time from onset 34 h), and allocated them to cilostazol or aspirin for 90 days (Lee et al., 2011). No statistically significant difference between the cilostazol and aspirin regimens was observed, including for the trial’s primary endpoint of death or dependency at 3 months (RR: 0.90, 95% CI: 0.66–1.22; Figure 9.12), and for progressing/recurrent ischaemic stroke (RR: 1.08, 95% CI: 0.74–1.08; see Figure 9.8), SICH (RR: 0.20, 95% CI: 0.01–4.12; see Figure 9.9), major extracranial haemorrhage (RR: 1.00, 95% CI: 0.14–7.01; see Figure 9.10), and mortality (RR: 0.96, 95% CI: 0.14–6.73; Figure 9.12).
Figure 9.12 Death at end of follow-up, head-to-head comparisons of early single antiplatelet agents.
Figure 9.11 Dependency or death at end of follow-up, head-to-head comparisons of early single antiplatelet agents.
In patients with acute ischaemic stroke or TIA early after onset (within 72 h) and of non-cardioembolic origin, dual antiplatelet therapy has been compared with mono antiplatelet therapy antithrombotic therapy in 17 randomized trial comparisons enrolling 13,982 patients (Wong et al., 2013; Johnston et al., 2018; Yang et al., 2018). The comparisons include 10 from trials confined to early-presenting patients and 7 from early-presenting subgroups of trials enrolling both early- and later-presenting patients.
Antiplatelet Regimens Tested
The most common monotherapy control comparator was aspirin alone, used in 13 trial arm comparisons enrolling 13,647 patients. Clopidogrel alone and dipyridamole alone were used as the monotherapy comparator in 2 trials each. The most common dual therapy regimen tested was clopidogrel plus aspirin, used in 10 randomized trial comparisons enrolling 12,823 patients. The combination of dipyridamole plus aspirin was evaluated in 6 randomized trial comparisons and cilostazol plus aspirin in 1 trial. The 2 largest trials, CHANCE and POINT, evaluated clopidogrel plus aspirin versus aspirin and together contributed 72% of the data (Wang et al., 2013; Johnston et al., 2018). The comparisons with the greatest number of patients were (1) clopidogrel plus aspirin versus aspirin (9 trials, 12,332 patients, 81% of all patients); (2) dipyridamole plus aspirin versus clopidogrel (1 trial, 1360 patients, 9% of all patients); and (3) dipyridamole plus aspirin versus aspirin (3 trials, 748 patients, 5% of all patients).