Aspirin for primary cardiovascular prevention in individuals with high levels of lipoprotein (a)? The jury is still out

This editorial refers to ‘Efficacy and safety of aspirin for primary prevention in adults with elevated lipoprotein(a) or genetic susceptibility via LPA gene variants: a systematic review and meta-analysis’, by R.A. Lopes et al. Lipoprotein(a) [Lp(a)] consists of an apolipoprotein(a) [apo(a)] covalently bound via a single disulfide bond to the apolipoprotein B-100 (apoB).1 Multiple large epidemiological studies have shown higher cardiovascular risk in patients with high levels of Lp(a) in both primary and secondary cardiovascular prevention settings,1 and large Mendelian randomization studies and genome-wide association studies suggest high probability of causality of high serum levels with cardiovascular events.1 Quoted mechanisms for the suggested causal link between Lp(a) levels and cardiovascular events are the induction and promotion of atherosclerosis, perhaps through the carrying of oxidized phospholipids that trigger vascular inflammation, and the induction of thrombosis. The structural similarity of apo(a) to plasminogen has indeed, since many years, raised the hypothesis of a higher thrombotic risk in individuals with high levels of this protein, related to inhibition of fibrinolysis.1 Either directly with a prothrombotic effect or indirectly through promotion of atherothrombosis, thrombotic events appear in any case to be the final link between high plasma levels of this molecule and adverse major cardiovascular events (MACE, conventionally defined as cardiovascular death, myocardial infarction, and stroke). As such, antithrombotic therapies in people with high levels of Lp(a) appear to be a logical approach to cardiovascular prevention. Aspirin, the oldest and time-honoured antithrombotic drug,2 has clearly an opportunity of indication in this setting, if it were not for the fact that the benefit–risk ratio in primary cardiovascular prevention continues to be a source of debate and uncertainty, reflected in international guidelines. It is now clear that the higher risk of bleeding related to aspirin use offsets benefits in the overall population of still healthy people never having suffered a manifestation of cardiovascular disease. It is, however, also quite clear that ‘primary cardiovascular prevention’ encompasses a huge range of cardiovascular risk, from young healthy people to the elderly with several cardiovascular risk factors, perhaps one day before they suffer from a myocardial infarction or an ischaemic stroke. As such, the qualifications of which populations to target with aspirin in primary cardiovascular prevention are essential, and many strategies have been proposed. One is to assess cardiovascular risk through risk calculators, e.g. the Systematic Coronary Risk Evaluation 2 (SCORE2) and its variants proposed by the European Society of Cardiology, and to reserve the use of aspirin to individuals beyond a certain threshold, which we have identified at a risk of MACE above 20/100 patient-10 years, with a further grey zone of possible use in classes of risk between 10 and 20/100 patient-10 years. At such levels of risk, the regression line relating MACE averted with aspirin diverges from the regression line relating excess bleeding events prospectively caused using aspirin. This separation of benefit (reduction of MACE) and of risk (induction of bleeding) suggests a favourable benefit–risk ratio.3 Another way to select a target population for aspirin use in primary cardiovascular prevention would be to use markers of the presence of already established atherosclerosis, such as the presence of carotid,4 aortic or coronary plaques, now detectable with carotid artery ultrasound or computed tomography angiography,5 or to use calcium score selecting patients with high such score6 or—but not yet demonstrated—those with a pathological ankle-brachial index, which indicates relevant aorto-ilio-femoral atherosclerosis. In this issue of the European Journal of Preventive Cardiology, Lopes et al. present a meta-analysis investigating the hypothesis that, in primary prevention, low-dose aspirin is associated with a significant benefit–risk trade-off in patients with high levels of Lp(a) or carrying gene variants predisposing to high levels of Lp(a).7 As of now, Lp(a) is not (yet) included in current risk calculators. The meta-analysis by Lopes et al. comprised 6 trials with a total of 6628 participants. Pooled random-effects showed no significant reduction in MACE with aspirin [hazard ratio (HR) = 0.60; 95% confidence interval (CI) 0.31–1.17; P = 0.10] but also no significant increase in clinical bleeding events (HR = 1.24; 95% CI 0.83–1.87; P = 0.18). Subgroup analyses showed a non-significant reduction in MACE restricting the analysis to subjects with elevated (>50 mg/dL) Lp(a) (HR = 0.63; 95% CI 0.16 to 2.49; P = 0.284) and statistically significant reduction among LPA rs3798220 carriers (HR = 0.37; 95% CI 0.19 to 0.71; P = 0.003). The authors interpret these findings as indicating that aspirin use did not confer a statistically significant cardiovascular benefit and was associated with numerically higher bleeding risk. The apparent reduction seen in rs3798220 carriers in the authors’ interpretation likely reflects small-sample bias. This study clearly does not refute the null hypothesis—and therefore cannot unequivocally support the hypothesis of a benefit for aspirin in individuals without previous manifestations of cardiovascular disease and qualified as high-risk based on high levels of Lp(a) or a genetic propensity to such high levels. The study, for which the authors have anyway to be congratulated, is, however, clearly not definitive, for several reasons, detailed here below. First, the study shows a numerically lower (40%) rate of MACE in aspirin users, at the cost of 24% higher rate of major bleeds. Although both are not statistically significant, they are still consistent with the hypothesis of a favourable trade-off, not here conclusively demonstrated for the wide confidence intervals around the mean estimate—an example of a type 2 statistical error—negating the original hypothesis because of insufficient sample size. Second, by looking at the rates of MACE in the individual studies listed in the meta-analysis, one notices that none of them has an absolute risk of such events >20/100 patient-10 years, suggesting that Lp(a) in the studies here included, perhaps in the absence of other risk factors, confers an ‘intermediate’—not a very high—level of risk, where the net effect of aspirin remains uncertain. Third, this concept is further reinforced by the observation that the level of risk across studies here included is quite heterogeneous (I2 ≈ 87%), suggesting that high levels of Lp(a) themselves do not identify a prognostically homogeneous population. Fourth, such heterogeneity among studies appears to be markedly reduced (I2 ≈ 37%) by excluding the study by Colantonio et al.8, characterized by a low risk. By excluding this study (leave-one-out analysis), the benefit signal for aspirin becomes statistically significant (HR = 0.50, 95% CI 0.27–0.94). Fifth, the adoption of a fixed threshold for Lp(a) levels (≥50 mg/dL) does not allow to distinguish risk across different levels of exposure, with higher Lp(a) levels possibly associated with a better net effect. Sixth, the benefit–risk estimates are here making, as—unfortunately—routinely done—one MACE (cardiovascular death, myocardial infarction or stroke) as equally relevant as one major bleeding, which is usually a reversible event for aspirin users, given the rarity of intracranial bleeding—the only type of irreversible event with persistent consequences associated with aspirin use. We have repeatedly emphasized the inappropriateness of such unweighted estimates of the benefit-risk evaluation.3,9 Finally, there was an apparent benefit in rs3798220 carriers: this is cautionarily downplayed but is still consistent with the original hypothesis of a conferred benefit in such patients. In essence, as in many other examples, the problem is not in the data but in their interpretation, and more studies on this topic are warranted. Several trials with different agents already proven as able to drastically reduce levels of Lp(a) are ongoing or planned, and one of them, the HORIZON trial with the antisense oligonucleotide pelacarsen,10 is going to be reported very soon this year. Should this or other trials prove effective in reducing cardiovascular risk associated with Lp(a), clearly such a therapeutic strategy addressing the root of the increased cardiovascular risk in such patients would be preferable. Indication to the use of aspirin, if proven by larger studies or meta-analyses, will then be confined to people unable to receive such innovative drugs. Raffaele De Caterina (Conceptualization, Validation, Writing—original draft [lead], Formal analysis, Writing—review & editing [equal]), and Gabriele Masini (Data curation, Validation [equal], Writing—review & editing [supporting]) No funding available for this commentary.