Aspirin for the Prevention of Cardiovascular Events in Patients With Peripheral Artery Disease:  A Meta-analysis of Randomized Trials FREE

Quiz Ref IDAlthough aspirin is effective in the prevention of cardiovascular events in patients with symptomatic coronary heart disease and cerebrovascular disease,^1- 3 its effect in patients with peripheral artery disease (PAD) is uncertain.^4- 5 In the largest investigation to date, the Antithrombotic Trialists' Collaboration (ATC) meta-analysis of 135 000 high-risk patients from 287 clinical trials demonstrated a reduction in myocardial infarction (MI), stroke, and death with antiplatelet therapy in all patients with symptomatic cardiovascular disease.^6- 7 In the subset of 42 PAD trials, which included 9214 patients and studied multiple antiplatelet agents compared with placebo, antiplatelet therapy was associated with a significant 23% reduction in cardiovascular events.⁷ However, nearly two-thirds of the PAD trials evaluated antiplatelet agents other than aspirin. It is possible that the overall benefit of antiplatelet therapy in PAD may have been driven by therapeutic regimens other than aspirin. The uncertainty about the benefit of aspirin in PAD was further raised by a meta-analysis of aspirin in PAD limited to studies in the ATC⁸ meta-analysis and by a recent large randomized controlled trial of aspirin therapy involving patients with diabetes and asymptomatic PAD that demonstrated no benefit of aspirin for reducing cardiovascular events.⁹

Despite the paucity of data supporting aspirin therapy for patients with PAD, major US and international guidelines cite the ATC meta-analysis as evidence to prescribe aspirin as first-line therapy to patients with PAD.^10- 11 However, the US Food and Drug Administration concluded that there was insufficient evidence to support a labeling indication for aspirin to treat PAD.¹² In addition, the recent Transatlantic InterSociety Consensus (TASC) II international guidelines for PAD gave aspirin a level C recommendation for patients with isolated PAD (without clinical evidence of coronary or cerebrovascular disease).¹³ To assess the benefit of aspirin in treating PAD, the current meta-analysis evaluated all available evidence from prospective, randomized controlled trials of aspirin alone or in combination with other antiplatelet drugs. This meta-analysis tested the null hypothesis that aspirin was not different from placebo or control in reducing the risk of the combined primary end point of nonfatal MI, nonfatal stroke, and cardiovascular death.

A comprehensive search was performed to identify randomized controlled trials in MEDLINE, the Cochrane Central Register of Controlled Trials (CENTRAL), EMBASE, Science Citation Index (SCI), Web of Science, and Social Science Citation Index (SSCI) in any language between 1966 and December 31, 2008, using the search terms aspirin, peripheral artery disease, peripheral arterial occlusive diseases, claudication, and randomized controlled trials, as well as combinations of these terms. Experts were questioned, bibliographies of retrieved articles were searched for other relevant studies, and major scientific meetings were monitored for the results of trials still under way at the time of the MEDLINE search. Data concerning study design, baseline patient characteristics, treatment, follow-up, and results were extracted from these reports. Additional data, when necessary, were derived from prior reports. In several studies, the primary outcome data were not part of the original publications, and therefore these data were obtained from the supplemental index of the Antithrombotic Trialists' Collaboration meta-analysis.¹⁴

Two investigators ( J.S.B., M.J.K.) independently evaluated studies for inclusion. Inclusion criteria were (1) prospective, randomized, controlled, studies either open-label or blinded; (2) assignment of PAD participants to aspirin treatment or a placebo or control group; and (3) available data on all-cause mortality, cardiovascular death, MI, stroke, and major bleeding. Trials with no events in any of the study groups were excluded. Any disagreements were resolved by discussion and consensus. The search was also cross-referenced to include aspirin trials involving patients with PAD noted in the most recent Antithrombotic Trialists' Collaboration meta-analysis.⁷ A total of 165 potentially eligible studies were identified and 107 were excluded because they were not randomized controlled trials (eg, review articles, editorials, letters to the editor, case reports, case-control studies, and meta-analyses). Of the randomized trials, 26 evaluated the effect of antiplatelet drugs other than aspirin and 14 were excluded because the population was not composed entirely of participants with PAD or there was no placebo or control group (Figure 1). Study quality was assessed for each trial using the Jadad score, a 5-point study quality assessment instrument.¹⁵

The primary outcome was the relative risk (RR) reduction of aspirin therapy on the composite end point of nonfatal MI, nonfatal stroke, and cardiovascular death in the population of patients who received any aspirin therapy (with or without dipyridamole). Secondary outcomes were all-cause mortality and each component of the primary end point. The primary safety outcome was the occurrence of major bleeding as defined by each study.

Data were analyzed using the intention-to-treat results from the included studies. Patient characteristics, aspirin dosages, and length of follow-up differed across studies, so the RR for each study was assumed to have a random offset from the population mean RR (ie, a random-effects model).¹⁶ Given the limited number of studies, trials were included even if 1 of the treatment groups did not have any cardiovascular events. In these cases a fixed value (0.5) was added to the patient groups with and without events to allow calculation of an RR. This adjustment tends to bias results toward the null.¹⁷ Secondary sensitivity analyses showed that results remained the same even if these studies were excluded.

The Cochran Q statistic and I² statistic were calculated to assess the degree of heterogeneity among the trials.¹⁸ The Q statistic did not indicate statistically significant heterogeneity for any end point. The I² statistic showed mild heterogeneity (<30%) for all end points except nonfatal MI compared between aspirin monotherapy and placebo or control, which showed moderate heterogeneity (I² = 46.2%). Therefore, although the studies had different characteristics, pooled estimates of the RR were judged reasonable. The primary result was presented as the Mantel-Haenszel pooled RR and 95% confidence interval (CI), calculated from the random effects model.

Subgroup analyses were performed to evaluate the effect of aspirin monotherapy vs placebo or control, aspirin therapy at currently recommended doses (75-325 mg/d); trials with a follow-up of more than a year; trials designed to assess cardiovascular events; trials that exclusively enrolled patients with both diabetes and PAD; and quality of the randomized trial (Jadad score ≥4). Statistical tests of the differences between the risk reductions in 2 subgroups were based on the standard error of the estimated pooled RRs calculated on each subgroup separately. Sensitivity analyses were conducted to further explore the robustness of the results. This was done to identify any study that may have exerted a disproportionate influence on the summary treatment effect by deleting 1 study at a time and repeating the analyses. Results obtained with a fixed-effects model were also compared with those obtained with a random-effects model. Funnel plots of the effect size vs the standard error of the log-transformed effect were generated for each end point to assess for publication bias.¹⁹

With 5000 patients, this meta-analysis had approximately 88% power to detect a 25% difference (from 10% to 7.5%) and 70% power to detect a 20% difference (from 10% to 8%) in the RR of cardiovascular death, MI, or stroke in the aspirin group vs placebo or control groups. The type 1 error rate was set at α = .05, and all tests were 2-sided.

All statistical analyses were performed using Review Manager version 5.0.17 (Nordic Cochrane Centre, Copenhagen, Denmark). Statistical tests were judged statistically significant if the 2-sided P value was less than .05.

Eighteen prospective randomized trials involving 5269 participants were identified for inclusion.^9,20- 35 Seven trials tested aspirin monotherapy vs placebo or control, 7 trials examined combined aspirin and dipyridamole vs placebo or control, and 4 trials had multiple arms (aspirin monotherapy, aspirin plus dipyridamole, and placebo). Of the 5269 participants, 2823 were randomized to aspirin therapy (of whom 1516 received aspirin monotherapy) and 2446 were randomized to placebo or control.

Details of the included studies are summarized in the Table. Duration of follow-up ranged from 10 days to 6.7 years. Aspirin doses ranged from 100 mg/d to 1500 mg/d for aspirin monotherapy and from 25 mg of aspirin and 75 mg of dipyridamole to 325 mg of aspirin and 75 mg of dipyridamole 3 times a day for each combination. Participants with PAD included patients with claudication, those undergoing percutaneous intervention or bypass surgery, and asymptomatic patients with an ankle brachial index of 0.99 or less. Two trials exclusively enrolled patients with both diabetes and PAD.^9,22

Figure 2 shows the effect of any aspirin vs placebo or control on the primary end point of combined cardiovascular events. A total of 251 (8.9%) cardiovascular events occurred among 2823 patients receiving any aspirin therapy compared with 269 (11.0%) events among 2446 control patients. The pooled RR showed a 12% reduction in cardiovascular event rates, which was not statistically significant (RR, 0.88; 95% CI, 0.76-1.04). Results for associations of aspirin therapy with the individual components of the primary end points revealed that the risk of nonfatal stroke was significantly lower in the aspirin group than in the placebo (52 events among 2823 patients [1.8%]) or in the control group (76 events among 2446 patients [3.1%]; RR, 0.66, 95% CI, 0.47-0.94; Figure 3). No significant differences were noted for any other secondary end point (Figure 4). Bleeding risk was not significantly increased (51 events among 2823 patients [1.8%] vs 44 events among 2446 patients [1.8%]; RR, 0.99; 95% CI, 0.66-1.50; Figure 4). However, bleeding was not formally adjudicated or reported in most studies, which limits the ability to establish the association of aspirin with major bleeding among participants with PAD.

The “Composite cardiovascular end points” panel in Figure 5 shows the effect of aspirin monotherapy on the primary end point. A total of 125 cardiovascular events occurred among 1516 patients (8.2%) receiving aspirin monotherapy compared with 144 events among 1503 patients (9.6%) in the placebo or control groups (RR, 0.75; 95% CI, 0.48-1.18). Aspirin monotherapy was associated with a significant reduction in the risk of nonfatal stroke (32 events among 1516 patients [2.1%] vs 51 events among 1503 patients [3.4%]; RR, 0.64; 95% CI, 0.42-0.99). No significant differences were identified for associations of aspirin monotherapy with other secondary outcomes (Figure 5).

Two of the trials comparing aspirin monotherapy with placebo used low-dose aspirin (100 mg/d),^9,21 These trials reported 112 cardiovascular events among 823 participants taking low-dose aspirin and 127 events among 819 patients assigned to the placebo group (13.6% vs 15.5%; RR, 0.64; 95% CI, 0.25-1.68). Although this benefit of low-dose aspirin was not statistically significant, the population studied was quite small and the 95% CI was wide, potentially limiting detection of important cardioprotective effects.

The Prevention of Progression of Arterial Disease and Diabetes (POPADAD)⁹ and VA-Cooperative²² trials exclusively enrolled patients with diabetes and PAD, and both demonstrated RRs on composite cardiovascular end points very close to 1.00 (Figure 2). Although the difference in risk reduction between these studies and the others was not statistically significant, a potential for differential effects in populations with or without diabetes cannot be ruled out. Importantly, these 2 studies contribute about 60% of the data for the analysis of the primary outcome measure; thus, the results should be interpreted in the context of an overall population with a high prevalence of diabetes. The POPADAD trial⁹ had the largest sample size and longest duration of follow-up. When analyses were repeated after excluding the POPADAD trial, the association between aspirin and the primary outcome of cardiovascular events yielded an effect size (146 events among 2185 patients taking aspirin [6.7%] vs 161 events among 1808 control patients [8.9%]; RR, 0.83; 95% CI, 0.67-1.02) similar in magnitude and direction to the overall results.

The primary findings of this meta-analysis did not differ in trials with follow-up periods of greater than 1 year^{9,21- 22,30,32- 35} (221 events among 1727 patients [12.8%] vs 241 events among 1533 patients [15.7%]; RR, 0.89; 95% CI, 0.76-1.05) or in trials specifically designed to assess cardiovascular events^9,21- 22 (148 events among 933 patients [15.9%] vs 167 events among 940 [17.8%]; RR, 0.92; 95% CI, 0.76-1.13). Similarly the primary results were unchanged if the trials with no events in 1 arm were excluded^24- 28 (243 events among 2222 patients [10.9%] vs 260 events among 2018 patients [12.9%]; RR, 0.88; 95% CI, 0.75-1.04) or if a fixed-effects inverse-variance analysis was used (RR, 0.88; 95% CI, 0.76-1.04). In the 6 trials that had a Jadad score of 4 or higher,^{9,20- 21,25,28,30} the primary results were similar (124 events among 1245 patients [10.0%] vs 138 events among 1112 patients [12.4%]; RR, 0.78; 95% CI, 0.53-1.14).

Funnel plots of the effect size vs the standard error of the log-transformed results were performed by plotting study precision (1 per standard error) against the prevention of cardiovascular events (primary end point). No skewed distribution was observed, suggesting no substantial evidence of publication bias (data not shown).

Quiz Ref IDResults of this meta-analysis demonstrated that for patients with PAD, aspirin therapy alone or in combination with dipyridamole did not significantly decrease the primary end point of cardiovascular events, results that may reflect limited statistical power. Based on the lower limit of the 95% CI, the current data rule out a 25% risk reduction in cardiovascular events with aspirin, but smaller levels of benefit, such as a 20% reduction, cannot be excluded with the available evidence. Quiz Ref IDAspirin therapy was associated with a significant reduction in the secondary end point of nonfatal stroke. No significant benefit was noted for all other secondary end points, including nonfatal MI, cardiovascular mortality, or all-cause mortality.

These findings contrast with literature that supports a more definitive role of aspirin in the treatment of symptomatic coronary heart disease and cerebrovascular disease.² There are several potential explanations for these findings. Compared with the abundance of randomized trials evaluating aspirin and other antiplatelet agents in the primary and secondary prevention of coronary heart disease and cerebrovascular disease, patients with PAD have been underrepresented in randomized trials. This is highlighted by examining the largest randomized trials of aspirin vs placebo in populations other than PAD including primary prevention of cardiovascular events in women (n = 39 878),³⁶ secondary prevention of coronary heart disease (n = 17 187),³⁷ and secondary prevention of cerebrovascular disease (n = 20 000).³⁸ In contrast, the single largest study of aspirin involving patients with PAD included only 1276 participants.⁹

Differences between combinations of aspirin with dipyridamole and aspirin doses across studies may have influenced the efficacy of aspirin. Experience with antiplatelet agents in other clinical settings supports varying efficacy of different antiplatelet medications in distinct patient populations. For example, although some studies demonstrate that a combination of aspirin and dipyridamole reduced risk of ischemic stroke in cerebrovascular disease,³⁹ this antiplatelet regimen is not more effective than aspirin alone for patients with coronary heart disease.³ Similarly, although warfarin in combination with aspirin prevents cardiovascular events among patients with coronary heart disease,⁴⁰ warfarin combined with aspirin is not efficacious for patients with PAD.⁴¹Quiz Ref IDRegarding aspirin dose, a lower dose of aspirin is at least as effective as higher aspirin doses in patients with symptomatic coronary heart disease and is associated with less harm, resulting in recommendations for a lower dose of aspirin for secondary prevention of cardiovascular disease.⁴² Current, US PAD guidelines support the use of low-dose aspirin despite the fact that the only 2 studies that evaluated a low aspirin dose for PAD were reported after publication of the guidelines.¹⁰ Of these 2 trials, the POPADAD study included PAD patients with diabetes who were otherwise asymptomatic for cardiovascular disease and reported no benefit for aspirin therapy with a hazard ratio of 0.98.⁹ The second trial (the Critical Leg Ischaemia Prevention Study [CLIPS]), was conducted in a population that included asymptomatic PAD patients (approximately 25%) and a high proportion of PAD patients with diabetes (approximately 75%).²¹ The CLIPS trial demonstrated a significant benefit for aspirin, with a 64% risk reduction in cardiovascular events in the aspirin-treated group. However the CLIPS trial planned a sample size of 2000 patients followed up for a mean of 2 years but terminated early due to difficulty recruiting patients and nontrial use of aspirin after only 366 participants were randomized.

Variation in patient phenotype may also influence the results of this meta-analysis, which combined patients with symptomatic and asymptomatic PAD; with or without diabetes, claudicants, and need for peripheral revascularization; and with mild, moderate, and severe PAD. Patients with PAD are at heightened cardiovascular risk because of increased atherothrombosis burden,^43- 44 endothelial dysfunction,⁴⁵ platelet activation,^46- 47 insulin resistance, and diabetes.⁴⁸ Thus, a number of mechanisms may explain the increased risk of cardiovascular events in patients with PAD. Whether this increased risk is similar in all types of PAD remains uncertain. Unfortunately, outcomes stratified by phenotype at enrollment were unavailable in many of the studies.

Quiz Ref IDPeripheral artery disease may represent a diffuse form of atherosclerosis with a high inflammatory burden^49- 51 and platelet activation⁵² that may be less responsive to aspirin than other cardiovascular diseases. Despite its irreversible platelet inhibition, aspirin is a weaker antiplatelet drug than other agents. In a prospective randomized trial, picotamide significantly reduced mortality in patients with diabetes and PAD compared with aspirin.⁵³ This may reflect picotamide's greater potency given a dual mechanism of action through inhibition of platelet thromboxane-A₂ synthase and antagonism of thromboxane-A₂ receptors.

Several studies suggest that aspirin delays the rate of PAD progression, reduces the need for lower extremity revascularization, and reduces graft failure in patients who have undergone lower extremity revascularization procedures.^26,30,54 Furthermore, because aspirin reduces cardiovascular morbidity and mortality in other high-risk populations and because it is inexpensive, many physicians and national and international guidelines recommend aspirin as first-line therapy for those with PAD.^10- 11 However, based on the current findings, there is a critical need for future comparative studies of aspirin and newer, more potent antiplatelet agents to test in PAD patients that are designed to assess long-term risks and outcomes. Only appropriately powered trials with adjudication of cardiovascular and bleeding events can provide definitive conclusions regarding the appropriate antiplatelet regimen for patients with PAD. A randomized, double-blind, placebo-controlled trial involving 3350 patients with PAD, the Aspirin in Asymptomatic Atherosclerosis (AAA) study (Trial registration number: ISRCTN 66587262), is currently under way and is likely to provide important information.

The major limitations of this meta-analysis reflect the limitations of published literature on aspirin for treating PAD. Many of these trials were small and of short duration, resulting in few major cardiovascular events. Although no significant benefit of any aspirin therapy was shown for the primary cardiovascular end point, the overall meta-analysis was underpowered to detect risk reductions less than 25%. Similarly, the meta-analysis was underpowered to determine the impact of aspirin on individual components of the primary composite end point. Despite combining multiple studies, the total number of events in this analysis was relatively small. Moreover, the pooled results include trials that were not originally intended to explore cardiovascular outcomes and did not centrally adjudicate cardiovascular events. Similarly, definitions of major bleeding varied across the trials. These variations in major bleeding definitions make it difficult to determine accurate measure of bleeding and approximate risk. Because many studies were performed more than 10 years ago, changes in the diagnosis and treatment of PAD over time may limit the applicability of these results to current patients. Furthermore, a description of patient-level data was not available making it difficult to ascertain whether specific subgroups of PAD patients may benefit from aspirin therapy.

This meta-analysis was conducted using a detailed protocol developed prior to study initiation and used a rigorous search for published and unpublished studies, along with explicit methods for study selection, data extraction, and data analysis. Nevertheless, the generalizability of all meta-analyses is limited by protocol heterogeneity and differences among study populations. A random-effects analysis was used to account for potential differential effects across studies. The results generalize to the treatment of PAD insofar as the included studies ultimately reflect the patients and therapeutic approaches representative of current clinical practice.^55- 56

This meta-analysis did not demonstrate a significant benefit of aspirin on cardiovascular events when compared with placebo or control. However the current evidence was insufficient to rule out small yet important benefits of aspirin (as suggested by the point estimate of a 12% risk reduction). Aspirin did, however, significantly reduce the risk of nonfatal stroke. In most trials included in this meta-analysis, major bleeding events were not formally assessed and therefore a full risk-benefit analysis was not performed. Larger prospective studies of aspirin and other antiplatelet agents are warranted among patients with PAD in order to draw firm conclusions about clinical benefit and risks.