Nutrition, Metabolism & Cardiovascular Diseases (2014) 24, 760e766

Available online at www.sciencedirect.com

Nutrition, Metabolism & Cardiovascular Diseases

j ournal homepage: www.e lsev ier .com/ locate/nmcd

B-vitamin levels and genetics of hyperhomocysteinemia are notassociated with arterial stiffness

S.C. van Dijk a,*, A.W. Enneman a, J. van Meurs a, K.M.A. Swart b, A.H. Ham a,J.P. van Wijngaarden c, E.M. Brouwer-Brolsma c, N.L. van der Zwaluw c,N.M. van Schoor b, R.A.M. Dhonukshe-Rutten c, L.C.P.G.M. de Groot c, P. Lips b,A.G. Uitterlinden a,d,e, H. Blom f, J.M. Geleijnse c, E. Feskens c, R.T. de Jongh g,Y.M. Smulders g,h, A.H. van den Meiracker a, F.U.S. Mattace-Raso a, N. van der Velde a,i

a Erasmus Medical Center, Department of Internal Medicine, Section of Geriatrics, P.O. Box 2040, 3000 CA Rotterdam, the Netherlandsb VU University Medical Center, EMGO Institute for Health and Care Research, Department of Epidemiology and Biostatistics, Amsterdam,the NetherlandscWageningen University, Division of Human Nutrition, Wageningen, the NetherlandsdNetherlands Consortium of Healthy Ageing, Rotterdam, the NetherlandseNetherlands Consortium of Healthy Ageing, Leiden, the Netherlandsf VU University Medical Center, Department of Clinical Chemistry, Metabolic Unit, Amsterdam, the Netherlandsg VU University Medical Center, Department of Internal Medicine, Amsterdam, the Netherlandsh Institute for Cardiovascular Research ICaR-VU, Amsterdam, the Netherlandsi Academic Medical Center, Department of Internal Medicine, Section of Geriatrics, Amsterdam, the Netherlands

Received 12 August 2013; received in revised form 11 January 2014; accepted 20 January 2014Available online 31 January 2014

KEYWORDSHomocysteine;Arterial stiffness;Pathophysiology;B-vitamin status;Genetics

* Corresponding author. Tel.: þ31 10 7E-mail address: s.c.vandijk@erasmusm

0939-4753/$ - see front matter ª 2014 Elseviehttp://dx.doi.org/10.1016/j.numecd.2014.01.008

Abstract Background and aims: Hyperhomocysteinemia is associated with arterial stiffness, butunderlying pathophysiological mechanisms explaining this association are to be revealed. Thisstudy was aimed to explore two potential pathways concerning the one-carbon metabolism. Apotential causal effect of homocysteine was explored using a genetic risk score reflecting an in-dividual’s risk of having a long-term elevated plasma homocysteine level and also associationswith B-vitamin levels were investigated.Methods and results: Baseline cross-sectional data of the B-PROOF study were used. In the cardio-vascular subgroup (n Z 567, 56% male, age 72.6 � 5.6 yrs) pulse wave velocity (PWV) was deter-mined using applanation tonometry. Plasma concentrations of vitamin B12, folate,methylmalonic acid (MMA) and holo transcobalamin (holoTC) were assessed and the genetic riskscore was based on 13 SNPs being associated with elevated plasma homocysteine. Associationswere examined using multivariable linear regression analysis. B-vitamin levels were not associ-ated with PWV. The genetic risk score was also not associated with PWV. However, the homo-cysteineegene interaction was significant (p < 0.001) in the association of the genetic riskscore and PWV. Participants with the lowest genetic risk of having long-term elevated homocys-teine levels, but with higher measured homocysteine levels, had the highest PWV levels.Conclusion: Homocysteine is unlikely to be causally related to arterial stiffness, because therewas no association with genetic variants causing hyperhomocysteinemia, whereas non-genetically determined hyperhomocysteinemia was associated with arterial stiffness. Moreover,the association between homocysteine and arterial stiffness was not mediated by B-vitamins.Possibly, high plasma homocysteine levels reflect an unidentified factor, that causes increasedarterial stiffness.ª 2014 Elsevier B.V. All rights reserved.

0 35979; fax: þ31 10 70 34768.c.nl (S.C. van Dijk).

r B.V. All rights reserved.

Genetics of hyperhomocysteinemia 761

Introduction

Figure 1 Simplified figure of the one-carbon metabolism. Vitamins,metabolites and enzymes which are involved in the one carbonmetabolism. Abbreviations: THF: tetrahydrofolate; SAM: S-adeno-sylmethionine; SAH: S-adenosylhomocysteine; MTHFR: methylenete-trahydrofolate reductase; MMA: methylmalonic acid; MMA-CoA:methylmalonyl-CoA; Suc-CoA: succinyl-CoA; FA: folic acid; B12:vitamin B12; B2: vitamin B2; B6: vitamin B6.

Hyperhomocysteinemia has been shown to be an impor-tant cardiovascular risk indicator, especially in the oldestold [4]. Arterial stiffness is considered to be a pre-clinicalstate of cardiovascular disease and recently we have re-ported an association between plasma homocysteine leveland arterial stiffness within an older, mild hyper-homocysteinemic population [20]. Whether this associa-tion is truly causal is not yet known, in particular sincetrials with B-vitamin supplementation aimed to reduceplasma homocysteine concentrations failed to demon-strate beneficial effects on cardiovascular outcomes [3].

A causal effect of homocysteine on the arterial stiffnessprocess can be evaluated by exploring the relation be-tween genetic determinants of elevated homocysteinelevels and arterial stiffness. Genetic polymorphisms are aninherited phenotype, being constant over time, reflectinglong-term elevated homocysteine levels. Furthermore,genotypes are in principle not modified by disease pro-cesses and are not affected by non-genetic confounding,and is referred to as the Mendelian randomization prin-ciple [16]. Recently, van Meurs et al. described a geneticrisk score of hyperhomocysteinemia (GRS Hcy), consistingof a combination of the most common single nucleotidepolymorphisms (SNPs), which are associated with highhomocysteine levels [22]. An association between this riskscore and arterial stiffness measurements would suggest apathophysiological link as explained above. Such a causaleffect of homocysteine could be initiated for example viathe combination of increased thrombogenecity, increasedoxidative stress and over-activation of redox-sensitive in-flammatory pathways [8,12,15,26]. Alternatively, homo-cysteine may not be causally associated with arterialstiffness. A direct effect of vitamin B12 or folate on arterialstiffness is also possible, since these vitamins are essentialin the one-carbon metabolism cycle and are stronglycorrelated with homocysteine. There are studies demon-strating an association of B-vitamins levels with bloodpressure [11,17] and atherosclerosis [5,14], but to this datereports regarding arterial stiffness are lacking.

We aim to explore underlying mechanisms of differentparts of the one-carbon metabolism cycle (Fig. 1), that mayunderlie the association between plasma homocysteineand arterial stiffness. Therefore, we investigated the as-sociation between the genetic risk score of hyper-homocysteinemia and arterial stiffness and also theassociation between B-vitamin levels and arterial stiffness.

Methods

Study population

The present study was conducted as a cross-sectionalbaseline analysis within the framework of the B-PROOF(B-vitamins for the Prevention of Osteoporotic Fractures)study. A detailed description of this randomized controlledtrial has been reported elsewhere [23]. In short, B-PROOF

is a multi-center, randomized, placebo controlled, double-blind trial including 2919 participants from three areas inthe Netherlands. Main inclusion criteria were age 65 yearsand older, and a mildly elevated homocysteine level(12e50 mmol/l). Fifty-one percent of the approachedpopulation could be included based on their homocysteinelevel. Main exclusion criteria were renal insufficiency(serum creatinine level > 150 mmol/l) and presence of amalignancy. All participants gave written informed con-sent before the start of the study. The Wageningen MedicalEthics Committee approved the study protocol, and theMedical Ethics committees of Rotterdam and Amsterdamgave approval for local feasibility.

At the Erasmus Medical Center (Rotterdam) and VUUniversity Medical Center (Amsterdam), a subsample ofparticipants underwent vascular measurements (nZ 567).Participants with cardiac arrhythmia were excluded fromthese additional measurements. During the measure-ments, participants were situated in supine position on aflat examination couch in a quiet laboratory room for atleast 5e10 min prior to the measurements, and the par-ticipants were not allowed to speak during the measure-ments. Use of alcohol or coffee during 12 h before themeasurements was prohibited.

Laboratory measurements

Venous blood samples were obtained in the morning,when the participants were in a fasted state, or had taken arestricted breakfast [23].

Measurements of folate and vitamin B12 statusSerum vitamin B12 and folate were measured usingimmunoelectrochemiluminescence assay (Elecsys 2010,Roche GmbH, Mannheim, Germany) (CV vitamin B12 5.1%

762 S.C. van Dijk et al.

at 125 pmol/l and 2.9% at 753 pml/l; CV folate: 5.9% at5.7 nmol/l and 2.8% at 23.4 nmol/l) [21]. Serum holo-transcobalamin (HoloTC) was determined by the AxSYManalyser (Abbott) (CV<8%) and serum methylmalomic acid(MMA) was measured by LC-MS/MS (CV < 9%) [6]. MMAalso reflects vitamin B12 status, next to serum vitamin B12level and homocysteine level and is considered as the mostrepresentative indicator of metabolic vitamin B(12) defi-ciency [1]. HoloTC is the fraction of vitamin B12 which isavailable for cells in the body, also referred to as activevitamin B12. HoloTC has a better diagnostic accuracy thanvitamin B12 in order to detect vitamin B12 deficiency [6].

HomocysteineFor total homocysteine analysis, a plasma EDTA tube wasstored in ice immediately after blood drawing [23], andsamples were processed within 4 h in order to prevent atemperature- and time-dependent increase in plasmahomocysteine [13]. Plasma homocysteine was measuredusing the Architect i2000 RS analyzer (VU UniversityMedical Center, intra assay CV Z 2%, inter assay CV Z 4%)and LC-MS/MS (Erasmus intra assay CV Z 5.5%, inter assayCV Z 1.3%). Outcomes of the centers did not differsignificantly at cross-calibration.

Serum creatinineSerum creatinine was measured with the enzymaticcolorimetric Roche CREA plus assay (CV Z 2%). Theestimated glomerular filtration rate (eGFR)) was calcu-lated with the formula of Modification of Diet in RenalDisease in ml/min/1.73 m2: 186 * (serum creatinine(mmol/l)/88.4) � 1.154 * age (years) � 0.203 * 0.742 (forfemales) [9].

Genotyping

DNA was isolated from buffycoats for genotyping. Allparticipants were genotyped using the Illumina Omni-express array (Illumina Inc., San Diego, CA, USA) accordingto the manufacturers’ protocol and quality standards. Thedataset was imputed to the HapMap v22 CEU referencepanel (z2.5 million single nucleotide polymorphisms[SNPs]). Hidden Markov Model-based algorithms wereused to infer unobserved genotypes probabilistically asimplemented in either MACH [10]. Imputation qualitycontrol metrics included the ratio of observed/expectedvariance of the allele dosage �0.01.

Genetic risk score of hyperhomocysteinaemia (GRS Hcy)The GRS Hcy was generated as described by van Meurset al. [22]. The method outlined by Horne et al. [7] wasused, where the individual’s GRS is equal to the sum of theexpected number of risk alleles at each SNP weighted bytheir effect sizes on plasma homocysteine (beta-co-efficients, obtained from the homocysteine meta-analysis).The meta-analysis identified SNPs of 13 independent lociexceeding the GWAS threshold (p < 5 � 10�8). Theseincluded 6 previously unreported loci in or near the genesMMACHC (p Z 2.1 � 10�9), SLC17A3 (1.0 � 10�8), GTPB10

(1.7 � 10�8), CUBN (p Z 7.5 � 10�10), HNF1A (1.2 � 10�12),and FUT2 (6.6 � 10�9). In addition, 7 loci previously re-ported were confirmed to be associated with tHcy levels ator near the MTHFR, MTR, CPS1, MUT, NOX4, DPEP1, and CBSgenes.

The MTHFR genotype was explored also separately,because of its strong individual association with hyper-homocysteinemia [24].

Clinical measures

Height was measured in duplicate to the nearest 0.1 cmusing a stadiometer, with the participant standing erectand wearing no shoes [23]. Weight was measured using acalibrated weighing device (SECA 761), with the partici-pant wearing light garments without shoes and emptypockets, to the nearest 0.5 kg [23]. Body Mass Index (BMI)was calculated as weight divided by squared height andexpressed as kg/m2. Self-reported medical history, alcoholintake and smoking habits were determined using aquestionnaire [23].

Blood pressure measurementPeripheral blood pressure at the time of vascular functiontests was measured once with a semi-automatic oscillo-metric device (Datascope Accurator Plus device, DatascopeCorp. New Jersey, USA) after at least 5 min of supine rest.Blood pressure measurements were conducted at the rightarm and measured in mmHg.

Applanation tonometryArterial tonometry was obtained from the right radial,right carotid and right femoral artery using the Sphyg-mocor device (Sphygmocor version 7.1, AtCor Medical,Sydney, Australia). Aortic pulse wave velocity (aPWV) wasmeasured with a three channel ECG recording andsimultaneously recording of the right carotid artery pulsewave form and subsequently the femoral artery pulsewaveform. The aPWV was calculated as the delay be-tween the femoral pulse wave and the carotid pulsewave, divided by the transit distance (intra CV Z 5%, interCV Z 8%). Transit distance was assessed with body sur-face measurement from the carotid artery to the femoralartery [19].

Statistical analysis

Normality of distributions was examined with histogramsand KolmogoroveSmirnov tests. If a variable was notnormally distributed, it was log-transformed. Associationsbetween B-vitamins and arterial stiffness measurementswere tested for linearity with curve estimation modeling.When linear relations had the best fit, we calculatedPearson correlation coefficients and subsequently testedthe associations using multivariable linear regressionanalysis. Potential confounders were evaluated in a step-wise model and included age, gender, study center, eGFR,MAP, heart rate, smoking, alcohol consumption, presenceof hypercholesterolemia and diabetes. Covariates were

Table 1 Population characteristics vascular subgroup B-PROOF(n Z 567).

Clinical characteristics

Age (yr) 72.5 � 5.6 (range: 65e98)Gender (male, %) 315 (55.6%)BMI (kg/m2) 27.0 � 3.7SBP (mmHg) 137.7 � 18.1DBP (mmHg) 77.3 � 9.6Pulse wave velocity (m/s) 14.3 � 4.5

Medical history of (n, %)CHD 61 (10.8%)Hypertension 211 (37.2%)TIA/Stroke 45 (7.9%)Diabetes Mellitus 65 (11.5%)Hypercholesterolemia 151 (26.6%)

Laboratory valuesHomocysteine (mmol/l) 14.2 [13.0e16.4]Vitamin B12 (pmol/l) 291.6 � 122.5Folate (nmol/l) 20.5 � 7.2MMA (mmol/l) 0.26 � 0.20HoloTC (pmol/l) 68.1 � 24.5eGFR (ml/min 1.73 m2) 91.3 � 35.9

GenotypingMTHFR genotype (n Z 508)CC (n, %) 226 (44.5%)CT (n, %) 205 (40.3%)TT (n, %) 77 (15.2%)GRS Hcy 0.90 � 0.18

Values are presented as number and percentage or as mean � SD orn (%), except for homocysteine: median (IQR) and age: mean � SDand (range).Abbreviations: BMI: body mass index; SBP: systolic blood pressure;DBP: diastolic blood pressure; PWV: pulse wave velocity; CHD:coronary heart disease; TIA: transient ischemic attack; MMA:methylmalomic acid; holoTC: holotranscobalamin; eGFR: estimatedglomerular filtration rate; MTHFR: Methylenetetrahydrofolatereductase, GRS Hcy: genetic risk score homocysteine.

Table 2 Linear regression analysis of the association between B-vitamin status and the genetic risk score of hyperhomocysteinemiawith PWV.

Model 1 Model 2

VitaminB12

�0.003 [�0.006; 0.0001] �0.002 [�0.005; 0.001]

Folate �0.022 [�0.079; 0.035] �0.027 [�0.079; 0.026]HoloTC �0.019 [�0.035; �0.002]* �0.010 [�0.026; 0.005]MMA 3.307 [1.209; 5.405] 1.600 [�0.391; 3.592]GRS Hcy �2.09 [�4.34; 0.23] �0.63 [�2.77; 1.52]

B-vitamin status: Model 1: crude; Model 2: adjusted for age, gender,study center, MAP, heart rate, eGFR.GRS Hcy: Model 1: crude; Model 2: adjusted for age, gender andstudy center.*p < 0.05. Abbreviations: as in Table 1.

Genetics of hyperhomocysteinemia 763

added to the final model as confounders if they caused achange of the point estimate (beta coefficient) of morethan 10% or were considered clinically relevant. Further-more, in a separate analysis, we tested whether homo-cysteine modified the association between B-vitaminlevels and PWV measurements in order to further explorea potential direct effect of homocysteine. If the interactionterm homocysteine e B-vitamins was significant(p < 0.05), a stratified analysis was done comparing par-ticipants with homocysteine concentrations under andabove the median. Because age was a significantly inter-acting in the association between homocysteine and PWV[20] we also tested the interaction between age and B-vitamin concentrations within the multivariable linearregression analysis and performed a stratified analysisbased on the mean age if this interaction term wassignificant.

Associations between GRS Hcy and PWV were alsotested using multivariable linear regression analysis afterconfirmation of linearity. Next, ANCOVA was used tocompare adjusted means of PWV values per quintile ofGRS Hcy. Furthermore, the interaction between homo-cysteine level and GRS Hcy was examined in order tofurther explore causality. Stratification was performedbased on homocysteine concentrations above and belowthe median if this interaction-term was significant(p < 0.05). Also we tested the interaction between age andGRS Hcy, because age was significantly interacting in theassociation homocysteine and PWV [20] and with this wecould explore the survival effect of the SNPs. The samepotential confounders mentioned above were used in astep-wise model.

Statistical analyses were performed using the statisticalsoftware package of SPSS version 20.0 (SPSS Inc, Chicago,Illinois, USA). P-values < 0.05 were considered to be sta-tistically significant.

Results

Characteristics of the study population (n Z 567) areshown in Table 1. The mean age of our population was72.5 � 5.6 years and 55.6% was male. The median homo-cysteine concentration was 14.2 [13.0e16.4] mmol/l.

Plasma vitamin B12, folate, MMA and holoTC were allinversely correlated with plasma homocysteine concen-trations (r Z �0.29, p < 0.001; r Z �0.30 p < 0.001;r Z 0.29, p < 0.001; r Z �0.36, p < 0.001 respectively).MMA and holoTC were correlated with PWV (r Z 0.13,p Z 0.002; r Z �0.10, p Z 0.03), whereas vitamin B12 andfolate were not.

Multivariate linear regression analysis showed thatfolate, vitamin B12, MMA and holoTC were not associatedwith PWV (Table 2). In the associations between vitaminB12, folate, MMA and HoloTC with PWV, the interactionwith homocysteine level was significant (p < 0.001,p < 0.001, p Z 0.04, p Z 0.004 respectively). However,after stratification for low and high levels of homocysteinewe did not find any associations between B-vitamin levelsand PWV.

Although the age e B-vitamin level interaction wassignificant within all associations of B-vitamins with PWV,age-stratified analysis did not show other associationsbetween different age groups (data not shown).

Homocysteine levels were higher in participants withthe MTHFR-TT genotype, compared with the MTHFR-CC

Figure 3 Association between GRS of hyperhomocysteinemia andplasma ln-homocysteine levels.

764 S.C. van Dijk et al.

participants and MTHFR-CT genotype, but these differ-ences were not significant (p Z 0.07 and p Z 0.26respectively). PWV levels between the MTHFR genotypesdid not differ (p for trend Z 0.08) (Fig. 2).

The GRS Hcy explained approximately 1% of the varia-tion in plasma homocysteine (r Z 0.10; p Z 0.02) (Fig. 3).

No associations were observed between GRS Hcy andPWV levels (Table 2). Nevertheless, the lowest quintile ofthe GRS Hcy had significant higher PWV levels comparedto the highest quintile (p Z 0.03) (PWV 14.2 m/s inquintile 1 compared to PWV 13.9 m/s in quintile 5; datanot shown).

Subsequently, we investigated the ln-homocysteine e

GRS Hcy interaction for the association between GRS Hcyand PWV. The interaction term was added to the modeland was highly significant (p Z 0.54 � 10�4). Figure 4demonstrates the estimated adjusted PWV means perquintile of GRS Hcy, stratified by plasma homocysteineconcentration. A trend of higher PWV levels was notice-able within the individuals with low genetic risk of higherhomocysteine levels and higher measured plasma homo-cysteine concentrations. However, the estimated means ofPWV per stratum of homocysteine concentrations werenot significantly different and the observed trend was alsonot significant (p Z 0.09 for participants with homocys-teine concentrations above the median).

Although the age e GRS Hcy interaction was signifi-cant in the association between GRS Hcy and PWV(p < 0.001), age-stratification analysis did not show anyassociations.

Discussion

Our study demonstrates that in the association betweenthe genetic risk score of hyperhomocysteinemia and PWV,the homocysteineegene interaction was significant.Therefore, the hypothesis that plasma homocysteine iscausally related to arterial stiffness was not confirmed.Furthermore, in multivariable analysis we found no

Figure 2 PWV levels according to MTHFR genotype Values aredepicted as Whiskers boxplots with estimated mean � SE with 95% CI.Data are adjusted for age, gender, study center, MAP and heart rate.Abbreviations as in Table 1.

associations between B-vitamins and arterial stiffness andtherefore the association between homocysteine andarterial stiffness is unlikely to be driven by B-vitaminlevels.

To our knowledge, this is the first study addressing theassociation between B-vitamin levels and arterial stiffnessin elderly. Despite the fact that folate has shown to affectendothelial function in vivo [25], we were not able todemonstrate an association of folate with PWV. Neitherwere vitamin B12, holoTC or MMA associated with thisparameter. All in all, our results suggest that B-vitaminsitself do not have a direct effect on the arterial stiffeningprocess, which is supported by the fact that homocysteineis mediating the association between B-vitamin levels and

Figure 4 PWV levels according to GRS Homocysteine Values aredepicted as estimated means � SE and are adjusted for age, gender,study center, MAP and heart rate. Lowest quintile indicates the lowestchance of having high Hcy levels and the highest quintile indicates thehighest chance of having high Hcy levels. Number of subjects are pre-sented in the bars. Abbreviations as in Table 1.

Genetics of hyperhomocysteinemia 765

PWV. Although, we should note we investigated theseassociations cross-sectionally.

Furthermore, we did not find evidence for a causal as-sociation between plasma homocysteine en arterial stiff-ness, as reflected by the lack of an association between theGRS Hcy and arterial stiffness. Although a significant dif-ference between the lowest and the highest quintile of thegenetic risk score in PWV level was observed, this trendwas opposite to the expected association and the differ-ences in PWV level between the quintiles were small.Thus, higher long-term plasma homocysteine concentra-tions according to the genetic risk score were associatedwith lower PWV levels. However, as was shown by strat-ification, most likely this inverse association was driven byparticipants with a low GRS Hcy, but with rather highmeasured plasma homocysteine levels. Since the plasmahomocysteineeGRS Hcy interaction significantly modifiedthe association between GRS Hcy and PWV, this indicatesthat the association between homocysteine and PWV isnot causal. If homocysteine would be a causal factor in thearterial stiffness pathway, one would expect a positiveassociation between GRS Hcy and PWV, instead of thenegative association observed. It might be speculated thatparticipants with genetically high homocysteine levelshave other, undefined protective mechanisms regardingarterial stiffness.

In our population, the GRS Hcy only explained about10% of the variance in plasma homocysteine level, whichaccords with a previous report [22]. This relatively smalleffect may explain why our hypothesis of a causal effect ofplasma homocysteine on arterial stiffness could not beconfirmed. Furthermore, as a consequence of the inclusioncriteria of the B-PROOF study, our population only con-sisted of elderly with (mildly) elevated homocysteinelevels and therefore is very selected. Because this GRS doesnot explain all the variance in homocysteine, the lack of anassociation does not definitely imply there is no causalrelation between homocysteine and arterial stiffness.Nevertheless, the fact that the homocysteineeGRS Hcyinteraction was clearly present in the association betweenthe GRS and PWV points more toward a non-causal rela-tionship: possibly an unidentified factor may explain theassociation between homocysteine and arterial stiffness,other than being a direct B-vitamin or homocysteineeffect.

Furthermore, the absent association between the GRSHcy and arterial stiffness may be due to the relatively smallsample size in terms of genetic analyses. Previously, wehave reported a small effect of plasma homocysteine onthe association with PWV [20] and since the GRS onlyexplains 1% of the plasma homocysteine level, a strongassociation is not likely to be expected. Also, there is stillthe possibility that existing cardiovascular disease itselfwould increase plasma homocysteine levels and PWV. Thisreverse causation has been mentioned previously [2,18].Nevertheless, the fact that the homocysteineeGRS Hcyinteraction was clearly present in the association betweenthe GRS and PWV points toward a non-causal relationship.Possibly a yet unidentified factor may explain the

association between homocysteine and arterial stiffness,other than being a direct B-vitamin or homocysteine ef-fect. For example, via oxidative stress and activation ofredox-inflammatory mechanisms, which lead to endothe-lial dysfunction or via thrombogenecity [8,12,15,26].

Despite the fact that a meta-analysis established anassociation between MTHFR and risk of cardiovasculardisease [24], our study could not find an association withthe MTHFR genotype and PWV, as being a preclinicalmarker for cardiovascular disease. Potentially the foundassociation with clinical outcomes in the meta-analyseswas driven by other factors than arterial stiffness.

Last, we want to mention that we did not include bloodpressure as an outcome variable, but only focused onarterial stiffness.

In conclusion, it is unlikely that homocysteine has acausal effect on arterial stiffness in older individuals. Thisis based on the lack of an association between geneticdeterminants of hyperhomocysteinemia and arterial stiff-ness and in particular on the significant homo-cysteineeGRS Hcy interaction. Furthermore, B-vitaminlevels were also not associated with measures of arterialstiffness. Homocysteine may rather be a risk indicator thana risk factor and potentially, high plasma homocysteinelevels reflect an unidentified factor, that causes increasedarterial stiffness at older age.

Funding

This study is supported and funded so far by TheNetherlands Organization for Health Research and Devel-opment (ZonMw, Grant 6130.0031), the Hague; unre-stricted grant from NZO (Dutch Dairy Association),Zoetermeer; Orthica, Almere; NCHA (Netherlands Con-sortium Healthy Ageing) Leiden/Rotterdam; Ministry ofEconomic Affairs, Agriculture and Innovation (project KB-15-004-003), the Hague; Wageningen University, Wage-ningen; VU medical center, Amsterdam; Erasmus MedicalCenter, Rotterdam. All organisations are based in theNetherlands. The sponsors have no role in the design orimplementation of the study, data collection, data man-agement, data analysis, data interpretation, or in thepreparation, review, or approval of the manuscript.

Acknowledgments

The authors would like to thank all B-PROOF participants,the total B-PROOF team, all co-workers, the endocrinologylaboratory for performing homocysteine analysis, the ge-netic laboratory for genotyping our participants and theVascular Clinical Research Units from Erasmus MC and VUUniversity Medical Center for their support and help withthe vascular function tests.

References

[1] Allen RH, Stabler SP, Savage DG, Lindenbaum J. Diagnosis ofcobalamin deficiency I: usefulness of serum methylmalonic acid

766 S.C. van Dijk et al.

and total homocysteine concentrations. Am J Hematol 1990;34:90e8.

[2] Brattstrom L, Wilcken DE. Homocysteine and cardiovascular dis-ease: cause or effect? Am J Clin Nutr 2000;72:315e23.

[3] Clarke R, Halsey J, Lewington S, Lonn E, Armitage J, Manson JE,et al. Effects of lowering homocysteine levels with B vitamins oncardiovascular disease, cancer, and cause-specific mortality: meta-analysis of 8 randomized trials involving 37 485 individuals. ArchIntern Med 2010;170:1622e31.

[4] de Ruijter W, Westendorp RG, Assendelft WJ, den Elzen WP, deCraen AJ, le Cessie S, et al. Use of Framingham risk score and newbiomarkers to predict cardiovascular mortality in older people:population based observational cohort study. BMJ 2009;338:a3083.

[5] Durga J, Bots ML, Schouten EG, Kok FJ, Verhoef P. Low concen-trations of folate, not hyperhomocysteinemia, are associated withcarotid intima-media thickness. Atherosclerosis 2005;179:285e92.

[6] Heil SG, de Jonge R, de Rotte MC, van Wijnen M, Heiner-Fokkema RM, Kobold AC, et al. Screening for metabolic vitaminB12 deficiency by holotranscobalamin in patients suspected ofvitamin B12 deficiency: a multicentre study. Ann Clin Biochem2012;49:184e9.

[7] Horne BD, Anderson JL, Carlquist JF, Muhlestein JB, Renlund DG,Bair TL, et al. Generating genetic risk scores from intermediatephenotypes for use in association studies of clinically significantendpoints. Ann Hum Genet 2005;69:176e86.

[8] Ke XD, Foucault-Bertaud A, Genovesio C, Dignat-George F, Lamy E,Charpiot P. Homocysteine modulates the proteolytic potential ofhuman arterial smooth muscle cells through a reactive oxygen spe-cies dependant mechanism. Mol Cell Biochem 2010;335:203e10.

[9] Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A moreaccurate method to estimate glomerular filtration rate from serumcreatinine: a new prediction equation. Modification of Diet inRenal Disease Study Group. Ann Intern Med 1999;130:461e70.

[10] Li M, Boehnke M, Abecasis GR. Efficient study designs for test ofgenetic association using sibship data and unrelated cases andcontrols. Am J Hum Genet 2006;78:778e92.

[11] Lim U, Cassano PA. Homocysteine and blood pressure in the ThirdNational Health and Nutrition Examination Survey, 1988e1994.Am J Epidemiol 2002;156:1105e13.

[12] Poddar R, Sivasubramanian N, DiBello PM, Robinson K,Jacobsen DW. Homocysteine induces expression and secretion ofmonocyte chemoattractant protein-1 and interleukin-8 in humanaortic endothelial cells: implications for vascular disease. Circu-lation 2001;103:2717e23.

[13] Refsum H, Smith AD, Ueland PM, Nexo E, Clarke R, McPartlin J,et al. Facts and recommendations about total homocysteine de-terminations: an expert opinion. Clin Chem 2004;50:3e32.

[14] Robinson K, Arheart K, Refsum H, Brattstrom L, Boers G, Ueland P,et al. Low circulating folate and vitamin B6 concentrations: riskfactors for stroke, peripheral vascular disease, and coronary arterydisease. European COMAC Group. Circulation 1998;97:437e43.

[15] Sipkens JA, Hahn N, van den Brand CS, Meischl C, Cillessen SA,Smith DE, et al. Homocysteine-induced Apoptosis in endothelialcells coincides with Nuclear NOX2 and peri-nuclear NOX4 activity.Cell Biochem Biophys; 2011.

[16] Smith GD, Ebrahim S. ‘Mendelian randomization’: can geneticepidemiology contribute to understanding environmental de-terminants of disease? Int J Epidemiol 2003;32:1e22.

[17] Sutton-Tyrrell K, Bostom A, Selhub J, Zeigler-Johnson C. High ho-mocysteine levels are independently related to isolated systolichypertension in older adults. Circulation 1997;96:1745e9.

[18] Ueland PM, Refsum H, Beresford SA, Vollset SE. The controversyover homocysteine and cardiovascular risk. Am J Clin Nutr 2000;72:324e32.

[19] Van Bortel LM, Laurent S, Boutouyrie P, Chowienczyk P,Cruickshank JK, De Backer T, et al. Expert consensus document onthe measurement of aortic stiffness in daily practice using carotid-femoral pulse wave velocity. J Hypertens 2012;30:445e8.

[20] van Dijk SC, Smulders YM, Enneman AW, Swart KM, vanWijngaarden JP, Ham AC, et al. Homocysteine level is associatedwith aortic stiffness in elderly: cross-sectional results from the B-PROOF study. J Hypertens 2013;31:952e9.

[21] van Driel LM, Eijkemans MJ, de Jonge R, de Vries JH, van Meurs JB,Steegers EA, et al. Body mass index is an important determinant ofmethylation biomarkers in women of reproductive ages. J Nutr2009;139:2315e21.

[22] van Meurs J, Pare G, Schwartz S. Common genetic loci influencingplasma homocysteine levels and their effect on risk of coronaryartery disease. Am J Clin Nutr; 2013. in press.

[23] van Wijngaarden JP, Dhonukshe-Rutten RA, van Schoor NM, vander Velde N, Swart KM, Enneman AW, et al. Rationale and designof the B-PROOF study, a randomized controlled trial on the effectof supplemental intake of vitamin B12 and folic acid on fractureincidence. BMC Geriatr 2011;11. 80-2318-11-80.

[24] Wald DS, Morris JK, Wald NJ. Reconciling the evidence on serumhomocysteine and ischaemic heart disease: a meta-analysis. PLoSOne 2011;6:e16473.

[25] Woo KS, Chook P, Lolin YI, Sanderson JE, Metreweli C,Celermajer DS. Folic acid improves arterial endothelial function inadults with hyperhomocystinemia. J Am Coll Cardiol 1999;34:2002e6.

[26] Zhang D, Jiang X, Fang P, Yan Y, Song J, Gupta S, et al. Hyper-homocysteinemia promotes inflammatory monocyte generationand accelerates atherosclerosis in transgenic cystathionine beta-synthase-deficient mice. Circulation 2009;120:1893e902.