Soluble Urokinase Plasminogen Activator Receptor: Genetic Variation and Cardiovascular Disease Risk in African Americans

Abstract

Background:

Soluble urokinase plasminogen activator receptor (suPAR) has emerged as an important biomarker of coagulation, inflammation, and cardiovascular disease (CVD) risk. The contribution of suPAR to CVD risk and its genetic influence in the African-American population have not been evaluated.

Methods:

We measured suPAR in 3,492 African-Americans from the prospective, community-based Jackson Heart Study (JHS). Cross-sectional associations of suPAR with lifestyle and CVD risk factors were assessed, whole-genome sequence data were used to evaluate genetic associations of suPAR, and relationships of suPAR with incident CVD outcomes and overall mortality were estimated over follow-up.

Results:

In Cox models adjusted for traditional CVD risk factors, eGFR, and C-reactive protein, each 1-standard deviation higher suPAR was associated with a 21% to 31% increased risk of incident coronary heart disease, heart failure, stroke, and mortality. In the GWAS, two missense (rs399145 encoding p.Thr86Ala, rs4760 encoding p.Phe272Leu) and two non-coding regulatory variants (rs73935023 within an enhancer element and rs4251805 within the promoter) of PLAUR (plasminogen activator, urokinase receptor) on chromosome 19 were each independently associated with suPAR and together explained 14% of suPAR phenotypic variation. The allele frequencies of each of the four suPAR-associated genetic variants differ considerably across African and European populations. We further show that PLAUR rs73935023 can alter transcriptional activity in vitro. We did not find any association between genetically-determined suPAR and CVD in JHS or a larger electronic medical record-based analyses of African-Americans or Whites.

Conclusions:

Our results demonstrate the importance of ancestry-differentiated genetic variation on suPAR levels and indicate suPAR is a CVD biomarker in African-American adults.

Introduction

The plasminogen activation system plays important roles in hemostasis and inflammation.^1–3 Several functions of the plasminogen activation system are mediated by the urokinase-type plasminogen activator receptor (uPAR), which is upregulated during conditions of injury and inflammation.^{2, 4, 5} In response to inflammatory mediators, uPAR can be cleaved or shed from the cell surface generating a soluble form, suPAR, that circulates with retained function in blood.⁶ Ligand binding of uPAR/suPAR by urokinase-type plasminogen activator (uPA) catalyzes activation of plasmin, which is involved in degradation of fibrin, coagulation factors and extracellular matrix components and activation of cytokines, growth factors, and matrix metalloproteases.^{1, 3, 7} uPAR/suPAR also binds cell surface receptors and other ligands independent of uPA and contributes to cell signaling, migration, and adhesion.^{2, 4, 5}

suPAR has gained considerable interest as a biomarker for adverse outcomes among patients with cardiovascular disease (CVD) or other chronic diseases.^{8, 9} Previous studies were mainly among individuals with pre-existing cardiometabolic conditions, rather than generally healthy individuals, and most evaluations of suPAR on CVD risk were in European-origin populations.⁸ There has been no comprehensive assessment of the epidemiology of suPAR or the prospective role of suPAR in CVD risk among community-dwelling African Americans. Moreover, very little information is available regarding the genetic factors contributing to inter-individual variation in suPAR,^{10, 11} particularly in non-European populations.

To investigate the relationships of suPAR with CVD risk among community-dwelling African Americans, we measured suPAR in samples collected at the baseline exam in 3,492 individuals from the prospective Jackson Heart Study (JHS). We evaluated suPAR’s associations with demographic, lifestyle, and CVD risk factors, genetic variants, prevalent subclinical vascular disease, and incident CVD events during follow-up.

Methods

The complete methods are available in the Supplemental Material in the Data Supplement. Data, including statistical code, from this manuscript are available to researchers who meet the criteria for access to confidential data. Access to JHS data is available through https://www.jacksonheartstudy.org/Research/Study-Data/Data-Access and through BioLINCC (https://biolincc.nhlbi.nih.gov/studies/jhs/). The JHS was approved by the Institutional Review Boards of Jackson State University, Tougaloo College, and the University of Mississippi Medical Center in Jackson. All participants provided written informed consent.

Results

Association of suPAR with demographic, lifestyle and CVD risk factors, and biomarkers of inflammation and coagulation

Characteristics of JHS participants at the baseline examination (n=3,492) are presented in Supplemental Table IV. suPAR was approximately normally distributed with a mean (SD) of 2246 pg/mL (795 pg/mL). suPAR was higher in women (mean =2366 pg/mL, SD=800 pg/mL) than men (mean 2046 pg/mL, SD=756 pg/mL) (p<0.0001) and higher with increasing age (p<0.0001). In cross-sectional analyses adjusted for sex and age, each 1 SD increase in suPAR was associated with lower levels of education and physical activity, with higher levels of SBP, triglycerides, BMI, waist circumference, current smoking, diabetes, and hypertension; and with lower HDL-cholesterol and eGFR (Table 1). In a multivariable adjusted model, age, sex, education, waist circumference, current smoking, SBP, hypertension, diabetes, and eGFR remained independently associated with suPAR, explaining ~29% of the suPAR phenotypic variation (Supplemental Table V). suPAR demonstrated modest correlation with factor VIII, fibrinogen, D-dimer, sCD14 and IP-10 (Spearman’s ρ~0.3) and somewhat less correlation with CRP, white blood cell and neutrophil count, and red cell distribution width (Spearman’s ρ~0.2) (Supplemental Table VI).

Table 1.

Associations of suPAR with demographic, lifestyle and cardiovascular disease risk factors in the JHS

	β (95% CI) per 1-SD increase in suPAR	P-Value
Age (years)	3.9 (3.4, 4.5)	<0.0001
Male sex (%)	−0.52 (−0.64, −0.41)	<0.0001
Education (≥ vs. < High school graduate)	−0.13 (−0.22, −0.04)	0.007
Global perceived stress (GPSS) score	0.05 (−0.09, 0.19)	0.50
Physical activity (Ideal vs. Intermediate or Poor)	−0.21 (−0.31, −0.10)	0.0001
ln-BMI (kg/m2)	0.02 (0.01, 0.03)	<0.0001
Waist circumference (cm)	2.5 (1.8, 3.2)	<0.0001
Current Smoking (%)	0.34 (0.24, 0.43)	<0.0001
Systolic blood pressure (mm Hg)	1.3 (0.67, 1.9)	<0.0001
Diastolic blood pressure (mm Hg)	−0.28 (−0.61, 0.05)	0.09
Hypertension (%)	0.33 (0.23, 0.44)	<0.0001
Total cholesterol (mg/dL)	−0.21 (−2.3, 1.9)	0.85
LDL cholesterol (mg/dL)	−0.17 (−2.0, 1.7)	0.86
HDL cholesterol (mg/dL)	−1.4 (−2.0, −0.82)	<0.0001
ln-Triglycerides (mg/dL)	0.05 (0.03, 0.08)	<0.0001
Fasting plasma glucose* (mg/dL)	0.03 (−0.35, 0.41)	0.87
Diabetes (%)	0.38 (0.29, 0.48)	<0.0001
eGFR (mL/min/1.73 m2)	−7.3 (−8.3, −6.3)	<0.0001

Models adjusted for age and sex (except for age and sex estimates). β represents the change associated with a 1-SD increase in suPAR. Body mass index (BMI) and triglycerides were natural log (ln) transformed. The standard deviation (SD) of suPAR was 795 pg/mL.

Associations of suPAR with measures of subclinical vascular disease and incident CVD events

In age- and sex-adjusted models, higher suPAR was associated with LVH, presence and severity of AAC and CAC, higher LVM index, and lower ABI (all p<0.005). In models additionally adjusted for CVD risk factors, the associations of suPAR with higher LVH (β= 0.23; 95% CI: 0.08, 0.38), LVM index (β=0.02; 95% CI: 0.009, 0.03), AAC score β=0.23 (95% CI: 0.07, 0.38) and lower ABI (β= −0.02; 95% CI: −0.009, −0.03) remained significant (Table 2).

Table 2.

Cross-sectional associations of suPAR with subclinical cardiovascular disease measures

		Model 1		Model 2
	N	Beta (95% CI)	p-value	Beta (95% CI)	p-value
Left Ventricular Hypertrophy (LVH)	2233	0.33 (0.18, 0.47)	<0.0001	0.23 (0.07, 0.38)	0.004
Any Aorto-iliac Calcium (AAC)	1936	0.27 (0.10, 0.43)	0.002	0.14 (−0.03, 0.30)	0.11
Any Coronary Artery Calcium (CAC)	1937	0.25 (0.11, 0.39)	0.0004	0.10 (−0.06, 0.26)	0.22
Left Ventricular Mass Index	2233	0.04 (0.03, 0.05)	<0.0001	0.02 (0.009, 0.03)	0.0009
AAC score	1936	0.42 (0.25, 0.59)	<0.0001	0.23 (0.07, 0.38)	0.004
CAC score	1937	0.30 (0.16, 0.44)	<0.0001	0.15 (0.007, 0.30)	0.04
Carotid intima media thickness (cIMT)	3317	0.007 (−0.002, 0.02)	0.13	−0.005 (−0.01, 0.005)	0.36
Ankle Brachial Index (ABI)	3099	−0.02 (−0.03, −0.01)	<0.0001	−0.02 (−0.03, −0.01)	<0.0001

Beta values are presented per 1-SD increase of suPAR (795 pg/mL). Model 1 adjusted for age and sex. Model 2 adjusted for age, sex, total cholesterol, HDL cholesterol, statin use, systolic blood pressure, blood pressure medication use, current smoking, and diabetes. Carotid IMT, left ventricular mass index, AAC and CAC were natural log transformed prior to analysis. The significance threshold corrected for multiple testing and defined as p<0.006.

Among the 3,492 participants, there were 645 deaths, 123 incident CHD events, 84 strokes, and 141 HF events during follow-up. The unadjusted Kaplan-Meier survival curves for each CVD outcome and mortality by suPAR quartile are presented in Supplemental Figures I–IV. In age- and sex-adjusted models, each 1-SD increase in suPAR was associated with an increased risk of CVD events and mortality (all p≤0.0001) (Table 3). Following adjustment for traditional CVD risk factors, eGFR and CRP, the association for CHD (HR: 1.24; 95% CI: 1.05, 1.46), stroke (1.24; 95% CI: 1.04, 1.47), HF (HR: 1.35; 95% CI: 1.22, 1.52), and overall mortality (HR: 1.32 (95% CI: 1.22, 1.42) remained significant (Table 3). Further adjustment of HF models for BNP attenuated the association but it remained statistically significant (HR: 1.29; 95% CI: 1.15, 1.46).

Table 3.

Associations of suPAR with incident cardiovascular disease events and mortality

			Model 1	Model 2	Model 3
Outcome	Events	N	HR (95% CI)	HR (95% CI)	HR (95% CI)
Coronary Heart Disease	123	2903	1.45 (1.28, 1.65)	1.24 (1.05, 1.46)	1.23 (1.04, 1.45)
Heart Failure	141	1921	1.44 (1.31, 1.58)	1.35 (1.20, 1.52)	1.35 (1.20, 1.52)
Stroke	84	2081	1.30 (1.14, 1.49)	1.24 (1.04, 1.47)	1.21 (1.02, 1.45)
All-Cause Mortality	645	3171	1.42 (1.31, 1.55)	1.32 (1.22, 1.42)	1.31 (1.22, 1.41)

Hazard ratios (HR) are for 1-SD increase of suPAR.

Model 1: Adjusted for age and sex.

Model 2: Coronary Heart Disease: Age, sex, systolic blood pressure, blood pressure medication use, total cholesterol, HDL cholesterol, statin use, eGFR, current smoking and diabetes;

Heart failure: Age, sex, systolic blood pressure, blood pressure medication use, eGFR, current smoking, diabetes, and history of cardiovascular disease, atrial fibrillation and left ventricular hypertrophy;

Stroke: Age, sex, systolic blood pressure, eGFR, blood pressure medication use, current smoking, diabetes, and history of cardiovascular disease, atrial fibrillation and left ventricular hypertrophy;

All-cause mortality: Age, sex, systolic blood pressure, blood pressure medication use, total cholesterol, HDL cholesterol, BMI, eGFR, current smoking and diabetes.

Model 3: Model 2 + CRP.

Genome-wide association study (GWAS) of suPAR

The age and sex-adjusted heritability of inverse normalized suPAR among related JHS participants was 0.466 (SE= 0.064; p=1.4×10⁻¹⁵). The GWAS of suPAR adjusted for age, sex, relatedness, and 10 PCs, showed no evidence for systematic inflation of results (λ=1.01). A single genomic region, containing the uPAR structural gene PLAUR (plasminogen activator, urokinase receptor) on chromosome 19q13, including 15 SNVs, was associated with suPAR (p<5×10⁻⁹) (Supplemental Figures V&VI). The sentinel SNV was rs399145 (p=2.2×10⁻⁷²), a PLAUR T>C missense variant resulting in a Thr86Ala substitution. Each copy of the rs399145 minor allele (MAF=0.11) was associated with 0.73 SD units (SE=0.04) higher ln-suPAR (Table 4). In JHS African Americans, we were unable to replicate two previously reported associations for suPAR conducted among Europeans: a promoter polymorphism of plasma kallikrein (KLKB1 rs4253238)¹⁰ and a common missense variant of PLAUR (rs2302524 encoding p.Lys220Arg)¹¹ (Supplemental Table VII). We suspect the previously reported rs2302524 coding variant association may represent an assay-dependent epitope-binding artefact (see Supplemental Results).

Table 4.

Single nucleotide variants identified by GWAS associated with suPAR

						Unconditioned analysis			Stepwise conditional analysis			MAF (1000 Genomes)
SNV	Annotation	Chr	build37	A1/A2	MAF (JHS)	p-value	β	95% CI	p-value	β	95% CI	Afr.	Eur.
rs399145 (Signal A)	p.Thr86Ala, PLAUR	19	44169522	T/C	0.11	2.0×10−78	0.73	0.65, 0.81	-	-	-	0.12	0.00
rs73935023 (Signal B)	intronic, PLAUR	19	44178955	T/C	0.04	1.1×10−16	−0.55	−0.69, −0.41	2.3×10−14	−0.50	−0.64, −0.36	0.03	0.00
rs4251805 (Signal C)	5’ UTR, PLAUR	19	44174441	C/T	0.08	1.8×10−14	−0.36	−0.46, −0.26	8.6×10−13	−0.34	−0.44, −0.24	0.09	0.03
rs4760 (Signal D)	p.Leu317Pro, PLAUR	19	44153100	A/G	0.03	3.9×10−04	0.27	0.11, 0.43	2.5×10−06	0.36	0.20, 0.52	0.00	0.16

Single nucleotide variants (SNVs) associated with suPAR before (unconditioned analysis) and after stepwise, forward-selection conditional analysis in participants of the Jackson Heart Study using ELISA-based suPAR levels (n=3492). Beta values and 95% confidence intervals (CI) represent the change in # of SD units of ln-transformed suPAR associated with a single copy of the minor (A2) allele. Minor allele frequencies (MAFs) are for Jackson Heart Study (JHS) participants, as well as African (Afr) and European (Eur) ancestry individuals in 1000 Genomes Phase 1. Chr, chromosome; UTR, untranslated region.

To identify additional SNVs within the PLAUR region independently associated with suPAR, we performed iterative, forward-selection conditional analysis and identified four distinct signals. After conditioning on rs399145 (signal A), the most significant SNV was rs73935023 (signal B, p_cond=2.3×10⁻¹⁴) located in the PLAUR promoter region. In the next rounds of conditional analysis, we identified rs4251805 (signal C, p_cond=8.6×10⁻¹³) located in the PLAUR 5’ untranslated region, and rs4760 (signal D, p_cond=2.5 ×10⁻⁰⁶), a PLAUR A>G missense variant resulting in a Leu272Pro substitution (Table 4 and Supplemental Figure VII). Together these four SNVs explained ~14% of the variation in suPAR. The minor alleles of rs399145 and rs73935023 (signals A and B) are present at an appreciable frequency only among African-ancestry individuals (MAF=0.0 in European-ancestry 1000 Genomes participants). rs4251805 (signal C) is also more common in African versus European-ancestry individuals (9% vs 3% in 1000 Genomes); in contrast, rs4760 (signal D) is more common in European versus African-ancestry individuals (16% vs 0% in 1000 Genomes) (Table 4).

Functional assessment of non-coding regulatory variants associated with suPAR

Next, we functionally characterized the sentinel variants and LD proxies in JHS (r²>0.7, Supplemental Table I) for two non-coding variant association signals represented by rs73935023 (signal B) and rs4251805 (signal C). In GTEx v8, a proxy for signal C, rs4251824, is the sentinel significant eQTL variant for PLAUR in skeletal muscle; and, a proxy for signal B, rs4251815, is the sentinel variant for a secondary significant distinct eQTL for PLAUR in whole blood.¹² Across relevant tissues, we identified variants with the highest functional annotation scores and tested them for allelic differences in transcriptional activity. Annotated variants included sentinel intronic variant rs73935023 (signal B), 5’UTR/promoter variant rs4251805 (signal C), and signal B proxies, rs4251808 and rs4251809. We cloned elements containing each of these three variant regions into luciferase reporter vectors and tested the effect of each cloned element on transcriptional activity in myoblasts, monocytes, and macrophages.

The element surrounding intronic variant rs73935023 (signal B) showed an allele-specific effect on enhancer activity with higher expression for the common T allele than the variant C allele in macrophage and myoblasts (Figure 1), consistent with the GWAS results in which the T allele was associated with higher suPAR levels and with eQTL evidence for the corresponding allele of a signal B proxy variant associated with higher PLAUR expression level in whole blood. The promoter element surrounding rs4251805 (signal C) and intronic element containing the signal B proxy variants rs4251809 and rs4251808 both showed higher transcriptional activity versus empty vector but no difference between alleles (Supplemental Figures VIII and IX).

Figure 1. PLAUR signal B variant rs73935023 is located within an enhancer element and exhibits allelic differences in transcriptional activity.

Reporter vectors containing a 370-bp fragment including either the T or C allele at rs73935023 showed significantly increased transcriptional activity compared to empty vector (EV) control in THP-1 monocytes (A), THP-1 differentiated macrophages (A), and C2C12 myoblasts (B). Bars represent means and standard errors of four independent clones. P-values show allelic differences from two-sided t-tests. Fwd, forward orientation; Rev, reverse orientation.

SNV-based tests of association with CVD events

With the identification of cis-acting genetic variants associated with suPAR levels, we explored the potential causal relationships between suPAR and clinical CVD events. We used a polygenic risk score (PRS) comprised of rs399145 (signal A), rs73935023 (signal B), rs4251805 (signal C), and rs4760 (signal D), to evaluate relationships with CVD risk. There was no evidence of association between genetically-determined suPAR and incident CHD, HF, or stroke in JHS (Supplemental Table VIII), but this analysis is based on relatively small numbers of CVD events. For this reason, we investigated PRS associations with subclinical CVD measures in JHS but did not observe any significant associations (Supplemental Table IX). We additionally evaluated the JHS-derived suPAR PRS in the setting of BioVU, an EHR-based resource that includes diagnosis codes for a large number of African Americans (n=15,123). We found no evidence of association between the suPAR PRS with diagnoses of ischemic heart disease, stroke, or congestive heart failure in either BioVU African Americans or BioVU European-ancestry individuals (n=70,337), or an ancestry-combined meta-analysis (Supplemental Table X). In the BioVU African American sample, assuming a genetic risk score explaining ~14% of the variation in suPAR, we had approximately 70% power to detect an odds ratio (OR) of 1.20 for ischemic heart disease and congestive heart failure and approximately 55% power to detect an OR of 1.20 for cerebrovascular disease, with alpha=0.05.

Discussion

Using a combination of epidemiologic, genomic, bioinformatic, and molecular approaches, we characterized the lifestyle, clinical, and genetic contributions to suPAR phenotypic variation in a community-based population of African American adults. Our results extend evidence for higher suPAR as a predictor of subclinical vascular disease, incident CVD events, and all-cause mortality (independent of CVD risk factors, eGFR, and CRP) to the general African American population. Further, we identified several ancestry-differentiated genetic variants associated with suPAR phenotypic variation among African-Americans. Although we did not observe associations between a suPAR polygenic risk score with incident CVD events, larger multi-ethnic studies may be required to adequately address the potential causal role of genetically-determined suPAR on CVD.

Epidemiology of suPAR and role of suPAR in CVD risk prediction

Our results from a community-based sample of African American adults are generally consistent with prior studies performed mainly in European-ancestry populations^13–18 showing higher suPAR levels are associated with female sex, increasing age, unhealthy lifestyles (smoking, sedentary behavior / obesity, alcohol intake), psychosocial factors, CVD risk factors (e.g., diabetes, hypertension, dyslipidemia), and lower eGFR. In JHS, demographic, behavioral and clinical factors accounted for ~29% of the suPAR phenotypic variance. We also confirmed that suPAR is moderately correlated with several biomarkers of innate and adaptive immune activation and coagulation. Compared to the widely used inflammation-sensitive biomarker CRP, suPAR is not an acute phase protein and is less susceptible than CRP to acute, short-term inflammatory fluctuations.⁸ Therefore, suPAR has been suggested as a useful prognostic biomarker of chronic inflammation and immune activation across a range of clinical settings.¹³

Several prior cross-sectional and longitudinal studies have implicated suPAR as a biomarker for atherosclerosis, clinical CVD events, and mortality in patients with preexisting diseases^14–21 or individuals from the general population.^22–26 The associations of higher suPAR with prevalent subclinical vascular disease measures (LVH, LVM index and AAC) and incident CHD, HF, stroke, and mortality in JHS following adjustment for CVD risk factors are consistent with prior studies and provide support for suPAR as a CVD biomarker in African Americans, independent of both eGFR and CRP. Together with the report of Botha et al. from Black South Africans showing suPAR predicts total and CVD mortality,¹⁹ our findings extend the importance of suPAR as a CVD biomarker to African-ancestry populations. Further, we add to recent literature that suPAR predicts incident HF independently of BNP and is associated with worse baseline LV function.²⁷

PLAUR gene variants and genetic regulation of suPAR

PLAUR expression is regulated transcriptionally and post-transcriptionally by various cytokines, hormones, and growth factors through cis- and trans-acting regulatory elements.²⁸ However, the specific genetic factors accounting for phenotypic variation in suPAR levels remain largely unknown and have been assessed previously only among Europeans.¹⁰ In JHS, we identified four independent SNVs in the PLAUR gene region on chromosome 19 that together explained ~14% of the variation in suPAR. Two of these genetic variants (rs399145 and rs73935023, signals A and B) are African-ancestry specific based on 1000 Genomes, one is European-ancestry specific (rs4760, signal D), and one (rs4251805, signal C) is more common in African compared with European populations.

Variants at the four signals could affect PLAUR function. Sentinel variants at signals A and D affect protein sequence, and signal C (rs4251805) is located in the PLAUR promoter, 118 bp upstream of the transcription start site. This promoter polymorphism was identified previously but was not known to have any functional or phenotypic consequences.²⁹ The promoter region overlapping rs4251805 (as well as the enhancer element overlapping rs73935023) in hematopoietic cells contains binding sites for several transcription factors that regulate inflammatory and immune responses including SP1, CEBPB, c-JUN, and STAT.³⁰ At signal B, we identified an allele-specific effect of rs73935023 on transcriptional activity in macrophages and myoblasts. Based on our assay results and eQTL evidence, the rs73935023 allele may affect PLAUR expression level in vivo. Our inability to detect allele-specific differences in transcriptional activity for the rs4251805 promoter variant may be due to the portion of the promoter tested or the strong promoter activity observed, which may mask subtler changes on PLAUR expression associated with allelic differences in rs4251805.

Circulating suPAR is a three-domain protein (D1, D2, D3) generated from membrane-bound uPAR by cleavage near the GPI anchor.³¹ The sentinel variants at signals A and D could impact circulating suPAR levels through post-transcriptional effects on protein secretion, stability, or sensitivity of either suPAR or membrane-bound uPAR to proteolytic cleavage. The signal A PLAUR rs399145 missense variant (p.Thr86Ala) is located within a protease-sensitive linker region (amino acids 84–94) that connects D1 and D2-D3. A conformational change in this linker region has been implicated in chemotaxis function of suPAR³² and susceptibility of suPAR vs. GPI-anchored uPAR to urokinase-mediated cleavage.³³ Alternatively, a structural change induced by the Thr86Ala substitution in this portion of the molecule may affect binding of suPAR to antibody or aptamer reagents used to measure suPAR in commercial assays. We evaluated this possibility further under the Supplemental Results and demonstrate that the particular suPAR assay can strongly influence the results for Thr86Ala (or other protein-altering variants such as rs2302524 p.Lys220Arg which is likely due to an epitope-binding artifact) in aptamer-based suPAR assays. This phenomenon has been observed in other plasma protein genetic studies,^{34, 35} underscoring the importance of confirming cis-pQTLs due to missense variants using orthogonal assay methods.

Causal relationship of genetically-determined suPAR to CVD

While suPAR is suggested as a useful prognostic CVD biomarker, it remains to be determined whether it plays a causal role in the disease process. Genetic variants, particularly those within the PLAUR structural gene associated with suPAR phenotypic variation, can be utilized in PRS or Mendelian randomization (MR) analyses to investigate causality. Our PRS analyses, however, did not provide evidence for associations of genetically-mediated suPAR levels with incident CVD risk in JHS or in the larger BioVU EHR database. Given these non-significant results, a formal MR analysis would not clarify our findings; if there is a true causal effect of genetically-mediated suPAR levels with incident CVD risk, we are not powered to detect it in JHS or BioVU. The analyses in BioVU European-Americans are limited by the low European population allele frequency of three of the PLAUR variants identified (signals A, B, and C) and therefore the absence of these three variants in larger Euro-centric PheWAS databases. It is also possible the common African rs399145 p.Thr86Ala variant represents an assay artefact rather than a true cis-pQTL (discussed further in the Supplemental Results), which would decrease our statistical power to detect an association between genetically determined suPAR and CVD events in JHS or BioVU African Americans. Although null, results in JHS and BioVU cannot exclude the possibility of causality. Mendelian randomization studies conducted in larger, ethnically diverse studies that are currently accruing³⁶ may be required to adequately assess the causal role of suPAR on atherosclerosis and CVD in African Americans.

In contrast to the three suPAR-associated variants that are more common among African populations, the fourth independent SNV associated with higher suPAR in JHS, rs4760 (signal D), is only common among Europeans. The rs4760 variant encodes a Leu272Pro missense substitution located within uPAR’s D3 domain and is the sentinel eQTL variant for PLAUR expression in blood in eQTLGen.³⁷ The recently reported association of rs4760 with lower WBC count and with alterations in the number and proportions of circulating neutrophils, monocytes, lymphocytes, eosinophils and basophils³⁸ may reflect differences in cellular or plasma levels of uPAR or functional differences in signal transduction related to uPAR’s role in inflammatory responses, immune cell adhesion, migration and chemotaxis, and/or hematopoietic cell differentiation.⁴

Conclusions

We have demonstrated the importance of ancestry-differentiated coding and regulatory genetic variation on circulating suPAR levels. Results from JHS suggest higher suPAR is a biomarker for incident CVD events and all-cause mortality independent of CVD risk factors, eGFR, and inflammation biomarkers in African American adults. Although we did not observe an association between suPAR-associated SNVs with CVD events, additional analyses incorporating multi-ethnic genetic data sets that include even larger numbers of African-ancestry participants may ultimately be required to address the question of whether suPAR is causally related to CVD. Collectively, our results highlight the importance of suPAR genetic studies in non-European populations.

Non-standard Abbreviations and Acronyms

AAC

aortic artery calcification

ABI

ankle brachial index

BNP

B-type natriuretic peptide

CAC

coronary artery calcification

cIMT

carotid intima media thickness

eGFR

estimated glomerular filtration rate

EHR

electronic health record

HF

heart failure

JHS

Jackson Heart Study

LD

linkage disequilibrium

LVH

left ventricular hypertrophy

LVM

left ventricular mass

MAF

minor allele frequency

MR

Mendelian randomization

PLAUR

plasminogen activator, urokinase receptor

PRS

polygenic risk score

RDW

red cell distribution width

SNV

single nucleotide variant

suPAR

soluble urokinase-type plasminogen activator receptor

TOPMed

Trans-Omics for Precision Medicine

WBC

white blood cell

WGS

whole-genome sequencing