Skip main navigation
×

Genetic Regulation of PCSK9 (Proprotein Convertase Subtilisin/Kexin Type 9) Plasma Levels and Its Impact on Atherosclerotic Vascular Disease Phenotypes

Originally publishedhttps://doi.org/10.1161/CIRCGEN.117.001992Circulation: Genomic and Precision Medicine. 2018;11:e001992

    Abstract

    Background:

    Inhibition of PCSK9 (proprotein convertase subtilisin/kexin type 9) is a novel strategy to treat hypercholesterolemia and reduce cardiovascular events. However, the potential role of circulating plasma PCSK9 concentrations as a diagnostic and predictive biomarker remains uncertain as of now. Here, we aimed to identify genetic variants associated with plasma PCSK9 and investigate possible causal effects on atherosclerotic vascular disease phenotypes.

    Methods:

    We performed the first genome-wide association study of plasma PCSK9 levels in a cohort of suspected and confirmed coronary artery disease (LIFE-Heart; n=3290).

    Results:

    Several independent variants at the PCSK9 gene locus were associated with circulating PCSK9 levels at genome-wide significance (lead SNP rs11591147, PCSK9-R46L; P=1.94×10−17). We discovered 4 independent PCSK9 SNPs explaining 4.4% of the variance of plasma PCSK9. In addition, we identified a genome-wide significant locus at chromosome 7p22.1 (rs6957201; P=7.01×10−9) and 7 suggestive hits (P<1×10−6). Using MR (Mendelian Randomization), we detected significant causal effects of circulating PCSK9 on coronary artery disease status and severity, carotid plaques, and intima-media thickness.

    Conclusions:

    Variants at the PCSK9 gene locus seem to be the major genetic determinants of plasma PCSK9 levels with 4 independent variants at the PCSK9 gene locus expressing allelic heterogeneity. The detected MR estimates support the hypothesis of a causal effect of PCSK9 on coronary artery disease and other vascular phenotypes. Other observed genetic associations for PCSK9 require validation in independent cohorts.

    Clinical Trial Registration:

    URL: http://www.clinicaltrials.gov. Unique Identifier: NCT00497887.

    Introduction

    See Editorial by Paquette and Baass

    Clinical Perspective

    PCSK9 (proprotein convertase subtilisin/kexin type 9) plays an essential role in regulating lipid metabolism. Functional inhibition of PCSK9 is a novel treatment strategy for hyperlipidemia. In our genome-wide association study, we identified 4 independent genetic variants at the PCSK9 locus at genome-wide significance, explaining 4.4% of the total variance of PSCK9. In an MR study (Mendelian randomization), we detected causal effects of PCSK9 plasma levels on coronary artery disease, number of coronary vessels with stenosis, and carotid artery plaques but not peripheral artery disease. This increases confidence that PCSK9 is a valid target not only to treat hyperlipidemia but also to reduce risk or severity of atherosclerotic vascular disease in carotid and coronary arteries.

    Elevated plasma levels of LDL-C (low-density lipoprotein cholesterol) are a major modifiable risk factor for atherosclerotic cardiovascular disease. Therapeutic interventions facilitating the hepatic uptake of LDL via the LDLR (LDL receptor) pathway are proven to reduce cardiovascular events.1 PCSK9 (proprotein convertase subtilisin/kexin type 9) plays an essential role in lipid metabolism as a key regulator of plasma LDL-C.2 The protein is mainly produced and secreted by the liver. It binds LDLR on the cell surface of hepatocytes and targets them for lysosomal degradation, resulting in increased LDL-C. Rare gain-of-function mutations in PCSK9 were identified to cause autosomal dominant hypercholesterolemia, whereas loss-of-function mutations in PCSK9 were found to be associated with low LDL-C and reduced risk of coronary artery disease (CAD).3,4 Moreover, genome-wide association analysis showed that common variants in the PCSK9 gene locus are associated with LDL-C and the risk for CAD.5,6 Therefore, inhibiting PCSK9 function has emerged as a promising strategy to treat hypercholesterolemia. Indeed, 2 monoclonal antibodies against PCSK9 (Alirocumab and Evolocumab) have been approved7,8 and shown to markedly reduce LDL-C in a wide range of patients. The use of evolocumab was also associated with a significant reduction of future cardiovascular events in a large-scale outcome trial.8

    Recent studies indicate that the functional effects of PCSK9 may not be limited to the LDLR pathway. PCSK9 was implicated to affect several other atherogenic risk factors, such as the metabolism of triglyceride-rich lipoproteins, degradation of the very-low-dense-lipoprotein receptor, inflammatory response, and glucose metabolism.2 In this context, circulating PCSK9 concentrations have gained interest as a potential biomarker for risk stratification. Current findings on the relation between PCSK9 plasma levels and cardiovascular disease are equivocal, but a meta-analysis of 7 studies reported that subjects with high PCSK9 levels have a 23% higher risk for total cardiovascular events when compared with patients with low PCSK9 levels.9

    Although considerable variation in circulating PCSK9 levels has been noted in different cohort studies,911 information on the genetic factors that contribute to this variability is limited. A recent GWAS (Genome-Wide Association Study) of PCSK9 plasma levels failed to detect genome-wide significant variants (genome-wide threshold P<5×10−8) in 2 cohorts of healthy, middle-aged Swedes, which might be because of limited statistical power given the moderate sample size (N=1215).10 In addition, causal links between PCSK9 plasma levels and atherosclerotic vascular disease (ASVD) phenotypes have not been studied using genetic instruments to date.

    In the present study, we performed genome-wide association analysis for circulating PCSK9 levels in a large cohort of patients receiving coronary angiography for suspected CAD (N=3290; LIFE-Heart study). Further, we analyzed allelic heterogeneity of the PCSK9 gene locus and applied MR (Mendelian Randomization) to identify causal effects of circulating PCSK9 levels on various atherosclerotic disease phenotypes including additional data of the large population-based LIFE-Adult study.

    Methods

    The data that support the findings of this study are available from the LIFE Research Center for Civilization diseases on qualified request. Requests for access to more detailed summary statistics, replication results, and analytic methods will be considered by the authors.

    Cohort Descriptions

    LIFE-Heart is an observational study of patients collected at the Heart Center of Leipzig, Germany. A total of ≈7000 patients were recruited with either suspected stable CAD or myocardial infarction. Study design and a detailed description of patients can be found elsewhere.12 For the present analysis, we only included patients with suspected stable CAD subjected to coronary angiography. PCSK9 plasma levels and genetic data were available for N=3358 of these patients.

    LIFE-Adult is a population-based cohort study of adult residents of the city of Leipzig, Germany. A total of ≈10 000 participants have been recruited and characterized including subclinical atherosclerosis phenotypes.13 For a total of 4985 LIFE-Adult participants, genetic data were available. These samples were used to improve power of the MR analysis.

    Both LIFE-Heart and LIFE-Adult meet the ethical standards of the Declaration of Helsinki and were approved by the Ethics Committee of the Medical Faculty of the University Leipzig, Germany (LIFE-Heart: Reg. No. 276-2005; LIFE-Adult: Reg. No. 263-2009-14122009). Written informed consents including agreement with genetic analyses were obtained from all patients of LIFE-Heart and participants of LIFE-Adult.

    Assessment of Atherosclerotic Phenotypes

    We analyzed 6 ASVD phenotypes at 3 locations: coronary arteries: CAD status and number of coronary vessels with stenosis (NVD50); carotid arteries: intima-media thickness (cIMT), and plaque score (PS; with values between 0 and 4 counting the number of vessels with plaque); and peripheral arteries: ankle-brachial index (ABI) and peripheral artery disease (PAD) status (defined as ABI<0.9). Coronary phenotypes are only available for LIFE-Heart. Details of the measurements are provided in the Data Supplement.

    Blood Lipid and PCSK9 Measurement

    Venous blood samples were taken before coronary angiography in patients of the LIFE-Heart study. Laboratory measurements were performed on the same day using an automated Roche Cobas 8000 Clinical Chemistry analyzer (Roche Diagnostics, Mannheim, Germany). Total cholesterol, LDL-C, and HDL-C (high-density lipoprotein cholesterol) were determined by homogeneous enzymatic colorimetric assays. ApoA1 and ApoB (apolipoproteins) were determined with Roche immunoturbidimetric assays. Total PCSK9 levels were analyzed in serum samples (previously stored at −80°C) using a commercial assay (Quantikine Human PCSK9 immunoassay; R&D Systems). Further details of PCSK9 measurement are given in the Data Supplement.

    SNP Genotyping and Imputation

    LIFE-Heart samples were genotyped using Affymetrix Axiom CEU1 or Affymetrix Axiom CADLIFE genome-wide SNP arrays. The latter essentially contains Axiom CEU as genome-wide backbone and an additional custom content of ≈62 500 SNPs from known CAD loci. Genotype calling was performed using Affymetrix Power Tools (v1.17.0 for Axiom CADLIFE and v1.16.1 for Axiom CEU) with their latest libraries (Axiom CADLIFE1, release 3 and Axiom Genome-Wide CEU 1 Array Plate, Analysis Files, release 6, respectively). LIFE-Adult samples were genotyped with Affymetrix Axiom CEU1 SNP array, and calling relied on Affymetrix Power Tools (version 1.17.0) with the same library as LIFE-Heart.

    An in-house pipeline performed SNP and sample quality control. Details can be found in the Data Supplement. For LIFE-Heart, 5700 (5688) samples and 504 593 (12 715) SNPs fulfilled all quality criteria for autosomal (X-chromosomal, respectively) analyses. For LIFE-Adult, 4985 (4978) samples and 532 875 (13 554) SNPs fulfilled all quality criteria.

    Imputation was performed using 1000 Genomes Phase 3, Version 514 (2015) as reference, SHAPEIT15 (version v2.r837) for phasing, and IMPUTE216 (version 2.3.2) for genotype estimation. For chromosome X, the same reference and software was used.

    SNPs with info score <0.5 or minor allele frequency <1% were filtered. A total of 9 882 017 SNPs were used for genome-wide analyses, 9 579 329 for autosomal, and 302 688 for X-chromosomal analysis.

    Statistical Analysis

    Genome-Wide Association Analysis

    PCSK9 and BMI were log-transformed for all analyses.

    Before any GWAS, we controlled several parameters for association with plasma PCSK9. Only sex, age, smoking, and statin therapy were significantly associated in a multivariate model. In addition, we found a significant sex–age interaction (P=0.0041), but decided against an inclusion of this covariate for GWA because it was expected that this interaction can hardly influence the SNP effect. Nevertheless, for the purpose of validation, we checked all genome-wide and suggestive GWA hits with this model and a model containing no covariates (univariate analysis). No relevant differences were found.

    We performed 2 genome-wide association analyses for PCSK9 levels in LIFE-Heart: the first GWAS included all patients (N=3290 patients with complete covariates), and the second was restricted to those without statin therapy (N=2022 patients with complete covariates, denoted as statin-free subset). Both analyses were executed with SNPTEST17 (version 2.5.2) using the additive frequentist model and expected genotype counts. We adjusted for sex, age, and current smoking in both analyses, and additionally, for lipid therapy in the first analysis. The threshold for genome-wide significance was set to P<5×10−8. Associations with P<1×10−6 were considered as suggestive and presented as list of top SNPs.

    X-chromosomal SNPs were analyzed assuming total X inactivation, that is, the gene doses of females are halved according to König et al.18 To consider possible alternatives, we tested X-chromosomal top hits with a model without X inactivation and analyzed sex–SNP interactions. No such interactions were found at genome-wide significance.

    Linkage disequilibrium between markers was calculated using data from 1000 Genomes Phase 3, Version 5 (2015)14 for European samples. Priority pruning of the top list was performed as follows: first, variants of the list of top SNPs which are in linkage disequilibrium with an association of higher significance were considered as tagged by the other variant if r2≥0.5. Then, to analyze the PCSK9 locus in detail, we used a stricter threshold of r2≥0.1.

    A comprehensive annotation was applied to all SNPs of our top list using the following bioinformatics resources: Ensemble,19 GWAS catalogue,20 expression quantitative trait loci data,21,22 and pathways from KEGG, GO, DOSE,23 and Reactome.24 Deleteriousness scores for nonsynonymous coding SNPs were calculated according to Kircher et al25 and Bendl et al.26 Further details are summarized in the Data Supplement.

    Relation of PCSK9 and ASVD

    Vascular phenotypes (CAD, NVD50, PS, cIMT, PAD, and ABI) were tested for association with PCSK9 levels and SNPs discovered in our GWAS. CAD and PAD status were analyzed by logistic regression, the ordered categories of NVD50 and PS by proportional odds logistic regression, and cIMT and ABI by linear regression.

    MR Analysis

    We performed an MR study using 4 independent PCSK9 variants with pairwise r2<0.1 as instrumental variables (IVs), plasma PCSK9 levels as exposure, and the vascular phenotypes as outcomes. Variants were assumed to satisfy MR assumptions for IVs as explained in the Data Supplement (the relevant directed acyclic graph is given in Figure I in the Data Supplement).

    LIFE-Heart was used to calculate the regression coefficients of IV on plasma PCSK9 and of IV on CAD and NVD50. LIFE-Heart and LIFE-Adult were used to calculate regression coefficients of IV on PS, cIMT mean, PAD, and ABI. We performed a fixed-effect model meta-analysis to generate combined estimates of IV on outcome. We then calculated the combined inverse-variance weighted causal estimate βIVW for each outcome.27 To test whether our positive causal results are confounded by pleiotropy, we applied the modified Q test.28 Test was negative for all outcomes.

    If not stated otherwise, statistical analyses were performed with the software R.29 For all analyses other than genome-wide SNP association, we applied a (nonadjusted) significance threshold of 5%, that is, our results are exploratory rather than confirmative.

    Results

    Relationships of Cardiovascular Risk Factors and Plasma Lipids to Circulating PCSK9 Levels

    The characteristics of the LIFE-Heart study population are summarized in Table 1. Plasma PCSK9 levels were associated with sex, age, BMI, diabetes mellitus, smoking status, hypertension, and statins in univariate analysis (Table I A in the Data Supplement). In a multivariate analysis, only sex, age, smoking status, and statins remained significant. Similar effect sizes of sex, age, and smoking were observed in the statin-free subset (Table I B in the Data Supplement). Male sex and age were associated with lower PCSK9 levels, that is, these factors have an opposite effect on PCSK9 than on CAD risk, whereas smoking and statins increased PCSK9. We observed a significant and qualitative interaction between sex and age (Figure II in the Data Supplement; Table IC and ID in the Data Supplement).

    Table 1. Characteristics of the LIFE-Heart Study Population

    ParameterAllMen vs WomenLipid vs No Therapy
    Men/women2167/11910.448
    Age, y61.7±10.8<0.001*<0.001
    Current smoker699<0.001*0.701
    Statin therapy12700.448
    BMI, kg/m229.7±5.00.083<0.001
    Type 2 diabetes mellitus10380.125<0.001
    Hypertension30240.721<0.001
    Systolic BP, mm Hg138.5±18.60.1720.074
    Diastolic BP, mm Hg83.5±10.90.477<0.001
    PCSK9, ng/mL230.9 (188.9–282.6)<0.001<0.001
    Cholesterol, mmol/L5.43±1.21<0.001<0.001
    HDL-C, mmol/L1.34±0.41<0.001<0.001
    LDL-C, mmol/L3.32±1.040.056<0.001
    ApoA, g/L1.51±0.29<0.0010.070
    ApoB, g/L1.00±0.280.223<0.001
    NVD50<0.001*<0.001
     01230
     0.5632
     1503
     2405
     3449
    Established CAD1357<0.001*<0.001
    cIMT, mm0.79±0.15<0.001*<0.001
    PS<0.001*<0.001
     01154
     1549
     2756
     3371
     4320
    PAD3550.127<0.001
    ABI1.08±0.19<0.001*<0.001

    All parameters are restricted to genotyped patients with PCSK9 measurement. For the continuous parameters, the arithmetic mean and SD is given. We tested men against women and statin treatment against no treatment. All binary parameters were tested with a χ2 test; continuous parameters were tested with a Mann–Whitney U test. NVD50 and PS were tested with proportional odds regression. ABI indicates ankle-brachial index; Apo, apolipoprotein; CAD, coronary artery disease; cIMT, carotid intima-media thickness; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; PAD, peripheral artery disease; and PS, plaque score.

    *More cases or higher values in men for binary or continuous parameters, respectively.

    More cases or higher values in the statin group for the binary parameters or continuous parameters, respectively.

    Known, in therapy or acute.

    PCSK9 levels were positively correlated with total cholesterol (Spearman ρ=0.16; P=2.06×10−19), LDL-C (ρ=0.10; P=8.66×10−9), ApoB (ρ=0.14; P=1.19×10−16), and ApoA1 (ρ=0.11; P=3.65×10−10) but not with HDL-C after controlling for sex, age, smoking, and statins (Table II in the Data Supplement). These correlations were stronger in the statin-free subset (Figure III in the Data Supplement).

    Genome-Wide Association Analysis of PCSK9 Plasma Levels

    To identify genetic variants associated with plasma PCSK9 levels, we performed GWA of all 3290 LIFE-Heart samples and then evaluated the consistency of the results in the statin-free subset (n=2022). In the primary GWAS, we identified a total of 31 variants (5 variants after pruning) with genome-wide significance. No inflation of test statistics was observed (λ=1.006). Manhattan plots and QQ plot are displayed in Figure 1 (statin-free subset analysis is given in Figure IV in the Data Supplement). Summary statistics and SNP annotations are displayed in Table III A in the Data Supplement (statin-free subset is given in Table III B in the Data Supplement).

    Figure 1.

    Figure 1. QQ plot and Manhattan plot.A, QQ plot of P values of the GWAS of all study subjects. SNPs with info score <0.8 are plotted as triangles, and those with minor allele frequency <0.05 are colored red. There is no inflation in our data (λ=1.006). B, Distribution of log-transformed P values of the GWAS of all study subjects. The bold line marks genome-wide significance (P=5×10−8). Two loci reached genome-wide significance, one at chromosome 1 (PCSK9) and the other one at chromosome 7 (FBXL18).

    Genetic Variants at the PCSK9 Locus

    The strongest associations with PCSK9 plasma level were observed on chromosome 1p32.3 at the PCSK9 locus. Here, 4 independent SNPs (r2<0.1) located within the gene or a distance of <60 kb were associated at genome-wide significance (Table 2; Tables III A and IV in the Data Supplement). A regional association plot of the locus is shown in Figure V A in the Data Supplement. The lead SNP rs11591147 (β=−0.315; P=1.94×10−17) is coding for the PCSK9-R46L missense mutation and was previously reported to associate with LDL-C5,30 and response to statin therapy.31 Further, the PCSK9-R46L allele was associated with lower plasma PCSK9 levels and CAD risk in candidate gene studies.11

    Table 2. Genome-Wide Significant SNPs of One of the GWAS

    SNP InfoStudy 1: All Patients (N=3290)Study 2: Statin-Free Patients (N=2022)
    FlagLocusLead SNPNearby Genes (distance)Info ScoreEffect/Other AlleleβSEP ValueExp VarEAFβSEP ValueExp VarEAF
    11p32.3rs11591147PCSK9 (0 kb)0.581T/G−0.3150.0371.94E-17*0.0180.012−0.3740.0465.91E-16*0.0300.012
    11p32.31:55520994PCSK9 (0 kb)0.761GCG/G0.0600.0083.04E-14*0.0150.5630.0700.0101.96E-12*0.0230.558
    11p32.3rs45448095PCSK9 (0 kb)0.683T/C−0.0850.0124.40E-12*0.0120.133−0.0820.0153.19E-080.0140.139
    01p32.3rs373507733PCSK9 (0 kb)0.581A/AT−0.0710.0112.24E-110.0110.234−0.0720.0132.61E-08*0.0140.242
    11p32.3rs2479409PCSK9 (0.57 kb)0.646A/G−0.0520.0097.41E-09*0.0090.640−0.0570.0113.19E-070.0120.640
    07p22.1rs6957201FBXL18 (0 kb) TNRC18 (10 kb)0.549C/T−0.1630.0287.01E-09*0.0090.971−0.1670.0351.70E-060.0110.970

    Model was adjusted for sex, age, current smoking, and statin treatment. EAF indicates effect allele frequency; ExpVar, explained variance; and Flag, SNPs used in Mendelian Randomization analysis.

    *Genome-wide significant and independent hits.

    Pairwise linkage disequilibrium (r2>0.3), tagged differently because of priority pruning.

    Associations with LDL-C and coronary heart disease were also reported for other identified SNPs at the PCSK9 locus.6,32,33 These variants are in strong linkage disequilibrium (r2>0.5) with several cis-eQTL-SNPs for PCSK9 (P=3.27×10−27;22 Tables IV and V in the Data Supplement), suggesting a potential functional effect via regulation of PCSK9 transcription.

    Association results were robust when restricting analysis to the statin-free subset. However, because of power loss and different priority pruning, only 3 SNPs reached genome-wide and 2 suggestive significance (regional association plot is given in Figure VI A in the Data Supplement). Of note, absolute values of β estimates increased for the majority of variants (Table 2; Table III B in the Data Supplement; Figure VII in the Data Supplement). Association results were also robust against additional adjustments about the observed age–sex interaction and testing the variants in a univariate model (Table VI in the Data Supplement).

    Because multiple SNPs in low linkage disequilibrium at the PCSK9 locus were associated with plasma PCSK9 levels (Figure 2), we analyzed this allelic heterogeneity in more detail. Of the 4 independent SNPs initially identified at this locus, 3 SNPs (rs11591147, rs45448095, and 1:55520994) remained significant when tested in a multivariate model. Together they explain 4.4% variance of plasma PCSK9. In a combined analysis of these SNPs and all covariates, the SNPs remained significantly associated, and the explained variance increased to 19.4%.

    Figure 2.

    Figure 2. Heat map of pairwise linkage disequilibrium (r2) for the SNPs at the PCSK9 locus. Gene range was added according to University of California, Santa Cruz Genes track (hg19). P values of the GWAS were shown according to the position in the gene, not in heat map. Colors mark the genome-wide significant SNPs with pairwise LD r2<0.1 (blue: rs11591147, green: rs45448095, magenta: rs2479409, and cyan: 1:55520994).

    FBXL18 Locus

    Besides the PCSK9 locus, we identified a second locus with genome-wide significance on chromosome 7 (Table 2; Figures V B and VI C in the Data Supplement). The lead SNP is rs6957201 (β=−0.163; P=7.01×10−9). However, the SNP has low minor allele frequency (3%) and inferior imputation quality (info score of 0.549) reducing its credibility. Association results were also consistent in the statin-free subset (β=−0.167; P=1.70×10−6) and robust against additional adjustment for the observed age/sex interaction (Table VI in the Data Supplement).

    Suggestive Loci

    Applying a suggestive significance threshold of P<1×10−6, we identified 3 loci when analyzing all patients (Table VI in the Data Supplement; regional association plots are given in Figure V in the Data Supplement). Those loci were at Xq27.3 (nearest gene SPANXN4), 22q12 (TOM1), and 13q13 (RPS12P24). Restricting GWA analysis to the statin-free subset revealed 4 suggestive loci (6q13 within KCNQ5, 18q12 near SYT4, 1p31 near IFI44, and 4p15; Table VI in the Data Supplement; Figure VI in the Data Supplement). Comprehensive annotation for all suggestive loci is provided in the Data Supplement.

    Replication of Previously Reported SNPs

    Chernogubova et al10 reported 6 SNPs at the PCSK9 locus with nominal significance. Of those 6, 4 were significant in our data, 2 even on genome-wide level. Outside the PCSK9 locus, we could not replicate the 19 reported hits (Table VII in the Data Supplement).

    Theusch et al34 reported 7 SNPs associated with the difference in PCSK9 levels before and after statin treatment. None of these SNPs showed significant association with PCSK9 plasma levels in our data (Table VII in the Data Supplement).

    In addition, we looked up known nonsynonymous coding mutations in the PCSK9 gene and tested them for association with circulating PCSK9 levels. Out of 50 reported mutations,35,36 only 5 were in our genotype data. All but one were significantly associated with PCSK9 plasma levels (Table VIII in the Data Supplement).

    Further details on replication analyses are documented in the Data Supplement.

    Associations With ASVD and MR

    We next analyzed the relationship between circulating PCSK9 and the 4 genome-wide significant PCSK9 SNPs with CAD and other atherosclerotic phenotypes available in LIFE-Heart (Table 3; Tables IX through XII in the Data Supplement).

    Table 3. Summary of MR Studies Using Inverse-Variance Weighted Method

    ASVDAll PatientsStatin-Free Patients
    CorrelationCausalityCorrelationCausality
    N (All)N (Subset)βSEP ValueβSEP ValueβSEP ValueβSEP Value
    CAD321919280.1920.1170.1010.9600.4860.048*−0.0890.1660.5891.1180.6410.081
    NVD50321919280.1850.1040.0751.2230.4340.005*−0.1280.1460.3821.5230.6330.016*
    PS3150+46421904+38170.3010.1040.004*0.6050.2660.023*0.1110.1440.4400.7690.2900.008*
    cIMT3088+47011863+3862−0.0190.0090.030*0.0410.0220.061−0.0400.0130.002*0.0450.0230.054
    PAD3220+44273220+44270.3980.1850.032*−0.5810.7080.412−0.1300.2820.646−1.1081.2490.375
    ABI3220+44273220+4427−0.0490.0116.36E-06*0.0060.0220.794−0.0300.0130.028*0.0230.0210.287

    Univariate simple regression results of PCSK9 and the different ASVDs are provided in the Correlation columns. MR-based causal effects of PCSK9 on the ASVDs are given in the Causality columns. MR results for coronary phenotypes were calculated in LIFE-Heart only. ABI indicates ankle-brachial index; ASVD, atherosclerotic vascular disease; CAD, coronary artery disease; cIMT, carotid intima-media thickness; PAD, peripheral artery disease; and PS, plaque score.

    *Significant P values (P<0.05).

    Association of PCSK9 Variants With Vascular Phenotypes

    We identified one PCSK9 SNPs (1:55520994) that was associated with the number of affected coronary arteries (NVD50) at nominal significance (P<0.05). It was also significantly associated with the presence of CAD in the statin-free subset. One of the other PCSK9 SNPs was associated with cIMT (rs2479409 for all patients). It was also associated with PAD in the statin-free subset (Table IX in the Data Supplement).

    We also analyzed the SNP associations with plasma lipid traits. The SNP 1:55520994 was associated with total cholesterol, LDL-C, and ApoB in both, all subjects, and the statin-free subset. Including patient with statins attenuated the associations with plasma lipids (Table IX in the Data Supplement).

    Correlation of PCSK9 Plasma Levels With ASVD

    PCSK9 plasma levels were significantly associated with CAD status and severity (P=0.034 and P=0.031, for CAD and NVD50, respectively) only in the statin-free subset (adjusted for all common risk factors: sex, age, smoking, hypertension, BMI, and type 2 diabetes mellitus). Including patients with statins reduced both the association and β estimators (Table X in the Data Supplement).

    In addition, circulating PCSK9 levels were significantly associated with all analyzed subclinical atherosclerosis parameters (cIMT, P=0.030; PS, P=0.004; PAD, P=0.032; ABI, P=6.36×10−6) in the univariate analysis of all subjects.

    MR With PCSK9 Variants as IV and ASVD Phenotypes as Outcomes

    Using the 4 independent (r2<0.1), genome-wide significant PCSK9 SNPs as IVs, we performed MR to analyze whether circulating PCSK9 plasma levels causally affect vascular phenotypes (Table 3).

    We found significant causal effects for both CAD status and severity (βIVW (CAD)=0.960, P=0.048; βIVW (NVD50)=1.223, P=0.005) in the entire cohort. In the statin-free subset analysis, only βIVW (NVD50) reached significance. For the analysis of carotid and peripheral artery phenotypes, we used the combined information of the LIFE-Heart and LIFE-Adult studies (total N=7647–7792). We detected significant causal effects on PS (βIVW (PS)=0.605; P=0.023) and a trend for cIMT (P=0.061). Effects were robust in the statin-free subset.

    We could not detect a significant causal effect for ABI or PAD in neither the full cohort nor the statin-free subset analysis. Scatterplots of SNP effects on PCSK9 levels and vascular phenotype are shown in Figures VIII and IX in the Data Supplement.

    Discussion

    PCSK9 plays an essential role in regulating lipid metabolism and functional inhibition of PCSK9 is a novel treatment strategy for hyperlipidemia. Here, we performed to date the largest genome-wide association study of circulating PCSK9 levels in 3290 patients of the LIFE-Heart study, a cohort focusing on CAD. We identified the PCSK9 locus as a strong and genome-wide significant genetic determinant of PCSK9 plasma levels and detected considerable allelic heterogeneity. In addition, we discovered a genome-wide significant locus on chromosome 7 within FBXL18 and several other suggestive loci that warrant independent validation. Availability of coronary angiography (in LIFE-Heart) and other atherosclerotic disease assessments (in LIFE-Heart and LIFE-Adult) allowed us to analyze the causal impact of circulating PCSK9 on these phenotypes for the first time. By MR, we detected causal relationships of PCSK9 for atherosclerosis at the coronary, and carotid arteries. These results provide further evidence that functional inhibition of PCSK9 may contribute to reduce ASVD risk.

    Nongenetic Factors Influencing PCSK9

    To analyze possible confounders of the observed associations, we first investigated factors influencing PCSK9 plasma concentration in detail. Statins treatment significantly increases PCSK9 levels via the SREBP-2 (sterol regulatory element–binding protein-2) pathway.37 Because 38% of the patients in LIFE-Heart received statins, we performed all analysis in the entire cohort (adjusted for statins) and in the statin-free subset. We found that the results of our GWAS were not confounded by statin treatment, suggesting that adjustment on statin use was sufficient for genetic association analyses. However, the partial correlations of PCSK9 to plasma lipid parameters were typically larger in the statin-free subset, indicating biased relations of PCSK9 to lipids under statin treatment. These results are in line with previous reports that statin use modifies the correlation between PCSK9 and plasma LDL-C.38

    Beyond statin treatment, age, sex, and smoking status were the strongest independent factors influencing PCSK9 plasma levels in our study population. These results are in accordance with observations of other cohort studies.10,11,3942 Interestingly, we observed an interaction between sex and age, independent of the other factors. In detail, PCSK9 levels increase in women until an age of 51 to 60 years and then decrease with a similar slope but remaining on higher values than in men. This suggests a menopausal effect on PCSK9 levels and is in agreement with previous findings reporting a sex-specific regulation of PCSK9.11,43 However, limiting factors of this finding are the small sample size of young women in our cohort and missing information on menopausal status and hormone replacement therapy. The interaction between age and sex had no effect on the genetic associations reported in our study.

    Genetic Factors Influencing Circulating PCSK9

    Plasma levels of PCSK9 show substantial variation over a wide range of concentrations, but little is known on the underlying genetic determinants. Candidate gene approaches identified only 1 low-frequency variant (PCSK9-R46L) that was robustly associated with lower PCSK9 plasma levels across several studies,10,11 whereas a GWAS approach to discover common variants associated with plasma PCSK9 levels did not detect any genome-wide significant associations.10 Using a considerably larger discovery cohort, we now identified several variants at the PCSK9 gene locus that were associated with circulating PCSK9 at genome-wide significance, both, in the entire cohort and in the statin-free subset. The strongest association was observed for SNP rs11591147, encoding the above-mentioned missense mutation PCSK9-R46L. This association is plausible as the mutation decreases the phosphorylation of Ser47, which is assumed to protect PCSK9 from proteolysis.35,44 Therefore, PCSK9 secretion into plasma decreases with rs11591147-T. We also detected 3 additional independent and multivariate significant SNPs, suggesting substantial allelic heterogeneity. Moreover, the PCSK9 locus harbored several other suggestive SNPs including the rare nonsynonymous coding mutations A53V, V474I, and E670G that correlated with circulating PCSK9 levels. Together our findings suggest that variants at the PCSK9 gene locus are the predominant genetic determinants for circulating PCSK9 levels in this population. Associations of the PCSK9 variants with plasma lipids (CHOL, LDL-C, and ApoB) were much weaker and likely mediated by the effect of PCSK9 on lipid metabolism.

    Besides the PCSK9 locus, we identified a second genome-wide significant locus in a gene-rich area at chromosome 7p22. The lead SNP was located in FBXL18, which encodes the F-box and leucine-rich repeat protein 18, whose function remains largely unknown. However, the SNP has inferior quality metrics. Follow-up studies especially replication in an independent population and functional assessments are necessary to validate this finding. Likewise, validation is also required for the suggestive hits (P<1×10−6) identified in the entire cohort (Xq27.3 near SPANXN4, 22q12 near TOM1, and 13q13 near RPS12P24) or the statin-free subset (6q13 within KCNQ5, 18q12 near SYT4, 1p31 near IFI44, and 4p15). Interestingly, suggestive associations between KCNQ5 and LDL-C were previously reported in the Framingham Heart Study Offspring Cohort,45 and we have recently reported suggestive associations between KCNQ5 and carotid atherosclerotic plaque burden.46

    PCSK9 Gene Variant and PCSK9 Plasma Levels in Relation to Vascular Phenotypes

    Recent MR studies have used PCSK9 SNPs as instruments to successfully show the causality of LDL-C on CAD or type 2 diabetes mellitus risk.4750 However, variants at the PCSK9 locus are much stronger associated with circulating PCSK9 protein levels. Hence, SNPs at the PCSK9 gene locus are well suited to perform MR analysis of the causal relationship of PCSK9 protein levels and ASVD phenotypes. This is of particular clinical interest because clinical applications for functional inhibitors of PCSK9 are currently explored.

    We confirmed the previously reported positive correlation of PCSK9 and PAD.51 In line with this, a negative correlation of PCSK9 and ABI was observed. However, in MR analysis, we found no causal effect.

    Despite the lack of a raw correlation of PCSK9 with CAD and NVD50 for both the complete data set and the statin-free subjects, we detected causal effects of PCSK9 on coronary and carotid phenotypes. We suppose that the strong confounding factors age and male sex blur the raw correlations. On one hand, these factors increase the risk for CAD and NVD50; on the other hand, they reduce PCSK9.

    Our study suggests that a genetic reduction of PCSK9 levels by 50% is associated with a reduction of CAD risk by 50%. Interestingly, the effect on carotid artery atherosclerosis was just half as strong as on CAD. On treatment strategies to inhibit PCSK9, the results of the MR study may not be overstated because the detected effect is a lifelong effect. Later onset of treatment with PCSK9 inhibitors might result in a more moderate effect.

    Study Limitations

    The present study has the following limitations. The GWAS was performed in a population at risk or with present CAD, that is, not within a population-based study. On the other hand, the availability of coronary angiography and vascular phenotyping allowed to analyze CAD severity in relation to PCSK9 and its genetic variants. The present study also included a high percentage of patients on statin treatment, which is known to increase PCSK9 levels. Therefore, all analyses were also performed in the statin-free subset and were robust.

    We have to acknowledge that MR as a method to show causality relies on assumptions that inherently cannot be proven. In the present case, we demonstrated plausibility of the assumptions by our considerations shown in Mendelian Randomization Analysis in the Data Supplement.

    Last, the ELISA assay used to determine plasma PCSK9 levels does not differentiate between active and inactive PCSK9. It is known that PCSK9 circulates as mature and furin-cleaved protein and that a fraction of PCSK9 is also lipoprotein bound.11 Possible differences in associations of these subfractions with genetics and phenotypes could not be assessed.

    Conclusions

    In conclusion, by genome-wide association for circulating PCSK9 levels, we identified 4 independent genetic variants at the PCSK9 locus at genome-wide significance, explaining 4.4% of the total variance of PSCK9. In addition, we describe 1 genome-wide significant locus on chromosome 7p22 and 7 suggestive loci associated with plasma PCSK9, which require independent replication and identification of the causal genes.

    We performed the first MR study to address causal effects of PCSK9 on ASVD phenotypes. We detected significant causal effects for atherosclerosis at the coronary and carotid arteries. This increases confidence in PCSK9 as a valid target to reduce risk or severity of ASVD.

    Acknowledgments

    We thank the participants of the LIFE-Heart and the LIFE-Adult studies. We thank Sylvia Henger for data quality control, Kay Olischer and Annegret Unger for technical assistance, and Kerstin Wirkner for running the LIFE-Adult study center.

    Footnotes

    *Drs Burkhardt and Scholz contributed equally to this work.

    http://circgenetics.ahajournals.org

    The Data Supplement is available at http://circgenetics.ahajournals.org/lookup/suppl/doi:10.1161/CIRCGEN.117.001992/-/DC1.

    Markus Scholz, PhD, Institute for Medical Informatics, Statistics and Epidemiology, Haertelstrasse 16-18, 04107 Leipzig, Germany, E-mail or Ralph Burkhardt, MD, Institute of Laboratory Medicine, Liebigstraße 27, 04103 Leipzig, Germany, E-mail

    References

    • 1. Rader DJ. New therapeutic approaches to the treatment of dyslipidemia.Cell Metab. 2016; 23:405–412. doi: 10.1016/j.cmet.2016.01.005.CrossrefMedlineGoogle Scholar
    • 2. Norata GD, Tavori H, Pirillo A, Fazio S, Catapano AL. Biology of proprotein convertase subtilisin kexin 9: beyond low-density lipoprotein cholesterol lowering.Cardiovasc Res. 2016; 112:429–442. doi: 10.1093/cvr/cvw194.CrossrefMedlineGoogle Scholar
    • 3. Abifadel M, Varret M, Rabès JP, Allard D, Ouguerram K, Devillers M, et al. Mutations in PCSK9 cause autosomal dominant hypercholesterolemia.Nat Genet. 2003; 34:154–156. doi: 10.1038/ng1161.CrossrefMedlineGoogle Scholar
    • 4. Cohen JC, Boerwinkle E, Mosley TH, Hobbs HH. Sequence variations in PCSK9, low LDL, and protection against coronary heart disease.N Engl J Med. 2006; 354:1264–1272. doi: 10.1056/NEJMoa054013.CrossrefMedlineGoogle Scholar
    • 5. Kathiresan S, Melander O, Guiducci C, Surti A, Burtt NP, Rieder MJ, et al. Six new loci associated with blood low-density lipoprotein cholesterol, high-density lipoprotein cholesterol or triglycerides in humans.Nat Genet. 2008; 40:189–197. doi: 10.1038/ng.75.CrossrefMedlineGoogle Scholar
    • 6. Nikpay M, Goel A, Won HH, Hall LM, Willenborg C, Kanoni S, et al. A comprehensive 1,000 Genomes-based genome-wide association meta-analysis of coronary artery disease.Nat Genet. 2015; 47:1121–1130. doi: 10.1038/ng.3396.CrossrefMedlineGoogle Scholar
    • 7. Robinson JG, Farnier M, Krempf M, Bergeron J, Luc G, Averna M, et al; ODYSSEY LONG TERM Investigators. Efficacy and safety of alirocumab in reducing lipids and cardiovascular events.N Engl J Med. 2015; 372:1489–1499. doi: 10.1056/NEJMoa1501031.CrossrefMedlineGoogle Scholar
    • 8. Sabatine MS, Giugliano RP, Keech AC, Honarpour N, Wiviott SD, Murphy SA, et al; FOURIER Steering Committee and Investigators. Evolocumab and clinical outcomes in patients with cardiovascular disease.N Engl J Med. 2017; 376:1713–1722. doi: 10.1056/NEJMoa1615664.CrossrefMedlineGoogle Scholar
    • 9. Vlachopoulos C, Terentes-Printzios D, Georgiopoulos G, Skoumas I, Koutagiar I, Ioakeimidis N, et al. Prediction of cardiovascular events with levels of proprotein convertase subtilisin/kexin type 9: a systematic review and meta-analysis.Atherosclerosis. 2016; 252:50–60. doi: 10.1016/j.atherosclerosis.2016.07.922.CrossrefMedlineGoogle Scholar
    • 10. Chernogubova E, Strawbridge R, Mahdessian H, Mälarstig A, Krapivner S, Gigante B, et al. Common and low-frequency genetic variants in the PCSK9 locus influence circulating PCSK9 levels.Arterioscler Thromb Vasc Biol. 2012; 32:1526–1534. doi: 10.1161/ATVBAHA.111.240549.LinkGoogle Scholar
    • 11. Lakoski SG, Lagace TA, Cohen JC, Horton JD, Hobbs HH. Genetic and metabolic determinants of plasma PCSK9 levels.J Clin Endocrinol Metab. 2009; 94:2537–2543. doi: 10.1210/jc.2009-0141.CrossrefMedlineGoogle Scholar
    • 12. Beutner F, Teupser D, Gielen S, Holdt LM, Scholz M, Boudriot E, et al. Rationale and design of the Leipzig (LIFE) Heart Study: phenotyping and cardiovascular characteristics of patients with coronary artery disease.PLoS One. 2011; 6:e29070. doi: 10.1371/journal.pone.0029070.CrossrefMedlineGoogle Scholar
    • 13. Loeffler M, Engel C, Ahnert P, Alfermann D, Arelin K, Baber R, et al. The LIFE-Adult-Study: objectives and design of a population-based cohort study with 10,000 deeply phenotyped adults in Germany.BMC Public Health. 2015; 15:691. doi: 10.1186/s12889-015-1983-z.CrossrefMedlineGoogle Scholar
    • 14. Auton A, Brooks LD, Durbin RM, Garrison EP, Kang HM, Korbel JO, et al. A global reference for human genetic variation.Nature. 2015; 526:68–74. doi: 10.1038/nature15393.CrossrefMedlineGoogle Scholar
    • 15. Delaneau O, Howie B, Cox AJ, Zagury JF, Marchini J. Haplotype estimation using sequencing reads.Am J Hum Genet. 2013; 93:687–696. doi: 10.1016/j.ajhg.2013.09.002.CrossrefMedlineGoogle Scholar
    • 16. Howie BN, Donnelly P, Marchini J. A flexible and accurate genotype imputation method for the next generation of genome-wide association studies.PLoS Genet. 2009; 5:e1000529. doi: 10.1371/journal.pgen.1000529.CrossrefMedlineGoogle Scholar
    • 17. Marchini J, Howie B, Myers S, McVean G, Donnelly P. A new multipoint method for genome-wide association studies by imputation of genotypes.Nat Genet. 2007; 39:906–913. doi: 10.1038/ng2088.CrossrefMedlineGoogle Scholar
    • 18. König IR, Loley C, Erdmann J, Ziegler A. How to include chromosome X in your genome-wide association study.Genet Epidemiol. 2014; 38:97–103. doi: 10.1002/gepi.21782.CrossrefMedlineGoogle Scholar
    • 19. Aken BL, Ayling S, Barrell D, Clarke L, Curwen V, Fairley S, et al. The Ensembl gene annotation system.Database (Oxford). 2016;2016:baw093. doi: 10.1093/database/baw093.Google Scholar
    • 20. Welter D, MacArthur J, Morales J, Burdett T, Hall P, Junkins H, et al. The NHGRI GWAS Catalog, a curated resource of SNP-trait associations.Nucleic Acids Res. 2014; 42(Database issue):D1001–D1006. doi: 10.1093/nar/gkt1229.CrossrefMedlineGoogle Scholar
    • 21. Kirsten H, Al-Hasani H, Holdt L, Gross A, Beutner F, Krohn K, et al. Dissecting the genetics of the human transcriptome identifies novel trait-related trans-eQTLs and corroborates the regulatory relevance of non-protein coding loci†.Hum Mol Genet. 2015; 24:4746–4763. doi: 10.1093/hmg/ddv194.CrossrefMedlineGoogle Scholar
    • 22. GTEx Consortium. Human genomics. The genotype-tissue expression (GTEx) pilot analysis: multitissue gene regulation in humans.Science. 2015; 348:648–660. doi: 10.1126/science.1262110.CrossrefMedlineGoogle Scholar
    • 23. Yu G, Wang L-G, Yan G-R, He Q-Y. DOSE: an R/Bioconductor package for disease ontology semantic and enrichment analysis.Bioinformatics. 2015; 31:608–609. doi: 10.1093/bioinformatics/btu684.CrossrefMedlineGoogle Scholar
    • 24. Yu G, He QY. ReactomePA: an R/Bioconductor package for reactome pathway analysis and visualization.Mol Biosyst. 2016; 12:477–479. doi: 10.1039/c5mb00663e.CrossrefMedlineGoogle Scholar
    • 25. Kircher M, Witten DM, Jain P, O’Roak BJ, Cooper GM, Shendure J. A general framework for estimating the relative pathogenicity of human genetic variants.Nat Genet. 2014; 46:310–315. doi: 10.1038/ng.2892.CrossrefMedlineGoogle Scholar
    • 26. Bendl J, Stourac J, Salanda O, Pavelka A, Wieben ED, Zendulka J, et al. PredictSNP: robust and accurate consensus classifier for prediction of disease-related mutations.PLoS Comput Biol. 2014; 10:e1003440. doi: 10.1371/journal.pcbi.1003440.CrossrefMedlineGoogle Scholar
    • 27. Burgess S, Dudbridge F, Thompson SG. Combining information on multiple instrumental variables in Mendelian randomization: comparison of allele score and summarized data methods.Stat Med. 2016; 35:1880–1906. doi: 10.1002/sim.6835.CrossrefMedlineGoogle Scholar
    • 28. Bowden J, Del Greco M F, Minelli C, Lawlor D, Sheehan N, Thompson J, et al. Improving the accuracy of two-sample summary data Mendelian randomization: moving beyond the NOME assumption.2017. https://doi.org/10.1101/159442. Accessed April 24, 2018.Google Scholar
    • 29. R Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing; 2016.Google Scholar
    • 30. Surakka I, Horikoshi M, Mägi R, Sarin AP, Mahajan A, Lagou V, et al.; ENGAGE Consortium. The impact of low-frequency and rare variants on lipid levels.Nat Genet. 2015; 47:589–597. doi: 10.1038/ng.3300.CrossrefMedlineGoogle Scholar
    • 31. Chasman DI, Giulianini F, MacFadyen J, Barratt BJ, Nyberg F, Ridker PM. Genetic determinants of statin-induced low-density lipoprotein cholesterol reduction: the Justification for the Use of Statins in Prevention: an Intervention Trial Evaluating Rosuvastatin (JUPITER) trial.Circ Cardiovasc Genet. 2012; 5:257–264. doi: 10.1161/CIRCGENETICS.111.961144.LinkGoogle Scholar
    • 32. Kathiresan S, Voight BF, Purcell S, Musunuru K, Ardissino D, Mannucci PM, et al. Genome-wide association of early-onset myocardial infarction with single nucleotide polymorphisms and copy number variants.Nat Genet. 2009; 41:334–41. doi: 10.1038/ng.327.CrossrefMedlineGoogle Scholar
    • 33. Schunkert H, König IR, Kathiresan S, Reilly MP, Assimes TL, Holm H, et al; Cardiogenics; CARDIoGRAM Consortium. Large-scale association analysis identifies 13 new susceptibility loci for coronary artery disease.Nat Genet. 2011; 43:333–338. doi: 10.1038/ng.784.CrossrefMedlineGoogle Scholar
    • 34. Theusch E, Medina MW, Rotter JI, Krauss RM. Ancestry and other genetic associations with plasma PCSK9 response to simvastatin.Pharmacogenet Genomics. 2014; 24:492–500. doi: 10.1097/FPC.0000000000000081.CrossrefMedlineGoogle Scholar
    • 35. Dron JS, Hegele RA. Complexity of mechanisms among human proprotein convertase subtilisin-kexin type 9 variants.Curr Opin Lipidol. 2017; 28:161–169. doi: 10.1097/MOL.0000000000000386.CrossrefMedlineGoogle Scholar
    • 36. Kotowski IK, Pertsemlidis A, Luke A, Cooper RS, Vega GL, Cohen JC, et al. A spectrum of PCSK9 alleles contributes to plasma levels of low-density lipoprotein cholesterol.Am J Hum Genet. 2006; 78:410–422. doi: 10.1086/500615.CrossrefMedlineGoogle Scholar
    • 37. Sahebkar A, Simental-Mendía LE, Guerrero-Romero F, Golledge J, Watts GF. Effect of statin therapy on plasma proprotein convertase subtilisin kexin 9 (PCSK9) concentrations: a systematic review and meta-analysis of clinical trials.Diabetes Obes Metab. 2015; 17:1042–1055. doi: 10.1111/dom.12536.CrossrefMedlineGoogle Scholar
    • 38. Welder G, Zineh I, Pacanowski MA, Troutt JS, Cao G, Konrad RJ. High-dose atorvastatin causes a rapid sustained increase in human serum PCSK9 and disrupts its correlation with LDL cholesterol.J Lipid Res. 2010; 51:2714–2721. doi: 10.1194/jlr.M008144.CrossrefMedlineGoogle Scholar
    • 39. Baass A, Dubuc G, Tremblay M, Delvin EE, O’Loughlin J, Levy E, et al. Plasma PCSK9 is associated with age, sex, and multiple metabolic markers in a population-based sample of children and adolescents.Clin Chem. 2009; 55:1637–1645. doi: 10.1373/clinchem.2009.126987.CrossrefMedlineGoogle Scholar
    • 40. Leander K, Mälarstig A, Van’t Hooft FM, Hyde C, Hellénius ML, Troutt JS, et al. Circulating proprotein convertase subtilisin/ kexin type 9 (PCSK9) predicts future risk of cardiovascular events independently of established risk factors.Circulation. 2016; 133:1230–1239. doi: 10.1161/CIRCULATIONAHA.115.018531.LinkGoogle Scholar
    • 41. Ridker PM, Rifai N, Bradwin G, Rose L. Plasma proprotein convertase subtilisin/kexin type 9 levels and the risk of first cardiovascular events.Eur Heart J. 2016; 37:554–60. doi: 10.1093/eurheartj/ehv568.CrossrefMedlineGoogle Scholar
    • 42. Zhu YM, Anderson TJ, Sikdar K, Fung M, McQueen MJ, Lonn EM, et al. Association of proprotein convertase subtilisin/ kexin type 9 (PCSK9) with cardiovascular risk in primary prevention.Arterioscler Thromb Vasc Biol. 2015; 35:2254–2259. doi: 10.1161/ATVBAHA.115.306172.LinkGoogle Scholar
    • 43. Ruscica M, Ferri N, Fogacci F, Rosticci M, Botta M, Marchiano S, et al. Circulating levels of proprotein convertase subtilisin/kexin type 9 and arterial stiffness in a large population sample: data from the Brisighella Heart Study.J Am Heart Assoc2017. doi: 10.1161/JAHA.117.005764.LinkGoogle Scholar
    • 44. Dewpura T, Raymond A, Hamelin J, Seidah NG, Mbikay M, Chrétien M, et al. PCSK9 is phosphorylated by a Golgi casein kinase-like kinase ex vivo and circulates as a phosphoprotein in humans.FEBS J. 2008; 275:3480–3493. doi: 10.1111/j.1742-4658.2008.06495.x.CrossrefMedlineGoogle Scholar
    • 45. Kathiresan S, Manning AK, Demissie S, D’Agostino RB, Surti A, Guiducci C, et al. A genome-wide association study for blood lipid phenotypes in the Framingham Heart Study.BMC Med Genet. 2007; 8(suppl 1):S17. doi: 10.1186/1471-2350-8-S1-S17.CrossrefMedlineGoogle Scholar
    • 46. Pott J, Burkhardt R, Beutner F, Horn K, Teren A, Kirsten H, et al. Genome-wide meta-analysis identifies novel loci of plaque burden in carotid artery.Atherosclerosis. 2017; 259:32–40. doi: 10.1016/j.atherosclerosis.2017.02.018.CrossrefMedlineGoogle Scholar
    • 47. Ference BA, Yoo W, Alesh I, Mahajan N, Mirowska KK, Mewada A, et al. Effect of long-term exposure to lower low-density lipoprotein cholesterol beginning early in life on the risk of coronary heart disease: a Mendelian randomization analysis.J Am Coll Cardiol. 2012; 60:2631–2639. doi: 10.1016/j.jacc.2012.09.017.CrossrefMedlineGoogle Scholar
    • 48. Jansen H, Lieb W, Schunkert H. Mendelian randomization for the identification of causal pathways in atherosclerotic vascular disease.Cardiovasc Drugs Ther. 2016; 30:41–49. doi: 10.1007/s10557-016-6640-y.CrossrefMedlineGoogle Scholar
    • 49. Rosenson RS, Koenig W. Mendelian randomization analyses for selection of therapeutic targets for cardiovascular disease prevention: a note of circumspection.Cardiovasc Drugs Ther. 2016; 30:65–74. doi: 10.1007/s10557-016-6642-9.CrossrefMedlineGoogle Scholar
    • 50. Schmidt AF, Swerdlow DI, Holmes MV, Patel RS, Fairhurst-Hunter Z, Lyall DM, et al; LifeLines Cohort study group; UCLEB consortium. PCSK9 genetic variants and risk of type 2 diabetes: a Mendelian Randomisation study.Lancet Diabetes Endocrinol. 2017; 5:97–105. doi: 10.1016/S2213-8587(16)30396-5.CrossrefMedlineGoogle Scholar
    • 51. Chao T-H, Chen I-C, Li Y-H, Lee P-T, Tseng S-Y. Plasma levels of proprotein convertase subtilisin/ kexin type 9 are elevated in patients with peripheral artery disease and associated with metabolic disorders and dysfunction in circulating progenitor cells.J Am Heart Assoc2016. doi: 10.1161/JAHA.116.003497.LinkGoogle Scholar