The Cardiovascular Benefit of Lp(a) Reduction: Not There Yet

Angela Pirillo; Alberico Luigi Catapano


Eur Heart J. 2020;41(44):4256-4258. 

Lipoprotein (a) [Lp(a)] is an independent, genetically determined risk factor for atherosclerotic cardiovascular disease (CVD), and Mendelian randomization studies are suggestive of a causal relationship between elevated Lp(a) and the risk of CVD.[1] However, while reducing LDL-cholesterol (LDL-C) levels is clearly associated with a reduction of CV risk in a number of randomized clinical trials, randomized evidence for the reduction of Lp(a) levels is still lacking. In fact, some of the LDL-C-lowering drugs reduce Lp(a), but the interpretation of the data is confounded by the concomitant decrease of LDL-C.

On top of these considerations, two Mendelian randomization studies have established that relatively large absolute reductions in Lp(a) (~100 mg/dL, or ~250 nmol/L; and ~65 mg/dL, or ~160 nmol/L, respectively) are required to achieve a CV risk reduction comparable with that observed with a 39 mg/dL (1 mmol/L) LDL-C reduction (20–25% relative risk reduction).[2,3] In agreement with these observations, an analysis of the HPS2-THRIVE trial showed that niacin–laropiprant lowered Lp(a) levels by 5 mg/dL (12 nmol/L) overall and 13.5 mg/dL (34 nmol/L) in the top Lp(a) quintile, and that these Lp(a) reductions resulted in reductions in coronary risk of ~2% overall and 6% in the top quintile by Lp(a) levels.[4] This contention is supported by the observation that, in Mendelian randomization studies, a 10 mg/dL genetically determined reduction of Lp(a) level is associated with a 5.8% lower risk of coronary heart disease, whereas a 10 mg/dL lower genetically predicted LDL-C level is associated with a 14.5% lower coronary heart disease risk.[2]

Current guidelines recommend that Lp(a) is measured at least once in life, to identify subjects with genetically determined high levels of Lp(a), and indicate that an Lp(a) ≥180 mg/dL, or ≥430 nmol/L, confers a CV lifelong risk similar to that of subjects with heterozygous familial hypercholesterolaemia.[5]

Statin therapy may increase Lp(a),[6] although this increase appears to be clinically relevant only in patients with high Lp(a) levels.[7] PCSK9 (proprotein convertase subtilisin/kexin type 9) monoclonal antibodies consistently reduce Lp(a) levels by 20–30%,[8] but the clinical value of this reduction in CV outcomes is still under scrutiny. The ODYSSEY OUTCOMES trial showed that alirocumab-mediated LDL-C reduction significantly reduced the occurrence of first major adverse cardiovascular events (MACE) in patients with a recent acute coronary syndrome (ACS), and also showed that Lp(a) was reduced by 5 mg/dL (23%).[9] The risk of MACE (coronary heart disease death, non-fatal myocardial infarction, ischaemic stroke, or hospitalization for unstable angina) was reduced by 15% compared with placebo.[9] Baseline levels of Lp(a) predicted the risk of MACE in the placebo group, and increased across Lp(a) baseline quartiles.[9] The occurrence of MACE or coronary heart disease death plus non-fatal myocardial infarction was in fact significantly higher in participants in the highest quartile of Lp(a) compared with those in the lowest quartile, and this association was even stronger when adjusted for baseline LDL-C levels.[9] Similar results were reported for first peripheral artery disease and thrombo-embolic events.[10]

As post-ACS patients are at very high risk of recurrent CV events, a clinically relevant question to be addressed is whether alirocumab added to their baseline lipid-lowering therapy prevents a higher number of total CV events compared with the number of first events, a finding reported also with intensive statin therapy and other drug classes.[11] In this issue of the European Heart Journal, a further analysis of the data from the ODYSSEY OUTCOMES trial is presented, which aimed at assessing the relevance of both Lp(a) baseline levels and alirocumab-induced Lp(a) changes in predicting total cardiovascular events.[12] The authors have included all CV events at follow-up, including CV death and total (first and recurrent) non-fatal CV events. A total of 48% of patients had a non-fatal first CV event and experienced multiple events during the trial.[12] In non-treated patients (placebo group) the risk of a CV event steadily increased across baseline Lp(a) quartiles, with a 60% adjusted higher risk for patients in the highest vs. lowest quartile. Overall, the alirocumab:placebo hazard ratio (HR) for a first CV event was 0.88, and the HR for total CV events was 0.85, with a significant linear trend across baseline Lp(a) quartiles (from 0.95 in the first quartile down to 0.75 in the fourth quartile).[12] The relative contribution of Lp(a) to the absolute risk reduction increased more than seven-fold across quartiles, from 0.5 events per 100 patient-years of follow-up in the first quartile to 3.7 events per 100 patient-years in the fourth quartile.[12]

Take home figure.

Pharmacological control of LDL-C and/or Lp(a) and clinical outcomes. The pharmacological approach with statins (which significantly reduce LDL-C levels without major effects on Lp(a)) or statin+PCSK9 mAbs (which reduces both LDL-C and Lp(a)) significantly reduces the incidence of cardiovascular events. Whether the specific inhibition of Lp(a) with mRNA-modulating approaches (which massively reduce Lp(a) without significant effect on LDL-C) translates into a cardiovascular benefit remains to be addressed.

Further relevant information provided by this analysis is that Lp(a) lowering induced by alirocumab predicted a reduced risk of total CV events independently of LDL-C levels, with each 5 mg/dL reduction in Lp(a) being predictive of a 2.5% relative reduction in total CV events.[12] In the alirocumab group, higher baseline Lp(a) levels were associated with greater absolute and relative risk reductions in total CV events: Lp(a) was reduced by 20 mg/dL (50 nmol/L) in the fourth quartile, in which Lp(a) contributes 39% of the joint predicted risk reduction (compared with no contribution in the first quartile).[12] Altogether these observations suggest that, in patients with Lp(a) ≥60 mg/dL (≥150 nmol/L), a relative reduction of at least 10% of their CV risk due to Lp(a) lowering would be expected. A pooled analysis of alirocumab phase III trials (not including the ODYSSEY OUTCOMES study) showed that, in patients not selected on the basis of elevated Lp(a), the CV benefit observed with PCSK9 inhibitors is mainly explained by the reductions of LDL-C rather than any Lp(a)-lowering effect;[13] however, in the subgroup of patients with higher baseline Lp(a) levels (≥50 mg/dL, ≥125 nmol/L), a significant association between reduction of Lp(a) and MACE was observed.[13] Thus, patients with a greater CV risk due to higher baseline levels of Lp(a) may benefit more from the treatment with PCSK9 inhibitors, due to their ability to also reduce Lp(a). This evidence supports the recommendation to test Lp(a) in people at high risk, allowing a better risk stratification.

As the authors correctly state among the limitations, the indirect measurement of the 'true' LDL-C reduction may introduce errors, and the validity of the observed correlations for events occurring in other CV districts remains to be addressed. Further, it remains a matter of debate whether the very low LDL-C levels achieved with PCSK9 inhibitors may contribute to the Lp(a)-mediated CVD risk. In fact, although LDL-C and Lp(a) are independently associated with CV risk, at low LDL-C levels, the risk associated with high Lp(a) levels is attenuated, at least in a context of primary prevention.[14] Further, the analysis of the total CV events deviates from the principle of randomization and therefore makes the present evidence less robust. The obvious solution will be an approach that specifically lowers Lp(a).

To date, no approved therapies to lower Lp(a) levels selectively are available. An antisense oligonucleotide targeting apo(a) mRNA is now under clinical evaluation, and it has the potential to allow robust Lp(a) lowering: the recent results of a phase II trial showed that Lp(a) could be reduced up to 80% at the highest dose (corresponding to ~190 nmol/L, or 75 mg/dL) in patients with elevated Lp(a) levels.[15] The potent Lp(a) lowering achieved with the antisense oligonucleotide appears also to be associated with a reduced inflammatory state, a finding not reported in patients treated with a PCSK9 monoclonal antibody (who, on the other hand, experienced a robust LDL-C reduction).[16] A phase II study is currently evaluating the efficacy, safety, and tolerability of an N-acetylgalactosamine-conjugated small interfering RNA (siRNA) in 240 subjects with Lp(a) >60 mg/dL (>150 nmol/L) (

Although the present data are quite suggestive, we will need to wait until data on CV events from these studies will be available to finally solve the question of whether decreasing Lp(a) (and by how much) contributes significantly.