Individualizing Aspirin Therapy for Prevention of Cardiovascular Events

In this issue of THE JOURNAL, He and colleagues¹ present the results of a meta-analysis on the risk of hemorrhagic stroke in patients treated with aspirin at the regimens currently prescribed for the prevention of carotid, coronary, or peripheral artery thrombotic occlusion. This review of a subset of 16 randomized controlled trials of aspirin for the prevention of cerebrovascular accidents, involving more than 55,000 patients, confirms that aspirin, even at the average dosage of 273 mg/d (range, 75-1500 mg/d), increases the risk of cerebral bleeding. Even though the main finding of this study is not new, important lessons can be drawn from this work. First, He et al¹ provide a quantitative estimate of the excess risk due to aspirin instead of the vague notion that the risk is increased. This is another example of a useful application of the meta-analytic approach for evaluation of adverse events, ie, to assess the unbiased excess hazards over their natural occurrence when no single trial can sort out excess risk because the trial is sized for showing an efficacy.² Furthermore, this estimate is obtained from the same patients as those in whom the benefit is measured, thus giving more relevance to the comparison of benefit and risk.

Second, the authors observed that the increase of the absolute risk of hemorrhagic stroke is not related to patient characteristics, such as age, sex, race, preexisting disease, hypertension, and hyperlipidemia. All are risk factors for cardiovascular events. Although the authors could not demonstrate it with certainty, their data suggest that the excess absolute risk of hemorrhagic stroke is constant and independent of the patient's profile and thus, of the absolute risk of cardiovascular event. Other systematic reviews have shown that the relative reduction of ischemic cardiovascular events with aspirin compared with placebo is constant over a wide range of absolute risk.³ In other words, the effect model for aspirin against ischemic events is linear and multiplicative, indicating that the absolute benefit is proportional to the patient's absolute risk of developing an ischemic cardiovascular event.

For the patient and the physician, the practical consequence of this fact is of great importance. For example, a physician is considering prescribing aspirin for 2 patients to prevent cardiovascular events. One patient is a 45-year-old man who has normal blood cholesterol and glucose levels, blood pressure, and weight and who does not smoke. Despite this absence of risk factors, the patient presented 1 month earlier with an uncomplicated inferior myocardial infarction. The other patient is a 65-year-old woman who has had hypertension since age 35 years, has diabetes, has had a previous myocardial infarction, smokes 2 packs of cigarettes a day, and reports breathlessness with exercise. Two weeks earlier, she presented with her second anterior myocardial infarction, along with signs of heart failure at admission that subsided after treatment. The former patient clearly has a very low (absolute) risk of recurrence of myocardial infarction, fatal or nonfatal stroke, or sudden death, whereas the latter patient has a much greater risk. Although tools are not yet available to predict the absolute cardiovascular risk with an acceptable precision in all clinical conditions, clinicians can achieve a qualitative scoring of their patients' risk.⁴ From a risk score established in 6160 unselected patients who survived more than 1 week after admission for an acute myocardial infarction, the predicted 1-year mortality risk would be 2% for the first patient and 30% for the second.⁵ Because the absolute benefit of aspirin is proportional to these risks, the second patient derives much more benefit from aspirin than the first patient, with expected absolute risk reductions of 0.3% and 4.8%, respectively. The excess risk of cerebral bleeding with aspirin is 0.12% for both patients, according to the meta-analysis of He et al.¹ This value of excess risk might appear to the physician, the patient, or both to be too high compared with the small reduction in mortality of 0.3% for the first patient. However, both physician and patient would agree that the predicted benefit outweighs the risk of cerebral bleeding for the second patient.

These values should be used only as an illustration because they have several important limitations, eg, the estimates were derived from data collected prior to the era of β-blockers and angiotensin-converting enzyme inhibitors (the 1-year mortality in post–myocardial infarction patients is lower today, which decreases the absolute risk reduction in patients with the same secondary risk factors as these 2 patients), and the 0.12% excess risk of cerebral bleeding covers more than 1 year. However, these estimates illustrate the logic of the process. Clinicians can identify with enough precision the patients for whom risk really outweighs benefit only through pooling individual data, unless markers of predisposition to cerebral bleeding are discovered.

For primary prevention, in which the risk of cardiovascular events is, on average, much less than in secondary prevention, the excess risk may well exceed the benefit of therapy. This is similar to use of class I antiarrhythmics in patients following myocardial infarction or anticoagulants in patients with atrial fibrillation.^{6 - 7} In low-risk patients, particularly in primary prevention, clinicians should be cautious in prescribing aspirin. Applying the Framingham equation to a 40-year-old man with hypertension, a cholesterol level of 5.17 mmol/L (200 mg/dL), a high-density lipoprotein cholesterol level of 2.59 mmol/L (100 mg/dL), a systolic blood pressure of 164 mm Hg, no left ventricular hypertrophy, and normal glycemia, and who is not a smoker, gives a predicted risk of 0.1% of having a cardiovascular event within 1 year, a rate that is lower than the anticipated excess risk of cerebral bleeding.⁸ These examples illustrate that physicians should think in terms of absolute risk and benefit rather than in terms of relative risk and benefit.

Although the benefit of aspirin varies with the patient risk factors for ischemic cardiovascular events, the meta-analysis by He et al¹ suggests that the increased risk of aspirin causing hemorrhagic stroke does not vary with these risk factors. Thus, the risks and benefits of aspirin therapy should be evaluated in light of each patient's individual risk profile for cardiovascular events.