Lipoprotein(a) association with residual risk: what has inflammation got to do with it?

This editorial refers to ‘C-reactive protein modifies lipoprotein(a)-related risk for coronary heart disease: the BiomarCaRE project’, by N. Arnold et al., https://doi.org/10.1093/eurheartj/ehad867.

Lipoprotein(a) [Lp(a)] is a genetic lipoprotein biomarker of increasing importance in coronary heart disease (CHD) risk, and has been shown to be a determinant of residual cardiovascular risk in secondary prevention populations treated with statins and aggressive lipid lowering. In fact, recent European guidelines and consensus statements recommend checking Lp(a) levels at least once in a lifetime for all adults, regardless of traditional CHD risk factors, in order to establish future Lp(a)-associated CHD risk and guide preventive therapies.¹ This, coupled with the fact that ∼20% of the world’s population is estimated to have unfavourably high Lp(a) levels, mandates a more in depth investigation of those most likely to benefit from intervention-specific Lp(a)-lowering therapies that are being tested in Phase III clinical outcomes trials.

In an attempt to identify those most likely to benefit from Lp(a)-lowering interventions, Arnold et al., in their study published in this issue of the European Heart Journal, sought to assess how inflammatory burden, as identified by high-sensitivity C-reactive protein (hsCRP), affects the Lp(a)-associated risk of either incident CHD or recurrent events in patients with known CHD. In their study, the authors used data from eight European population cohort studies from the BiomarCARE project, with 65 661 participants falling into a primary prevention category and 6017 forming a secondary prevention cohort for analysis. These were prospective studies with a median follow-up time of 9.8 (3282 first events) and 13.8 years (1373 recurrent events), respectively. Lp(a) levels were measured in two central laboratories using an automated turbidimetric mass-based assay not affected by apo(a) heterogeneity.² It is worth noting that the highest quintile category had a mean Lp(a) of 24.7 and 28.1 for primary and secondary prevention cohorts, respectively. Based on current consensus statements, Lp(a) levels <30 mg/dL by mass-based assay are considered normal, levels of 30–50 mg/dL suggest intermediate risk/grey zone, and Lp(a) > 50 mg/dL suggests high risk,¹ making the relevance of the other subcategories studied by Arnold et al. less clinically meaningful. At baseline, the authors confirmed that, as expected, the correlation between hsCRP and Lp(a) in either cohort is low. In the entire population, Lp(a) was associated with increased risk of CHD [hazard ratio (HR) 1.44, 95% confidence interval (CI) 1.30–1.60 for the top vs. the bottom quintile) in those with and without known CHD, and in those with and without persistent inflammation as assessed by hsCRP levels. Similarly, hsCRP was also associated with increased CHD risk in the entire cohort as well as in those with and without known CHD.

When assessing CHD risk associations as stratified by hsCRP using a cut-off point of 2 mg/L, the authors found robust associations for Lp(a) and CHD risk in the primary prevention cohort irrespective of hsCRP levels. In those with known CHD, recurrent cardiac events were only found to be associated with elevated Lp(a) in individuals with persistent inflammation identified as hsCRP ≥2 mg/L (HR 1.34, 95% CI 1.03–1.76) which was not statistically significant in those with hsCRP <2 mg/L (HR 1.29, 95% CI .98–1.71) or hsCRP <1 mg/L (HR 1.14, 95% CI 0.75–1.73). However, the similarity of the HR point estimates and the large overlap in the 95% CI suggest that any difference in Lp(a)-associated residual risk by baseline hsCRP levels is likely to be small. This is further corroborated with the similar Lp(a)-associated residual risks with recurrent events when Lp(a) was analysed per standard deviation (not quintiles) among individuals with hsCRP ≥2 vs. < 2 mg/L, with HRs of 1.09 and 1.13, respectively (P-interaction of .53).

Previously, the ACCELERATE trial (n = 10 503) reported that hsCRP modulated Lp(a)-related cardiovascular disease (CVD) risk only in a secondary prevention cohort when hsCRP was ≥2 mg/L.³ Another study in patients with known CHD undergoing percutaneous coronary intervention (n = 10 424)⁴ also demonstrated higher risk in those with hsCRP ≥2 mg/L and Lp(a) ≥30 mg/dL compared with lower hsCRP levels. The current authors’ results identifying Lp(a)-associated CHD risk as being independent of hsCRP levels in primary prevention participants were also similar to those from the large Copenhagen General Population Study (n = 68 090).⁵ In a recent analysis of the primary prevention Multi-Ethnic Study of Atherosclerosis (MESA)⁶ (n = 4679), no association with CVD risk was seen at any level of Lp(a) or hsCRP independently; however, when hsCRP was ≥2 mg/L, a significant CVD risk was observed with Lp(a) of 50–99.9 mg/dL (HR 1.36) and Lp(a) ≥ 100 mg/dL (HR 2.09). These data suggest that categorization of cohorts into primary vs. secondary prevention groups based on documented CVD events may not be the optimal way to assess Lp(a)-associated CVD risk in the highest risk patients. Taken together, the aggregate findings so far suggest that a possible interplay of Lp(a) and inflammation changes over the life span either as people progress from a primary to a secondary prevention stage or in terms of overall risk burden, and open up the avenue for further precision-based approaches to inform treatment strategies.

The study by Arnold et al. represents the largest and most comprehensive analysis of the interaction between hsCRP and Lp(a)-associated CHD risk, which is a major strength of the study. Yet, as noted by the authors, these results should be viewed as hypothesis-generating. As also noted by the authors, the current study had several potential limitations. One important potential limitation is its generalizability. Apart from ethnic homogeneity due to solely European cohorts, the study populations had very low usage of aspirin (5.3% overall and 32.6% in secondary prevention) and lipid-lowering therapy (3.9% and 15.6%, respectively) unlike contemporary populations. In addition, Lp(a) levels were classified into quintiles using mean levels, although given the skewed distribution of Lp(a), medians or log-transformed values may have been more statistically suitable. Due to the skewed distribution of Lp(a), these values were then cubic root transformed per investigator discretion. The spline analysis in the secondary prevention cohort suggested that the risk of Lp(a) with CHD events was not monotonic or graded as there was a blunted risk association for individuals with both high Lp(a) and high hsCRP which may reflect diminished statistical power in the highest risk category of both biomarkers.

Clinically, the relationship between inflammation and Lp(a) is complex (Graphical Abstract). While lower Lp(a) levels are seen in severe inflammatory conditions such as sepsis and severe burns, increased Lp(a) levels are seen in the acute phase response and other inflammatory conditions.¹ Mechanistically, it has been demonstrated that Lp(a) has proinflammatory and procalcific properties due to the oxidized phospholipid content of Lp(a).^1,7 High Lp(a) levels can initiate atherosclerosis in the arterial walls, promote a proinflammatory gene expression profile in vascular and aortic valve cells, and modulate cytokine [e.g. interleukin (IL)-6 and IL-1β] and chemoattractant release, and transendothelial monocyte migration.⁸ There is a growing body of evidence implicating the role of oxidized phospholipids carried by Lp(a) particles as the stimulants of release of proinflammatory cytokine such as IL-1β, tumour necrosis factor-α (TNF-α), and IL-6.⁷ IL-6 expression also appears to contribute in part to Lp(a) levels, with IL-6 blockade resulting in a modest decrease in Lp(a) in patients with elevated hsCRP in the RESCUE trial.⁹ IL-6 may be part of a complex dynamic between Lp(a)-mediated production and potential pathological positive feedback on Lp(a) production itself in certain higher risk milieus. Glycolysis is a driver of Lp(a)-related inflammation, suggesting that Lp(a)-oxidized phospholipids actively modulate inflammation in an energy-dependent process.¹

Graphical Abstract.

Lipoprotein(a)-related inflammatory residual risk. Abbreviations: CHD, coronary heart disease; CRP, C-reactive protein; hsCRP, high-sensitivity CRP; Lpa, lipoprotein(a); IL-6, interleukin-6

Several critical questions remain regarding Lp(a), inflammation, and CVD risk. Notably we have still yet to prove that decreasing Lp(a) levels improves CVD outcomes, irrespective of baseline inflammation. We also do not know whether there will be a significant difference in outcome data depending on the methodology used to quantify Lp(a) (molar vs. mass-based assays). It may be that hsCRP, while easily measurable in clinical practice, is not the ideal biomarker for assessing Lp(a)-related residual inflammatory risk. Thus far, only plasmapheresis affords substantial Lp(a) reductions until newer small interfering RNA (siRNA) and ASO (antisense oligonucleotide) therapies that decrease Lp(a) production become available.¹⁰ Whether these will have an effect on CVD outcomes remains to be seen.

While the current study provides corroboration for the interesting dichotomy between hsCRP’s relationship with Lp(a)-associated risk and subpopulations of Europeans with and without CHD, the link between causation and pathophysiological targets remains to be elucidated. Understanding the pathophysiological processes that result in stable to unstable plaque progression and how lifelong Lp(a) elevation affects this process either via a direct effect on inflammation or as a result of chronic inflammatory effects is an exciting area that will further help identify therapeutic interventions and the optimal population towards whom these interventions should be directed. So is the inflammatory modulation of Lp(a)-associated CHD risk as measured by hsCRP a red herring or the real McCoy? Only time and further investigation will tell.

Declarations

Disclosure of Interest

All authors declare no disclosure of interest for this contribution.

Funding

Z.M.F. is funded by the Massachusetts Life Sciences 2023 First Look Award. S.M. was supported by research grants from the National Institute of Diabetes and Digestive and Kidney Diseases (DK112940) and the National Heart, Lung, and Blood Institute (R01 HL117861, R01HL134811 and K24 HL136852, R01HL134168, 1R01HL143227, and R01HL 160799). The funding sources had no role in the current manuscript, and the opinions expressed here are solely those of the authors.