Outcomes following aspirin discontinuation among baseline users in contemporary primary prevention aspirin trials: A meta-analysis

Based on available evidence, including the results of 3 trials published in 2018 (ASPREE, ASCEND, and ARRIVE)^1-3, guidelines no longer recommend low-dose aspirin for primary prevention of atherosclerotic cardiovascular disease (ASCVD) in most adults. Indeed, US guidelines now advise against the use of aspirin for primary prevention among adults aged over 70 years. However, the majority of participants in these 3 trials were not previously taking aspirin. Therefore, these trials predominantly tested the effect on ASCVD outcomes of randomization to initiate aspirin vs. placebo among aspirin naïve adults. This raises a key clinical question, should aspirin be discontinued in persons already being treated with it?

Compared with continuing aspirin, results from observational studies suggest a 28% higher ASCVD risk among adults characterized as primary prevention who discontinue aspirin in the absence of a clear indication.⁴ Determining whether this observed higher risk after aspirin discontinuation is causal or simply associated with confounding requires randomized data.

Among the 47,140 participants in the 3 randomized trials published in 2018 assessing low-dose aspirin versus placebo for primary ASCVD risk prevention, approximately 15% (7,222) were regularly taking aspirin prior to enrolment (the ARRIVE trial excluded persons taking aspirin at trial screening²). These 7,222 participants provide an opportunity for exploratory analyses to evaluate outcomes based on randomization to continue or stop aspirin among baseline users. In a previously published subgroup analysis, the ASPREE trial authors studied 1,714 participants who were taking aspirin at least twice weekly before enrolment. Those randomized to stop aspirin and take placebo had a trend to increased risk for ASCVD events compared with continuing aspirin (55/841 [6.5%] vs. 44/873 [5.0%]; HR 1.32, 95% CI 0.89 to 1.97), though these analyses are based on only 99 total events yielding uncertain precision of the estimates, with these limitations acknowledged by the authors.⁵

In this study, trial-level meta-analyses were performed of the subgroup of participants taking aspirin at baseline in both the ASPREE and ASCEND trials (n=7222).^{1, 3} Both trials were IRB-approved and all participants provided written informed consent for research. De-identified data were obtained from the primary publications of both trials.^{1, 3} Data that support the study findings are available from the corresponding author. Outcomes were the composite of fatal or non-fatal ASCVD for efficacy and major bleeding for safety (see the Figure for more details on outcome definitions in each trial). Random-effects models with restricted maximum likelihood were used for meta-analyses. I² statistics assessed for heterogeneity across trials.

Figure: Meta-analytic Effect of Stopping vs Continuing Aspirin on ASCVD (Panel A) and Major Bleeding (Panel B) among Baseline Aspirin Users.

Panel A: ASCVD was defined as a composite of fatal coronary heart disease, nonfatal myocardial infarction, fatal or nonfatal stroke, or hospitalization for heart failure in ASPREE and a composite of nonfatal myocardial infarction, nonfatal stroke (excluding confirmed intracranial hemorrhage) or transient ischemic attack, or death from any vascular cause (excluding confirmed intracranial hemorrhage) in ASCEND. Panel B: Major Bleeding was defined as a composite of hemorrhagic stroke, symptomatic intracranial bleeding, or clinically significant extracranial bleeding in ASPREE and as a composite of any confirmed intracranial hemorrhage, sight-threatening bleeding event in the eye, gastrointestinal bleeding, or any other serious bleeding (i.e., a bleeding event that resulted in hospitalization or transfusion or that was fatal) in ASCEND.

Higher risk was found for the composite of fatal or non-fatal ASCVD among baseline aspirin users who were randomized to stop aspirin and take placebo vs. those randomized to continue aspirin (450/3609 [12.5%] vs. 374/3613 [10.4%]; HR 1.21, 95% CI 1.05 to 1.39). The risk for major bleeding was not significantly different among baseline aspirin users who were randomized to stop vs. continue aspirin (123/3609 [3.4%] vs. 142/3613 [3.9%]; HR 0.86, 95% CI 0.68 to 1.10). For both outcomes, there was no evidence of significant trial-level heterogeneity (Figure).

The results of this trial-level meta-analysis suggest that individuals treated with aspirin for primary prevention who discontinue aspirin may be at higher subsequent risk for ASCVD events but without a significantly lower risk for major bleeding. The 21% relative excess risk for incident ASCVD events observed among baseline aspirin users randomized to discontinue aspirin in this meta-analysis is similar to previously reported observational data.⁴ In addition, the lack of a significantly lower risk of major bleeding among individuals who were taking and then stopped aspirin appears consistent with a clinically relevant background survivor bias with respect to major aspirin-related bleeding, since these persons entered the trials having avoided prior major aspirin-associated bleeding complications. The present results suggest that data from primary prevention trials testing the hypothesis of initiating aspirin vs placebo in adults who are aspirin-naïve may not apply to the clinical scenario of aspirin discontinuation among persons already taking aspirin. Interestingly, only one of the 3 trials reported in 2018 demonstrated a benefit of aspirin on ASCVD outcomes,³ and this trial included far more participants who were taking aspirin prior to trial entry than the other two null trials. ^{1, 2}

Limitations of these meta-analyses include the analysis of summary trial-level data from just 2 trials (instead of individual patient-level data); the fact that outcome definitions in the trials differed slightly (Figure); the relatively small number of events; and the fact that neither trial demonstrated statistical interaction on the effect of aspirin randomization according to baseline aspirin use (though neither trial was powered to do so).

These results are hypothesis-generating but are the best available data. Pending further evidence, it seems reasonable that adults already safely treated with low-dose aspirin for primary prevention should be continued on it - unless significant risk factors for aspirin-related bleeding develop in which case a risk: benefit discussion should occur.

Non-standard Abbreviations and Acronyms

ARRIVE

Aspirin to Reduce Risk of Initial Vascular Events

ASCEND

A Study of Cardiovascular Events iN Diabetes

ASCVD

Atherosclerotic Cardiovascular Disease

ASPREE

ASPirin in Reducing Events in the Elderly

CI

Confidence Interval

HR

Hazard Ratio