Repeat Measures of Lipoprotein(a) Molar Concentration and Cardiovascular Risk

Abstract

Background:

When indicated, guidelines recommend measurement of lipoprotein(a) for cardiovascular risk assessment. However, temporal variability in lipoprotein(a) is not well understood, and it is unclear if repeat testing may help refine risk prediction of coronary artery disease (CAD).

Objectives:

We examined the stability of repeat lipoprotein(a) measurements and the association between instability in lipoprotein(a) molar concentration with incident CAD.

Methods:

We assessed the correlation between baseline and first follow-up measurements of lipoprotein(a) in the UK Biobank (n=16,017 unrelated individuals). The association between change in lipoprotein(a) molar concentration and incident CAD was assessed among 15,432 participants using Cox proportional-hazards models.

Results:

Baseline and follow-up lipoprotein(a) molar concentration were significantly correlated over a median of 4.42 years (IQR=3.69-4.93; Spearman rho=0.96, p-value <0.0001). The correlation between baseline and follow-up lipoprotein(a) molar concentration were stable across time between measurements of <3 (rho=0.96), 3-4 (rho=0.97), 4-5 (rho=0.96), and >5 years (rho=0.96). While there were negligible-to-modest associations between statin use and changes in lipoprotein(a) molar concentration, statin usage was associated with a significant increase in lipoprotein(a) among individuals with baseline levels ≥70 nmol/L. Follow-up lipoprotein(a) molar concentration was significantly associated with risk of incident CAD (hazard ratio [95% CI] per 120 nmol/L: 1.32 [1.16-1.50], p-value=0.0002). However, the delta between follow-up and baseline lipoprotein(a) molar concentration was not significantly associated with incident CAD independent of follow-up lipoprotein(a) (p-value=0.98).

Conclusions:

These findings suggest that, in the absence of therapies substantially altering lipoprotein(a), a single accurate measurement of lipoprotein(a) molar concentration is an efficient method to inform CAD risk.

CONDENSED ABSTRACT

The temporal variability in lipoprotein(a), especially in the context of response to statin therapy and risk of coronary artery disease (CAD), is inadequately understood. Here, we studied up to 16,017 unrelated UK Biobank participants to assess the stability of repeat lipoprotein(a) measurements and the association between lipoprotein(a) instability with incident CAD. In most cases, baseline and follow-up measurements of lipoprotein(a) molar concentration was highly correlated regardless of time between testing. Lipoprotein(a) instability was not significantly associated with incident CAD. Together these findings suggest that a single accurate measurement of lipoprotein(a) molar concentration is an efficient method to inform CAD risk.

INTRODUCTION

Lipoprotein(a) is a potentially causal risk factor for multiple cardiovascular diseases (1, 2). Lipoprotein(a) levels are 75-95% heritable and predominately determined by copy number variants in the kringle IV type 2 domain of LPA and single-nucleotide variants within and surrounding the LPA gene (3, 4). Mendelian randomization studies suggest that lipoprotein(a) is a causal contributor to coronary artery disease (CAD) (5), heart failure (6), ischemic stroke (7), aortic stenosis (8), and all-cause mortality (9), and ongoing randomized clinical trials are testing these hypotheses (10-12). It remains unclear to what extent the features of lipoprotein(a) contribute to atherothrombosis (1), but the molar concentration of lipoprotein(a) has been suggested to fully explain the cardiovascular risk associated with elevated lipoprotein(a) (13, 14).

Routine lipid indices, comprised of total cholesterol, low-density lipoprotein cholesterol (LDL-C), triglycerides, and high-density lipoprotein cholesterol vary based on genetic, environmental, lifestyle, clinical, and stochastic factors. The longitudinal nature of these biomarkers have emerged as important predictors of cardiovascular risk (15-17). In particular, the concept of “cholesterol-years”, which arose from the study of patients with familial hypercholesterolemia, suggests that cardiovascular outcomes are directly related to the years in which an individual is exposed to elevated LDL-C (analogous to “pack-years” and lung cancer risk) (17, 18). These findings highlight two important concepts: 1) understanding an individual’s longitudinal trajectory of LDL-C exposure may help refine cardiovascular risk prediction, and 2) lowering LDL-C levels early in life for individuals’ at high cardiovascular risk is likely a more effective strategy for primary prevention (19).

Whether the longitudinal trajectory of lipoprotein(a) can similarly refine cardiovascular disease risk is incompletely understood. A notable proportion of participants in the IONIS-APO(a) Rx and IONIS-APO(a)-L Rx antisense oligonucleotide clinical trials allocated to placebo displayed temporal variability in lipoprotein(a) molar concentration ≥±25% for at least one time point (20). It has also been suggested that statin therapy may increase lipoprotein(a) levels and it is unclear whether understanding an individual’s lipoprotein(a) response to statin therapy should be monitored (1, 21, 22).

Here we used individual level data from up to 16,017 UK Biobank participants with repeat lipoprotein(a) measurements to investigate the following questions in real-world contemporary cohorts: 1) how stable is a measurement of lipoprotein(a), 2) do statins have an appreciable effect on lipoprotein(a) molar concentration, and 3) is instability in lipoprotein(a) molar concentration associated with cardiovascular risk?

METHODS

UK Biobank cohort.

The UK Biobank is a prospective observational study of approximately 500,000 volunteer adults aged 40-69 years recruited from 22 sites across the United Kingdom between 2006-2010 with follow-up ongoing (23). The UK Biobank protocol was approved by the Northwest Multi-Center Research Ethics Committee, and all study participants provided written informed consent. Secondary use of data for this study was approved by the Massachusetts General Hospital Institutional Review Board 2013P001840 under UK Biobank application 7089.

This study included participants of 3^rd degree relatedness or less that had a baseline and follow-up measurement of lipoprotein(a) (Figure 1). Biochemical measurements, physical exam measurements, and medical histories were assessed at the time of baseline and follow-up measurements of lipoprotein(a). Self-reported ethnicities were categorized as Mixed, African/Black, European/White, East Asian, South Asian, and Unknown.

Figure 1. Flow diagram of how the study cohort was used for analyses.

Unrelated participants from the UK Biobank with baseline and follow-up measurements of lipoprotein(a) were included in this study to assess the stability of repeat lipoprotein(a) measurements. The majority of these participants without prevalent coronary artery disease, were studied to assess the association between lipoprotein(a) instability and risk of incident coronary artery disease.

Coronary artery disease (CAD), lipoprotein(a) (Lp[a]).

Lipoprotein(a) measurements and calculation of estimated glomerular filtration rate.

For UK Biobank participants, the molar concentration of lipoprotein(a) was measured at baseline and, for a random subset, follow-up using an immune-turbidimetric method with polyclonal antibodies targeting apolipoprotein(a) epitopes (Denka Seiken, Coventry, UK), which minimizes the impact of apo(a) size variability on imprecision to concentration measurements (24). Use of cholesterol-lowering medications were assessed by verbal interview with a trained UK Biobank staff member at the time of sample collection for baseline and follow-up lipoprotein(a) measurements.

Estimated glomerular filtration rate was calculated using the Chronic Kidney Disease Epidemiology Collaboration equation and serum creatinine (25).

Definition of cardiovascular events.

For UK Biobank participants, cardiovascular events were defined using International Statistical Classification of Diseases and Related Health Problems and Office of Population Censuses and Surveys Classification of Interventions and Procedures codes (Supplemental Table 1) (26). CAD was defined as non-fatal myocardial infarction, fatal myocardial infarction, or coronary revascularization. Composite atherosclerotic cardiovascular disease events included non-fatal myocardial infarction, non-fatal stroke, fatal myocardial infarction, or fatal stroke. Incident events were defined as the first event occurring between the date of follow-up lipoprotein(a) measurement and the end of follow-up of March 31^st, 2020. Participants were censored at death or loss to follow-up. We performed a sensitivity analysis using the date of baseline lipoprotein(a) measurement lipoprotein(a) measurement and the end of follow-up of March 31^st, 2020.

Statistical analyses.

Analyses were performed using R version 4.0.2 (R Core Team [2020]). Correlation between baseline and follow-up blood biomarkers, such as lipoprotein(a) levels, were assessed by Spearman correlation and linear regression. For the comparison of plasma lipoprotein levels, normally distributed data were analyzed with Mann-Whitney U test or Kruskal-Wallis test (post hoc test: Dunn’s test with Benjamini and Hochberg correction). The mean percent change of biomarkers was calculated as: $\frac{follow-up-baseline}{baseline}\ast 100$.

We assessed the risk of incident cardiovascular events using Cox proportional-hazards models with lipoprotein(a) molar concentration using the covariates of sex, age, age squared, and the first 4 principal components of ancestry derived from genome-wide genotyping in the UK Biobank using the “survival” package version 3.2-7 (27). When modeling continuous lipoprotein(a) molar concentration we used a natural cubic spline and, when explicitly stated, Cox proportional-hazards models included the additional covariate of delta lipoprotein(a) levels (follow-up – baseline). Proportional hazards were assessed by visual inspection of Schoenfeld residuals and Schoenfeld tests.

Statistical significance was claimed when two-sided p-values were <0.05 unless otherwise stated.

RESULTS

Participant characteristics.

The characteristics of 16,017 UK Biobank participants at the time of baseline and follow-up measurement of lipoprotein(a) molar concentration are displayed in the Table. The mean age of Biobank participants was 58.2 years at baseline (SD: 7.4 years) and 62.5 years at follow-up (SD: 7.4 years). Most participants were of White ethnicity (97.7%) and 50.2% of participants were female. In general, individuals with both a baseline and follow-up measurement of lipoprotein(a) molar concentration tended to have similar enrollment characteristics to individuals with only a baseline measurement (Supplemental Table 1). However, individuals with both a baseline and follow-up lipoprotein(a) molar concentration measurement were more likely to be of White or European ethnicity, less likely to be current smokers, and less likely to be female relative to individuals with only a baseline measurement.

Table 1.

Cohort characteristics at the time of baseline and follow-up measurement of lipoprotein(a) molar concentration.

Characteristic	Measure	Baseline	Follow-up
N	no.	16017	16017
Age - years	mean (SD)	58.2 (7.4)	62.5 (7.4)
Sex - female	no. (%)	8039 (50.2)	-
Ethnicity - White	no. (%)	15647 (97.7)	-
Ethnicity - South Asian	no. (%)	81 (0.5)	-
Ethnicity - Black/Caribbean	no. (%)	68 (0.4)	-
Ethnicity - East Asian	no. (%)	56 (0.4)	-
Ethnicity - Mixed	no. (%)	54 (0.3)	-
Ethnicity - Unknown	no. (%)	111 (0.7)	-
Lipoprotein(a) - nmol/L	median (IQR) / n	19.50 (7.56-72.50)	20.40 (7.70-77.50)
LDL-C - mmol/L	mean (SD) / n	3.54 (0.85) / 15848	3.54 (0.89) / 15386
Triglycerides - mmol/L	median (IQR) / n	1.41 (1.18-1.68) / 15900	1.47 (1.23-1.77) / 15405
HDL-C - mmol/L	mean (SD) / n	1.45 (0.38) / 14636	1.52 (0.40) / 13473
Hemoglobin A1c	median (IQR) / n	35.0 (32.6-37.5) / 15115	35.7 (33.4-38.2) / 11249
C-reactive protein (mg/L)	median (IQR) / n	1.17 (0.59-2.37) / 15869	1.21 (0.61-2.40) / 15371
Cholesterol-lowering medication	no. (%) / n	2621 (16.4) / 15941	3793 (23.8) / 15932
Hypertension	no. (%) / n	3893 (24.3) / 15997	4639 (29.0) / 15994
Diabetes mellitus	no. (%) / n	651 (4.1) / 15994	899 (5.6) / 15989
Body mass index - kg/m2	no. (%) / n	26.9 (4.5) / 15982	26.9 (4.5) / 15992
Current smoker	no. (%) / n	1017 (6.4) / 15979	731 (4.6) / 15972
Myocardial infarction	no. (%)	312 (1.95)	412 (2.57)
Coronary revascularization	no. (%)	270 (1.69)	381 (2.38)
Stroke	no. (%)	432 (2.70)	543 (3.39)

Low-density lipoprotein cholesterol (LDL-C), high density lipoprotein cholesterol (HDL-C).

Stability of repeat lipoprotein(a) measurements.

The median time between baseline and follow-up measurements of lipoprotein(a) molar concentration was 4.42 years (interquartile range [IQR]: 3.69-4.93 years; Supplemental Figure 1). Median lipoprotein(a) molar concentration was 19.50 nmol/L at baseline (IQR: 7.56-72.50 nmol/L) and 20.40 nmol/L at follow-up (IQR: 7.70-77.50 nmol/L) and there was a strong correlation between baseline and follow-up measurements (Spearman correlation rho: 0.96, p-value: <0.0001; Figure 2A). Summary statistics and correlation between baseline and follow-up lipoprotein(a) measurements were comparable for individuals that had time intervals between measurements of : < 3 years (rho: 0.96, p-value: <0.0001), 3-4 years (rho: 0.97, p-value: <0.0001), 4-5 years (rho: 0.96, p-value: <0.0001), and greater than 5 years (rho: 0.96, p-value: <0.0001; Supplemental Table 2). The Spearman correlation between baseline and follow-up measurements of other blood biomarkers used in cardiovascular risk assessment were more modest among UK Biobank participants: 0.72 for total cholesterol, 0.70 for LDL-C, 0.63 for triglycerides, 0.85 for high-density lipoprotein cholesterol, 0.64 for C-reactive protein, and 0.75 for hemoglobin A1c (all p-values: <0.0001; Supplemental Figure 2). Despite the relative stability of lipoprotein(a) molar concentration, some participants displayed differences between follow-up and baseline lipoprotein(a) measurements that were quite large (i.e., >120 nmol/L, Figure 2B-C). There was also a tendency for follow-up lipoprotein(a) measurements to be greater than baseline lipoprotein(a) measurements. For instance, 8.66% of individuals had a follow-up lipoprotein(a) measurement at least 25 nmol/L greater than their baseline, while only 4.12% of individuals had a follow-up lipoprotein(a) measurement at least 25 nmol/L less than their baseline (Figure 2D).

Figure 2. Repeat measurements of lipoprotein(a) molar concentration are relatively stable.

(A) The correlation between baseline and follow-up lipoprotein(a) molar concentration is depicted for UK Biobank participants. Waterfall plots show: (B) the absolute delta of lipoprotein(a) molar concentration for each UK Biobank participant (follow-up – baseline) and (C) the mean percent change in lipoprotein(a) molar concentration. Red dotted lines depict difference in lipoprotein(a) molar concentration ≥ 120 nmol/L or mean percent changes greater than 25%. (D) The percentage of UK Biobank participants versus the size of delta lipoprotein(a) levels is shown.

Lipoprotein(a) [Lp(a)].

Given that previous work has suggested a potential association between lipoprotein(a) levels and kidney function (28), we tested the strength in correlation between the mean percent change in lipoprotein(a) and estimated glomerular filtration rate. The rho value obtained from the Spearman’s rank correlation between the mean percent change in lipoprotein(a) and mean percent change in eGFR was -0.026 (p-value: 0.001). However, the proportion of variance explained by this association was minimal (R²: 2.8x10⁻⁵, p-value=0.23; Supplemental Figure 3). There was also only a weak correlation between baseline or follow-up log lipoprotein(a) molar concentration and the respective baseline or follow-up estimated glomerular filtration rate (rho [p-value]: −0.040 [<0.0001] and −0.022 [<0.0001], respectively; Supplemental Figure 3).

Association between lipoprotein(a) molar concentration and use of cholesterol-lowering medication.

There is mixed evidence about the effect of cholesterol-lowering medications such as statins on the molar concentration of lipoprotein(a) and the downstream influence on cardiovascular risk. Here, we sought to compare mean percent change in lipoprotein(a) molar concentration between participants using and not using statin medication. As a positive control we assessed the change in LDL-C levels associated with use of cholesterol-lowering medication. Relative to individuals naïve to statin medication at baseline or follow-up, individuals starting statin medication after baseline displayed a significant decrease in the mean percent change in LDL-C levels (Kruskal-Wallis p-value: <0.0001; Dunn-Test with Benjamin-Hochberg correction p-value: <0.0001; Supplemental Figure 4). In contrast, there was no significant difference in the mean percent change in lipoprotein(a) molar concentration between participants starting statin medication after baseline relative to participants not naïve to statins at baseline and follow-up, discontinuing cholesterol-lowering medication prior to follow-up, or continued statin use at baseline and follow-up (Kruskal-Wallis p-value: 0.25; Figure 3A). There was also no significant difference in the mean percent change in lipoprotein(a) molar concentration between individuals that started specific statins prior to follow-up versus those that remained statin naïve at follow-up (Kruskal-Wallis p-value: 0.11; Figure 3B). Among participants naïve to statin exposure at baseline, there were modest, but significant differences in the mean percent change in lipoprotein(a) molar concentration in those starting statin therapy versus those remaining statin naïve at follow-up based on the lipoprotein(a) categories of <70 nmol/L (median [IQR]: 0.00% [−23.4 - 27.4%] vs 6.06% [−15.0 - 32.2%], Mann-Whitney U p-value: <0.0001), 70-120 nmol/L (11.9% [−4.19 - 37.1%] vs 6.87% [−7.31 - 25.7%], p-value: 0.04), or >=120 nmol/L (11.7% [−3.95-30.0%] vs 2.72% [−9.36 - 15.7%], p-value: <0.0001; Figure 3C). Similar results were observed when comparisons were made for absolute changes in lipoprotein(a) molar concentration (Supplemental Figure 5).

Figure 3. Association between statin use and changes in lipoprotein(a) molar concentration.

(A) Boxplots depict the median and interquartile range of the mean percent change in lipoprotein(a) molar concentration associated with statin medication use by UK Biobank participants. For individuals naïve to stain medication at baseline, boxplots depict the median and interquartile range in the mean percent change of lipoprotein(a) molar concentration for UK Biobank participants (B) newly starting specific statin treatments prior to follow-up biomarker measurement and (C) lipoprotein(a) molar concentration based on thresholds of baseline lipoprotein(a) (only 7/16,017 [0.04%] of individuals had a mean percent change greater than 2,000). The line within the boxplots represents the median. The upper and lower edges of the boxplots represent the 1^st and 3^rd quartile, respectively. The whiskers represent the 1^st quantile – 1.5 x interquartile range or 3^rd quartile + 1.5 x interquartile range, while individual points represent outliers that are extreme values to these ranges. Numeric values above the boxplots depict the n per group. The dotted black lines depict a mean percent change of 0 (no change). Lipoprotein(a) [Lp(a)].

Association of instability in lipoprotein(a) molar concentration with incident coronary artery disease.

To assess whether changes in lipoprotein(a) molar concentration are associated with risk of incident CAD, we grouped UK Biobank participants without prevalent CAD based on lipoprotein(a) thresholds of <70 (~30 mg/dL), 70-150, and >= 150 nmol/L (~50 mg/dL) (29). We defined lipoprotein(a) instability as follow-up lipoprotein(a) measurements that differed from baseline measurements by >10% . There were modest differences in baseline lipoprotein(a) molar concentration between individuals defined as having stable versus unstable lipoprotein(a) at the thresholds of <70 (median [IQR]: decrease 12.7 [6.40-27.2] nmol/L, stable 12.9 [6.49-27.5], increase 10.4 [4.96-22.9], p-value: <0.0001), 70-150 (decrease 108 [88.7-108] nmol/L, stable 111 [91.4-131], increase 108 [89.2-128], p-value: 0.27), and >=150 nmol/L (decrease 203 [171-259] nmol/L, stable 204 [174-256], increase 191 [168-239], p-value: <0.0001). The median follow-up time for CAD events after measurement of follow-up lipoprotein(a) molar concentration were 7.15 (IQR: 6.94-7.39), 7.15 (IQR: 6.94-7.39) , and 7.14 years (IQR: 6.94-7.39) for lipoprotein(a) groups of <70, 70-150, and >=150 nmol/L, respectively. There were no significant differences in the association between lipoprotein(a) instability and incident CAD among individuals with a baseline lipoprotein(a) molar concentration >=150 nmol/L that had a decreased (reference group), stable (HR [95% CI]: 0.74 [0.41-1.34], p-value: 0.32), or increased follow-up lipoprotein molar concentration (HR [95% CI]: 0.80 [0.43-1.49], p-value: 0.0.49; Figure 4). Similar results were observed among individuals with baseline lipoprotein(a) molar concentrations <70 and 70-150 nmol/L (Figure 4).

Figure 4. Association of lipoprotein(a) instability with risk of incident coronary artery disease.

Forest plots show the adjusted hazard ratios for incident coronary artery disease for UK Biobank participants with decreased, stable, or increased follow-up lipoprotein(a) molar concentration based on the thresholds of <70, 70-150, and >=150 nmol/L. Follow-up lipoprotein(a) molar concentrations that were >10% different from baseline measurements were considered unstable. Hazard ratios were adjusted for age, age squared, sex, and the first 4 principal components of ancestry.

Lipoprotein(a) [Lp(a)].

When follow-up lipoprotein(a) was modelled as a continuous variable, follow-up lipoprotein(a) molar concentration was significantly associated with risk of incident CAD (HR [95% CI] per 120 nmol/L increase in lipoprotein(a) molar concentration: 1.32 [1.15-1.50], p-value: <0.0001, 396 events). However, the delta between follow-up and baseline lipoprotein(a) measurements did not display a significant association with risk of incident CAD when assessed alone (HR [95% CI] per 120 nmol/L increase in delta lipoprotein(a): 1.54 [0.90-2.62], p-value: 0.11) or when additionally adjusted for follow-up lipoprotein(a) molar concentration (HR [95% CI] per 120 nmol/L increase in delta lipoprotein(a): 1.00 [0.60-1.67], p-value: 0.99).

DISCUSSION

Here we use a cohort of up to 16,017 participants from the UK Biobank with baseline and follow-up measurements of lipoprotein molar concentration performed on a standardized assay to show that the molar concentration of lipoprotein(a) is relatively stable regardless of cholesterol-lowering medication use. Although some individuals did display relatively large differences between follow-up and baseline lipoprotein(a) measures, these differences did not associate with increased risk of incident CAD since these changes largely persisted within the abnormal range, i.e. among individuals with elevated lipoprotein(a). (Central Illustration).

Central Illustration. Repeat measurements of lipoprotein(a) molar concentration and incident coronary artery disease.

This study assessed the relationship between repeat lipoprotein(a) measurements and coronary artery disease in up to 16,017 unrelated UK Biobank participants. This work suggests that modest changes in repeat lipoprotein(a) measurements are common, but do not associate with incident coronary artery disease.

Coronary artery disease (CAD).

Our study advances the field by 1) reporting on repeat lipoprotein(a) measurements among the general population, rather than the setting of a randomized clinical trial, and 2) assessing the influence of statins on the lipoprotein(a) molar concentration rather than lipoprotein(a) mass. Our findings suggest that statin therapy has a largely negligible influence on lipoprotein(a) molar concentration. However, there are important differences in study design between this work and those of clinical trial trials that may be relevant for interpreting some of these discrepant interstudy results (21). Specifically, a change in lipoprotein(a) levels following statin therapy may: 1) be dependent on the statin itself rather than a class effect, 2) only be appreciable with high potency statins such as atorvastatin and rosuvastatin, or 3) only be appreciable among individuals with higher baseline levels of lipoprotein(a). This study assessed the association between statin therapy and lipoprotein(a) levels among a volunteer population from the United Kingdom. It is reasonable to expect that UK Biobank participants were treated with less potent statins and have lower levels of lipoprotein(a) than contemporary clinical trial populations in which participants are generally required to have prevalent cardiovascular disease as inclusion criteria and are enriched for individuals with elevated lipoprotein(a) (i.e., primary versus secondary prevention) (21, 30). Alternatively, it is possible that this work’s assessment of lipoprotein(a) molar concentration, rather than lipoprotein(a) mass, could explain the difference in study results. The Denka Seiken assay used in this study to measure lipoprotein(a) molar concentration displays high concordance with enzyme-linked immunosorbent assays and has an estimated coefficient of variation of ~5% (31), but is known to overestimate lipoprotein(a) levels in samples containing large apo(a) isoforms (32, 33). However, similar to the ONIS-APO(a) Rx and IONIS-APO(a)-L Rx antisense oligonucleotide clinical trials, it is unlikely that assay variability alone can explain the large temporal changes in lipoprotein(a) molar concentration observed in small proportion of individuals.

This work suggests that most currently used medications and lifestyle factors have negligible-to-modest effects on lipoprotein(a) levels and are unlikely to have a meaningful effect on risk of lipoprotein(a)-associated cardiovascular events (34, 35). Elevated levels of lipoprotein(a) are a strong residual risk factor for cardiovascular events among patients optimally treated to LDL-C targets (22) and a one-time measurement of lipoprotein(a) may be helpful to refine cardiovascular risk prediction (24, 36, 37). However, our results suggest that repeat monitoring of lipoprotein(a) levels is unlikely to be clinically useful for understanding an individual patient’s residual risk of an incident cardiovascular event in the context of primary prevention. Specifically, individuals with more notable changes in absolute lipoprotein(a) levels tend to be those at the extremes of the distribution of lipoprotein(a) and the consequential relative changes in lipoprotein(a) for these individuals are more modest (i.e. an individual with a baseline lipoprotein(a) of 300 nmol/L increasing by 25 nmol/L rather than an individual with a baseline lipoprotein(a) of 20 nmol/L increasing by 25 nmol/L).

Secondary analyses of the Outcomes: Evaluation of Cardiovascular Outcomes After an Acute Coronary Syndrome During Treatment With Alirocumab trial (ODYSSEY OUTCOMES) for secondary prevention of CAD suggest that lowering-lipoprotein(a) with a proprotein convertase subtilisin/kexin type 9 inhibitor can modestly reduce the risk of recurrent CAD events (38, 39). Additionally, emerging antisense oligonucleotide therapies targeting the LPA gene hold promise for more effectively lowering lipoprotein(a) levels by up to ~80% and will be tested for efficacy in phase 3 clinical trials of cardiovascular risk reduction over the next several years (11) (Assessing the Impact of Lipoprotein (a) Lowering With TQJ230 on Major Cardiovascular Events in Patients With CVD; unique identifier: NCT04023552). If treatment strategies such as these, which effectively lower lipoprotein(a), become commonly used in clinical practice, it may become important to understand the longitudinal trajectory of lipoprotein(a) to accurately assess the cardiovascular risk associated with lipoprotein(a).

In conclusion, this study found that statin therapy was not associated with significant changes in the molar concentration of lipoprotein(a) and that changes in lipoprotein(a) were associated with negligible differences in risk of incident cardiovascular events. These findings suggest that longitudinal measurements of lipoprotein(a) are likely unnecessary for cardiovascular risk assessment in the context of primary prevention because the molar concentration of lipoprotein(a) is generally stable, regardless of statin use. It is possible that these conclusions could be different for individuals with prevalent atherosclerotic cardiovascular disease who are at higher risk of recurrent cardiovascular events.

Study limitations.

This study has some limitations worthy of consideration. First, most participants in the UK Biobank study are middle-age adults of White ethnicity and the generalizability of these findings to other age and ethnic groups will require further research. Second, this study has limitations associated with the ascertainment bias of volunteer recruitment. Participants of the UK Biobank are, on average, “healthier” than the general population (38). Third, the use of cholesterol-lowering medication was determined by verbal interview and the doses of statin medication used by participants are unknown (i.e., for a participant using atorvastatin, it is unknown whether their dosing was 10, 20, 40, or 80 mg/day). Fourth, cardiovascular events were defined by diagnosis and operation codes. Fifth, analyses of cardiovascular events had a relatively small number of events and should be interpreted cautiously because of the limited precision of the effect estimates.

CLINICAL PERSPECTIVES.

Competency in Patient Care:

Unlike the temporal variability of many traditional cardiovascular risk factors, the molar concentration of lipoprotein(a) is stable in most individuals and repeat measurements of lipoprotein(a) do not refine the estimated risk of incident coronary artery disease.

Translational Outlook:

Further research is needed to refine clinical practice algorithms that utilize a single accurate measurement of lipoprotein(a) molar concentration in detection and management of patients at risk of coronary artery disease.

ABBREVIATIONS

CAD

Coronary artery disease

LDL-C

low-density lipoprotein cholesterol