Lipoprotein (a), Arterial Inflammation, and PCSK9 Inhibition

Jean-Claude Tardif; Eric Rhéaume; David Rhainds; Marie-Pierre Dubé


Eur Heart J. 2019;40(33):2782-2784. 

Lipoprotein(a) [Lp(a)] is composed of an LDL particle in which apolipoprotein B is covalently bound to apolipoprotein(a) by a disulphide bond (Figure 1). The Lp(a) particle is associated with multiple pro-atherogenic, pro-thrombotic, and pro-inflammatory effects.[1] The latter are mediated by the increased oxidized phospholipid content in Lp(a), increased cytokine and interleukin expression and release, and heightened monocyte chemotaxis.[1] A meta-analysis of observational clinical studies has shown a linear relationship between Lp(a) concentrations >75 nmol/L (30 mg/dL) and cardiovascular events, while genetic studies have linked genetically elevated Lp(a) with the risk of myocardial infarction.[2,3] However, elevation of this biomarker to the status of actionable risk factor would require demonstration that therapeutic alteration of the Lp(a) level affects cardiovascular endpoints.

Figure 1.

Lipoprotein (a) characteristics.

In this issue of the European Heart Journal, Stiekema et al. report the results of a clinical study conducted to determine the effect of the PCSK9 (proprotein convertase subtilisin/kexin type 9) inhibitor evolocumab on arterial wall inflammation in 129 patients with an elevated plasma Lp(a) concentration [125 nmol/L (50 mg/dL) or above].[4] Patients also needed to have a fasting LDL-C level of ≥2.6 mmol/L and a target to background ratio of at least 1.6 measured with [18F]fluoro-deoxyglucose positron emission tomography (FDG-PET) in the most diseased segment of an index vessel (carotid artery or thoracic aorta). Despite reductions of 61% and 14% in LDL-cholesterol (LDL-C) and Lp(a) concentrations, respectively, with monthly subcutaneous injections of evolocumab 420 mg, no significant change over time in arterial inflammation was observed between treatment groups. The change from baseline to 16 weeks in the primary endpoint of the most diseased segment target to background ratio was –8.3% and –5.3% in the evolocumab and placebo groups, respectively (P = 0.18). The decrease in the primary endpoint observed in both study arms may have represented at least in part regression to the mean, considering that patients were selected based on a target to background ratio value of at least 1.6 at baseline. Numerical trends in favour of evolocumab compared with placebo were also observed in exploratory PET endpoints.

These results raise the question of the reasons for the absence of significant benefits of evolocumab on PET endpoints in this study. The first consideration is the utlization of FDG-PET, which is used clinically for the identification of metabolically active foci.[5] FDG is a positron tracer associated with a glucose analogue that accumulates in cells at a rate that is directly proportional to their glycolysis rate.[6] Macrophages contribute to this cellular FDG uptake because of their high basal glycolytic rate and dependence on an external glucose source due to the lack of glycogen storage.[7,8] FDG uptake has been shown to correlate with the extent of macrophage infiltration in carotid plaques of patients scheduled for carotid endarterectomy.[9] Several groups have also reported that the arterial uptake of FDG is reduced by statin therapy.[10] While these characteristics appear to support the use of FDG-PET for assessing arterial inflammation, cellular assays have suggested that FDG uptake might actually reveal macrophage hypoxia and cytokine-activated smooth muscle cells.[11] Thus, the meaning of statistically significant changes in arterial FDG uptake or lack thereof is not entirely certain. Secondly, the reproducibility of PET measurements and their changes over time could also affect the ability to reach statistical significance. Although the authors mention that repeated analyses were performed for determination of intraobserver and interobserver variability, these results were unfortunately not provided to readers.

Thirdly, the follow-up duration of 16 weeks might have been too short to allow detection of a treatment effect on PET endpoints with evolocumab. Given that statins have yielded significant effects on FDG-PET imaging at 12 weeks of therapy, this possibility appears less likely. Fourthly, the type of patients included in this study warrants comments. The median concentration of high-sensitivity C-reactive protein was 1.1 mg/L at baseline, indicating the lack of detectable systemic inflammation in the study population. However, this plasma biomarker was previously shown not to be correlated with FDG-PET endpoints.[12]

Fifthly, the selection of patients with elevated Lp(a) concentrations—approximately 200 nmol/L or 80 mg/dL—at baseline is perhaps a design issue of greater importance that represented a double-edged sword for this clinical study testing a PCSK9 inhibitor. Indeed, such patients are at higher cardiovascular risk than those with lower concentrations. Furthermore, Lp(a) has potential pro-inflammatory effects that were described above. PCSK9 inhibition also reduces plasma Lp(a) concentrations, at least in part through enhanced particle clearance.[13] In contrast, the Lp(a)-reducing effect of evolocumab is blunted in patients with elevated values at baseline, to ~15% in previous studies.[14] The smaller apo(a) isoforms present in patients with high Lp(a) concentrations may be less efficiently cleared by the LDL receptor, although other mechanisms may be at play. That observation may at first lead us to question the rationale for studying this population, but the investigators' hope was that robust LDL-C and modest Lp(a) reductions would yield demonstrable benefits in patients with elevated plasma concentrations of Lp(a) at baseline. The mean placebo-corrected reduction in Lp(a) was 13.9% with evolocumab in the current study, corresponding to a median absolute change of 28.0 nmol/L (11.2 mg/dL). The median on-treatment Lp(a) concentration of 188.0 nmol/L (75.2 mg/dL) in the evolocumab group remained high and may have contributed to the lack of significant reduction in arterial inflammation.

Randomization was stratified by background statin therapy in the current study, and 54% of patients were taking statins. Although the treatment difference in the PET primary endpoint was –5.6% (P = 0.045) in patients who received statins and 0.1% (P = 0.98) in those who did not, the interaction between study treatment (evolocumab or placebo) and statin use was not significant (P = 0.21). Whether this result represents a real phenomenon in an underpowered analysis cannot be determined in this study. Of note, a meta-analysis of statin outcome trials involving >29 000 patients observed that Lp(a) concentrations were more strongly associated with cardiovascular risk in patients assigned to statins than those receiving placebo.[15] In addition, a study investigating the genetic causes of residual cardiovascular risk in statin-treated patients found the LPA gene encoding apolipoprotein(a) to be associated with coronary events during therapy, independently of the extent of LDL-C lowering.[16]

A Mendelian randomization study has suggested that a reduction of Lp(a) of 250 nmol/L (or 100 mg/dL) may be required to achieve cardiovascular benefits.[17] Lifelong differences in Lp(a) concentrations evaluated in Mendelian randomization studies may not reflect the cardiovascular responses to more rapid reductions in such levels induced by therapy initiated when disease is already established. Therefore, the clinical effects of large reductions of up to 80% in Lp(a) concentration with a targeted antisense oligonucleotide, administered in addition to background statin therapy, should be determined in future trials.[18] Nevertheless, the results of the current study do not support the use of a PCSK9 inhibitor to reduce arterial inflammation and Lp(a) in patients with elevated baseline concentrations.