Lipoprotein(a), Oxidized Phospholipids, and Coronary Artery Disease Severity and Outcomes

Abstract

Background:

Lipoprotein(a) [Lp(a)] and oxidized phospholipids (OxPLs) are each independent risk factors for atherosclerotic cardiovascular disease (ASCVD). The extent to which Lp(a) and OxPLs predict coronary artery disease (CAD) severity and outcomes in a contemporary, statin-treated cohort is not well established.

Objectives:

To evaluate the relationships between Lp(a) particle concentration and OxPLs associated with apolipoprotein B (OxPL-apoB) or apolipoprotein(a) [OxPL-apo(a)] with angiographic CAD and cardiovascular outcomes.

Methods:

Among 1098 participants referred for coronary angiography in the CASABLANCA study (Catheter Sampled Blood Archive in Cardiovascular Diseases, ClinicalTrials.gov NCT00842868), Lp(a), OxPL-apoB and OxPL-apo(a) were measured. Logistic regression estimated risk of multi-vessel coronary stenoses by Lp(a)-related biomarker level. Cox proportional hazards regression estimated risk of major adverse cardiovascular events (MACE; coronary revascularization, non-fatal MI, non-fatal stroke, and cardiovascular death) in follow-up.

Results:

Median [IQR] Lp(a) was 26.45 nmol/L [11.39-89.49]. Lp(a), OxPL-apoB, and OxPL-apo(a) were highly correlated (Spearman R≥0.91 for all pairwise combinations). Lp(a) and OxPL-apoB were associated with multi-vessel CAD. Odds of multi-vessel CAD per doubling of Lp(a), OxPL-apoB, and OxPL-apo(a) were 1.10 (95% CI 1.03-1.18, P=0.006), 1.18 (95% CI 1.03-1.34, P=0.01), and 1.07 (95% CI 0.99-1.16, P=0.07), respectively. All biomarkers were associated with cardiovascular (CV) events. Hazard ratios for MACE per doubling of Lp(a), OxPL-apoB, and OxPL-apo(a) were 1.08 (95% CI 1.03-1.14, P=0.001), 1.15 (95% CI 1.05-1.26, P=0.004), and 1.07 (95% CI 1.01-1.14, P=0.02), respectively.

Conclusions:

In patients undergoing coronary angiography, Lp(a) and OxPL-apoB are associated with multi-vessel CAD. Lp(a), OxPL-apoB, and OxPL-apo(a) are associated with incident CV events.

Condensed abstract:

Isoform-independent Lp(a) particle number and oxidized phospholipids associated with apo(a) and apoB were measured among patients referred for coronary angiography. Lp(a) and oxidized phospholipids were highly correlated with one another but not with established cardiovascular risk factors or lipid parameters. In adjusted models, Lp(a) and OxPL-apoB were significantly associated with angiographic multi-vessel CAD, and Lp(a), OxPL-apoB, and OxPL-apo(a) were associated with major adverse cardiovascular events.

Introduction

Despite the widespread use of low-density lipoprotein cholesterol (LDL-C) lowering therapies, the risks of first and recurrent coronary artery disease (CAD) events remain high across populations¹. Elevated lipoprotein(a) [Lp(a)] has emerged as an important risk factor for first and recurrent CAD events². Lp(a) is a low-density lipoprotein-like particle comprising both apolipoprotein B100 and apolipoprotein(a) [apo(a)]. Apo(a) is encoded by the LPA gene, and serum Lp(a) is uniquely highly heritable^3-7. Furthermore, Lp(a)-associated alleles at the LPA locus strongly associate with CAD risk^8-11.

Despite the abundance of preclinical data supporting a causal role for Lp(a) and CAD, there are currently no approved therapies for lowering lipoprotein(a) concentrations and CAD risk. Protein subtilisin/kexin type 9 serine protease (PCSK9) monoclonal antibodies, which are approved for reducing LDL-C concentrations and CAD risk, decrease Lp(a) by ~15- 30% versus placebo, which may further contribute to improvements in CAD prognosis based on post hoc analyses^12-15. More potent therapeutics lowering Lp(a) are currently in late-stage clinical development to prospectively test the hypothesis that Lp(a) lowering reduces CAD risk^16-18.

A major proposed mechanism for Lp(a)-related CAD risk is via oxidized phospholipids (OxPLs), which are principally carried by Lp(a) in both the liquid phase and bound to apo(a)¹⁹. Importantly, ~85% of OxPLs precipitated by antibodies to apoB are also associated with apo(a), and 30-70% of OxPLs are extractable from isolated Lp(a), suggesting that Lp(a) is the preferential plasma carrier of these pro-atherogenic particles^{20, 21}. Depending on the acuteness of the clinical scenario, patient population, ethnicity and Lp(a) isoform size, the correlation between OxPL-apoB and Lp(a) may range from R=0.30-0.90^{20, 22}.

Although the role of Lp(a) as a risk factor for cardiovascular (CV) disease is well-established, the relative role of Lp(a) plasma concentrations versus OxPL-apoB in angiographic disease and risk prediction remains undefined. Both are inversely associated with the number of apo(a) kringle-IV2 repeats and positively associated with peripheral atherosclerosis progression, prevalent symptomatic atherosclerotic cardiovascular disease, and coronary atherosclerosis among patients presenting for coronary angiography from cohorts ascertained more than 25 years ago^22-24. The degree to which Lp(a) accounts for the residual risk of cardiovascular events in statin-treated patients is debated, with some evidence that its predictive ability depends on the degree of LDL-C reduction achieved^{25, 26}. In a contemporary high-risk cohort treated with more widespread potent LDL-C-lowering therapies, with low average LDL-C, we sought to examine the extent to which coronary atherosclerosis, as well as incident cardiovascular disease events, are attributable to Lp(a) concentrations, OxPL-apoB, and OxPL-apo(a).

We employed a novel isoform-independent assay using monoclonal antibody LPA-KIV9, which targets the kringle IV9 present in a single copy in apo(a), to more precisely quantify Lp(a) particle number in participants of the Catheter Sampled Blood Archive in Cardiovascular Diseases (CASABLANCA) cohort^27-29. Further, we measured OxPLs separately associated with apoB and apo(a). This enabled further investigation between Lp(a) and OxPLs with both anatomical disease and cardiovascular events in follow-up. We hypothesized that Lp(a) and OxPLs would predict coronary artery disease severity at baseline and adverse cardiovascular outcomes in follow-up.

Methods

Study procedures were approved by the Mass General Brigham Institutional Review Board (IRB 2008P001076, IRB 2021P002474) and conducted in accordance with the Declaration of Helsinki.

The rationale and design of the CASABLANCA study (ClinicalTrials.gov NCT00842868) has been published. The study was designed to identify circulating biomarkers predictive of short and long-term adverse events in individuals undergoing coronary or peripheral angiography. This is a convenience sample of patients who underwent angiography at Massachusetts General Hospital between 2008 and 2011; study participants were approached prior to their procedure and consented for enrollment^{30, 31}. Pre- and post-procedural blood samples were collected from a centrally inserted arterial catheter (femoral or radial), placed on ice, spun immediately in a refrigerated centrifuge, aliquoted and frozen at −80 degrees until the present analysis. Demographic and medical data were collected, and detailed anatomic characterization of baseline CAD and peripheral arterial disease burden were performed. Individuals were followed for both cardiovascular and non-cardiovascular outcomes for a median of 4 years. A total of 1251 individuals were enrolled, of whom 1098 underwent coronary angiography and were included in this study.

Data acquisition and follow-up.

After informed consent, clinical and demographic data were collected for each participant, as previously described²⁹.

CAD burden was based on visual assessment of stenosis severity by the operator performing the procedure. Operators were unaware of the results of biomarker testing. In each of left main (LM), left anterior descending (LAD), left circumflex (LCx), and right coronary arteries (RCA), the most severe stenosis was recorded. We considered the accepted standard of greater than or equal to 70% stenosis in LAD, LCx, or RCA or greater than or equal to 50% stenosis in LM as significant stenoses. We considered two angiographic CAD outcomes: 1) ‘Any CAD’: the presence of at least one ≥70% coronary stenosis in LAD, LCx, or RCA or ≥50% LM stenosis, and 2) ‘Multi-vessel CAD’: the presence of at least two ≥70% coronary stenoses in LAD, LCx, or RCA, or LM stenosis ≥50% and any number of ≥70% stenoses in LAD, LCx, or RCA. In a secondary analysis we quantified coronary artery disease burden using the modified Duke Coronary Artery Disease Index (Duke Index), which is a 13-level scale (0-100) accounting for all coronary stenoses >=50%, with score increasing by both increasing severity stenosis and greater overall vessel involvement³².

Endpoints in follow-up were gathered via medical record review and telephone interviews with patients and/or their providers and adjudicated using strict charter definitions³⁰. Death was also determined by the Social Security Death Index or death announcements. We defined MACE (the primary clinical outcome) as a composite of cardiovascular death, non-fatal MI, non-fatal stroke, or coronary revascularization in follow-up.

Biomarker testing.

Using first-thaw samples from pre-procedure blood, we measured isoform-independent Lp(a) particle number, OxPL-apoB, and OxPL-apo(a) for 1098 CASABLANCA participants who underwent coronary angiography, with or without peripheral angiography, at baseline. Details of the assays used are outlined in the Online Methods.

Statistical analysis.

To obtain normal distributions, all biomarker levels were log base 2 transformed prior to analysis; as a result, effect estimates reflect the risk associated with a doubling of Lp(a), OxPL-apoB, or OxPL-apo(a) concentrations. In presenting baseline characteristics, we divided the cohort into Lp(a) bins of 0-49 nmol/L, 50-99 nmol/L, 100-149 nmol/L, and ≥150 nmol/L and compared the above parameters across these groups using chi-square test for categorical variables and Kruskal-Wallis for continuous variables.

Study covariates included age, sex, smoking status (ever/never), statin prescription at baseline, LDL-C, high-density lipoprotein cholesterol (HDL-C), systolic blood pressure (SBP), prevalent hypertension, prevalent diabetes mellitus (type 1 or type 2), and body-mass index, all assessed at baseline. We tested the association between each study covariate and each Lp(a)-related biomarker using univariate linear regression.

We tested the correlation among each of the Lp(a)-related biomarkers [Lp(a), OxPL-apoB, and OxPL-apo(a)] and with traditional lipid parameters, including LDL-C, HDL-C, non-HDL-C, total cholesterol, apoB, and triglycerides using Spearman correlation testing. In sensitivity analysis, we repeated Spearman correlation testing stratified by baseline statin prescription. We tested the association between LDL-C and apoB and both angiographic outcomes and MACE in follow-up using a linear regression model fully adjusted for the study covariates.

Odds of ‘Any CAD’ or ‘Multi-vessel CAD’ were estimated per doubling of Lp(a), OxPL-apoB, and OxPL-apo(a) using logistic regression. Association analyses were both unadjusted and adjusted for the aforementioned covariates. In secondary analysis, linear regression estimated the change in Duke Index per doubling of Lp(a), OxPL-apoB, and OxPL-apo(a) in both unadjusted and adjusted models.

Hazard of incident MACE, as defined above, was estimated per doubling of Lp(a), OxPL-apoB, and OxPL-apo(a) using Cox proportional hazards regression (R package ‘survival’, version 3.3-1, 2022-03-03)³³. The proportional hazards assumption was assessed by the Schoenfeld test and met. Models were both unadjusted and adjusted for the aforementioned covariates. In secondary analyses, we explored MACE sub-components as separate outcomes in survival analyses. Furthermore, we performed survival analysis stratified by the presence of baseline ‘Any CAD’ or ‘Multi-vessel CAD’; in models adjusted for the presence of any baseline obstructive CAD; and stratified by the occurrence of percutaneous coronary intervention at the index angiogram. Finally, because the coronary revascularization subcomponent of MACE may include both planned and unplanned revascularizations, we repeated the survival analysis excluding individuals who had early revascularization at <30 days from index angiogram (N=145).

Given that Lp(a) and OxPLs may be altered acutely at the time of myocardial infarction (MI), we performed sensitivity analysis repeating the above analyses for ‘Any CAD’, ‘Multi-vessel CAD’, and MACE after excluding 212 (19.3%) of individuals who underwent coronary angiography for an acute coronary syndrome (ACS: acute MI or unstable angina).

To explore the significance of previously reported clinical thresholds of Lp(a) in mg/dL, we converted Lp(a) particle concentration in nmol/L to estimated mass in mg/dL using both Lp(a)/2.0 and Lp(a)/2.5 conversion factors, then compared the probabilities of ‘Any CAD’ and ‘Multi-vessel CAD’ at baseline and MACE in follow-up across Lp(a) strata 0-29 mg/dL, 30-49 mg/dL, 50-69 mg/dL, and ≥ 70 mg/dL using Fisher’s exact test.

All statistics were performed by using R software, version 4.0 (R Foundation for Statistical Computing, Vienna, Austria)³⁴. P values are 2-sided, with a value <0.05 considered significant.

Results

Baseline characteristics

Baseline characteristics of the 1098 study participants are displayed in Table 1. Mean age was 66.3 years (SD 11.4), 778 (70.9%) were male, and 1022 (93.1%) were white. A minority of individuals (N=214, 19.5%) underwent coronary angiography for an acute coronary syndrome (acute MI or unstable angina). The most common indication for catheterization was an abnormal stress test (N=421, 38.3%) (Online Table 1). Prevalent CAD was present for 565 (51.5%) participants, of whom 302 (27.5%) had prior percutaneous coronary intervention (PCI), 194 (17.7%) had prior CABG, and 260 (23.7%) had prior MI. Statins were prescribed at baseline in 797 (72.6%), and mean LDL-C was 83.4 (SD 31.3) mg/dL. Lp(a)-related biomarkers followed left-skewed distributions (Online Figure 1). Medians [interquartile ranges, IQR] for Lp(a), OxPL-apoB, and OxPL-apo(a) were 26.45 nmol/L [11.39-89.49], 3.80 nmol/L [2.71-8.43], and 11.69 [4.54-36.97] nmol/L, respectively. Across strata of Lp(a) (0-49, 50-99, 100-149, and ≥150 nmol/L), statistically significant differences were observed for CAD, heart failure, family history of premature CAD, and baseline statin prescription. Mean baseline LDL-C was similar across strata (82.4, 85.2, 84.0, and 85.8 mg/dL for strata 0-49, 50-99, 100-149, and ≥150 nmol/L Lp(a), respectively).

Table 1.

Baseline characteristics.

		nmol/L Lp(a)
	All	<50	50-100	100-150	>150	P
N	1098	703	150	85	160
Demographics
Age (mean (SD))	66.3 (11.4)	66.2 (11.4)	67.5 (11.4)	64.1 (11.2)	66.8 (11.7)	0.17
Male (%)	778 (70.9)	494 (70.3)	113 (75.3)	61 (71.8)	110 (68.8)	0.58
Race/Ethnicity (%)						0.22
American/African	27 (2.5)	14 ( 2.0)	6 ( 4.0)	1 ( 1.2)	6 ( 3.8)
Asian/Pacific	11 (1.0)	5 ( 0.7)	2 ( 1.3)	3 ( 3.5)	1 ( 0.6)
Caucasian	1022 (93.1)	657 (93.5)	136 (90.7)	79 (92.9)	150 (93.8)
Hispanic	22 (2.0)	16 ( 2.3)	3 ( 2.0)	1 ( 1.2)	2 ( 1.2)
Native American	2 (0.2)	1 ( 0.1)	0 ( 0.0)	1 ( 1.2)	0 ( 0.0)
Other (or unknown)	14 (1.3)	10 ( 1.4)	3 ( 2.0)	0 ( 0.0)	1 ( 0.6)
Past Medical History
Current smoker (%)	163 (14.8)	112 (15.9)	19 (12.7)	16 (18.8)	16 (10.0)	0.15
Hypertension (%)	813 (74.0)	516 (73.4)	117 (78.0)	55 (64.7)	125 (78.1)	0.09
SBP, mm Hg (mean (SD))	137.3 (22.5)	137.5 (22.5)	137.0 (22.4)	136.62 (22.53)	136.7 (22.8)	0.97
Coronary artery disease(%)	565 (51.5)	353 (50.2)	75 (50.0)	43 (50.6)	94 (58.8)	0.26
Prior PCI (%)	302 (27.5)	193 (27.5)	38 (25.3)	18 (21.2)	53 (33.1)	0.2
Prior CABG (%)	194 (17.7)	117 (16.6)	22 (14.7)	14 (16.5)	41 (25.6)	0.04
Prior MI (%)	260 (23.7)	162 (23.0)	37 (24.7)	21 (24.7)	40 (25.0)	0.93
Dyslipidemia(%)	724 (66.1)	441 (62.8)	107 (71.8)	62 (72.9)	114 (71.2)	0.03
Diabetes mellitus (%)	296 (27.0)	195 (27.7)	33 (22.0)	21 (24.7)	47 (29.4)	0.43
Atrial Fibrillation (%)	211 (19.2)	128 (18.2)	33 (22.0)	14 (16.5)	36 (22.5)	0.44
Heart Failure (%)	216 (19.7)	122 (17.4)	28 (18.7)	21 (24.7)	45 (28.1)	0.01
CVA/TIA (%)	116 (10.6)	80 (11.4)	12 ( 8.0)	10 (11.8)	14 ( 8.8)	0.53
COPD (%)	191 (17.4)	120 (17.1)	26 (17.3)	14 (16.5)	31 (19.4)	0.91
CKD (%)	143 (13.0)	84 (11.9)	25 (16.7)	11 (12.9)	23 (14.4)	0.44
FH of premature CAD (%)						0.03
No	721 (65.7)	473 (67.3)	104 (69.3)	51 (60.0)	93 (58.1)
Unknown	6 (0.5)	4 ( 0.6)	0 ( 0.0)	2 ( 2.4)	0 ( 0.0)
Yes	371 (33.8)	226 (32.1)	46 (30.7)	32 (37.6)	67 (41.9)
Baseline lipids
Statin use (%)	797 (72.6)	490 (69.7)	113 (75.3)	59 (69.4)	135 (84.4)	0.002
LDL-C, mg/dL (mean (SD))	83.4 (31.3)	82.4 (31.8)	85.2 (31.1)	83.95 (29.10)	85.8 (30.2)	0.53
HDL-C, mg/dL (mean (SD))	45.1 (15.8)	44.6 (16.3)	43.9 (14.1)	47.93 (15.54)	47.3 (15.4)	0.06
TG, mg/dL (mean (SD))	130.6 (83.5)	133.2 (85.6)	140.2 (103.9)	119.52 (61.74)	116.8 (58.4)	0.14
TC, mg/dL (mean (SD))	153.4 (39.0)	153.0 (39.8)	152.1 (40.6)	153.62 (34.43)	156.5 (36.2)	0.76
non-HDL-C, mg/dL. (mean (SD))	108.4 (37.7)	108.5 (38.5)	108.8 (42.4)	105.67 (31.16)	108.8 (33.1)	0.93
Lp(a), nmol/L (median [IQR])	26.5 [11.4, 89.5]	14.4 [7.5, 24.9]	71.8 [58.0, 87.1]	119.43 [109.83, 136.06]	221.6 [182.0, 274.6]	<0.001
OxPL-apo(a), nmol/L (median [IQR])	11.7 [4.5, 37.0]	5.7 [3.0, 10.6]	31.6 [26.1, 37.8]	46.14 [40.07, 49.69]	55.0 [47.6, 62.0]	<0.001
OxPL-apoB, nmol/L (median [IQR])	3.8 [2.7, 8.4]	3.0 [2.5, 3.7]	7.3 [5.9, 8.7]	10.15 [8.97, 11.74]	14.7 [12.3, 18.3]	<0.001
Log2-Lp(a) (median [IQR])	4.7 [3.5, 6.5]	3.8 [2.9, 4.6]	6.2 [5.9, 6.4]	6.9 [6.8, 7.1]	7.8 [7.5, 8.1]	<0.001
Log2-OxPL-apo(a) (median [IQR])	3.5 [2.2, 5.2]	2.5 [1.6, 3.4]	5.0 [4.7, 5.2]	5.5 [5.3, 5.6]	5.8 [5.6, 6.0]	<0.001
Log2-OxPL-apoB (median [IQR])	1.9 [1.4, 3.1]	1.6 [1.3, 1.9]	2.9 [2.6, 3.1]	3.3 [3.2, 3.6]	3.9 [3.6, 4.2]	<0.001

Baseline characteristics of all 1098 CASABLANCA participants who underwent coronary angiography at enrollment and stratified by Lp(a) bins in nmol/L (<50, 50-99, 100-149, and >=150 nmol/L). Statistical comparisons between Lp(a) strata were performed using ANOVA for continuous variables and Chi-square for counts.

CABG = coronary artery bypass grafting; CAD = coronary artery disease; CKD = chronic kidney disease; COPD = chronic obstructive pulmonary disease; CVA/TIA = cerebrovascular accident/transient ischemic attack; FH = family history; HDL-C = high density lipoprotein cholesterol; LDL-C = low density lipoprotein cholesterol; Lp(a) = lipoprotein(a); MI = myocardial infarction; OxPL = oxidized phospholipid; PCI = percutaneous coronary intervention; SBP = systolic blood pressure; TC = total cholesterol; TG = triglycerides.

Correlation between Lp(a) biomarkers and standard lipid parameters and risk factors

We observed a high degree of correlation between Lp(a) and oxidized phospholipids (R=0.91 with OxPL-apoB and R=0.95 with OxPL-apo(a), P<0.001) and between the two OxPL biomarkers (R=0.93, P<0.001) (Central Illustration, Figure 1, Online Figure 2). In contrast, there was minimal correlation between any Lp(a)-related biomarker and traditional lipid parameters, apart from a weak positive correlation between Lp(a) and apoB (R=0.07, P=0.02) and weak negative correlations between Lp(a), OxPL-apoB, and OxPL-apo(a) and triglycerides. Correlations among Lp(a)-related biomarkers and between Lp(a)-related biomarkers and traditional lipid parameters were not importantly changed according to baseline statin use (Online Table 2).

Central illustration. Lp(a)-related Biomarkers and Coronary Artery Disease Burden and Cardiovascular Outcomes.

Lipoprotein(a) [Lp(a)] and oxidized phospholipids associated with apolipoprotein B (OxPL-apoB) and apolipoprotein(a) [OxPL-apo(a)] were measured among CASABLANCA participants undergoing coronary angiography at baseline. Top left panel: histograms demonstrating population distribution of Lp(a), OxPL-apoB, and OxPL-apo(a) among 1098 individuals. Bottom left panel: pairwise correlations between Lp(a), OxPL-apoB, and OxPL-apo(a) within the cohort. Blue line is the linear relationship. Top right panel: Prevalence of angiographic ‘Multi-Vessel CAD’ (≥ 2 severe coronary stenoses) at baseline angiography according to quintile of Lp(a), OxPL-apo(a), or OxPL-apoB. Bottom right panel: Cox proportional hazards regression estimated the risk of major adverse cardiovascular events in follow-up per doubling of Lp(a), OxPL-apoB, or OxPL-apo(a) in unadjusted models and models adjusted for the study covariates.

CAD = coronary artery disease; CI = confidence interval; Lp(a) = lipoprotein(a); OxPL-apo(a) = oxidized phospholipids on apo(a); OxPL-apoB = oxidized phospholipids on apoB.

Figure 1. Correlation heatmap between Lp(a) biomarkers and traditional lipid parameters.

Spearman correlations between log₂-transformed Lp(a), OxPL-apo(a), and OxPL-apoB, and between each Lp(a)-related biomarker and traditional lipid parameters. Cells are highlighted with ‘*’ when meeting P-value threshold of P<0.05.

ApoB = apolipoprotein B; HDL-C = high density lipoprotein cholesterol; LDL-C = low density lipoprotein cholesterol; Lp(a) = lipoprotein(a); OxPL = oxidized phospholipid; TG = triglycerides; total-C = total cholesterol.

We did not observe statistically significant associations between Lp(a), OxPL-apoB, or OxPL-apo(a) and traditional ASCVD risk factors (Online Table 3). However, we observed that statin prescription at baseline was associated with increased concentrations of all three Lp(a)-related biomarkers.

We observed a negative association between LDL-C and apoB concentrations and both angiographic outcomes ‘Any CAD’ and ‘Multi-vessel CAD’ and MACE in follow-up in fully adjusted models (Online Table 4).

Association between Lp(a), OxPL-apoB, and OxPL-apo(a) and angiographic coronary atherosclerosis

Any coronary artery stenosis (‘Any CAD’) was present in 666 (60.7%) participants, and multiple coronary artery stenoses (‘Multi-vessel CAD’) were present in 409 (37.2%) participants. Crude prevalence rates by estimated Lp(a) mass bins are presented in Table 2. Prevalence of ‘Multi-vessel CAD’ increased at thresholds of 70 mg/dL and 50 mg/dL when using nmol/L / 2.0 and nmol/L / 2.5 correction factors, respectively (P<0.001). In both unadjusted and adjusted logistic regression models, there was no statistically significant association between any of the three Lp(a) biomarkers and the presence of ‘Any CAD’. In contrast, for ‘Multi-vessel CAD’, there were significant associations with both Lp(a) and OxPL-apoB and a borderline association with OxPL-apo(a). For doubling of Lp(a), OxPL-apoB, and OxPL-apo(a) in adjusted analyses, odds ratios were 1.10 (95% CI 1.03-1.18, P=0.006), 1.18 (95% CI 1.03-1.34, P=0.01), and 1.07 (95% CI 0.99-1.16, P=0.07) (Central Illustration, Figure 2, Online Table 5). When OxPL-apoB was added to the model for Lp(a), the Lp(a) effect was preserved and the OxPL-apoB effect attenuated. A similar pattern was observed when OxPL-apo(a) was added to the fully adjusted Lp(a) model (Online Table 6). After excluding individuals who underwent angiography for ACS, Lp(a) and OxPLs remained associated with ‘Multi-vessel CAD’ but not ‘Any CAD’ in fully adjusted models (Online Table 7).

Table 2.

Rates of coronary stenosis and MACE stratified by Lp(a) in mg/dL.

		nmol/L Lp(a) / 2.0
Lp(a) stratum	Total	Any CAD	Multi-vessel CAD	MACE
0-29 mg/dL	750	444 (59%)	257 (34%)	277 (37%)
30-49 mg/dL	103	70 (68%)	40 (39%)	42 (41%)
50-69 mg/dL	71	41 (58%)	21 (30%)	33 (46%)
>=70 mg/dL	174	111 (64%)	91 (52%)	97 (56%)
P-value		0.28	<0.001	<0.001
		nmol/L Lp(a) / 2.5
Lp(a) stratum	Total	Any CAD	Multi-vessel CAD	MACE
0-29 mg/dL	785	470 (60%)	270 (34%)	292 (37%)
30-49 mg/dL	121	75 (62%)	39 (32%)	51 (42%)
50-69 mg/dL	65	39 (60%)	33 (51%)	37 (57%)
>=70 mg/dL	127	82 (65%)	67 (53%)	69 (54%)
P-value		0.35	<0.001	<0.001

Lp(a) in nmol/L was converted to mg/dL using both Lp(a) / 2.0 and Lp(a) / 2.5 conversion factors. Rates of ≥ 1 severe coronary stenosis at baseline (‘Any CAD’), ≥ 2 severe coronary stenoses at baseline (‘Multi-vessel CAD’), and MACE in follow-up were calculated for each of four Lp(a) stratum. Cochran-Armitage trend test compared rates across strata.

Lp(a) = lipoprotein(a); MACE = major adverse cardiovascular events; mg/dL = milligrams per deciliter; nmol/L = nanomoles per liter.

Figure 2. Association between Lp(a), OxPL-apo(a), and OxPL-apoB and angiographic CAD.

A) Logistic regression estimated the odds of one or more coronary stenoses (‘Any CAD’) or two or more coronary stenoses (‘Multi-vessel CAD’) per doubling of Lp(a), OxPL-apo(a), and OxPL-apoB in both unadjusted models and models adjusted for age, body-mass index, diabetes mellitus, high density lipoprotein cholesterol, hypertension, low density lipoprotein cholesterol, sex, systolic blood pressure, smoking status, and statin use. Results are plotted as odds ratio with 95% confidence interval. (B) Prevalence of ‘Multi-vessel CAD’ for quintiles of Lp(a), OxPL-apoB, and OxPL-apo(a).

CAD = coronary artery disease; Lp(a) = lipoprotein(a); OxPL-apo(a) = oxidized phospholipid on apo(a); OxPL-apoB = oxidized phospholipid on apoB.

In secondary analyses, quantifying coronary disease burden using the Duke Index, we observed that Lp(a) was associated with a significant increase in Duke Index in the unadjusted model. For each doubling of Lp(a), there was a 0.91 (95% CI 0.17-1.66) point increase in Duke Index (P=0.02). The associations did not reach statistical significance across the remaining biomarkers, though effects were directionally consistent with the primary analysis (Online Table 8).

Effect of Lp(a), OxPL-apoB, and OxPL-apo(a) on risk of incident MACE

Participants were followed for a median [IQR] of 4.2 [3.4–4.4] years. There were 449 events in follow-up (40.9% of participants), occurring at median 118 days from the index angiogram. Events were most commonly coronary revascularization (N=182) and acute MI (N=174) and less commonly CV death (N=69) or ischemic cerebrovascular accident (CVA) (N=24). Early events tended to be driven by coronary revascularization, occurring at median 12.5 days. In contrast, acute MI occurred at median 197 days, and cardiovascular death occurred at median 484 days. Crude prevalence rates by estimated Lp(a) mass bins are presented in Table 2. Prevalence of MACE in follow-up increased at thresholds of 70 mg/dL and 50 mg/dL when using nmol/L / 2.0 and nmol/L / 2.5 correction factors, respectively (P<0.001). Lp(a) biomarkers were associated with MACE in both unadjusted and adjusted Cox models, with the strongest effect observed per doubling of OxPL-apoB (Central Illustration, Figure 3, Online Table 9). For each doubling of Lp(a), OxPL-apoB, and OxPL-apo(a) in adjusted models, hazard ratio (HR) for MACE was 1.08 (95% CI 1.03-1.14, P=0.001), 1.15 (95% CI 1.05-1.26, P=0.004), and 1.07 (95% CI 1.01-1.14, P=0.02), respectively. Hazards for the MACE outcome were materially unchanged when additionally adjusting for prevalent CAD status (prior MI, CABG, or PCI) (Online Table 10). As was observed with the prevalent disease outcome, upon addition of OxPLs to the Lp(a) survival model the Lp(a) effect was preserved and the OxPL effect attenuated (Online Table 11). After excluding individuals who underwent angiography for ACS, Lp(a) and OxPLs remained associated with MACE in fully adjusted models (Online Table 12).

Figure 3. Association between Lp(a), OxPL-apoB, and OxPL-apo(a) and MACE.

Cox proportional hazards regression estimated the hazard of MACE in follow-up per doubling of Lp(a), OxPL-apo(a), and OxPL-apoB in both unadjusted models and models adjusted for age, body-mass index, diabetes mellitus, high density lipoprotein cholesterol, hypertension, low density lipoprotein cholesterol, sex, systolic blood pressure, smoking status, and statin use. Results are plotted as hazard ratio with 95% confidence interval.

CI = confidence interval; Lp(a) = lipoprotein(a); MACE = major adverse cardiovascular events; OxPL-apo(a) = oxidized phospholipid on apo(a); OxPL-apoB = oxidized phospholipid on apoB.

We explored survival analyses separately for the MACE components. There were 288 coronary revascularizations; 174 acute MIs; 130 cardiovascular deaths; and 40 ischemic CVAs in follow-up. Effects were most consistent for revascularization and CVA (Online Figure 3). MACE effects for all three biomarkers yielded similar magnitudes upon stratification by the presence of baseline ‘Any CAD’ or ‘Multi-vessel CAD’ (Online Figure 4) or by the occurrence of PCI at the index coronary angiogram (Online Figure 5).

When we excluded individuals who underwent early revascularization (<30 days) from index angiogram, associations between Lp(a), OxPL-apoB, and OxPL-apo(a) and MACE were unchanged compared to the overall cohort in both unadjusted and adjusted models (Online Table 13).

Discussion

In a contemporary cohort of patients undergoing coronary angiography for both urgent and non-urgent indications, elevated Lp(a) and OxPLs increased the risks of both multiple coronary stenoses at baseline and cardiovascular events in follow-up. In contrast, LDL-C and apoB were negatively associated with all outcomes, which likely reflects indication bias from higher risk patients being prescribed more intense LDL-C-lowering regimens. These findings also highlight challenges in using LDL-C and apoB for secondary risk prediction. Importantly, Lp(a) and OxPLs were not predicted by traditional lipid parameters and were not correlated with traditional cardiovascular risk factors. We found that Lp(a)-related biomarkers did not significantly associate with ‘Any CAD’ but only ‘Multi-vessel CAD’ consistent with prior associations of Lp(a) with atherosclerosis and clinically meaningful events.

Our study has implications for current clinical care and for future studies of Lp(a), OxPLs, and Lp(a)-lowering therapies. First, elevated Lp(a) is common among patients presenting for coronary angiography. We observe that 1 in 3 have Lp(a) greater than the customary clinical threshold of 30mg/dL, and 1 in 6 have Lp(a) greater than 70mg/dL, the cutoff used as an enrollment criterion in the Lp(a)HORIZON clinical trial (Assessing the Impact of Lipoprotein(a) Lowering with Pelacarsen (TQJ230) on Major Cardiovascular Events in Patients with CVD, ClinicalTrials.gov Identifier NCT04023552). The high prevalence of this risk factor in a high risk population is in support for routine screening³⁵.

Second, we found that MACE risk did not depend on whether individuals had established CAD at baseline or whether PCI was performed at the index angiogram. Whereas the Lp(a)HORIZON trial is treating patients with established CAD, our results highlight that high-risk primary prevention patients also have high event risks attributable to Lp(a).

Third, multiple Lp(a)-related biomarkers measured within a single cohort enable new insights into the relationship between Lp(a) and CAD. We differentiate OxPLs associated separately with apo(a) (Lp(a)-specific) and apoB (predominantly Lp(a) but also associated with other apoB-containing lipoproteins). In parallel, we employ an isoform-independent assay for Lp(a) quantification, which yields Lp(a) particle number rather than Lp(a) mass. This assay is apo(a) isoform size independent, and there is notably an inverse relationship between apo(a) isoform size with OxPL concentration³⁶. We observe a higher degree of correlation between OxPLs and Lp(a) compared to prior reports with isoform-dependent Lp(a) assays^{20, 23, 37}, indicating that apo(a) size may not be a sufficient determinant of OxPL concentration. However, though the difference between the biomarkers was not large, OxPL-apoB may be the strongest predictor of both prevalent coronary atherosclerosis and MACE, aligned with prior reports for related outcomes³⁸. Larger studies are required to understand whether OxPL-apoB provides prognostic utility beyond Lp(a). Lp(a)-lowering therapies in late state investigation potently reduce OxPL-apoB¹⁶; these studies may provide insight into whether OxPL-apoB lowering reduces CV risk.

Fourth, all three Lp(a)-biomarkers predicted the presence of multiple coronary stenoses at baseline. Lp(a) is associated with the presence of atherosclerotic disease of many vascular territories, including coronary, peripheral arterial, and cerebrovascular^39-42, and less consistently with coronary calcium^43-45; these associations are also seen with OxPLs^{23, 24} There is, however, conflicting evidence that Lp(a) and OxPLs predict the degree or overall burden of coronary atherosclerosis. In this high-risk cohort, we found that Lp(a)-related biomarkers were associated with the presence of more severe CAD.

Fifth, both Lp(a) and OxPLs are associated with incident major adverse cardiovascular events in follow-up. Most events were due to coronary revascularization and acute MI, with a small proportion from cardiovascular death or ischemic CVA. As with prevalent CAD, the association with MACE was numerically stronger for OxPL-apoB compared to OxPL-apo(a) or Lp(a). Coronary revascularization tended to occur early in follow-up, whereas acute MI was delayed. It is debated to what extent Lp(a) drives cardiovascular risk through a pro-thrombotic state leading to atherothrombosis, potentially via its homology with fibrinogen, versus plaque accumulation and atherosclerosis, and to what extent elevated Lp(a) levels predict events after MI and PCI^{25, 46-48}. More recent data links elevated Lp(a) to target lesion revascularization (TLR)⁴⁹, though we are unable to directly address this with our data. However, our finding that Lp(a) and OxPLs drive CAD risk both early and late in follow-up is consistent with both mechanisms of insidious plaque accumulation and early plaque instability after PCI or MI.

It was also noted that statin use was associated with higher Lp(a) and related OxPL biomarkers. This is consistent with prior observational studies finding that statins are associated with 10-30% higher Lp(a) and OxPLs^{50, 51} but does not negate the outsized beneficial effects of LDL-C-lowering. Furthermore, this may be magnified in observational study where more potent LDL-C-lowering is given to patients with more severe ASCVD, who are enriched for increased Lp(a)-related biomarker concentrations.

Study Limitations

Our study has several strengths and limitations. The CASABLANCA cohort is carefully phenotyped, and we were able to measure Lp(a) biomarkers from high quality first thaw samples. Of note, samples are arterial, and the consistency with venous samples is assumed though not empirically established. Importantly, CASABLANCA is predominantly male and White, such that our conclusions may not generalize to other populations. Though our study was not designed to directly test the hypothesis that OxPLs contribute to CAD pathogenesis via Lp(a), the high correlation between OxPLs and Lp(a) particle number and our finding of heightened risk due to elevated OxPLs suggested further study of OxPL-related risk is warranted. Finally, the CASABLANCA cohort was enrolled at a tertiary care hospital and included only individuals referred for angiography; thus, the pre-test probability of obstructive CAD is higher than would be expected in a non-angiography-based community cohort. However, the low LDL-C distribution more closely reflects contemporary practice patterns compared to prior OxPL cardiovascular outcome studies.

Conclusions

In a high-risk coronary angiography cohort, isoform-independent Lp(a) particle number and OxPLs associated with apoB and apo(a) were highly correlated and associated with the presence of severe coronary atherosclerosis at baseline and adverse cardiovascular outcomes in follow-up. Clinical trials of Lp(a)-lowering therapies will ultimately determine the extent to which Lp(a)-related residual risk can be mitigated.

Perspectives.

Competency in Medical Knowledge:

Lp(a) is a risk factor for atherosclerotic cardiovascular disease associated with severity and outcomes in patients with coronary artery disease independent of LDL-cholesterol and other established clinical predictors.

Translational Outlook:

Specific Lp(a) lowering therapies are under investigation in clinical trials. At present, identification of individuals with elevated Lp(a) levels can refine cardiovascular risk assessment, and in the future may form a basis for primary and secondary prevention treatments that complement statin therapy.

Abbreviations:

OxPL

oxidized phospholipid

ApoB

apolipoprotein B

Apo(a)

apolipoprotein(a)

ASCVD

atherosclerotic cardiovascular disease

CAD

coronary artery disease

Lp(a)

lipoprotein(a)

MACE

major adverse cardiovascular events

CVA

cerebrovascular accident