lable at ScienceDirect

Atherosclerosis 240 (2015) 98e104

Contents lists avai

Atherosclerosis

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

APOA5 variants predispose hyperlipidemic patients to atherogenicdyslipidemia and subclinical atherosclerosis

Montse Guardiola a, b, Montserrat Cof�an c, d, Isabel de Castro-Oros e, Ana Cenarro e,Núria Plana a, b, Philippa J. Talmud f, Lluís Masana a, b, Emilio Ros c, d, Fernando Civeira e,Josep Ribalta a, b, *

a Unitat de Recerca en Lípids i Arteriosclerosi, Hospital Universitari Sant Joan de Reus, Institut d'Investigacions Sanit�aries Pere Virgili,Universitat Rovira i Virgili, Reus, Spainb CIBER Diabetes y Enfermedades Metab�olicas (CIBERdem), Instituto de Salud Carlos III (ISCIII), Spainc Lípid Clinic, Endocrinology & Nutrition Service, Institut d'Investigacions Biom�ediques August Pi Sunyer, Hospital Clínic, Barcelona, Spaind CIBER Fisioptaologia de la Obesidad y Nutrici�on (CIBERobn), ISCIII, Spaine Hospital Universitario Miguel Servet, Instituto Investigaci�on Sanitaria de Arag�on, Zaragoza, Spainf Centre of Cardiovascular Genetics, Institute of Cardiovascular Science, University College London, London, United Kingdom

a r t i c l e i n f o

Article history:Received 20 November 2014Received in revised form24 February 2015Accepted 3 March 2015Available online 9 March 2015

Keywords:APOA5PolymorphismsNMRIMTAtherogenic dyslipidemia

* Corresponding author. Unitat de Recerca en LípiUniversitari Sant Joan de Reus, Institut d'InvestigaUniversitat Rovira i Virgili, Sant Llorenç 21, 43201 Re

E-mail address: josep.ribalta@urv.cat (J. Ribalta).

http://dx.doi.org/10.1016/j.atherosclerosis.2015.03.0080021-9150/© 2015 Elsevier Ireland Ltd. All rights rese

a b s t r a c t

Background: Triglycerides (TG) are the initiators of the metabolic changes leading to the atherogenicdyslipidemia, which is a major inducer of atherosclerosis as a result of quantitative and qualitativechanges in lipoprotein subclass distributions. We hypothesized that variation at the of APOA5 gene locus,encoding apoAV, a key regulator of TG levels, significantly affect lipoprotein subclass distributions towarda more atherogenic pattern in both hyperTG patients and dyslipemic patients.Methods: We recruited four hundred and twenty-two subjects attending a Lipid Clinic, prior to lipid-lowering treatment. We genotyped two APOA5 variants, rs662799 (-1131T>C) and rs3135506 (S19W).Circulating lipoproteins were determined by nuclear magnetic resonance (NMR). Intima-media thickness(IMT) was evaluated using B-mode ultrasound.Results: Carriers of the rare alleles of rs662799 and rs3135506 compared to common allele homozygotes,had a significantly proatherogenic profile of the VLDL and LDL subclasses, resulting in increased con-centrations of the proatherogenic subclasses, large VLDLs (þ133%, p < 0.001) and small LDLs (þ34%,p ¼ 0.014). Significant changes in smaller HDL (þ71%, p ¼ 0.032), as well as an 18% decrease in large HDL(p ¼ 0.046), were also been observed. This atherogenic NMR subclass distribution was significantlyassociated with increased carotid IMT.The observed effects were significantly stronger in patients with a BMI � 25 kg/m2 and in male andfemale patients with a waist circumference �90 cm or �85 cm, respectively.Conclusion: In a dyslipemic population, genetic variants of APOA5 modulate lipoprotein subclass distri-butions, inducing an atherogenic profile associated with IMT defined subclinical atherosclerosis.

© 2015 Elsevier Ireland Ltd. All rights reserved.

1. Introduction

Plasma lipid levels play a pivotal role in the pathogenesis ofatherosclerosis and are a major predictor of coronary artery disease(CAD) [1]. Hence, LDL cholesterol is the focus of most strategies

ds i Arteriosclerosi, Hospitalcions Sanit�aries Pere Virgili,us, Spain.

rved.

directed at reducing cardiovascular risk. Triglycerides (TG) are alsoimportant in this regard, not only because they are a risk factor foratherosclerosis [2] but because they are also the initiators of themetabolic changes leading to an atherogenic lipoprotein profile, theso-called atherogenic dyslipidemia. The rationale behind the pre-sent investigation is that atherogenic changes induced by TG occurat concentrations as low as 1.7 mmol/L; the level at which smalldense LDL become predominant [3]. Thus small dense LDL areclinically relevant not only for hypertriglyceridemic patients butalso for those with other dyslipidemias.

M. Guardiola et al. / Atherosclerosis 240 (2015) 98e104 99

The condition that best reflects the pernicious role of TG isatherogenic dyslipidemia, a metabolic disturbance characterized byhypertriglyceridemia and low HDLc that entails an increased car-diovascular disease risk [4e6]. It is a feature of obesity, type 2diabetes mellitus, and metabolic syndrome, conditions with a highprevalence globally [7,8]. It has been suggested that high plasma TGlevels modulate the size and number of certain lipoprotein sub-classes, triggering an imbalance that promotes increase in circu-lating proatherogenic small dense LDL particles and cholesterol-rich remnant particles, as well as a decrease in anti-atherogenicHDL particles.

To gain insight into this scenario two additional issues must beaddressed. The first involves the factors that predispose an indi-vidual to increased TG; the second is our capacity to detect TG-induced lipoprotein changes via traditional lipid and lipoproteinmeasurements.

Regarding the first issue, the apolipoprotein A5 (APOA5) gene,encoding apoAV, is one of the major genetic determinants ofcirculating TG levels. In animal models, an inverse relationshipbetween hepatic APOA5 expression and plasma TG levels has beendescribed [9]. In humans, both single-gene and genome-wide as-sociation studies in different populations, have confirmed thatAPOA5 is one of the strongest genes influencing TG concentrations[10e13]. It has also been reported that APOA5 variants are associ-atedwith other lipid parameters, LDL and HDL, suggesting that theyare not limited to determining TG levels, but also play an importantrole in the overall regulation of lipid metabolism.

To date, several mechanisms of apoAV action have been pro-posed, including plasma TG removal, either by stimulating TG hy-drolysis and stabilizing the lipoprotein lipase (LPL) active dimer, orfacilitating the attachment between TG-rich lipoproteins and gly-cosylphosphatidylinositol anchored high density lipoproteinbinding protein 1 (GPIHBP1) [14,15], as well as accelerating hepaticuptake of TG-rich lipoproteins and their remnants by heparansulfate proteoglycan (HSPG) and LDL receptor (LDLR) familymembers. Intracellular effects of apoAV on VLDL production andsecretion by the liver may also explain the effect of apoAV on TGlevels, but the evidence supporting this notion is weaker [16].

Regarding lipidmeasurements, it has been known for some timethat lipid constituents of major lipoprotein classes (such ascholesterol and TG), may be further characterized into subclassesusing various techniques, among them nuclear magnetic resonance(NMR), a method that helps quantify lipoprotein subclasses [17].The profile of these subclasses represents an important prognosticfactor in both the manifestation and the progression of CAD.

Therefore, the objective of this study was to analyze the effectsof APOA5 variants on lipid profiles, lipoprotein subclass sizes andnumbers, and carotid atherosclerosis, determined by measures ofintima media thickness (IMT) in hyperlipidemic subjects recruitedfrom a lipid clinic.

2. Subjects and methods

2.1. Study subjects

Subjects were recruited at 3 Lipid Clinics (Hospital UniversitariSant Joan de Reus, Hospital Clínic of Barcelona and Hospital Uni-versitario Miguel Servet of Zaragoza) in Spain [18]. The study beganin 2005, when clinical, analytical and sonographic methodologieswere standardized among centers. All patients �17 years of agewith a clinical diagnosis of familial hyperlipidemia were includedand provided informed consent to participate in a protocolapproved by the ethical review boards of each of the participatinginstitutions. Within 2e6 weeks of their first visit, all participantshad venipuncture to collect fasting blood samples and underwent a

carotid ultrasound according to a predefined protocol. We studied422 untreated subjects.

Isolated primary hypercholesterolemia was diagnosed in sub-jects with off-treatment LDLc levels above the age- and sex-specific95th percentile of a Spanish reference population [19], as well as TGlevels below 5.17 mmol/L. The diagnosis of primary hyper-triglyceridemia was based on the presence of either combinedhyperlipidemia or isolated hypertriglyceridemia in untreated pa-tients whose serum cholesterol and TG concentrations were abovethe sex- and age-specific 90th percentiles for the Spanish popula-tion. The criterion for combined hyperlipidemia was having serumtotal apolipoprotein B levels �120 mg/dL, and for isolated hyper-triglyceridemia this was the presence of high TG alone. Secondarycauses of hyperlipidemia were excluded in all subjects. A controlgroup consisting of healthy, unrelated male and female volunteers,aged 18e75 years, who underwent a medical examination at theHospital Miguel Servet of Zaragoza was also studied. Exclusioncriteria for control subjects included a personal or parental historyof CAD or dyslipidemia, an existing acute illness, or the use of drugscapable of influencing either glucose or lipid metabolism.

Eighteen participants did not report data on BMI and 10 par-ticipants did not report data on waist circumference so they werenot included into the analysis.

2.2. Laboratory measurements

Fasting blood for biochemical profiles was drawn after patientswere off hypolipidemic drug treatment for at least 4 weeks.Cholesterol and TG levels were determined by standard enzymaticmethods. HDLc was measured by a precipitation technique. LDLcwas estimated with the Friedewald equation, except in sampleswith triglycerides �3.5 mmol/L, when it was measured in thed ¼ 1.063 g/mL fraction separated by density gradient ultracentri-fugation. Apolipoprotein (apo) B and lipoprotein(a) levels weredetermined by immunoturbidimetry (Unimate 3, Roche, Basel,Switzerland).

2.3. Nuclear magnetic resonance lipoprotein profile measurements

Both lipoprotein subclass particle concentrations and theaverage sizes of lipoprotein particles were measured by protonNMR spectroscopy (LipoScience, Inc., Raleigh, North Carolina), aspreviously described [20]. The particle concentrations for lipo-protein subclasses of different size were obtained directly usingthe measured amplitudes of their spectroscopically distinct lipidmethyl group NMR signals. Weighted-average lipoprotein particlesizes were derived from the sum of the diameter of each subclassmultiplied by its relative mass percentage based on the amplitudeof its methyl NMR signal. The concentrations of the followingsubclasses were measured: small LDL (diameter 18.0e21.2 nm),large LDL (21.2e23.0 nm), intermediate-density lipoprotein (IDL)(23.0e27.0 nm), large high-density lipoprotein (HDL)(8.8e13.0 nm), medium HDL (8.2e8.8 nm), small HDL(7.3e8.2 nm), large very low-density lipoprotein (VLDL) (>60 nm),medium VLDL (35.0e60.0 nm), and small VLDL (27.0e35.0 nm).VLDL and LDL particle concentrations are expressed in nmol/L, andHDL, in mmol/L.

2.4. Carotid intima media thickness (IMT) measurements

For carotid sonography, we used at each center an AcusonSequoia instrument (Siemens Medical Solutions, Erlangen, Ger-many) equipped with a linear array ultrasound transducer (L7,5e12 MHz). Scanning and image analysis procedures were stan-dardized as previously described [21]. In summary, scans were

Table 1Subjects' description.

Normolipidemia (n ¼ 79) Hypertriglyceridemia (n ¼ 125) Hypercholesterolemia (n ¼ 216)

Gender (_/\) 33/46 88/37 96/120Age (years) 36.42 (14.44) 48.42 (11.98) 46.87 (13.67)BMI (kg/m2) 24.31 (4.71) 27.88 (3.42) 25.38 (3.90)Total cholesterol (mmol/L) 4.39 (0.56) 7.58 (1.61) 7.21 (1.42)Triglycerides (mmol/L) 0.71 (0.34) 3.69 (2.63) 0.99 (0.36)HDL cholesterol (mmol/L) 1.37 (0.37) 1.14 (0.29) 1.47 (0.41)Non-HDL (mmol/L) 3.02 (0.45) 6.44 (1.55) 5.73 (1.32)LDL cholesterol (mmol/L) 2.70 (0.42) 5.12 (1.70) 5.29 (1.30)Apolipoprotein A (mmol/L) 48.95 (9.41) 46.62 (8.14) 50.19 (9.84)Apolipoprotein B (mmol/L) 1.51 (0.28) 2.96 (0.77) 2.72 (0.65)

Data are expressed as a mean (SD).

Table 2APOA5 genotypes vs. traditional lipid parameters.

-1131T>C S19W

TT (n ¼ 359) TC/CC (n ¼ 63) p SS (n ¼ 354) SW/WW (n ¼ 68) p

Gender (_/\) 181/177 37/27 NS 182/171 36/33 NSAge (years) 45.31 (13.91) 45.62 (14.66) NS 45.08 (13.75) 46.60 (15.17) NSBMI (kg/m2) 25.90 (4.10) 26.33 (4.36) NS 25.90 (4.24) 26.32 (3.54) NSTotal cholesterol (mmol/L) 6.73 (1.78) 7.13 (1.88) NS 6.76 (1.80) 6.93 (1.79) NSTriglycerides (mmol/L) 1.68 (2.00) 2.09 (1.55) 0.014 1.56 (1.59) 2.67 (3.03) 0.002HDL cholesterol (mmol/L) 1.37 (0.41) 1.26 (0.30) 0.032 1.36 (0.38) 1.33 (0.50) NSLDL cholesterol (mmol/L) 4.67 (1.61) 5.07 (1.85) NS 4.74 (1.64) 4.60 (1.68) NSApolipoprotein A (mmol/L) 49.12 (9.59) 47.73 (8.27) NS 48.97 (9.37) 48.39 (9.47) NSApolipoprotein B (mmol/L) 2.54 (42.59) 2.82 (0.99) NS 2.59 (0.83) 2.57 (0.73) NS

Data are expressed as a mean (SD).p value is adjusted for age, gender and BMI.

M. Guardiola et al. / Atherosclerosis 240 (2015) 98e104100

performed from a fixed lateral angle. The far walls of the threebilateral carotid segments were visualized: the right and leftcommon carotid arteries, carotid bifurcations, and internal carotidarteries. Sonographic variables of interest included the maximumIMT in any of the 6 carotid segments (maxIMT), the mean of themaximum IMT at each of the 6 carotid areas (mean-maxIMT), andthe mean IMT for the 6 carotid segments (mean IMT). All pro-cedures were performed by trained certified sonographers. Highresolution images of each of the segments were saved as DICOMstills during the diastole of the vessel. The recordings were subse-quently stored on CD for offline analyses. The scans from eachcenter were analyzed by a certified ultrasound reader in Zaragoza.Semiautomated edge-tracking software (eTRACK, AMC VascularImaging, Amsterdam, The Netherlands) was used. The ultrasoundreader was blinded to the demographic and clinical information ofthe subjects.

2.5. APOA5 genotyping

DNA was isolated from EDTA blood samples following standardprotocols. We genotyped the rs662799 (-1131T>C) and rs3135506(S19W) APOA5 variants using TaqMan Technology (Applied Bio-systems) and the 7900HT Sequence Detection machine (AppliedBiosystems) using the 7900HT Sequence Detection machine(Applied Biosystems), as reported previously [22].

2.6. Statistical analyses

Statistical analyses were carried out using SPSS software,version 17.0. The Chi-square (c2) test was used to test for Har-dyeWeinberg equilibrium. Unpaired t-test was used to comparelipid, apolipoprotein and IMT data amongst genotypes. Not nor-mally distributed variables were log-transformed before analysis.The results are expressed as means (SD). All analyses were adjusted

for age, gender and BMI. A Bonferroni correction for multiplecomparisons was applied. Statistical significance was accepted atthe p ¼ 0.05 level.

3. Results

3.1. Subjects' description

We studied 422 subjects. As we predicted that APOA5 genotypeswouldmodulate lipoprotein profiles in all types of dyslipidemia, wepooled all study subjects. To assess individual dyslipidemias,however, we divided the subjects into different groups (Normoli-pidemia, Hypercholesterolemia and Hypertriglyceridemia). Table 1shows the clinical and biochemical parameters of each group.Hypertriglyceridemic and hypercholesterolemic patients wereolder, and hypercholesterolemic subjects presented with increasedtotal and LDLc levels, whereas hypertriglyceridemic subjects(including thosewith combined hyperlipidemia) had higher TG andhigher total and LDLc levels.

3.2. APOA5 genotypes and conventional lipid parameters

The frequencies of minor alleles -1131C and 19W of were 0.07and 0.08, respectively which are similar to those observed inCaucasian populations [23]. The genotype distributions were inHardyeWeinberg equilibrium. We pooled heterozygotes and rareallele homozygotes to increase statistical power. APOA5 genotypesshowed no difference in gender distribution, mean age or BMI.Table 2 shows levels of conventional lipids by APOA5 in the wholecohort.

Compared to common allele homozygotes, carriers of rare al-leles for -1131C and 19W had 24% (p ¼ 0.014), and 71% (p ¼ 0.002)higher TG levels respectively, and -1131C carriers had borderline 8%(p ¼ 0.032) lower HDLc levels.

Table 3APOA5 genotypes vs. NMR lipid parameters.

-1131T>C APOA5 S19W APOA5

TT (n ¼ 359) TC/CC (n ¼ 63) p SS (n ¼ 354) SW/WW (n ¼ 68) p

VLDL & Chylo total particles (nmol/L) 83.64 (51.71) 104.95 (63.15) 0.032 82.84 (50.37) 105.69 (66.41) 0.006VLDL large & Chylo particles (nmol/L) 3.22 (7.10) 5.90 (10.35) NS 2.98 (6.43) 6.96 (11.99) <0.001VLDL median particles (nmol/L) 28.61 (30.52) 38.89 (37.71) NS 27.23 (29.00) 44.46 (41.21) <0.001VLDL small particles (nmol/L) 51.81 (27.96) 60.16 (34.51) NS 52.63 (27.94) 54.27 (33.98) NSIDL particles (nmol/L) 65.33 (67.81) 80.48 (68.89) NS 66.68 (68.89) 72.00 (64.26) NSLDL particles (nmol/L) 1581.60 (585.24) 1852.47 (739.79) 0.014 1605.62 (613.60) 1697.07 (637.12) NSLDL large particles (nmol/L) 702.52 (401.72) 667.57 (484.23) NS 716.31 (408.77) 606.67 (439.03) NSLDL small particles (nmol/L) 813.75 (688.29) 1104.41 (822.79) 0.010 822.64 (685.01) 1018.35 (853.87) NSLDL median small particles (nmol/L) 157.52 (131.96) 220.60 (166.64) 0.002 159.17 (131.69) 203.78 (170.78) 0.024LDL very small particles (nmol/L) 656.22 (558.68) 883.79 (658.99) 0.014 663.47 (555.69) 814.59 (685.81) NSHDL particles (mmol/L) 30.39 (5.51) 30.23 (5.76) NS 30.40 (5.56) 30.07 (5.37) NSHDL large particles (mmol/L) 7.03 (3.88) 5.74 (3.77) 0.046 6.95 (3.70) 6.24 (4.77) NSHDL median particles (mmol/L) 1.56 (2.89) 2.27 (3.56) NS 1.49 (2.78) 2.55 (3.88) 0.032HDL small particles (mmol/L) 21.80 (5.24) 22.22 (5.24) NS 21.96 (5.31) 21.28 (4.71) NSVLDL size (nm) 46.41 (9.38) 47.26 (9.82) NS 45.89 (8.28) 49.96 (13.56) <0.001LDL size (nm) 21.34 (1.00) 21.05 (1.09) 0.062 21.35 (0.96) 21.04 (1.25) 0.022HDL size (nm) 9.01 (0.60) 8.84 (0.55) 0.067 9.00 (0.59) 8.92 (0.62) NS

Data are expressed as a mean (SD).p value is adjusted for age, gender and BMI.Total LDL particle concentrations are the sum of the intermediate-density lipoprotein, large LDL, and small LDL subclass concentrations which correspond to the sum ofmediansmall and very small particles.

M. Guardiola et al. / Atherosclerosis 240 (2015) 98e104 101

3.3. APOA5 genotypes and NMR lipoprotein subclasses

To further characterize the effects of APOA5 variants on lipidprofiles, we evaluated their association with concentrations of li-poprotein subclasses determined by NMR (Table 3).

The -1131T>C polymorphismwas significantly associated with a25% increase in the number of total VLDL and chylomicron particles(p ¼ 0.032).

C allele carriers also had significantly higher number of total LDLparticles (p ¼ 0.014). In further detail this association with LDL waslimited to the smaller particles: 35% increase in small LDL(p¼ 0.010); 40% increase in medium small LDL (p¼ 0.002) and 34%increase in very small LDL particles (p ¼ 0.014). Patients carryingthe minor allele for the -1131T>C variant also presented with an18% decrease in their levels of large HDL particles (p ¼ 0.046).

Regarding the S19W genotype, carriers of the minor allele pre-sented with a 27% increase in the number of total VLDL andchylomicron particles (p ¼ 0.006). In further detail, this associationwas limited to the largest VLDL particles: 133% increase in largeVLDL and chylomicron particles (p < 0.001), as well as 63% increasein median VLDL particles (p < 0.001).

Carriers of the rare 19W allele presented with 23% increasedmedian small LDL particles (p ¼ 0.024).

Moreover, patients carrying theminor allele of the S19W variantalso presented with a 71% increase in medium HDL particle levels(p ¼ 0.032).

3.4. Influence of adiposity on the association of APOA5 genotypeswith NMR lipoprotein subclasses

We performed the following sub-analysis based on BMI andwaist circumference stratification: BMI<25 kg/m2 (n ¼ 173) andBMI � 25 kg/m2 (n ¼ 231). The significant associations describedbetween APOA5 variants and lipid parameters were limited almostentirely to overweight patients (Supplemental data, Table S1 andS2).

We also divided the cohort into two groups of abdominalobesity: waist circumference <90 cm in men (n ¼ 64) and <85 cmin women (n ¼ 109), and waist circumference �90 cm in men(n ¼ 153) and �85 cm in women (n ¼ 86) (Supplemental data,

Table S3 and S4). Again only patients with largest waist circum-ference showed associations between APOA5 gene markers andlipid parameters.

3.5. NMR subclasses, APOA5 genotypes and carotid IMT

Total VLDL and chylomicron, and small VLDL particle concen-trations were increased by 55% (p ¼ 0.028) and 40% (p ¼ 0.012),respectively in the top IMT tertile compared with the bottom tertile(respectively). VLDL remnants (IDL particles) were almost 2 timeshigher in patients with the highest IMT values (p ¼ 0.006), andsignificant increases in total LDL particle concentrations betweenthe first and third IMT tertiles were also noted (p ¼ 0.002). Addi-tional lipoprotein particles showed associations with maxIMT ter-tiles (summarized in Fig. 1). Regarding APOA5 genotypes, the rare19W allele was associated with higher internal carotid thickness(Fig. 2) (p ¼ 0.028), and this effect was limited to patients withhigher waist circumference (Table S5).

4. Discussion

Increased serum TG levels are an independent risk factor forcardiovascular disease because they induce quantitative and qual-itative atherogenic changes in TG-rich lipoproteins, as well as inboth LDL and HDL particles [24]. The aim was to investigatewhether atherogenic changes induced by TG occur at concentra-tions as low as 1.7 mmol/L (the level at which small dense LDLparticles become more numerous) [3], were clinically relevant, notonly for hypertriglyceridemic patients, but also for all hyper-lipidemic patients.

We show that APOA5 variants (-1131C/T and S19W) were asso-ciated with increased TG levels and an atherogenic lipoproteinprofile characterized by a shift toward smaller LDL and HDLparticles.

These changes were present in the entire dyslipidemic popu-lation and are not confined only to hyperTG patients. We observeda trend towards an atherogenic lipoprotein profile (in relation tothe -1131C allele) among hypercholesterolemic patients with verymoderate TG concentrations. However because of the limitedsample size these results did not to reach statistical significance

Fig. 1. NMR lipid parameters vs. maxIMT tertiles. Mean values of parameters from the NMR-lipoprotein profile with a statistically significant association with tertiles of maxIMT. Pvalues are adjusted for age, gender and BMI. Error bars represent standard deviation.

Fig. 2. APOA5 S19W genotype vs. meanIMT. Mean values of internal carotid mean andmaximum values according to S19W APOA5 variant. P values are adjusted for age,gender and BMI. Error bars represent standard deviation.

M. Guardiola et al. / Atherosclerosis 240 (2015) 98e104102

(data not shown).The atherogenic lipoprotein subclass changes associated with

APOA5 variants showed statistically significantly subclinicalatherosclerosis, which was exacerbated by excess body weight andincreased waist circumference.

4.1. Triglyceride-driven changes in LDL and HDL by NMR

Atherogenic changes in LDL and HDL particle size are notdetectable by routine lipid analyses, but can be uncovered by li-poprotein NMR analysis.

This lipoprotein profile is a common feature of metabolic dis-turbances associated with an increased cardiovascular risk, such asinsulin resistance, diabetes mellitus, lupus erythematosus [25], andother conditions. It is well known that the onset of this atherogenicprofile is due to increased fatty acid flux from adipose tissue, whichinduces increased synthesis and secretion of TG-rich particles(VLDL). These VLDLs are larger, contain more TG, and possess agreater proportion of apoCIII, the inhibitor of LPL, compared withapoCII, which results in impaired LPL activity and deficient TG hy-drolysis. In this situation remnant particles are poorly recognizedby both the LDL-receptor (LDLR) and LRP receptor (LDLR-relatedprotein), which are responsible for their clearance from the circu-lation. This increases the time that these lipoproteins remain incirculation and allows for the formation of more remnant-likeparticles. Due to the actions of CETP (cholesteryl ester transferprotein) and HL (hepatic lipase), these particles become small anddense LDL (sdLDL), which are highly susceptible to oxidation andproatherogenic, as they are associated with at least a 3-fold in-crease in CHD risk [26]. CETP and HL also promote changes causinglarge cholesterol-rich HDL particles to become TG-rich andcholesterol poor. The hydrolysis of TG induces rapid changes thatyield much smaller alpha HDL particles, which are subject to renalexcretion due to their small sizes.

4.2. The APOA5 promotes TG-driven LDL and HDL changes

The data from the lipoprotein subclasses obtained by NMR

M. Guardiola et al. / Atherosclerosis 240 (2015) 98e104 103

provide a better understanding of the effects of APOA5 variants onlipoprotein profiles. NMR analyses revealed significant associationsamongst the APOA5 variants and the number and sizes of chylo-microns and VLDL, LDL and HDL particles. Rare APOA5 alleles wereassociated with an atherogenic profile characterized by elevationsof the levels of large chylomicrons and VLDL particles, and conse-quently high levels of small dense LDL. This also leads to a reductionin large HDL particles.

Our studies largely concur with those of previous studies.Talmud et al. [27] studied the relationship between APOA5 poly-morphisms and lipid profiles obtained via preparative ultracentri-fugation in coronary patients and found an association between the-1131T>C polymorphism and increased VLDL levels, but notincreased apoB levels. The S19 allele was associated with anincreased size of IDL particle, which are precursors of sdLDL. Laiet al. [28] analyzed the relationship between APOA5 poly-morphisms and NMR lipoprotein subclasses in the FraminghamOffspring Study and reported that -1131T>C and S19W variantswere associatedwith increased levels of larger VLDL subclasses. In alargemeta-analysis of 101 studies withmore than twenty thousandcoronary heart disease cases and > thirty-five thousand controls,the -1131T>C polymorphism was also associated with an athero-genic profile of lipoproteins as determined by NMR; however, noeffects on LDL subclasses were noted [2].

Our results support the importance of a genetic predispositionto hypertriglyceridemia as a risk factor for atherosclerosis. APOA5variants are associated with increased TG levels and changes in thesizes and concentrations of LDL and HDL subparticles. We haveshown how these lipoprotein characteristics are more commonamong subjects in the highest tertiles of IMT. Moreover, the APOA519W allele was significantly associated with increased IMT. Theseresults concur with data from Elosua et al. [29] who showed that-1131T>C and S19W variants were both significantly associatedwith increased common carotid artery IMT in obese participants, inthe Framingham Offspring Study. Our results for 19S allele are inagreement with these.

4.3. Lipoprotein effects of APOA5 variants are not limited tohypertriglyceridemic subjects

With our limited sample size we did not have the statisticalpower for subgrouping disease types, however our results do showthat the atherogenic changes in LDL and HDL are present not only inhypertriglyceridemic subjects, but also in subjects with isolatedhypercholesterolemia. In this group, which has a modest mean TGconcentration, changes similar to those associated with APOA5variants are also observed (data not shown).

Abdominal obesity is associated with high plasma TG and withlow plasma HDLc levels [30]; therefore, we were interested in apotential link between APOA5 and obesity in our subjects. Inter-estingly, we found that the relationships between APOA5 genemarkers and lipid parameters were more significant if we consid-ered only overweight subjects (defined as a BMI� 25 kg/m2), whichmay be explained in part by overweight patients having a moredyslipidemic profile than patients with lower BMI (data notshown), a finding that confirms the hypothesis that the effects ofAPOA5 depend largely on the metabolic background of the subject.This relationship with obesity was previously reported by Evans l[31], who showed that -1131T<C was associated with higher TGlevels and lower HDLc levels in a group of dyslipidemic patientsattending a lipid clinic but only observed in those with aBMI � 25 kg/m2.

In summary, we show that APOA5 variants predispose to higherTG concentrations and also to more atherogenic LDL and HDLparticles not detectable by routine analyses. These changes have

clinical importance as they are associated with increased subclin-ical atherosclerosis, and are observed not only in hyper-triglyceridemic subjects but also in those with isolatedhypercholesterolemia, particularly subjects who are overweight orhave large waist circumferences.

Conflict of interest

None.

Acknowledgments

This study was supported by CIBERDEM (CIBER de Diabetes yEnfermedades Metab�olicas Asociadas), FIS PI12/01087 and RETIC(RIC RD12/0042/0055), which are of ISCIII (Instituto de Salud CarlosIII). PJT is supported by the British Heart Foundation (RG08/008).

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.atherosclerosis.2015.03.008.

References

[1] Third Report of the National Cholesterol Education Program (NCEP) ExpertPanel on Detection, Evaluation, and treatment of high blood cholesterol inadults (adult treatment panel III) final report, Circulation 106 (2002)3143e3421.

[2] N. Sarwar, M.S. Sandhu, S.L. Ricketts, A.S. Butterworth, E. Di Angelantonio,S.M. Boekholdt, et al., Triglyceride-mediated pathways and coronary disease:collaborative analysis of 101 studies, Lancet 375 (2010) 1634e1639.

[3] A.D. Sniderman, G.R. Dagenais, B. Cantin, J.P. Despres, B. Lamarche, Highapolipoprotein B with low high-density lipoprotein cholesterol and normalplasma triglycerides and cholesterol, Am. J. Cardiol. 87 (2001) 792e793. A8.

[4] S.M. Grundy, Small LDL, atherogenic dyslipidemia, and the metabolic syn-drome, Circulation 95 (1997) 1e4.

[5] J.P. Despres, I. Lemieux, Abdominal obesity and metabolic syndrome, Nature444 (2006) 881e887.

[6] M. Adiels, S.O. Olofsson, M.R. Taskinen, J. Boren, Diabetic dyslipidaemia, Curr.Opin. Lipidol. 17 (2006) 238e246.

[7] E.S. Ford, C. Li, G. Zhao, W.S. Pearson, A.H. Mokdad, Prevalence of the meta-bolic syndrome among U.S. adolescents using the definition from the Inter-national Diabetes Federation, Diabetes Care 31 (2008) 587e589.

[8] S. Valdes, P. Botas, E. Delgado, F. Alvarez, F.D. Cadorniga, Population-basedincidence of type 2 diabetes in northern Spain: the Asturias Study, DiabetesCare 30 (2007) 2258e2263.

[9] L.A. Pennacchio, M. Olivier, J.A. Hubacek, J.C. Cohen, D.R. Cox, J.C. Fruchart, etal., An apolipoprotein influencing triglycerides in humans and mice revealedby comparative sequencing, Science 294 (2001) 169e173.

[10] J. Ribalta, L. Figuera, J. Fernandez-Ballart, E. Vilella, M. Castro Cabezas,L. Masana, et al., Newly identified apolipoprotein AV gene predisposes to highplasma triglycerides in familial combined hyperlipidemia, Clin. Chem. 48(2002) 1597e1600.

[11] M. Guardiola, R. Ferre, J. Salazar, C. Alonso-Villaverde, B. Coll, S. Parra, et al.,Protease inhibitor-associated dyslipidemia in HIV-infected patients is stronglyinfluenced by the APOA5-1131T->C gene variation, Clin. Chem. 52 (2006)1914e1919.

[12] J.A. Hubacek, Apolipoprotein A5 and triglyceridemia. Focus on the effects ofthe common variants, Clin. Chem. Lab. Med. 43 (2005) 897e902.

[13] S. Kathiresan, O. Melander, C. Guiducci, A. Surti, N.P. Burtt, M.J. Rieder, et al.,Six New Loci Associated With Blood Low-density Lipoprotein Cholesterol,High-density Lipoprotein Cholesterol or Triglycerides in Humans, vol. 40,2008, pp. 189e197.

[14] A.P. Beigneux, R. Franssen, A. Bensadoun, P. Gin, K. Melford, J. Peter, et al.,Chylomicronemia with a mutant GPIHBP1 (Q115P) that cannot bind lipo-protein lipase, Arterioscler. Thromb. Vasc. Biol. 29 (2009) 956e962.

[15] P. Gin, A.P. Beigneux, B. Davies, M.F. Young, R.O. Ryan, A. Bensadoun, et al.,Normal binding of lipoprotein lipase, chylomicrons, and apo-AV to GPIHBP1containing a G56R amino acid substitution, Biochim. Biophys. Acta 1771(2007) 1464e1468.

[16] X.-Y. Zheng, S.-P. Zhao, H. Yan, The role of apolipoprotein A5 in obesity andthe metabolic syndrome, Biol. Rev. Camb. Philos. Soc. 88 (2013) 490e498.

[17] J.D. Otvos, E.J. Jeyarajah, D.W. Bennett, Quantification of plasma lipoproteinsby proton nuclear magnetic resonance spectroscopy, Clin. Chem. 37 (1991)377e386.

[18] E. Jarauta, R. Mateo-Gallego, R. Gilabert, N. Plana, M. Junyent, E. de Groot,A. Cenarro, L. Masana, E. Ros, F. Civeira, Carotid atherosclerosis and lipoprotein

M. Guardiola et al. / Atherosclerosis 240 (2015) 98e104104

particle subclasses in familial hypercholesterolaemia and familial combinedhyperlipidaemia, Nutr. Metab. Cardiovasc. Dis. 22 (7) (2012) 591e597.

[19] J.A. Gomez-Gerique, J.A. Gutierrez-Fuentes, M.T. Montoya, A. Porres, A. Rueda,A. Avellaneda, et al., Lipid profile of the Spanish population: the DRECE (dietand risk of cardiovascular disease in Spain) study. DRECE study group, Med.Clin. (Barc.) 113 (1999) 730e735.

[20] E.J. Jeyarajah, W.C. Cromwell, J.D. Otvos, Lipoprotein particle analysis by nu-clear magnetic resonance spectroscopy, Clin. Lab. Med. 26 (2006) 847e870.

[21] E. De Groot, S.I. van Leuven, R. Duivenvoorden, M.C. Meuwese, F. Akdim,M.L. Bots, et al., Measurement of carotid intima-media thickness to assessprogression and regression of atherosclerosis, Nat. Clin. Pract. Med. 5 (2008)280e288.

[22] J. Wang, M.R. Ban, B.A. Kennedy, S. Anand, S. Yusuf, M.W. Huff, et al., APOA5genetic variants are markers for classic hyperlipoproteinemia phenotypes andhypertriglyceridemia, Nat. Clin. Pract. Cardiovasc. Med. 5 (2008) 730e737.

[23] M.-J. Ariza, M.-A. S�anchez-Chaparro, F.-J. Bar�on, A.-M. Hornos, E. Calvo-Bonacho, J. Rioja, et al., Additive effects of LPL, APOA5 and APOE variantcombinations on triglyceride levels and hypertriglyceridemia: results of theICARIA genetic sub-study, BMC Med. Genet. 11 (2010) 66.

[24] J.W. Skeggs, R.E. Morton, LDL and HDL enriched in triglyceride promoteabnormal cholesterol transport, J. Lipid Res. 43 (2002) 1264e1274.

[25] M. Gonz�alez, J. Ribalta, G. Vives, S. Iftimie, R. Ferr�e, N. Plana, et al., Nuclearmagnetic resonance lipoprotein subclasses and the APOE genotype influencecarotid atherosclerosis in patients with systemic lupus erythematosus,

J. Rheumatol. 37 (2010) 2259e2267.[26] S. Ip, A.H. Lichtenstein, M. Chung, J. Lau, E.M. Balk, Systematic review: asso-

ciation of low-density lipoprotein subfractions with cardiovascular outcomes,Ann. Intern. Med. 150 (2009) 474e484.

[27] P.J. Talmud, S. Martin, M.R. Taskinen, M.H. Frick, M.S. Nieminen,Y.A. Kesaniemi, et al., APOA5 gene variants, lipoprotein particle distribution,and progression of coronary heart disease: results from the LOCAT study,J. Lipid Res. 45 (2004) 750e756.

[28] C.Q. Lai, S. Demissie, L.A. Cupples, Y. Zhu, X. Adiconis, L.D. Parnell, et al., In-fluence of the APOA5 locus on plasma triglyceride, lipoprotein subclasses, andCVD risk in the Framingham Heart Study, J. Lipid Res. 45 (2004) 2096e2105.

[29] R. Elosua, J.M. Ordovas, L.A. Cupples, C.Q. Lai, S. Demissie, C.S. Fox, et al.,Variants at the APOA5 locus, association with carotid atherosclerosis, andmodification by obesity: the Framingham Study, J. Lipid Res. 47 (2006)990e996.

[30] S. Sam, S. Haffner, M.H. Davidson, R.B. D'Agostino, S. Feinstein, G. Kondos, etal., Relationship of abdominal visceral and subcutaneous adipose tissue withlipoprotein particle number and size in type 2 diabetes, Diabetes 57 (2008)2022e2027.

[31] D. Evans, A. Buchwald, F.U. Beil, The single nucleotide polymorphism-1131T>C in the apolipoprotein A5 (APOA5) gene is associated with elevatedtriglycerides in patients with hyperlipidemia, J. Mol. Med. (Berl.) 81 (2003)645e654.