Angiopoietin-like protein 4 significantly predicts future cardiovascular events in coronary patients

lable at ScienceDirect

Atherosclerosis 237 (2014) 632e638

Contents lists avai

Atherosclerosis

journal homepage: www.elsevier .com/locate/atherosclerosis

Angiopoietin-like protein 4 significantly predicts future cardiovascularevents in coronary patients

Axel Muendlein a, 1, Christoph H. Saely a, b, c, 1, Andreas Leiherer a, b, Peter Fraunberger d,Elena Kinz a, b, Philipp Rein a, c, Alexander Vonbank a, c, Daniela Zanolin a, b,Cornelia Malin a, Heinz Drexel a, b, c, e, *

a Vorarlberg Institute for Vascular Investigation and Treatment (VIVIT), Feldkirch, Austriab Private University of the Principality of Liechtenstein, Triesen, Liechtensteinc Department of Medicine and Cardiology, Academic Teaching Hospital Feldkirch, Feldkirch, Austriad Medical Central Laboratories, Feldkirch, Austriae Drexel University College of Medicine, Philadelphia, PA, USA

a r t i c l e i n f o

Article history:Received 11 July 2014Received in revised form17 October 2014Accepted 19 October 2014Available online 22 October 2014

Keywords:Coronary artery diseaseCardiovascular eventsBiomarkersAngiopoietin-like protein 4Single nucleotide polymorphismsMetabolic syndrome

* Corresponding author. Department of MedicineTeaching Hospital Feldkirch, Carinagasse 47, A-6807 F

E-mail address: [email protected] (H. Drexel).1 Both authors contributed equally.

http://dx.doi.org/10.1016/j.atherosclerosis.2014.10.0280021-9150/© 2014 Published by Elsevier Ireland Ltd.

a b s t r a c t

Background: Angiopoietin-like protein 4 (ANGPTL4) has been associated with cardiometabolic disordersincluding dyslipidemia and atherosclerosis in animal studies; in humans, however, its impact onmetabolic traits and cardiovascular risk remains unclear. Methods: We examined the association ofplasma ANGPTL4 levels with the metabolic syndrome (harmonized consensus definition), with angio-graphically determined coronary artery disease (CAD), and with the risk of future cardiovascular eventsin a cohort of 490 patients undergoing coronary angiography for the evaluation of stable CAD. Inaddition, we investigated the influence of the tagging single nucleotide polymorphisms (SNPs)rs4076317, rs2278236, rs1044250, and rs11672433 as well as variant rs116843064 (E40K) of the ANGPTL4gene on cardiovascular risk in a larger sample of 983 angiographied coronary patients including theabove mentioned 490 subjects. Results: Plasma ANGPTL4 was significantly higher in patients with themetabolic syndrome than in subjects without the metabolic syndrome (26.0 ± 19.4 ng/ml vs.22.2 ± 19.7 ng/ml; p ¼ 0.008). No significant association was found between ANGPTL4 and angio-graphically characterized coronary atherosclerosis. Prospectively, however, plasma ANGPTL4 significantlypredicted future cardiovascular events both univariately (HR1.45 [1.16e1.82], p ¼ 0.001) and afteradjustment for standard cardiovascular risk factors (1.26 [1.01e1.58]; p ¼ 0.045). Concordantly,rs4076317, rs2278236, and rs1044250 significantly affected the risk of future cardiovascular events(adjusted HRs 0.70 [0.54e0.90]; p ¼ 0.005, 0.76 [0.61e0.94]; p ¼ 0.012, and 1.30 [1.03e1.62]; p ¼ 0.025,respectively). Conclusions: We conclude that plasma ANGPTL4 levels as well as ANGPTL4 variantssignificantly predict cardiovascular events independently of conventional cardiovascular risk factors.

© 2014 Published by Elsevier Ireland Ltd.

1. Introduction

Recently, orphan ligands with structural similarity to angio-poietins were identified in the systemic circulation, which havebeen designated angiopoietin-like proteins (ANGPTLs). Members ofthe ANGPTL-family, in particular ANGPTL4, have been shown toregulate angiogenesis and to be involved in lipid, glucose, and en-ergy metabolism [1,2].

and Cardiology, Academiceldkirch, Austria.

In mice, Angptl4 is predominantly expressed in adipose tissueand, therefore, has been classified as an adipokine [3,4]. In humans,ANGPTL4 is ubiquitously expressed and occurs at higher levels inthe liver, placenta, small intestine and heart as well as in adiposetissue, albeit with drastic interindividual variations [3,4].

The human ANGPTL4 gene is located on chromosome 19p13.3,has seven exons, and encodes a 406-amino-acid glycoproteinwith amolecular mass of 50 kDa [4]. ANGPTL4 proteins form oligomersand occur in glycosylated and cleaved isoforms [5]. Cleavage isperformed by several types of proprotein convertases in the cellularmatrix or in blood plasma, thereby releasing its carboxylfibrinogen-like domain (C-terminal fragment) and an N-terminalcoiled-coil domain [5e7]. Cleavage appears to be tissue dependent:the liver secretes the cleaved N-terminal fragment, whereas

A. Muendlein et al. / Atherosclerosis 237 (2014) 632e638 633

adipose tissue secrets the full length form [4]. Both, full length andtruncated form are found in plasma, with the N-terminal and fulllength protein undergoing oligomerization [6,8].

Oligomerized Angptl4 is an inhibitor of lipoprotein lipase (LPL)in mice, thus suppressing the clearance of circulating triglycerides(TG) [6,8,9]. In humans, ANGPTL4 concentration has been positivelycorrelated with TG in patients with the metabolic syndrome [10],but not all studies confirmed such a correlation [11,12]. A significantnegative correlation between HDL-cholesterol and ANGPTL4 hasbeen described in men [10,12]. Therefore, ANGPTL4 could beassociated with disorders of lipid metabolism. This is also sup-ported by human genetic data including genome wide associationstudies [13,14]. ANGPTL4 thus could also be involved in the devel-opment of cardiovascular disease.

Most studies on the association between ANGPTL4 and cardio-vascular risk have investigated single nucleotide polymorphisms(SNPs) within the ANGPTL4 locus. Resequencing of the ANGPTL4gene has identified a relatively rare, non-synonymous variant(rs116843064, c.118G > A, E40K), present in approximately 3% ofpeople of European ancestry [14]. The variant has been associatedwith significantly lower plasma TG and higher HDL-cholesterol[14,15]. However, studies investigating the impact of this varianton cardiovascular risk yielded conflicting results, showing either anincreased or a decreased risk of coronary heart disease in patientswith the E40K variant [15,16].

So far, the influence of plasma ANGPTL4 levels on future coro-nary heart disease was assessed only in healthy middle-aged menfrom the general population-based Northwick Park Heart Study II(NPHSII) [12]; no significant prospective impact of ANGPTL4 levelson cardiovascular risk was found in this primary prevention study.However, the influence of ANGPTL4 levels or ANGPTL4 SNPs onfuture cardiovascular events may differ in patients at higher car-diovascular risk.

The primary aim of our study therefore was to determine theimpact of ANGPTL4 plasma levels as well as of ANGPTL4 taggingSNPs and rs116843064 (E40K) on future cardiovascular events in ahigh-risk cohort of patients undergoing coronary angiography forthe evaluation of suspected or established stable coronary arterydisease (CAD).

2. Patients and methods

2.1. Study subjects

In brief, we enrolled 983 consecutive Caucasian patients, whowere referred to elective coronary angiography for the evaluation ofestablished or suspected stable CAD at the academic teachinghospital Feldkirch, a tertiary care centre inWestern Austria (state ofVorarlberg). Out of these 983 patients, plasma samples of 490 pa-tients (245 with and 245 without significant coronary stenoseswith lumen narrowing �50%, matched for sex) were selected forthe determination of ANGPTL4 levels. In addition, DNA samplesobtained of all patients of the cohort were available for genotyping.According to National Cholesterol Education Programme ATP-IIIcriteria (NCEP-ATPIII [17]), the metabolic syndrome (MetS) wasdiagnosed when at least three of the following criteria applied:waist circumference >102 cm in men and >88 cm in women, tri-glycerides �150 mg/dl (1.7 mmol/l), high density lipoprotein (HDL)cholesterol <40 mg/dl (1.0 mmol/l) in men and <50 mg/dl(1.3 mmol/l) in women, blood pressure �130/�85 mm Hg, orfasting glucose �100 mg/dl (5.6 mmol/l). Coronary angiographywas performed using the Judkin's technique and the severity ofstenosis was assessed by visual inspection by a team of two in-vestigators, who were blinded to serologic assays as describedpreviously [18]. In short, coronary artery stenoses with any

narrowing lumen were defined as any coronary artery disease(CAD) and stenoses with lumen narrowing �50% were defined assignificant CAD. Detailed information on the recruitment protocoland the determination of baseline subject characteristics is given inthe Supplemental appendix. The present study has been approvedby the Ethics Committee of the University of Innsbruck and writteninformed consent was given by all participants.

2.2. Prospective study

During a mean (±standard deviation) follow-up period of3.5 ± 1.1 years, fatal as well as non-fatal cardiovascular events wererecorded. The primary study endpoint was a composite of coronarydeath (fatal myocardial infarction, sudden cardiac death, mortalityfrom congestive heart failure due to CAD), fatal ischemic stroke,non-fatal myocardial infarction, non-fatal ischemic stroke and needfor aorto-coronary bypass surgery, percutaneous transluminalcoronary angioplasty, or revascularization in the carotid or pe-ripheral arterial beds. Revascularization procedures were consid-ered as end points unless they were planned as a consequence ofthe baseline examination and therefore were not “future” events.Time and causes of death were regularly obtained from a nationalsurvey (Statistik Austria, Vienna, Austria) or from hospital records.

2.3. Plasma ANGPTL4 levels

Plasma ANGPTL4 levels were determined using a commercialANGPTL4 enzyme-linked immunosorbent assay (ELISA) kit (Avis-cera Bioscience, CA, USA; catalog no. SK00309-01). According tomanufacturer's information the assay shows a sensitivity of 0.3 ng/ml, an intra-assay coefficient of variation of 4e6%, and an inter-assay coefficient of variation of 8e10%. Precision and repeatabilityof this used assay are comparable with other ANGPTL4 ELISAsavailable and used in the literature [12,19].

2.4. SNP genotyping

Four SNPs were selected according to their previously describedroles as tagging SNPs covering almost all common genetic variantswithin the ANGPTL4 gene: rs4076317 (c.207C > G), rs2278236(c.547 þ 378G > A), rs1044250 (c.797C > T; T266M), andrs11672433 (c.1167G > A; P389¼) [20,21]. Indeed, according toHapMap SNP database [22], release 27, rs2278236, rs1044250, andrs11672433 capture 100% of variants with a minor allelefrequency � 0.05 within the ANGPTL4 gene locus (as described indetail in the Supplemental part). Further, rs116843064 (c.118G > A,E40K) was included in the present study due to its previouslydescribed effect on ANGPTL4 function [8], plasma lipid levels[14,15], and cardiovascular risk [15,16].

Genotyping of said variants was carried out by the 5’ nucleaseassay using TaqMan® MGB probes on a LightCycler® 480 Real-TimePCR System (F. Hoffmann-La Roche Ltd, Basel, Switzerland). Taq-Man® MGB probes together with corresponding PCR primers wereprovided by the Assay-on-demand™ service (Life Technologies, SanFrancisco, CA) for rs4076317, rs2278236, rs1044250, andrs11672433 analysis and by the Assay-by-design service™ (LifeTechnologies, San Francisco, CA) for rs116843064. Used primer andprobe sequences for rs116843064 analysis were reported previ-ously [15]. The 50 nuclease assay was performed in a 6 ml volume,comprising 10e30 ng genomic DNA, 1� TaqMan® Universal PCRMaster Mix (Life Technologies), and 1� primer/probe mix underthe following amplification conditions: 10 min at 95 �C and 40cycles at 92 �C for 15 s and 60 �C for 1 min. Reference controls aswell as non template controls were included in each run. Genotypes

A. Muendlein et al. / Atherosclerosis 237 (2014) 632e638634

were automatically determined by LightCycler® software 1.5 fol-lowed by a visual control of accurate genotype classification.

2.5. Statistics

Hazard ratios (HRs) and 95% confidence intervals of the HRswere derived from univariate and multivariate Cox proportionalhazardsmodels; continuous variables were z-transformed for theseanalyses. Survival curves were generated using the KaplaneMeiermethod and compared using the Log-Rank-Mantel-Cox-tests. Cor-relation analyses were performed calculating non-parametricSpearman rank correlation coefficients. Associations between ge-netic variants and categorical baseline characteristics were testedfor statistical significance with the Chi-squared test. In addition,analysis of variance models (ANOVA) as well as analysis of covari-ance models (ANCOVA) were built using a general linear modelapproach. The KolmogoroveSmirnov test was used as a test fornormality. Non-normally distributed variables were log-transformed before they were entered into parametric models.Results are given as mean (±standard deviation) if not denotedotherwise. P-values <0.05 were considered significant. These ana-lyses were performed with SPSS 20.0 for Windows (SPSS, Inc.,Chicago, IL). Observed numbers of each genotype were comparedwith those expected for samples in HardyeWeinberg equilibriumusing the Chi-Squared test with one degree of freedom. All deter-mined genotype frequencies were in HardyeWeinberg equilibrium.Haplotype frequencies were evaluated by the PHASE 2.1 program[23,24]. PHASE 2.1 outputs several pairs of haplotype for each in-dividual, if more than one haplotype pair is estimated. Eachhaplotype pair is assigned a posterior probability of certainty. Thepair with the highest probability for an individual was used in Coxproportional hazard models to assess hazard ratios of first cardio-vascular events.

Table 1Baseline characteristics.

Subjects analyzedfor ANGPTL4 levels(n ¼ 490)

Subjects analyzedby genotyping(n ¼ 983)

Age (years)a 65.3 ± 10.9 65.4 ± 10.6Male gender, n (%) 247 (50.4) 642 (65.3)Body mass index (kg/m2)a 27.6 ± 4.5 27.6 ± 4.4Waist circumference (cm) 98.6 ± 12.0 98.9 ± 11.9History of smoking, n (%) 265 (54.1) 584 (59.4)Hypertension, n (%) 378 (77.1) 792 (80.6)Any CAD, n (%) 357 (72.9) 799 (81.3)Significant CAD, n (%) 245 (50.0) 549 (55.8)Metabolic syndrome, n (%) 147 (30.0) 311 (31.6)Type 2 diabetes, n (%) 112 (23.0) 235 (24.1)Fasting glucose (mg/dl)a 103.9 ± 30.8 106.4 ± 34.7Fasting insulin (mU/l)a 11.5 ± 10.8 11.8 ± 10.4HOMA-IRa 3.3 ± 5.0 3.5 ± 4.5CRP(mg/dl)a 0.39 ± 0.60 0.42 ± 0.72Total cholesterol (mg/dl) 200 ± 47 196 ± 47LDL-cholesterol (mg/dl) 130 ± 42 128 ± 41HDL-cholesterol (mg/dl)a 59 ± 18 57 ± 17Triglycerides (mg/dl)a 138 ± 90 138 ± 88Aspirin, n (%) 324 (66.1) 687 (69.9)Statins, n (%) 216 (44.1) 452 (46.0)Fibrates, n (%) 7 (1.4) 15 (1.5)Beta blockers, n (%) 255 (62.0) 546 (55.5)ACE inhibitors, n (%) 145 (29.6) 309 (31.4)AT-II receptor blockers, n (%) 46 (9.4) 103 (10.5)

a Variables not normally distributed; normality was assessed with the Kolmo-goroveSmirnov test. CAD was defined as the presence of any lumen narrowing atangiography; significant CAD was diagnosed in the presence of coronary stenoses�50%. HOMA-IR, homeostasis model for the assessment insulin resistance; CRP, C-reactive protein; ACE, angiotensin converting enzyme, AT-II, angiotensin-II.

Statistical power analysis was performed using the Power andSample Size Program [25] under an additive model of inheritanceon the basis of minor allele frequencies of 25% on average ofrs4076317, rs2278236, rs1044250, and rs11672433 given in theNCBI SNP database (http://www.ncbi.nlm.nih.gov/snp/) and at analpha fault of 0.05 with a power of 80%. Due to its low minor allelefrequency (http://www.ncbi.nlm.nih.gov/snp/), rs116843064 wasexcluded from power analysis. Assuming a median survival time of3.3 years of controls (carriers of the common alleles) and a medianfollow up period of 3.4 years, a priori power analysis indicated that900 patients would be required to demonstrate hazard ratios of0.81 or 1.25 for the respective rare alleles.

3. Results

3.1. Characteristics of the study population

Clinical and biochemical characteristics of patients subjected toANGPTL4 levels determination and of patients subjected to geno-typing are shown in Table 1. Overall, the characteristics of our pa-tients were representative for patients undergoing coronaryangiography for the evaluation of CAD, with a high prevalence ofT2DM (23.0% and 24.1% in patients subjected to ANGPTL4 levels andin patients subjected to genotyping, respectively), MetS (30.0% and31.6%, respectively), hypertension (77.1% and 80.6%, respectively),and history of smoking (54.1% and 59.4%, respectively).

3.2. Association of plasma angiopoietin-like protein 4 with clinicaland biochemical subject characteristics

Correlations between plasma ANGPTL4 and continuous baselinecharacteristics are summarized in Table 2. Plasma ANGPTL4correlated positively with age, waist circumference, fasting glucose,insulin, homeostasis model for the assessment of insulin resistance(HOMA) index, and C-reactive protein (CRP), and negatively withHDL-cholesterol; the correlation between plasma ANGPTL4 and TGfell short of statistical significance (p ¼ 0.051).

Plasma ANGPTL4 levels were significantly increased in patientswith the MetS compared to subjects without the MetS(26.0 ± 19.4 ng/ml vs. 22.2 ± 19.7 ng/ml; p¼ 0.008). The associationof ANGPTL4 levels with the MetS proved significant after adjust-ment for age, sex, and BMI (F ¼ 4.26; p ¼ 0.039). However, afterfurther adjustment for variables significantly correlating withplasma ANGPTL4 (as given in Table 2 and including TG) ANCOVAshowed that ANGPTL4was not an independentmarker for theMetS(F¼ 0.099; p¼ 0.753). No significant impact of T2DM, sex, smoking,

Table 2Correlation between plasma ANGPTL4 levels and continuous baseline subjectcharacteristics.

R p-value

Age (years) 0.190 <0.001Body mass index (kg/m2) 0.076 0.094Waist circumference (cm) 0.101 0.030Fasting glucose (mg/dl) 0.125 0.006Fasting insulin 0.100 0.033HOMA-IR 0.113 0.016CRP(mg/dl) 0.103 0.023Total cholesterol (mg/dl) �0.027 0.548LDL-cholesterol (mg/dl) �0.042 0.352HDL-cholesterol (mg/dl) �0.119 0.008Triglycerides (mg/dl) 0.088 0.051

R ¼ Spearman's Rank correlation coefficient.HOMA-IR, homeostasis model for the assessment insulin resistance; CRP, C-reactiveprotein.

A. Muendlein et al. / Atherosclerosis 237 (2014) 632e638 635

or use of medications on plasma ANGPTL4 was observed (alladjusted p-values >0.05).

3.3. Association of plasma angiopoietin-like protein 4 withangiographically determined coronary atherosclerosis

Circulating ANGPTL4 did not differ significantly between pa-tients with significant CAD (23.3± 21.2 ng/ml) and subjects withoutsignificant CAD at angiography (23.3 ± 18.0 ng/ml; p¼ 0.344). Also,ANGPTL4 levels in patients with any visible lumen narrowing atangiography did not differ from those of patients with normalcoronary arteries (23.4 ± 22.0 ng/ml vs. 23.3 ± 18.8 ng/ml;p ¼ 0.270).

3.4. Power of plasma angiopoietin-like protein 4 to predict futurecardiovascular events

During follow-up, 79 (16.1%) first cardiovascular eventsoccurred among the 490 patients in whom ANGPTL4 levels weredetermined. In these patients, T2DM, HDL-cholesterol, and fastingglucose were significantly associated with event-free survival(HR ¼ 2.01 [1.27e3.18; p ¼ 0.003, HR ¼ 0.75 [0.59e0.95],p ¼ 0.017, and HR ¼ 1.30 [1.07e1.58]; p ¼ 0.008, respectively). Theimpact of the MetS on the incidence of future vascular events didnot reach statistical significance (HR ¼ 1.41 [0.89e2.23];p ¼ 0.142).

Fig. 1 shows KaplaneMeier curves for event-free survival ac-cording to tertiles of ANGPTL4. Event rates were highest (23.2%) inpatients in the third tertile, intermediate (16.9%) in the secondtertile and lowest (11.0%) in the first tertile, ptrend ¼ 0.001. Univar-iate hazard ratios were 1.57 [1.17e2.09] and 1.61 [0.87e2.96] forANGPTL4 tertiles 3 and 2 when compared to tertile one(ptrend ¼ 0.002). Also as a continuous variable ANGPTL4 signifi-cantly predicted the risk of cardiovascular events univariately(HR ¼ 1.45 [1.16e1.82]; p ¼ 0.001) and after adjustment for age,gender, T2DM, BMI, hypertension, HDL-cholesterol, TG, CRP, waistcircumference, fasting glucose, insulin, and HOMA index (HR¼ 1.26[1.01e1.58]; p ¼ 0.045). Significant association between impact of

Fig. 1. Kaplan Meier survival curves for event-free survival according to ANGPTL4tertiles. 1st tertile ranges from 2.3 ng/ml to 13.1 ng/ml (low tertile); 2nd tertile rangesfrom 13.8 ng/ml to 25.9 ng/ml (intermediate tertile); 3rd tertile ranges from 26.0 ng/mlto 157.6 ng/ml (high tertile) plasma ANGPTL4. Ptrend was calculated by Log-Rank test.Events were defined as vascular deaths, non-fatal myocardial infarctions, non-fatalstrokes, and the necessity of surgical intervention (coronary artery bypass graftings,percutaneous coronary interventions, or non-coronary revascularization).

ANGPTL4 on future events still proved significant after additionaladjustment for the MetS (HR ¼ 1.26 [1.00e1.57]; p ¼ 0.047).

3.5. Association of ANGPTL4 single nucleotide polymorphisms withbaseline parameters

Clinical or biochemical parameters with respect to includedANGPTL4 SNPs are given in Supplemental tables 2aee. Carriers ofthe rare allele of rs116843064 showed significantly higher HDL-cholesterol levels (p < 0.001). No further significant associationsbetween included ANGPTL4 SNPs and baseline parameters(including lipid parameters, T2DM, CRP, and angiographicallydetermined coronary atherosclerosis) were found. Also, as shownin Table 3, the association between genetic variants and ANGPTL4plasma levels were not statistical significant.

3.6. Association of ANGPTL4 tagging single nucleotidepolymorphisms with the risk of future cardiovascular events

A total of 170 first vascular events occurred in the 983 patientssubjected to genotyping was performed. The impact of genotypedvariants on the risk of future cardiovascular events is summarizedin Table 4 and respective KaplaneMeier survival curves are dis-played in Supplemental fig. 1aee. Rare alleles of rs4076317 andrs2278236 (G-allele and A-allele, respectively) showed a protectiveeffect on future cardiovascular risk, whereas patients with the rareallele of rs1044250 (T-allele) were at an increased risk of vascularevents. Also in a Cox regression model adjusted for age, gender,T2DM, BMI, hypertension, HDL-cholesterol, TG, CRP, waistcircumference, fasting glucose, insulin, and HOMA index the threegenetic variants showed a significant impact on future cardiovas-cular events. Association between rs116843064 and future cardio-vascular risk was not statistical significant.

Estimated frequencies of haplotypes derived from these fourSNPs and their associations with future cardiovascular risk areshown in Table 5. Haplotype analysis resulted in five commonhaplotypes. Each individual haplotype was compared to the pooledremaining haplotypes. Only haplotypes either carrying frequentalleles of rs4076317, rs116843064, and rs2278236 and the minorallele of rs1044250 (CGGT-haplotype) or carrying minor alleles ofrs4076317 and rs2278236 and the frequent alleles of rs116843064and rs1044250 (GGAC-haplotype) showed a significant impact onvascular events. Adjustment for age, gender, T2DM, BMI, hyper-tension, HDL-cholesterol, TG, CRP waist circumference, fastingglucose, insulin, and HOMA index did not influence associationbetween haplotypes and cardiovascular risk. Notably, only thehaplotype carrying the rare allele of variant rs1044250 and thecommon allele of rs116843064 (CGGT-haplotype) was associated

Table 3ANGPTL4 plasma levels with respect to included ANGPTL4 SNPs.

SNP Genotype p-value

AA AB BB

rs4076317(c.-207C > G)

23.6 ± 17.9(n ¼ 211)

22.0 ± 18.4(n ¼ 206)

23.1 ± 25.7(n ¼ 46)

0.155

rs116843064(c.118G > A, E40K)

22.8 ± 19.0(n ¼ 450)

23.7 ± 17.4(n ¼ 20)

e

(n ¼ 0)0.732

rs2278236(c.547 þ 378G > A)

23.8 ± 17.6(n ¼ 135)

23.0 ± 19.5(n ¼ 221)

21.3 ± 19.4(n ¼ 114)

0.163

rs1044250(c.797C > T; T266M)

21.6 ± 17.9(n ¼ 219)

23.1 ± 18.7(n ¼ 205)

27.7 ± 23.7(n ¼ 47)

0.102

rs11672433(c.1167G > A; P389¼)

23.3 ± 19.9(n ¼ 355)

21.0 ± 15.1(n ¼ 105)

26.5 ± 23.4(n ¼ 8)

0.553

AA, homozygous for the common allele; AB, heterozygous; BB, homozygous for therare allele.

Table 4ANGPTL4 SNPs as predictors of future cardiovascular events.

ANGPTL4 HR [95%CI]* p-value* HR [95%CI]** p-value**

rs4076317 (c.-207C > G) 0.70 [0.54e0.90] 0.005 0.70 [0.54e0.90] 0.005rs116843064 (c.118G > A, E40K) 1.14 [0.56e2.32] 0.713 1.37 [0.67e2.80] 0.395rs2278236 (c.547 þ 378G > A) 0.76 [0.61e0.94] 0.011 0.76 [0.61e0.94] 0.012rs1044250 (c.797C > T; T266M) 1.29 [1.03e1.61] 0.025 1.30 [1.03e1.62] 0.025rs11672433 (c.1167G > A; P389¼) 0.94 [0.69e1.28] 0.695 0.89 [0.65e1.21] 0.449

*Unadjusted; **Adjusted for age, gender, type 2 diabetes mellitus, hypertension, body mass index, HDL-cholesterol, triglycerides, C-reactive protein, waist circumference,fasting glucose, insulin, and homeostasis model for the assessment of insulin resistance index. HR, hazard ratio; CI, confidence interval.

A. Muendlein et al. / Atherosclerosis 237 (2014) 632e638636

with an increased cardiovascular risk pointing to a causal effect ofvariant rs1044250 independent from rs116843064.

Association between these haplotypes and clinical or biochem-ical subject characteristics are given in Supplemental tables 3aee. Ahighly significant association between elevated HDL-cholesterollevels and by the haplotype carrying the rare allele of variantrs116843064 (CAGT-haplotype) was observed (p ¼ 0.001). Further,haplotype CGAC was significantly associated with decreased TGlevels (p ¼ 0.008). However, neither haplotype CAGT nor haplotypeCGAC significantly influenced future cardiovascular risk.

SNP rs11672433 was not associated with cardiovascular diseasein either association model. Further, rs11672433 was only in weaklinkage disequilibrium with other genetic variants included in thestudy (as shown in Supplemental table 1). Therefore, rs11672433was not included in haplotype analysis.

4. Discussion

In the present work, we report a positive association of circu-lating ANGPTL4 levels as well as of ANGPTL4 SNPs with futurecardiovascular risk in angiographically characterized coronarypatients.

The association between Angptl4 and cardiometabolic traits hasbeen extensively studied in animal models [3,9,26e28] but humandata are scarce. Indeed, besides our study, only a single investiga-tion, the NPHSII study, has evaluated the impact of circulatingANGPTL4 levels on cardiovascular risk so far [12]. In that study,Smart-Halajko et al. reported no significant association betweenANGPTL4 levels and coronary heart disease. However, only healthymiddle-agedmenwere included in the NPHSII study and, therefore,characteristics of this study population differ from those in ourstudy, namely a high-risk population of angiographied coronarypatients. These differences may also account for the difference incirculating ANGPTL4 levels between the two study populations: Inthe NPHSII study median ANGPTL4 levels were 7.7 ng/ml and it wasmentioned that 92% of the study subjects displayed ANGPTL4 levelsbelow 20 ng/ml [12]. Median ANGPTL4 levels in our study were20.0 ng/ml and, therefore, 50% of our patients showed ANGPTL4plasma levels equal or greater than 20 ng/ml. Thus, ANGPTL4plasma levels in our study were considerably higher compared tothe NPHSII study. Therefore, it can be hypothesized that mostsubjects included in the NPHSII study did not reach the thresholdvalue of ANGPTL4 to predict future cardiovascular risk. Adequatelypowered prospective studies including well-characterized subjectsare needed to clarify the role of circulating ANGPTL4 on cardio-vascular risk in different populations.

Our study showed that ANGPTL4 levels were associated withcharacteristics of the MetS as well as with the MetS itself. The as-sociation of plasma ANGPTL4 with traits of the MetS maycontribute to the impact of elevated ANGPTL4 levels on cardio-vascular risk, in particular the associationwith lowHDL-cholesterollevels [29e32]. However, it does not fully explain it: ANGPTL4levels remained associated with vascular events after adjustmentfor these metabolic traits as well as for the MetS itself or other

cardiovascular risk factors. That said, from these data it could not beexcluded that the observed association between ANGPTL4 levelsand cardiovascular risk was mediated by metabolic traits unrec-ognized in our study. To further clarify this issue, in addition to thedetermination of circulating ANGPTL4 levels, genotyping of taggingSNPs of the ANGPTL4 gene was performed.

In our study, SNPs rs4076317, rs2278236, and rs1044250 weresignificantly associated with future cardiovascular events. Howev-er, none of these SNPs revealed significant correlation withbiochemical or clinical data. Consequently, the association betweenthese genetic variants and cardiovascular disease remained signif-icant after adjustment for cardiovascular risk factors pointing to anindependent impact of these SNPs on future cardiovascular events.Haplotype analysis provided no additional cardiovascular risk otherthan that seen with the haplotype defined by the minor allele ofrs1044250. This may point to a functional role of rs1044250 or avariant highly correlated with it.

Paradoxically, rs1044250 in previous investigations had beenlinked to decreased rather than increased ANGPTL4 activity [15]. Itshould be noted, that ANGPTL4 may show diverse physiologicalfunctions in different tissues and different ANGPTL4 isoforms canexert differential effects in the human body [5]. In the bloodstream,oligomerized N-terminal and full-length ANGPTL4 protein inhibitsthe activity of blood lipoprotein lipase thus influencing lipidmetabolism [3e9]. However, in endothelial cells the C-terminus ofANGPTL4 represents a master regulator of vascular permeabilityand angiogenesis, as reviewed by Guo et al. [33]. The C-terminus ofANGPTL4 alone has been shown to be sufficient to suppressangiogenesis through inhibiting the Raf/MEK/ERK signalingpathway in endothelial cells [34]. Angiogenesis is the predominantform of neovascularization in atherosclerosis. Neovascularization inearly atherosclerosis is associated with inflammation and lipiddeposition, leading to plaque destabilization and rupture [35]. Ithas been hypothesized that ANGPTL4 inhibition (e.g. caused by aloss-of-function mutation) could promote revascularization andthus plaque instability [15]. Further, it has been shown that vascularpermeability and myocardial infarction severity were increased inangptl4-deficient mice [36].

Variant rs1044250 is located in the C-terminal region ofANGPTL4 and is conserved across human, rat, and mouse species,supporting a functional role [6,15]. Thus, it may be assumed thatrs1044250 may influence cardiovascular risk by decreasingANGPTL4 activity in endothelial cells increasing angiogenesis(leading to plaque destabilization and rupture) and vascularpermeability. This assumption is supported by the fact that circu-lating ANGPTL4 protein, but not rs1044250 influenced metabolictraits in our study.

Furthermore, haplotype analysis showed that variant rs1044250influenced cardiovascular risk independent from rs116843064,which has been previously associated with coronary heart disease[15,16]. Variant rs116843064 is located in the N-terminal region ofthe ANGPTL4 gene and decreases oligomer formation of theANGPTL4 protein [7] resulting in decreased lipoprotein lipase in-hibition activity effecting plasma lipid profile. Indeed, rs116843064

Table 5Association of ANGPTL4 haplotypes with future vascular events.

Haplotypes Frequency (%) HR [95%CI]* p-value* HR [95%CI]** p-value**

rs4076317C > G rs116843064 G > A rs2278236 G > A rs1044250C > T

G G A C 31.0 0.70 [0.54e0.90] 0.005 0.70 [0.55e0.91] 0.007C G G T 29.2 1.31 [1.04e1.64] 0.021 1.28 [1.02e1.62] 0.034C G G C 20.0 1.10 [0.85e1.42] 0.470 1.08 [0.83e1.40] 0.575C G A C 16.6 0.97 [0.72e1.30] 0.830 1.00 [0.74e1.35] 0.998C A G T 1.9 1.21 [0.60e2.44] 0.598 1.43 [0.71e2.90] 0.321

Haplotype frequencies were estimated by PHASE 2.1; each individual haplotype was compared to pooled remaining haplotypes in Cox regression analysis: * unadjusted;**adjusted for age, gender, type 2 diabetes mellitus, hypertension, body mass index, HDL-cholesterol, triglycerides, C-reactive protein, waist circumference, fasting glucose,insulin, and HOMA index. HR, hazard ratio; CI, confidence interval.

A. Muendlein et al. / Atherosclerosis 237 (2014) 632e638 637

and plasma ANGPTL4 showed similar effects on plasma lipid levelsin our study. However, a significant association betweenrs116843064 and future cardiovascular risk could not be realized,probably to the rarity of the variant implicating limited statisticalpower. Further studies are needed to elucidate the impact ofrs116843064 on cardiovascular risk in high-risk patients.

Vascular events such as myocardial infarction or stroke areprecipitated by plaque rupture and thrombus formation [37,38] andthus may not optimally reflect the atherogenicity of a risk marker.This phenomenon could also be observed in our study. Despite thesignificant impact of plasma ANGPTL4 levels or ANGPTL4 SNPs onfuture vascular events, we did not find any significant associationsbetween the circulating protein or its genetic variants and angio-graphically determined coronary atherosclerosis. It should benoted, that the impact of Angptl4 on atherosclerosis has beeninvestigated only in mice so far. In an initial study Angptl4 deficientmice on an ApoE deficient background developed less atheroscle-rotic lesions on a chow diet compared with control mice [26]. Onthe other hand, a more recent study in E3L mice found that re-combinant Angptl4 significantly decreased uptake of oxidized low-density lipoprotein by macrophages suppressing foam cell forma-tion and that overexpression of Angptl4 reduced atherosclerosisdevelopment [39]. However, our data do not support an impact ofANGPTL4 in coronary atherogenesis in humans.

Our study has strengths and limitations. By design, our studypopulation of angiographied coronary patients is selected; our re-sults therefore are not necessarily applicable to the general popu-lation. However, the high-risk patient population we chose toinvestigate is of particular clinical interest. Further, we cannotexclude that some associations not found to be significant in ourstudy may have reached statistical significance with a larger pop-ulation or a longer follow-up period. Notably, however, studies onANGPTL4 in humans are scarce and our study represents the firstobservation linking ANGPTL4 to future cardiovascular event risk incoronary patients so far.

In conclusion, our study demonstrates a predictive impact ofplasma ANGPTL4 as well as of genetic variants of the ANGPTL4 geneon cardiovascular risk in coronary patients.

Because the only published human study on circulatingANGPTL4 pointed against a role for the molecule in the develop-ment of cardiovascular disease, it is important that our resultsdemonstrate such a role. Therefore, the molecule should not beneglected and deserves further studies to clarify the molecularbackground behind these findings.

Acknowledgments

The present project was supported by funds of the Oes-terreichische Nationalbank (Oesterreichische Nationalbank, Anni-versary Fund, project number: 14159). We thank Dr. Eva-MariaBrandtner (Vorarlberg Institute for Vascular Investigation andTreatment, Feldkirch, Austria) for linguistic revision of our

manuscript. We are grateful to Mag. Gabriela Dür and the Vor-arlberger Landesregierung (Bregenz, Austria), to Franz Rauch andthe Vorarlberger Industriellenvereinigung (Bregenz, Austria), to Dr.Peter Woess and the Vorarlberger Aerztekammer (Dornbirn,Austria), as well as to Dr. Peter Fraunberger and the Institute forClinical Chemistry at the Academic Teaching Hospital Feldkirch(Feldkirch, Austria) for continuously supporting our researchinstitute.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.atherosclerosis.2014.10.028.

References

[1] T. Hato, M. Tabata, Y. Oike, The role of angiopoietin-like proteins in angio-genesis and metabolism, Trends Cardiovasc. Med. 18 (2008) 6e14.

[2] Y. Oike, M. Akao, Y. Kubota, T. Suda, Angiopoietin-like proteins: potential newtargets for metabolic syndrome therapy, Trends Mol. Med. 11 (2005)473e479.

[3] S. Mandard, F. Zandbergen, E. van Straten, et al., The fasting-induced adiposefactor/angiopoietin-like protein 4 is physically associated with lipoproteinsand governs plasma lipid levels and adiposity, J. Biol. Chem. 281 (2006)934e944.

[4] P. Zhu, Y.Y. Goh, H.F. Chin, S. Kersten, N.S. Tan, Angiopoietin-like 4: a decade ofresearch, Biosci. Rep. 32 (2012) 211e219.

[5] C. Grootaert, W.T. Van de, W. Verstraete, M. Bracke, B. Vanhoecke, Angio-poietin-like protein 4: health effects, modulating agents and structure-function relationships, Expert Rev. Proteomics 9 (2012) 181e199.

[6] H. Ge, G. Yang, X. Yu, T. Pourbahrami, C. Li, Oligomerization state-dependenthyperlipidemic effect of angiopoietin-like protein 4, J. Lipid Res. 45 (2004)2071e2079.

[7] X. Lei, F. Shi, D. Basu, et al., Proteolytic processing of angiopoietin-like protein4 by proprotein convertases modulates its inhibitory effects on lipoproteinlipase activity, J. Biol. Chem. 286 (2011) 15747e15756.

[8] W. Yin, S. Romeo, S. Chang, N.V. Grishin, H.H. Hobbs, J.C. Cohen, Geneticvariation in ANGPTL4 provides insights into protein processing and function,J. Biol. Chem. 284 (2009) 13213e13222.

[9] K. Yoshida, T. Shimizugawa, M. Ono, H. Furukawa, Angiopoietin-like protein 4is a potent hyperlipidemia-inducing factor in mice and inhibitor of lipoproteinlipase, J. Lipid Res. 43 (2002) 1770e1772.

[10] D. Stejskal, M. Karpisek, H. Reutova, et al., Angiopoietin-like protein 4:development, analytical characterization, and clinical testing of a new ELISA,Gen. Physiol. Biophys. 27 (2008) 59e63.

[11] M.R. Robciuc, E. Tahvanainen, M. Jauhiainen, C. Ehnholm, Quantitation ofserum angiopoietin-like proteins 3 and 4 in a Finnish population sample,J. Lipid Res. 51 (2010) 824e831.

[12] M.C. Smart-Halajko, M.R. Robciuc, J.A. Cooper, et al., The relationship betweenplasma angiopoietin-like protein 4 levels, angiopoietin-like protein 4 geno-type, and coronary heart disease risk, Arterioscler. Thromb. Vasc. Biol. 30(2010) 2277e2282.

[13] S. Kathiresan, C.J. Willer, G.M. Peloso, et al., Common variants at 30 locicontribute to polygenic dyslipidemia, Nat. Genet. 41 (2009) 56e65.

[14] S. Romeo, L.A. Pennacchio, Y. Fu, et al., Population-based resequencing ofANGPTL4 uncovers variations that reduce triglycerides and increase HDL, Nat.Genet. 39 (2007) 513e516.

[15] P.J. Talmud, M. Smart, E. Presswood, et al., ANGPTL4 E40K and T266M: effectson plasma triglyceride and HDL levels, postprandial responses, and CHD risk,Arterioscler. Thromb. Vasc. Biol. 28 (2008) 2319e2325.

[16] A.R. Folsom, J.M. Peacock, E. Demerath, E. Boerwinkle, Variation in ANGPTL4and risk of coronary heart disease: the Atherosclerosis Risk in CommunitiesStudy, Metabolism 57 (2008) 1591e1596.

A. Muendlein et al. / Atherosclerosis 237 (2014) 632e638638

[17] Expert panel third report of the National Cholesterol Education Program(NCEP) expert panel on detection, evaluation, and treatment of high bloodcholesterol in adults (adult treatment panel III) final report, Circulation 106(2002) 3143e3421.

[18] H. Drexel, F.W. Amann, J. Beran, et al., Plasma triglycerides and three lipo-protein cholesterol fractions are independent predictors of the extent ofcoronary atherosclerosis, Circulation 90 (1994) 2230e2235.

[19] S. Kersten, L. Lichtenstein, E. Steenbergen, et al., Caloric restriction and ex-ercise increase plasma ANGPTL4 levels in humans via elevated free fatty acids,Arterioscler. Thromb. Vasc. Biol. 29 (2009) 969e974.

[20] V. Legry, S. Bokor, D. Cottel, et al., Associations between common geneticpolymorphisms in angiopoietin-like proteins 3 and 4 and lipid metabolismand adiposity in European adolescents and adults, J. Clin. Endocrinol. Metab.94 (2009) 5070e5077.

[21] H. Staiger, F. Machicao, R. Werner, et al., Genetic variation within theANGPTL4 gene is not associated with metabolic traits in white subjects at anincreased risk for type 2 diabetes mellitus, Metabolism 57 (2008) 637e643.

[22] A haplotype map of the human genome, Nature 437 (2005) 1299e1320.[23] M. Stephens, N.J. Smith, P. Donnelly, A new statistical method for haplotype

reconstruction from population data, Am. J. Hum. Genet. 68 (2001) 978e989.[24] M. Stephens, P. Donnelly, A comparison of bayesian methods for haplotype

reconstruction from population genotype data, Am. J. Hum. Genet. 73 (2003)1162e1169.

[25] W.D. Dupont, W.D. Plummer Jr., Power and sample size calculations. A reviewand computer program, Controll. Clin. Trials 11 (1990) 116e128.

[26] H. Adachi, Y. Fujiwara, T. Kondo, et al., Angptl 4 deficiency improves lipidmetabolism, suppresses foam cell formation and protects against athero-sclerosis, Biochem. Biophys. Res. Commun. 379 (2009) 806e811.

[27] L. Lichtenstein, J.F. Berbee, S.J. van Dijk, et al., Angptl4 upregulates cholesterolsynthesis in liver via inhibition of LPL- and HL-dependent hepatic cholesteroluptake, Arterioscler. Thromb. Vasc. Biol. 27 (2007) 2420e2427.

[28] A. Xu, M.C. Lam, K.W. Chan, et al., Angiopoietin-like protein 4 decreases bloodglucose and improves glucose tolerance but induces hyperlipidemia and he-patic steatosis in mice, Proc. Natl. Acad. Sci. U. S. A. 102 (2005) 6086e6091.

[29] H. Drexel, S. Aczel, T. Marte, et al., Is atherosclerosis in diabetes and impairedfasting glucose driven by elevated LDL cholesterol or by decreased HDLcholesterol? Diabetes Care 28 (2005) 101e107.

[30] H. Drexel, S. Aczel, T. Marte, A. Vonbank, C.H. Saely, Factors predicting car-diovascular events in statin-treated diabetic and non-diabetic patients withcoronary atherosclerosis, Atherosclerosis 208 (2010) 484e489.

[31] P. Rein, C.H. Saely, S. Beer, A. Vonbank, H. Drexel, Roles of the metabolicsyndrome, HDL cholesterol, and coronary atherosclerosis in subclinicalinflammation, Diabetes Care 33 (2010) 1853e1855.

[32] C.H. Saely, L. Koch, F. Schmid, et al., Adult Treatment Panel III 2001 but notInternational Diabetes Federation 2005 criteria of the metabolic syndromepredict clinical cardiovascular events in subjects who underwent coronaryangiography, Diabetes Care 29 (2006) 901e907.

[33] L. Guo, S.Y. Li, F.Y. Ji, et al., Role of Angptl4 in vascular permeability andinflammation, Inflamm. Res. 63 (2014) 13e22.

[34] Y.H. Yang, Y. Wang, K.S. Lam, et al., Suppression of the Raf/MEK/ERK signalingcascade and inhibition of angiogenesis by the carboxyl terminus ofangiopoietin-like protein 4, Arterioscler. Thromb. Vasc. Biol. 28 (2008)835e840.

[35] P.R. Moreno, K.R. Purushothaman, M. Sirol, A.P. Levy, V. Fuster, Neo-vascularization in human atherosclerosis, Circulation 113 (2006) 2245e2252.

[36] A. Galaup, E. Gomez, R. Souktani, et al., Protection against myocardialinfarction and no-reflow through preservation of vascular integrity byangiopoietin-like 4, Circulation 125 (2012) 140e149.

[37] V. Fuster, L. Badimon, J.J. Badimon, J.H. Chesebro, The pathogenesis of coro-nary artery disease and the acute coronary syndromes (2), N. Engl. J. Med. 326(1992) 310e318.

[38] V. Fuster, L. Badimon, J.J. Badimon, J.H. Chesebro, The pathogenesis of coro-nary artery disease and the acute coronary syndromes (1), N. Engl. J. Med. 326(1992) 242e250.

[39] A. Georgiadi, Y. Wang, R. Stienstra, et al., Overexpression of angiopoietin-likeprotein 4 protects against atherosclerosis development, Arterioscler. Thromb.Vasc. Biol. 33 (2013) 1529e1537.