Aspirin for primary prevention: time to reconcile discrepancies

Fifteen years ago, on the wake of consolidated opinions about the advantageous use of aspirin in secondary prevention of coronary heart disease, cerebrovascular disease, post-bypass and carotid surgery, essential thrombocythaemia, polycythaemia vera, pre-eclampsia, and thrombophilic syndromes, aspirin popularity as a therapy in cardiovascular disease was so common that a rushed extension of its use to primary cardiovascular prevention was predicated. The hype was cooled down by a series of inconsistent results, culminating, in 2018, in those of three megatrials, ASPREE,1 ARRIVE,2 and ASCEND,3 all formally inconclusive in showing significant benefit in the balance between ischaemic events prevented and bleeding events procured. Subsequent more accurate interpretation of ASCEND in diabetic subjects3 and the favourable results later observed in TIPS-2 with aspirin added either to placebo or to a polypill combination of ramipril, atenolol, hydrochlorothiazide, and simvastatin4 have again raised perplexities over non-nuanced conclusions. In 2014, we had proposed a ‘weighted meta-regression’ of trials available at that time of the aspirin benefit as a function of baseline cardiovascular risk, suggesting a relationship of the benefit/risk balance with baseline risk risk.5 Here, we re-propose a similar analysis, now incorporating all the most recent evidence. Here, the effect of aspirin in relation to cardiovascular risk was assessed using a univariate linear meta-regression model. For each study, the 10-year risk of major adverse cardiovascular events (MACE: death, myocardial infarction, and stroke and cardiovascular death, myocardial infarction, and stroke) in the control group was calculated and included in the model as the independent variable. The relative benefit or harm from aspirin vs. placebo was estimated as the difference in absolute risk per 100 subjects-year and served as the dependent variable for the meta-regression. For each endpoint—total MACE or total major bleeding—this value was computed as the difference between the risk/years of follow-up per 100 subjects in the control group and the risk/years in the aspirin group. Two linear regression analyses depict the relative reduction of ischaemic events and the relative increase in bleeding as functions of cardiovascular risk. Each study was weighted for the number of subjects included. Slopes of the regression lines were compared using the Student’s t-test; upper and lower limits of the 95% confidence interval—for both MACE and total major bleeds—were estimated with a point-by-point approach for each meta-regression line based on the adjusted difference in absolute risk per 100 subjects-year.5 Results in the Figure 1 show that the regression line depicting the benefit—protection from MACE, however expressed—significantly (two-sided P < 0.05) increases as a function of cardiovascular risk in the control arm of the studies. Most relevant, however, and contrary to a widely held belief, the line depicting risk—occurrence of major bleeding—while also increasing with risk (a finding expected because several risk factors such as age and comorbidities are related to both ischaemic and bleeding events), does so less steeply, to the point that the two lines more and more diverge as risk increases (Figure 1), approaching levels found in the most recent prevention trials in secondary cardiovascular prevention. The figure also illustrates the considerable gap in evidence from trials in the range of cardiovascular risk estimated at 20/100 patients at 10 years and beyond. The effect of aspirin on major adverse cardiovascular events [(A) MACE-1—all-cause death + non-fatal myocardial infarction + non-fatal stroke; (B) MACE-2: cardiovascular death + non-fatal myocardial infarction + non-fatal stroke] and on major bleeding events in primary prevention trials (for a complete list, see5 plus1–3 plus4) in relation to cardiovascular risk, assessed using univariable linear meta-regression models. Two separate linear regression analyses were calculated for each panel to examine both the relative reduction of major adverse cardiovascular events (blue line) and the relative increase in bleeding (red line) as functions of cardiovascular risk. The upper and lower limits of the 95% for both major cardiovascular events and total major bleeding events were estimated for each meta-regression (dotted blue lines for major adverse cardiovascular events and dotted red lines for major bleeding events). Each panel includes two separate weighted least squares meta-regression models. The first depicts the association between baseline cardiovascular risk and the absolute risk reduction in the composite major adverse cardiovascular events outcome; the second depicts the association between the same baseline risk and the absolute risk change in major bleeding events. This dual analysis enables a simultaneous evaluation of both cardiovascular benefit and bleeding risk as functions of baseline event rate. Weighted least squares regression accounts for heterogeneity in the precision of effect estimates across studies. Weights were calculated as the inverse of the squared standard error (1/SE2) of the absolute risk change.5 The standard error for each study was derived analytically based on the event rate and the sample size, so that both the variability of the outcome and the number of participants contributed to the weight. This method therefore assigns greater weight to studies with larger sample sizes and more precise estimates, improving efficiency and validity of the regression. The weighted least squares approach also addresses potential heteroscedasticity by ensuring that studies contribute to the model proportionally to the reliability of their estimates. All analyses were performed using SPSS (IBM SPSS Statistics for Windows, Version 29.0.1.0, IBM Corp, Armonk, NY), and a two-sided P-value < 0.05 was considered statistically significant. While detection of significant asymptomatic atherosclerosis, such as plaques in the carotid, lower limb or coronary arteries, or a high calcium score, is already considered a condition where the use of aspirin ‘might’ be considered, the graphs shown in the Figure 1 indicate that a window of opportunity also exists for subjects without overt evidence of atherosclerosis, provided that a calculated cardiovascular risk exceeds a projected number of MACE > 20/100 patients at 10 years, but also most likely between 10 and 20/100 patients at 10 years, complementing the 2022 US Preventive Services Task Force Recommendation Statement.6 Limitations of this analysis are the reliance upon aggregated data and not on data of individual patients; in not fully capturing the decrease in baseline cardiovascular risk observed in recent years for similar levels of risk factors likely due to the wider implementation of preventive—mostly pharmacological—measures, including a larger and larger use of statins; in the imprecision of risk factor estimates; and in not capturing the likely variability in the benefit-risk ratio in older7 or frail patients, or those with renal failure, diabetes, or other comorbidities. Some other considerations would favour, however, continued use of aspirin in high-risk classes in primary prevention: The reversible nature of most bleeding—except the rare intracranial haemorrhage, for which a 25% increase with aspirin has been estimated. Reversible bleeding has been so far inappropriately weighted similarly against the always irreversible nature of a myocardial infarction or of an ischaemic stroke The largely preventable risk of bleeding with the concomitant use of proton pump inhibitors, to be advocated certainly in the elderly taking aspirin, but possibly also earlier in susceptible patients The ‘potentially favourable’ effect of aspirin on tumorigenesis, not captured in the relatively short-term follow-ups of cardiovascular studies The advantage demonstrated in TIPS-2 in addition to other preventive strategies, including a statin4 We certainly advocate the more and more advertised ‘shared decision-making’, explaining to patients that aspirin use has uncertainties and that aspirin should always be in any case given in addition to lifestyle measures; adequate blood pressure control, important to reduce the risk of intracranial bleeding; and the larger use of lipid-lowering medications, all measures having less problematic benefit-risk balances. This analysis, however, advocates the responsible use of aspirin in medium- to high-risk primary prevention. Alerts should always be offered on measures limiting the risk of bleeding—most notably intake with or soon after a meal, wider use of proton pump inhibitors, and recommendations to avoid non-steroidal anti-inflammatory drugs as much as possible. We suggest, however, that a non-negligible proportion of patients coming to the attention of cardiologists/general practitioners with risk factors but without a previous vascular event and even without evidence of overt atherosclerosis should be offered the chance of aspirin protection when exceeding the above-described risk thresholds considering the very artificial boundary between primary and secondary cardiovascular prevention. Authorship: R.D.C. contributed to the conception and design of the work. M.M. contributed to the data acquisition and performed the analysis. R.D.C. supervised the analysis. Both R.D.C. and M.M. contributed to the interpretation of data. R.D.C. drafted the manuscript. M.M. critically revised the manuscript. Both gave final approval and agree to be accountable for all aspects of work ensuring integrity and accuracy. The data underlying this article are available in the article and in the references provided.