74463150

doi: 10.1111/j.1365-2796.2011.02465.x

Type 2 diabetes mellitus interacts with obesity and commonvariations in PLTP to affect plasma phospholipid transferprotein activity

R. P. F. Dullaart1, M. Vergeer2, R. de Vries1, P. J. W. H. Kappelle1 & G. M. Dallinga-Thie2

Fromthe1DepartmentofEndocrinology,UniversityMedicalCenterGroningen,UniversityofGroningen,Groningen;and2DepartmentsofVascularandExperimentalVascularMedicine,AcademicMedicalCenterAmsterdam,Amsterdam,TheNetherlands

Abstract. DullaartRPF,VergeerM,deVriesR,KappellePJWH, Dallinga-Thie GM (University Medical CenterGroningen, University of Groningen, Groningen; andAcademic Medical Center Amsterdam, Amsterdam;The Netherlands). Type 2 diabetes mellitus interactswith obesity and common variations in PLTP to affectplasma phospholipid transfer protein activity.J InternMed2012;271: 490–498.

Background. Phospholipid transfer protein (PLTP) is anemerging cardiometabolic riskmarker that is impor-tant in high-density lipoprotein (HDL) and triglycer-ide metabolism. Plasma PLTP activity is elevated intype 2 diabetes mellitus, whereas glucose may regu-latePLTP gene transcription in vitro. Of interest, com-mon PLTP variations that predict cardiovascular dis-ease have been identified recently. We investigatedwhether the diabetic state is able to amplify relation-ships between obesity and PLTP gene variations withcirculatingPLTPlevels.

Subjects and methods. Plasma PLTP activity (using aphospholipid vesicles–HDL system), PLTP gene score[number of PLTP activity-decreasing alleles based ontwo tagging polymorphisms (rs378114 and rs60-65904)] andwaist circumference were determined intwoDutch cohorts comprising 237patientswith type2diabetesand78controlsubjects.

Results. Patients with diabetes were more obese(P < 0.001 for prevalence of increased waist circum-ference) and had 13% higher plasma PLTP activity(P < 0.001). PLTP gene score was not different in dia-betic and control subjects (P = 0.40). PLTP activity

was highest in patients with diabetes with an en-largedwaist and lowest incontrol subjectswithanor-mal waist circumference (P < 0.001). Multiple linearregressionanalysis revealedapositive interactionbe-tween diabetes status and waist circumference onPLTP activity (b = 0.200, P = 0.005). Furthermore,diabetes status (b = )0.485, P = 0.046) or HbA1c(b = )0.240, P = 0.035) interacted with PLTP genescore toaffectPLTPactivity.

Conclusions.Type2diabetesandenlargedwaist circum-ference interact to impact on plasma PLTP activity.Diabetes may also amplify the association betweenplasma PLTP activity and common PLTP gene varia-tions. Our findings support the hypothesis that dia-betes–environment and diabetes–gene interactionsgovernplasmaPLTPactivity.

Keywords: obesity, phospholipid transfer protein,phospholipid transfer protein gene, type 2 diabetesmellitus,waist circumference.

Abbreviations: anova,analysisofvariance;apo,apolipo-protein; AU, arbitrary unit; BMI, body mass index;CETP, cholesteryl ester transfer protein; CRP, high-sensitivity C-reactive protein; DALI study, DiabetesAtorvastatin Lipid Intervention study; EDTA, ethy-lenediaminetetraacetic acid; HbA1c, glycated hae-moglobin; HDL, high-density lipoprotein; MetS, met-abolic syndrome; NCEP-ATP-III, national cholesteroleducation programme adult treatment panel III;PLTP, phospholipid transfer protein; SNP, single-nucleotide polymorphism; VLDL, very low-densitylipoprotein

Introduction

Phospholipid transferprotein (PLTP)belongs to the li-pid transfer ⁄ lipopolysaccharidebinding protein fam-ily and is expressed in several tissues and cell sys-

tems including liver, macrophages and adiposetissue [1–4].PLTP isahigh-density lipoprotein (HDL)-associated lipid transfer protein that is able to trans-fer phospholipids towardsHDLduring lipolysis of tri-glyceride-rich lipoproteins and to remodel HDL into

490 ª 2011 The Association for the Publication of the Journal of Internal Medicine

Original Article |

different sized (both small and large) particles [1–4].Studies in mice support the notion that PLTP in-creases hepatic very low-density lipoprotein (VLDL)production [5, 6]. Moreover, PLTP exchanges a-tocopherol between lipoproteins and is associatedwithpro-inflammatoryproteins inplasma [7,8].

Although the potential role of PLTP in the develop-ment of atherosclerosis has not been unequivocallyestablished, available evidence from human studiesmostly supports the possibility that elevated plasmaPLTP activity is related positively to (subclinical) ath-erosclerosis [2, 3, 9–12]. Genetic variations in PLTPthat are associated with plasma PLTP activity, lipidphenotype and obesity-related variables, as well ascardiovascular disease, have been identified recently[13–16].

During thepast fewyears, evidencehasaccumulatedto support the possibility that plasmaPLTPactivity iselevated in insulin-resistant states, such as type 2diabetes mellitus, obesity and metabolic syndrome(MetS) [3, 4, 17–22]. Circulating PLTP activity isclosely related to glucose and lipid homoeostasis.PlasmaPLTP increases inparallel with increments incirculating free fatty acid levels and decreases inresponse to acipimox administration, insulin infu-sion and weight loss [3, 4, 18, 19, 23–27]. Of note, ithas been reported that high glucose stimulates PLTPpromoter activity in vitro, possibly via nuclear hor-mone receptor-dependent mechanisms [28]. Takentogether, these findings raise the hypothesis that thediabetic state may influence relationships of obesityand common genetic variations in PLTP with plasmaPLTPactivity levels inhumans.

Given an emerging role of PLTP as a possible cardio-metabolic risk factor, we sought to determinewhether the diabetic state interacts with obesity toimpact on plasmaPLTP activity.We also investigatedwhether thepresenceof diabetesmellitusmayampli-fy the observed association between plasma PLTPactivity and common PLTP gene variations. Thesepossibilities were tested in two cohorts, comprisingtype 2 diabetic and control subjects, inwhichwe pre-viously identified two PLTP-tagging single-nucleotidepolymorphisms (SNPs) that are associatedwith plas-maPLTPactivity [16].

Subjects and methods

Subjects included in the present study were partici-pants from two previous studies: the Groningencase–control study, which was originally designed to

examine whether intima–media thickness is relatedto plasma PLTP activity [10]; and the Diabetes Ator-vastatin Lipid Intervention (DALI) study, a double-blind, placebo-controlled, randomized, multicentrestudy,whichevaluated the effect of atorvastatinon li-pid metabolism, endothelial function, coagulationand inflammatory markers [29, 30]. The protocolswere approved by the medical ethics committees ofthe participating centres. All participants providedwritten informedconsent.

The inclusion and exclusion criteria for the Gronin-gen and DALI studies have been described in detailelsewhere [10, 29, 30]. Inbrief, theparticipants of theGroningen study were recruited by advertisement inlocal newspapers. Subjects with and without previ-ously diagnosed type 2 diabetes mellitus (definedaccording toWorldHealthOrganizationcriteria)wereincluded. Smokers and subjects using lipid-loweringdrugs, insulin and thiazolidones were excluded fromthe Groningen cohort. Patients with diabetes partici-pating in theDALI studyhadadiabetes durationof atleast 1 year, were free of clinically manifest cardio-vascular disease at entry and had a glycated haemo-globin (HbA1c) level <10%. When applicable, lipid-lowering drugs were withdrawn at least 8 weeks be-fore the start of the run-in phase. Premenopausalwomenwere excluded.Plasma lipid inclusion criteriawere levels of fasting total cholesterol between 4.0and 8.0 mmol L)1 and triglycerides between 1.5 and6.0 mmol L)1. Further exclusion criteria for partici-pation in either cohort were clinically manifest car-diac disease (history of myocardial infarction, coro-nary intervention or major ischaemia on anelectrocardiogram) and pregnancy. For the presentstudy, we included all subjects who had completedata with respect to obesitymeasures and PLTP genescore. Consequently, a total of 78 control subjectsand 67 patients with diabetes from the Groningencohort and 170 patients with diabetes from the DALIcohortwere included.

Blood pressure was measured using routine clini-cal procedures. Body mass index (BMI) was calcu-lated as the ratio between weight and heightsquared (in kg m)2). Waist circumference was mea-sured as the smallest girth between the rib cageand iliac crest, and waist ⁄hip ratio was measuredas the waist circumference divided by the hip cir-cumference. The revised national cholesterol edu-cation programme adult treatment panel III (NCEP-ATP-III) criteria were applied for classification ofMetS, and a waist circumference >102 cm for menand >88 cm for women [31] was used to indicate

R. P. F. Dullaart et al. | Diabetes interacts with obesity and PLTP variation to affect PLTP

ª 2011 The Association for the Publication of the Journal of Internal Medicine 491

Journal of Internal Medicine, 2012, 271; 490–498

central obesity. Venous blood was obtained afteran overnight fast.

Laboratory analyses

Venous bloodwas collected in ethylenediaminetetra-acetic acid (EDTA)-containing tubes (1.5 mg mL)1),which were placed on ice immediately. Plasma wasobtained by centrifugation at 1400 g for 15 min at4 �C. Plasma glucose and HbA1c were measuredshortly after blood collection. Plasma samples formeasurement of PLTPactivity, lipids andapolipopro-teins were stored at )80 �C until analysis. In bothstudies, total levels of cholesterol and triglyceridesweremeasured using routine enzymatic colorimetricassays, andHDLcholesterolwasmeasuredwith adi-rect homogeneousmethod [10,29].Non-HDLcholes-terol was calculated as the difference between totalandHDL cholesterol.High-sensitivity C-reactive pro-tein (CRP) was assayed by nephelometry (Dade Beh-ring, Marburg, Germany) in the Groningen cohort[22] and by enzyme immunoassay (Dako, Copenha-gen,Denmark) in theDALI study [32]. ApolipoproteinE (apoE) was measured by nephelometry (Wako,Osaka,Japan) [30].

Plasma PLTP activity was measured with a phos-pholipid vesicles–HDL system, using [14C]-labelleddipalmitoyl phosphatidylcholine as previously de-scribed [10, 32, 33]. Briefly, plasma samples (1 lL)were incubated with [14C]-labelled phosphatidylcho-line vesicles and excess pooled normal HDL for45 min at 37 �C. The method is specific for PLTPactivity; the phospholipid transfer-promoting prop-erty of cholesteryl ester transfer protein (CETP) doesnot interfere with the assay. Levels of plasma PLTPactivity vary linearly with the amount of plasmaadded to the incubation system.PLTPactivitywas re-lated to the activity in human reference pooled plas-ma andwas expressed in arbitrary units (AU; 100AUcorresponds to 13.6 lmol phosphatidylcholinetransferred per mL per hour). The inter-assay coeffi-cient of variation of themeasurement of PLTPactivityis5%.

Variation in PLTPwas determined using a gene scorethat is based on two common PLTP-tagging SNPs[rs378114 (c.330)117G>A) and rs6065904(c.705+256C>T)], as described in detail elsewhere[16]. ThePLTP score represents thenumber (from0 to4) of plasma PLTP activity-decreasing alleles (forrs378114: presence of 0, 1 or 2 G alleles; forrs6065904: presence of 0, 1 or 2 T alleles) and hasbeen found to inversely predict plasma PLTP activity

in both the Groningen and the DALI cohorts [16].A higher PLTP gene score is also linearly associatedwith less PLTP mRNA expression in human liverspecimens [16].

Statistical analysis

spss 18 was used for data analysis. Results are ex-pressed as mean ± SD or median (interquartilerange) unless stated otherwise. Differences in vari-ables between diabetic and nondiabetic subjectswere determined by unpaired t-tests or Mann–Whit-ney U tests where appropriate. Differences betweendiabetic and nondiabetic subjects with and withoutan enlarged waist circumference were determined byone-way analysis of variance (anova) with subse-quent Bonferroni correction for multiple compari-sons. Logarithmically transformed triglyceride val-ues were used in regression analyses. Between-group differences in proportions were assessed bychi-square analysis. Multivariate linear regressionanalysis was applied to determine differences be-tween diabetic and control subjects after adjustmentfor age and sex. Multivariate linear regression analy-ses were also carried out to determine the indepen-dent contribution of variables to plasma PLTP activ-ity. Additionally, interaction terms between diabetesstatus,HbA1c,PLTP gene score andobesity variableswere calculated. To this end, a distributionof centredto the mean was made for continuous variables bysubtracting the individual value of the variable ofinterest from its groupmean value. Interaction termswere considered to be statistically significant at two-sidedP-values<0.1 [34].Otherwise, the levelof signif-icancewassetat two-sidedP < 0.05.

Results

A total of 237 patients with diabetes and 78 controlsubjects were included in the study. Patients withdiabetic were older, had higher systolic and diastolicblood pressure and were more obese than controlsubjects, but the male ⁄ female ratio was similar be-tween thegroups (Table1).Averagediabetesdurationwas 9 ± 7 years. Fifty-two patients with diabetes(22%) were current smokers (P < 0.001 compared tocontrol subjects, because of exclusion of smokers inthe Groningen study). Nineteen of the patients withdiabetes were treated with diet alone. Oral blood glu-cose-lowering drugs alone (metformin and ⁄or sulfo-nylurea) were used by 130 patients, whereas 43 pa-tients used insulin alone and 45 used insulin incombination with oral blood glucose-lowering drugs(mainlymetformin). A total of210 (89%)patientswith




diabetes fulfilled the criteria forMetS compared to 21control subjects (27%,P < 0.001).

Asshown inTable1,systolicanddiastolicbloodpres-sure, BMI, waist circumference, waist ⁄hip ratio, fast-ingplasmaglucose,HbA1candCRP levelswerehigh-er in patients with diabetes than in control subjects.Plasmatotal cholesterolwas similar betweendiabeticand control subjects. Non-HDL cholesterol, triglyce-rides and apoE levelswere higher, whereasHDL cho-lesterol was lower in patients with diabetes (Table 1).All these differences remained significant afteradjustment for age, sex and smoking. Plasma PLTPactivity was on average 13% higher in patients withdiabetes; this difference remained significant after

adjustment for age, sex, smoking and study cohort(P < 0.001). The distribution of PLTP gene score wasnot different in diabetic compared to control subjects(Table 1; median gene score (interquartile range): 2(1–3) inbothdiabeticandcontrol subjects).

As shown in Fig. 1, plasma PLTP activity was higherin patients with diabetes with an enlarged waist(n = 174), compared to patients with diabetes with anormal waist circumference (n = 63), controlsubjects with an enlarged waist (n = 18) and controlsubjects with a normal waist circumference (n = 60)(anova, P < 0.001). These differences remained sig-nificant after adjustment for age, sex, smoking andstudy cohort (anova, P < 0.001), after additional

Table 1 Clinical characteristics, plasma lipids, phospholipid transfer protein (PLTP) activityandvariation inPLTP in237 type2dia-

beticand78controlsubjects

Type2diabetic

subjects (n = 237)

Control

subjects (n = 78) P-value P-value*

Age (years) 59 ± 8 56 ± 9 0.003

Sex (m ⁄ f) 129 ⁄108 44 ⁄34 0.87

Systolicbloodpressure (mmHg) 151 ± 21 130 ± 18 <0.001 <0.001

Diastolicbloodpressure (mmHg) 87 ± 9 82 ± 10 <0.001 <0.001

BMI (kgm)2) 30.2 ± 5.1 26.0 ± 4.1 <0.001 <0.001

Waist circumference (cm) 105 ± 14 90 ± 13 <0.001 <0.001

Enlargedwaist 174 (73%) 18(23%) <0.001

Waist ⁄hipratio 0.98 ± 0.10 0.90 ± 0.08 <0.001 <0.001

Glucose (mmolL)1) 10.0 ± 3.1 5.7 ± 1.0 <0.001 <0.001

HbA1c(%) 7.9 ± 1.3 5.4 ± 0.4 <0.001 <0.001

CRP(mgL)1) 2.77 (1.29–5.43) 1.36 (0.51–2.61) <0.001 <0.001

Total cholesterol (mmolL)1) 5.80 ± 0.96 5.70 ± 1.03 0.43 0.86

Non-HDLcholesterol (mmolL)1) 4.70 ± 0.99 4.22 ± 1.11 <0.001 0.004

HDLcholesterol (mmolL)1) 1.11 ± 0.31 1.48 ± 0.42 <0.001 <0.001

Triglycerides (mmolL)1) 2.20 (1.76–2.97) 1.26 (0.89–2.02) <0.001 <0.001

ApoE (gL)1) 0.041 ± 0.011 0.040 ± 0.010 0.027 0.021

PLTPactivity (AU) 107 ± 17 95 ± 11 <0.001 <0.001

PLTPgenescore 0:n = 16 0:n = 1 0.40

1:n = 59 1:n = 20

2:n = 92 2:n = 30

3:n = 51 3:n = 21

4:n = 19 4:n = 6

BMI, body mass index; CRP, high-sensitivity C-reactive protein; HDL, high-density lipoprotein; SNPs, single-nucleotidepolymorphisms.Dataaregivenasmean (SD),median (interquartile range)ornumber (%).*P-values fordifferencebetweentype2diabeticandcontrol subjectsadjusted forage,sexandsmoking.PLTPgenescore:derivedfromtwotaggingSNPs (rs378114andrs6065904)associatedwithplasmaPLTPactivity.




adjustment for PLTP gene score (anova, P < 0.001)andalso after excluding the88patientswithdiabetesusing insulin (n = 227,anova,P < 0.001).

Multiple linear regression analysis was carried out todetermine the extent to which plasma PLTP activitywas governed by PLTP gene score, diabetes statusand obesity. In initial models that included eitherBMI orwaist circumference, itwas found that plasmaPLTP activity was related to BMI (b = 0.179,P = 0.002) or waist circumference (b = 0.255,P < 0.001), independently of the presence of diabetes(P < 0.001,datanot shown).WhenBMIandwaist cir-cumference were included together, PLTP activitywas found to be determined by waist circumferencealone (b = 0.258, P = 0.002) and not by BMI(b = 0.01, P = 0.95). Therefore, waist circumferencewasselectedas theprimaryobesityvariable forevalu-atingdiabetes–obesity interactions to impactonplas-ma PLTP. As shown in Table 2, plasma PLTP activitywas independentlyand inversely related toPLTPgenescoreandpositively related todiabetes statusanden-larged waist circumference after adjustment for age,sex, smoking, CRP and study cohort (model 1). Wenext examined whether the presence of diabetesmodified the relationship between plasma PLTPactivity and enlarged waist circumference. As shownin model 2, there was a positive interaction betweendiabetes status and waist circumference on PLTPactivity.WhenHbA1c, non-HDL cholesterol, triglyce-rides andapoE levelswere added to themodels, PLTP

activity remained independently related to enlargedwaist (model 3), and there was still an interaction be-tween diabetes status and waist circumference onPLTP activity (model 4). When patients with diabetesusing insulin were excluded, the interaction betweendiabetes status and waist circumference on PLTPactivity also remained significant (b = 0.213, P =0.013 and b = 0.180, P = 0.038 without and withadditional adjustment for HbA1c, non-HDL choles-terol, triglycerides and apoE, respectively; data notshown). In further analyses, there was also an inter-action between diabetes status and waist ⁄hip ratioonPLTPactivity (model 2:b = 0.219,P < 0.001;mod-el4:b = 0.215,P = 0.001,datanotshown).

Finally, we determined whether diabetes status andthe degree of chronic hyperglycaemia, as reflected bytheHbA1c level, interactedwithPLTPgenescore toaf-fect plasma PLTP activity (Table 3). Remarkably, dia-betes status showed an interaction with PLTP genescore on PLTP activity after adjustment for age, sex,smoking, CRP, study cohort and enlarged waist cir-cumference (model 1), as well as after additionaladjustment for non-HDL cholesterol, triglyceridesand apoE (model P < 0.001 compared to control sub-jects, because of exclusion of smokers in the Gronin-gen study3). Thus, thepresence of diabetes amplifiedtheassociationbetweenplasmaPLTPactivityand thenumber of PLTP activity-decreasing alleles. Thisinteraction was essentially unaltered after exclusionof patients with diabetes using insulin (model 1:b = )0.449, P = 0.085; model 3: b = )0.418, P =0.098, data not shown). Likewise, interactionsbetween the HbA1c level and the PLTP gene score onPLTPactivitywereobserved inanalysiswithout (mod-el 2; graphically depicted in Fig. 2) andwith (model 4)additional adjustment for non-HDL cholesterol, tri-glyceridesandapoE.

Discussion

In the present study, we have demonstrated for thefirst time that the presence of diabetes strengthensthe relationship between plasma PLTP activity andenlarged waist circumference independently of theinfluence of common PLTP variations. Furthermore,our results suggest that the association betweenplasma PLTP activity and variations in PLTP is modi-fied by the diabetic state, meaning that plasma PLTPactivity is more strongly affected in individuals withdiabetes mellitus for a given PLTP gene score. Suchan interaction was also observed when using theHbA1c level as ameasure of the degree of chronic hy-perglycaemia. Taken together, our findings support

120

P < 0.001

P = 0.021

P < 0.001

80

100

PLT

P a

ctiv

ity (

AU

)

0A B C D

Fig. 1 Plasma phospholipid transfer protein activity accord-ing to diabetes status and enlarged waist circumference.Data aremeans ± SEM. (a) Nondiabetic subjectswith normalwaist (n = 60); (b) nondiabetic subjects with enlarged waist(n = 18); (c) type 2 diabetic subjects with normalwaist (n = 63); (d) type 2 diabetic subjects with enlargedwaist (n = 174). anova,P < 0.001.




the hypothesis that diabetes–environment and dia-betes–gene interactions are involved in the regulationofcirculatingPLTPactivity.

In this study, plasmaPLTPactivitywasmore stronglyrelated to waist circumference than to BMI. Of note,bothBMIandwaist circumferencearecorrelatedwithvisceral abdominal as well as with subcutaneousabdominal fat tissue [35]. Therefore, these obesitymeasures do not fully discriminate between relation-ships between plasma PLTP activity and either obes-ity per se or a centrally distributed adiposity pattern.Of further note, PLTP is expressed in visceral as wellas in subcutaneous adipose tissue, although PLTPmRNA has been found to be related positively to BMIin subcutaneousbut not in visceral adipose tissue [3,36].Nonetheless, amorecentrallydistributedobesity

patternmayberegardedasamajordriving force lead-ing to increasing free fatty acid delivery to the liverand hence to disturbances in hepatic lipid homoeo-stasis [37], which are likely to include elevations inplasma PLTP activity [22]. We did not specifically as-sess visceral and subcutaneous fat areas. Yet, weconsider the present finding that the diabetic stateinteracted positively with waist circumference to af-fectPLTPactivity robust,as this interactionwas inde-pendent of plasma CRP, which is known to be ele-vated in (central) obesity [38], and was replicatedusingthewaist ⁄hipratioasameasureofobesity.

The present observation that the diabetic state and,in alternative analyses, the HbA1c level interact withthe PLTP gene score to modulate circulating PLTPactivity clearly supports the hypothesis that the ex-

Table 2 Multiple linear regressionmodelsshowing independentrelationshipsbetweenplasmaphospholipid transferprotein (PLTP)

activity and PLTP gene score, diabetes status and waist circumference, and interaction between diabetes status and enlarged

waist in237patientswith type2diabetesand78control subjects

Model1 Model2 Model3 Model4

b P-value b P-value b P-value b P-value

Age )0.120 0.019 )0.098 0.057 )0.10 0.053 )0.087 0.094

Sex (menvs.women) )0.073 0.162 )0.137 0.015 )0.067 0.21 )0.117 0.043

Smoking (yes ⁄no) 0.127 0.022 0.123 0.025 0.128 0.021 0.122 0.028

LnCRP 0.035 0.54 0.01 0.94 0.001 0.99 )0.022 0.71

DALIvs.Groningencohort )0.152 0.026 )0.129 0.058 )0.011 0.90 0.002 0.98

PLTPgenescore )0.280 <0.001 )0.284 <0.001 )0.265 <0.001 )0.271 <0.001

Diabetesmellitus (yes ⁄no) 0.155 0.021 0.173 0.010 0.101 0.17 0.113 0.125

Enlargedwaist circumference (yes ⁄no) 0.208 0.001 0.076 0.32 0.176 0.006 0.070 0.38

Diabetes–waistcircumference interaction 0.200 0.005 0.162 0.028

HbA1c 0.174 0.040 0.183 0.030

Non-HDLcholesterol )0.021 0.74 )0.026 0.68

Lntriglycerides 0.059 0.44 0.039 0.61

ApoE 0.184 0.004 0.189 0.003

CRP, high-sensitivity C-reactive protein; DALI, Diabetes Atorvastatin Lipid Intervention; HDL, high-density lipoprotein; SNPs,single-nucleotidepolymorphisms.PLTP gene score: derived from two tagging SNPs (rs378114and rs6065904) associatedwith plasmaPLTPactivity. b: standard-ized regression coefficient. Logarithmically transformed values of triglycerides and C-reactive protein (CRP) levels are used intheanalyses.Allmodelsareadjusted forage, sex, smoking,CRPandstudycohort (DALIcohort,Groningencohort).OtherstatisticaldeterminantsofPLTPactivity included inthemultivariate linear regressionanalyses:Model1:diabetesstatus,enlargedwaistcircumference,PLTPgenescore;Model 2: diabetes status, enlargedwaist circumference,PLTP gene score and interaction betweenwaist circumference anddia-betesstatus;Model3:diabetesstatus,enlargedwaistcircumference,PLTPgenescore, glycatedhaemoglobin (HbA1c),non-high-density lipo-protein (HDL)cholesterol, ln triglyceridesandapolipoproteinE (apoE);Model 4: diabetes status, enlarged waist circumference, PLTP gene score, HbA1c, non-HDL cholesterol, ln triglycerides, apoEand interactionbetweenwaist circumferenceanddiabetesstatus.




tent to which genetic variation in PLTP affects circu-latingPLTPactivity levelsmaybeamplifiedby the de-greeof chronichyperglycaemia in vivo. ThePLTP genescore used here is strongly associated with hepaticmRNA expression [16]. Earlier studies have demon-strated that high glucose stimulates PLTP promoteractivity in vitro [28]. Accordingly, PLTPmRNA expres-sion is modulated by nuclear receptors, such as theperoxisome proliferator-activated receptor, the liverX-receptor and the farnesoid X-receptor [28, 39, 40].In the present study, we did not aim to delineate themechanisms whereby exposure to hyperglycaemiamaymodifyPLTPexpression.

Several other aspects of our study should be ad-dressed. In view of the possibility that PLTPactivity ishigher in cigarette smokers [41], and given the exclu-

sion of smokers in the Groningen cohort, we tookaccount of smoking status in the multivariableanalyses. Indeed, current smoking was found to beassociated with higher plasma PLTP levels in thisstudy. Furthermore, the notion that PLTP haspro-inflammatoryproperties [8, 42]providesanaddi-tional argument to evaluate whether or not interac-tions between diabetes and obesity on circulatingPLTP activity are confounded by enhanced chronicsubclinical inflammation. We also included the plas-ma level of apoE, which has been considered to beinvolved in activating PLTP [30, 43] and in stimulat-ing theabilityofPLTPtoremodelHDLparticles [44].Astrong correlation between plasma PLTP activity andapoE was observed that was independent of plasmaapoB-containing lipoproteins and obesity. Of note,interactionsbetweendiabetes statusandobesity and

Table 3 Multiple linear regressionmodels showing relationshipsbetweenplasmaphospholipid transfer protein (PLTP) activity and

PLTP gene score, diabetes status and HbA1c, and interactions between PLTP gene score and diabetes status and HbA1c in 78

nondiabeticpatientsand237type2diabeticsubjects

Model1 Model2 Model3 Model4

b P-value b P-value b P-value b P-value

Age )0.125 0.014 )0.130 0.027 )0.106 0.042 )0.106 0.041

Sex (menvs.women) )0.072 0.167 )0.075 0.130 )0.066 0.22 )0.068 0.21

Smoking (yes ⁄no) 0.126 0.022 0.127 0.020 0.122 0.027 0.125 0.024

LnCRP 0.047 0.41 0.025 0.66 0.021 0.71 0.007 0.90

DALIvs.Groningencohort )0.152 0.026 )0.064 0.83 )0.080 0.28 )0.004 0.96

PLTPgenescore 0.153 0.49 )0.272 <0.001 0.104 0.64 )0.263 <0.001

Diabetesmellitus (yes ⁄no) 0.370 0.004 0.101 0.14 0.356 0.006 0.104 0.16

HbA1c 0.393 0.002 0.355 0.008

Diabetes–PLTPgenescore interaction )0.485 0.046 )0.423 0.084

HbA1c–PLTPgenescore interaction )0.240 0.035 )0.203 0.078

Enlargedwaistcircumference (yes ⁄no) 0.209 0.001 0.196 0.001 0.192 0.002 0.182 0.004

Non-HDLcholesterol )0.006 0.92 )0.015 0.81

Lntriglycerides 0.064 0.41 0.059 0.44

ApoE 0.178 0.005 0.177 0.005

DALI,DiabetesAtorvastatinLipid Intervention;HDL,high-density lipoprotein;SNPs, single-nucleotidepolymorphisms.PLTP gene score: derived from two tagging SNPs (rs378114and rs6065904) associatedwith plasmaPLTP activity. b: standard-ized regression coefficient. Logarithmically transformed values of triglycerides and C-reactive protein (CRP) levels are used intheanalyses.Allmodelsareadjusted forage,sex, smoking,CRPandstudycohort (DALIcohort,Groningencohort).OtherstatisticaldeterminantsofPLTPactivity included in themultivariate linear regressionanalyses:Model 1:PLTPgenescore, diabetes status, enlargedwaist circumference and interactionbetweenPLTP gene score anddiabetesstatus.Model2:PLTPgenescore,diabetesstatus,HbA1c, enlargedwaist circumferenceand interactionofPLTPgenescorewithHbA1c.Model 3: PLTP gene score, diabetes status, enlargedwaist circumference, non-HDL cholesterol, ln triglycerides, apolipoproteinE (apoE)and interactionbetweenPLTPgenescoreanddiabetesstatus.Model4:PLTPgenescore,diabetesstatus,glycatedhaemoglobin (HbA1c), enlargedwaistcircumference,non-high-density lipo-protein (HDL)cholesterol, ln triglycerides,apoEand interactionbetweenPLTPgenescoreandHbA1c.




betweendiabetes statusandPLTPgenescore toaffectplasma PLTP activity were not essentially influencedby inclusionofapoEinthemultivariableanalysis.

The current findings are based on an analysis of twocohorts combined. In the Groningen cohort, insulin-treated patients with diabetes were excluded [10],whereas a considerable number of DALI participantsused insulin [29]. Furthermore,mild tomoderate hy-pertriglyceridaemia was a selection criterion for theDALI study [29], which probably explains the highprevalence of enlarged waist circumference amongpatientswithdiabetes. For these reasons,we tookac-count of the origin of the study cohort inallmultivari-able analyses. In this regard, it is relevant that allinteractions remained essentially unaltered afterexclusion of insulin-treated patients. Moreover, theanalyses concerning PLTP gene variation that weused here are based on a recently described PLTPgene score [16]. This approach was chosen in theexpectation that statistical power to detect diabetesstatus–PLTP gene and HbA1c–PLTP interactions wasbetter compared to that using each PLTP gene varia-tion separately. Finally, it should be appreciated thatthe cross-sectional design is a limitation of the pres-entstudy.

In conclusion, our findings support the hypothesisthat diabetes–environment and diabetes–gene inter-

actions are involved in the regulation of plasmaactivity of PLTP, an emerging cardiometabolic riskfactor. The present observations impact on the sup-position that improvingmetabolic control andweightreductionmay coordinately and perhaps even syner-gistically affect lipoprotein metabolism via PLTPregulation.

Conflict of interest statement

Noconflictsof interestweredeclared.

References

1 van Tol A. Phospholipid transfer protein.Curr Opin Lipidol2002;

13:135–9.

2 Vergeer M, Dallinga-Thie GM, Dullaart RPF, van Tol A. Evalua-

tion of phospholipid transfer protein as a therapeutic target.

FutLipidol2008;3:327–35.

3 Tzotzas T, Desrumaux C, Lagrost L. Plasma phospholipid trans-

fer protein (PLTP): review of an emerging cardiometabolic risk

factor.ObesRev2009;10:403–11.

4 Borggreve SE, de Vries R, Dullaart RPF. Alterations in high-den-

sity lipoprotein metabolism and reverse cholesterol transport in

insulin resistance and type 2 diabetes mellitus: role of lipolytic

enzymes, lecithin:cholesterol acyltransferase and lipid transfer

proteins.EurJClin Invest2003;33:1051–69.

5 JiangXC,QinS,QiaoC etal.ApolipoproteinBsecretion andath-

erosclerosis are decreased in mice with phospholipid-transfer

proteindeficiency.NatMed2001;7:847–52.

6 Lie J, de CromR, vanGent T et al. Elevation of plasma phospho-

lipid transfer protein in transgenic mice increases VLDL secre-

tion.JLipidRes2002;43:1875–80.

7 Schneider M, Verges B, Klein A et al. Alterations in plasma vita-

min E distribution in type 2 diabetic patients with elevated plas-

ma phospholipid transfer protein activity. Diabetes 2004; 53:

2633–9.

8 CheungMC, Vaisar T,HanX,Heinecke JW, Albers JJ. Phospho-

lipid transfer protein in human plasma associates with proteins

linked to immunity and inflammation. Biochemistry 2010; 49:

7314–22.

9 Schlitt A, Bickel C, Thumma P et al. High plasma phospholipid

transfer protein levels asa risk factor for coronaryarterydisease.

ArteriosclerThrombVascBiol2003;23:1857–62.

10 de Vries R, Dallinga-Thie GM, Smit AJ, Wolffenbuttel BHR, van

Tol A, Dullaart RPF. Elevated plasma phospholipid transfer pro-

teinactivity isadeterminantofcarotid intima-media thickness in

type2diabetesmellitus.Diabetologia2006;49:398–404.

11 Schgoer W, Mueller T, Jauhiainen M et al. Low phospholipid

transferprotein (PLTP) isa risk factor forperipheral atherosclero-

sis.Atherosclerosis2008;196:219–26.

12 Schlitt A,BlankenbergS,BickelC etal.PLTPactivity is a risk fac-

tor for subsequent cardiovascular events in CAD patients under

statin therapy: the AtheroGene study. J Lipid Res 2009; 50:

723–9.

13 Engler MB, Pullinger CR, Malloy MJ et al. Genetic variation in

phospholipid transfer protein modulates lipoprotein profiles in

hyperalphalipoproteinemia.Metabolism2008;57:1719–24.

14 Bosse Y, Bouchard L, Despres JP, Bouchard C, Perusse L, Vohl

MC. Haplotypes in the phospholipid transfer protein gene are

130

120

110

100

90

800

1PLTP gene score

PLT

P a

ctiv

ity (

AU

)

23

4

4.50 – 5.50

HbA

1c (%

) cat

egor

y5.60 – 6.70

6.80 – 8.00

8.10 – 9.20

9.30 – 12.20

Fig.2 Graphical presentation of the interaction between theHbA1c level (presented in categories) and PLTP gene score[number of phospholipid transfer protein (PLTP) activity-decreasing alleles] on plasma PLTP activity (data derivedfrom Table 3, model 2; interaction term: b = )0.240, P =0.035).




associated with obesity-related phenotypes: the Quebec Family

Study. Int JObes (Lond)2005;29:1338–45.

15 Jarvik GP, Rajagopalan R, Rosenthal EA et al.Genetic and non-

genetic sources of variation in phospholipid transfer protein

activity.JLipidRes2010;51:983–90.

16 Vergeer M, Boekholdt SM, SandhuMS et al.Genetic variation at

the phospholipid transfer protein locus affects its activity and

high-density lipoprotein size and is a novel marker of cardiovas-

culardiseasesusceptibility.Circulation2010;122:470–7.

17 Riemens S, van Tol A, Sluiter W, Dullaart RPF. Elevated plasma

cholesteryl ester transfer in NIDDM: relationships with apolipo-

protein B-containing lipoproteins and phospholipid transfer

protein.Atherosclerosis1998;140:71–9.

18 Riemens SC, van Tol A, Sluiter WJ, Dullaart RPF. Plasma phos-

pholipid transfer protein activity is lowered by 24-h insulin and

acipimox administration: blunted response to insulin in type 2

diabeticpatients.Diabetes1999;48:1631–7.

19 Riemens SC, van Tol A, Sluiter WJ, Dullaart RPF. Plasma phos-

pholipid transfer protein activity is related to insulin resistance:

impaired acute lowering by insulin in obese Type II diabetic

patients.Diabetologia1998;41:929–34.

20 KaserS,SandhoferA, FogerB etal. Influenceof obesity and insu-

lin sensitivity on phospholipid transfer protein activity. Diabeto-

logia2001;44:1111–7.

21 Dullaart RPF, de Vries R, Dallinga-Thie GM, van Tol A, Sluiter

WJ.Plasmacholesterylester transferproteinmassandphospho-

lipid transfer protein activity are associated with leptin in type 2

diabetesmellitus.BiochimBiophysActa2007;1771:113–8.

22 Dullaart RPF, de Vries R, Dallinga-Thie GM, Sluiter WJ, van Tol

A. Phospholipid transfer protein activity is determined by type 2

diabetes mellitus and metabolic syndrome, and is positively

associated with serum transaminases. Clin Endocrinol (Oxf)

2008;68:375–81.

23 Riemens SC, Van Tol A, Sluiter WJ, Dullaart RPF. Acute and

chronic effects of a 24-hour intravenous triglyceride emulsion

challenge on plasma lecithin: cholesterol acyltransferase, phos-

pholipid transfer protein, and cholesteryl ester transfer protein

activities.JLipidRes1999;40:1459–66.

24 Murdoch SJ, KahnSE, Albers JJ, Brunzell JD, Purnell JQ. PLTP

activity decreases with weight loss: changes in PLTP are associ-

ated with changes in subcutaneous fat and FFA but not IAF or

insulinsensitivity.JLipidRes2003;44:1705–12.

25 LaimerMW,EnglJ,TschonerAetal.Effectsofweight lossonlipid

transfer proteins in morbidly obese women. Lipids 2009; 44:

1125–30.

26 Tzotzas T, Dumont L, Triantos A et al.Early decreases in plasma

lipid transfer proteins during weight reduction. Obesity (Silver

Spring)2006;14:1038–45.

27 Oomen PH, van Tol A, Hattori H, Smit AJ, Scheek LM, Dullaart

RPF.Humanplasmaphospholipid transfer protein activity is de-

creased by acute hyperglycaemia: studies without and with hy-

perinsulinaemia in Type 1 diabetes mellitus. Diabet Med 2005;

22:768–74.

28 Tu AY, Albers JJ. Glucose regulates the transcription of human

genes relevant to HDL metabolism: responsive elements for

peroxisome proliferator-activated receptor are involved in the

regulation of phospholipid transfer protein. Diabetes 2001; 50:

1851–6.

29 Diabetes Atorvastin Lipid Intervention (DALI) Study Group. The

effect of aggressive versusstandard lipid loweringbyatorvastatin

on diabetic dyslipidemia: the DALI study: a double-blind, ran-

domized, placebo-controlled trial in patientswith type 2diabetes

anddiabeticdyslipidemia.DiabetesCare2001;24:1335–41.

30 Dallinga-Thie GM, van Tol A, Hattori H, Rensen PC, Sijbrands

EJ. Plasmaphospholipid transfer protein activity is decreased in

type 2 diabetes during treatment with atorvastatin: a role for

apolipoproteinE?Diabetes2006;55:1491–6.

31 GrundySM,CleemanJI, Daniels SR et al.; AmericanHeart Asso-

ciation;NationalHeart,Lung,andBloodInstitute.Diagnosisand

management of the metabolic syndrome: an American Heart

Association ⁄ National Heart, Lung, and Blood Institute Scientific

Statement.Circulation2005;112:2735–52.

32 vandeReeMA,HuismanMV, PrincenHM,MeindersAE, KluftC;

DALI-StudyGroup.StrongdecreaseofhighsensitivityC-reactive

proteinwithhigh-doseatorvastatin inpatientswith type2diabe-

tesmellitus.Atherosclerosis2003;166:129–35.

33 Speijer H, Groener JE, van Ramshorst E, van Tol A. Different

locations of cholesteryl ester transfer protein and phospholipid

transfer protein activities in plasma. Atherosclerosis 1991; 90:

159–68.

34 Selvin S. Statistical Analysis of Epidemiological Data. New York:

OxfordUniversitypress,1996.

35 Fox CS, Massaro JM, Hoffmann U et al. Abdominal visceral and

subcutaneous adipose tissue compartments: association with

metabolic risk factors in the Framingham Heart Study. Circula-

tion2007;116:39–48.

36 Dusserre E, Moulin P, Vidal H. Differences in mRNA expression

of the proteins secreted by the adipocytes in human subcutane-

ous and visceral adipose tissues. Biochim Biophys Acta 2000;

1500:88–96.

37 Frayn KN, Williams CM, Arner P. Are increased plasma non-

esterified fatty acid concentrations a risk marker for coronary

heart disease and other chronic diseases? Clin Sci (Lond) 1996;

90:243–53.

38 Schmidt MI, Duncan BB. Diabesity: an inflammatory metabolic

condition.ClinChemLabMed2003;41:1120–30.

39 Cao G, Beyer TP, Yang XP et al. Phospholipid transfer protein is

regulated by liver X receptors in vivo. J Biol Chem 2002; 277:

39561–5.

40 Goldstein JL, BrownMS, Liang G. LXR-SREBP-1c-phospholipid

transfer protein axis controls very low density lipoprotein (VLDL)

particle size.JBiolChem2010;285:6801–10.

41 Dullaart RPF, Hoogenberg K, Dikkeschei BD, van Tol A. Higher

plasma lipid transfer protein activities and unfavorable lipopro-

tein changes in cigarette-smoking men. Arterioscler Thromb

1994;10:1581–5.

42 TanKC, ShiuSW,WongY, TamS. Plasmaphospholipid transfer

protein activity and subclinical inflammation in type 2 diabetes

mellitus.Atherosclerosis2005;178:365–70.

43 Janis MT, Metso J, Lankinen H et al. Apolipoprotein E activates

the low-activity form of human phospholipid transfer protein.

BiochemBiophysResCommun2005;331:333–40.

44 Settasatian N, Barter PJ, Rye KA. Remodeling of apolipoprotein

E-containing spherical reconstituted high density lipoproteins

byphospholipidtransferprotein.JLipidRes2008;49:115–26.

Correspondence:R.P.F.Dullaart,MD, PhD,Department ofEndocri-

nology, University Medical Centre Groningen, PO Box 30001,

9700RBGroningen,TheNetherlands.

(fax:+31503619392;e-mail: [email protected]).




Copyright of Journal of Internal Medicine is the property of Wiley-Blackwell and its content may not be copied

or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.

However, users may print, download, or email articles for individual use.

74463150

Documents