Serum Iron Parameters, HFE C282Y Genotype, and Cognitive Performance in Older Adults: Results From the FACIT Study

1312

C

OGNITIVEfunctiondeclineswithage,eveninother- wisehealthyindividuals.Inthelastdecades,identify-ingbiologicdeterminantsofage-relatedcognitivedecline has become an increasingly important goal of aging re-search.Theputativeroleofnutritionalfactorsinmodifying the risk of cognitive impairment has widely been studied (1–3). In this respect, disturbed iron homeostasis, which may be genetically or environmentally determined, has been coined as one suspected risk factor for age-related cognitivedecline(4).

Iron deficiency is the most common nutritional defi-ciency worldwide, with a prevalence of 5%–20% among adultsinWesterncountries(5–8).Adepletionofbodyiron stores may result from insufficient dietary intake, ham-pered absorption, or excessive losses (9,10). In addition, regular whole blood donation may also lower body iron concentrations(8).

Asaconstituentofvariousenzymes,ironisessentialfor several physiological functions, including oxygen trans-portandDNAsynthesis(11).Thebraininparticularhas a high demand for iron not only because it is the main oxygen-consumingorganinthehumanbodybutalsobe-cause many neurobiologic processes, including myelina-tion, neurotransmitter synthesis, and synaptic plasticity, are iron dependent (12–14). Consequently, one of the major symptoms of iron deficiency is reduced cognitive performance(4,10,13,14).

Although the effects of iron deficiency on cognitive functioninghavefrequentlybeeninvestigatedinchildren and young adults (15–19), research performed in older personsisratherlimited(20–22).Irondeficiencyinolder individualsmaybeduetopoordietaryhabits,gastrointes-tinal malignancies, or diseases characterized by chronic inflammation(9,23–25).

SerumIronParameters,HFEC282YGenotype,and

CognitivePerformanceinOlderAdults:ResultsFromthe

FACITStudy

OlgaJ.G.Schiepers,

1

_{MartinP.J.vanBoxtel,}

1

_{RenateH.M.deGroot,}

1,2,3

_{JelleJolles,}

1,2

_

WimL.A.M.deKort,

4

_{DorineW.Swinkels,}

5

_{FransJ.Kok,}

6

_{PetraVerhoef,}

6,7,8

_{andJaneDurga}

6,7,9

1_{SchoolforMentalHealthandNeuroscience(MHeNS)/EuropeanGraduateSchoolforNeuroscience(EURON),Departmentof} PsychiatryandNeuropsychology,MaastrichtUniversity,TheNetherlands. 2_{AZIREResearchInstitute,FacultyofPsychologyandEducation,VUUniversityAmsterdam,TheNetherlands.} 3_{CentreforLearningSciencesandTechnologies,OpenUniversity,Heerlen,TheNetherlands.} 4_{SanquinBloodBankSouthEastRegion,Nijmegen,TheNetherlands.} 5_{DepartmentofLaboratoryMedicine,LaboratoryofClinicalChemistry441,RadboudUniversityNijmegenMedicalCentre,} TheNetherlands. 6_{DivisionofHumanNutrition,WageningenUniversity,TheNetherlands.} 7_{TopInstituteFoodandNutrition,Wageningen,TheNetherlands.} 8_{UnileverResearchandDevelopment,Vlaardingen,TheNetherlands.} 9_{CognitiveSciencesGroup,NutritionandHealthDepartment,NestléResearchCentre,Lausanne,Switzerland.} AddresscorrespondencetoOlgaJ.G.Schiepers,MSc,DepartmentofPsychiatryandNeuropsychology,MaastrichtUniversity,P.O.Box616,6200MD Maastricht,TheNetherlands.Email:olga.schiepers@np.unimaas.nl Althoughironhomeostasisisessentialforbrainfunctioning,theeffectsofironlevelsoncognitiveperformanceinolder individualshavescarcelybeeninvestigated.Inthepresentstudy,serumironparametersandhemochromatosis(HFE) C282Ygenotypeweredeterminedin818olderindividualswhoparticipatedina3-yearrandomized,placebo-controlled double-blindtrialexaminingtheeffectsoffolicacidoncarotidintima-mediathickness.Allparticipantshadslightlyel-evatedhomocysteinelevelsandwerevitaminB12replete.Cognitivefunctioningwasassessedatbaselineandafter3 yearsbymeansofaneuropsychologicaltestbattery.Atbaseline,increasedserumferritinwasassociatedwithdecreased sensorimotorspeed,complexspeed,andinformation-processingspeedandincreasedserumironwasassociatedwith decreasedsensorimotorspeed.Cognitiveperformanceover3yearswasnotassociatedwithHFEC282Ygenotypeoriron parameters.Inconclusion,serumironparametersdonotshowastraightforwardrelationshipwithcognitivefunctioning, althoughelevatedironlevelsmaydecreasecognitivespeedinolderindividualssusceptibletocognitiveimpairment.

Key Words: Cognitiveperformance—Ironparameters—HFE—Longitudinalstudy—Olderadults. Received April 20, 2010; Accepted July 30, 2010

Elevatedbodyironlevelsmightalsoimpairbrainfunc-tioning.Becausebodyironconcentrationstendtoincrease with age, particularly in women who reached the meno-pause(26),increasedironconcentrationsaremorefrequent inolderpersonsascomparedwithyoungadults;elevated iron levels may be present in up to 20% of community-dwellingindividualsaged50yearsorolder(6,8,27).

Elevatedironconcentrationsmayalsobeobservedincar-riersoftheC282Ymutationofthehemochromatosisgene (HFE),whichcausesincreasedironabsorption(26,28).The HFEC282Ymutationhasaprevalenceofaround10%in communities of Northern European descent (28,29). This mutationhasbeenassociatedwithneurodegenerativedisor-ders (12,30), including Parkinson’s disease (31) and Al-zheimer’sdisease(32,33).Itshouldbenoted,however,that a number of studies did not find any evidence for such a relationship in patients with Alzheimer’s disease (34) or Parkinson’sdisease(35).Inaddition,ithasalsobeensug- gestedthattheHFEC282Ymutationmighthaveaprotec-tiveroleinAlzheimer’sdisease(36).

Despite the fact that both iron deficiency and elevated ironlevelsfrequentlyoccurinolderindividualsandareas-sociated with negative effects on cognitive functioning, onlyasmallnumberofstudieshaveinvestigatedtheeffects ofironparametersoncognitiveperformanceinlaterstages of life. The results of these cross-sectional studies are mixed.WhereasGaoandcolleagues(20)didnotfindany associations between plasma iron and cognitive function-ing,Lamandcolleagues(22)foundaninvertedU-shaped relationshipbetweenserumironandcognitivefunctioning inmenandaninverserelationshipinwomen.Inaddition, LaRueandcolleagues(21)reportedthatserumtransferrin was positively correlated with cognitive performance in olderindividuals.

The objective of the present study was to examine the cross-sectionalandlongitudinalassociationsbetweeniron parametersandcognitivefunctioninginalargesampleof healthyolderadults.Thisstudyinvestigatedlinearandqua- draticassociationsbetweenserumironparametersandcog-nitiveperformanceaswellastherelationshipbetweenthe HFEC282Ymutationandcognitiveperformance. Methods Study Population Thestudypopulationconsistedof818menandwomenwho participatedintheFolicAcidandCarotidIntima-MediaThick-nessstudy.Thisrandomized,double-blindplacebo-controlled trialwasoriginallydesignedtoinvestigatetheeffectsof3-year folicacidsupplementationontheriskofcardiovasculardis-easeasmeasuredbycarotidintima-mediathickness(37).The studysampleincludedarelativelylargeproportionofblood donors(54%),asparticipantswererecruitedfrombloodbank registriesaswellasfrommunicipalregistries.Allparticipants

were aged between 50 and 70 years, and, specifically for women,hadreachedthemenopauseatleast2yearsprior. Exclusioncriteriawereplasmatotalhomocysteineconcentra-tionslessthan13mmol/Lorgreaterthan26mmol/L,useof B-vitaminsupplementsordrugsthatcouldaffectatheroscle-roticprogression(eg,lipid-loweringorhormonereplacement therapies),orself-reportedintestinaldisease.Wealsoexcluded individualswithelevatedhomocysteine concentrations due tofactorsotherthansuboptimalfolateconcentrations,in-cluding serum vitamin B12 concentrations less than 200 pmol/L,self-reportedmedicaldiagnosisofrenalorthyroid disorders,orself-reporteduseofmedicationsthatinfluence folate metabolism. The Medical Ethics Committee at WageningenUniversityapprovedthestudy,andallpartici-pantsgavewritteninformedconsent.

Cognitive Functioning

Cognitivefunctioningonthedomainsofmemory,senso-rimotor speed, complex speed, information-processing speed, and word fluency was assessed at baseline and at 3-yearfollow-upbymeansofaneuropsychologicaltestbat-tery, consisting of theVisualVerbal Word Learning Task (38), the Stroop Color–Word Interference Test (39), the Concept Shifting Test (40), the Letter Digit Substitution Test (41), and the Verbal Fluency Test (42). A detailed descriptionofthecognitivetestbatteryandthemethodused forcreatingcognitiveperformanceindicesbasedontheraw testscorescanbefoundelsewhere(37). Blood Measurements Theironparametersmeasuredweretotalserumiron;to-taliron-bindingcapacity,whichisameasureoftheserum concentration of the iron transport protein transferrin (43,44);transferrinsaturation,whichisexpressedasthera-tio (×100%) of serum iron concentration and total iron-bindingcapacity;serumferritin,anindicatoroftotalbody ironstores(45);andnon–transferrin-boundiron(46,47). Atbaselineandat3-yearfollow-up,fastingvenousblood sampleswerecollected,centrifugedwithin2hours,andthe serumsupernatantwasstoredinmultiplealiquotsat−80°C. Within15monthsofstorage,sampleswerethawedforserum measurements.Serumironandtotaliron-bindingcapacity weremeasuredusingHitachi747(RocheDiagnostics,Basel, Switzerland).SerumferritinwasdeterminedontheImmulite 1ofDPC(DiagnosticProductsCorporation,LosAngeles,CA) using a two-site immunometric assay (reference values: 15–190mg/Lforpostmenopausalwomenand15–280mg/L formen).Non–transferrin-boundiron,whichwasmeasured atbaselineonly,wasdeterminedbyafluorescence-based one-step chelation method (48). Serum high-sensitivity C-reactiveproteinwasdeterminedatbaselinewithELISA usingpolyclonalantibodies(Dako,Glostrup,Denmark). Anautomatedhematologyanalyzer(Sysmex,Hamburg, Germany)wasusedtomeasureserumhemoglobin.For

activeblooddonors,bloodsampleswerecollectedatleast 6weeksafterthemostrecentblooddonation.

Genotyping

GenomicDNAwasextractedfromwholebloodsamples using a Qiamp 96 DNA blood kit (Qiagen, Venlo, The Netherlands). DNA samples were stored at −80°C until furtheranalysis.HFEC282Ygenotypewasdeterminedby an automated method using minor-groove–binding DNA oligonucleotides (MGB probes) (49). The presence of a C282Y allele was confirmed by conventional polymerase chain reaction with restriction fragment length polymor-phismanalysis(50,51).ApolipoproteinE(APOE)genotype wasdeterminedasdescribedelsewhere(37).

Other Measurements

Levelofeducation,measuredatbaselinebyclassifying formalschoolingaccordingtotheDutcheducationalsystem, was categorized into low, middle, or high, that is, corre-spondingtoprimaryeducation,juniorvocationaltraining, andseniorvocationaloracademictraining,respectively(52). Alcoholconsumption(gramsperday)andcurrentsmoking (yesorno)wereascertainedatbaselinebymeansofself-reportquestionnaires,whichwerereviewedbyatrained research assistant. Body mass index (BMI; kilograms per squaremeter)wascalculatedfromheightandweight,and physicalactivitywasestimatedusingthePhysicalActivity ScalefortheElderly(53).

Statistical Analysis

Cross-sectional analyses.—The cross-sectional associa-tions between iron parameters and cognitive functioning were assessed by means of hierarchical linear regression analysis.The iron parameters considered relevant in rela- tiontocognitiveperformanceweretotalserumiron,ferri-tin, and non–transferrin-bound iron. These parameters representthedifferentsourcesofironintheblood,thatis, totalcirculatingiron,areflectionoftotalstoredbodyiron, andcirculatingironnotboundtotheplasmatransportpro-teintransferrin,respectively(45–47).Transferrinsaturation andtotaliron-bindingcapacitywerenotincludedinthere- gressionanalysesnotonlybecausetheyareindirectmea-suresofbodyironlevelsbutalsobecauseincludingthese variables in the statistical models would have introduced multicollinearity.

Separateregressionmodelswerefittedforthefivecogni-tiveperformanceindices.Thecovariatesage,sex,levelof education, alcohol consumption, smoking, BMI, physical activity, C-reactive protein, hemoglobin, and APOE E4 carrier status were entered in Step 1, followed by the iron parametersinStep2.Similarregressionanalyseswereper-formedtoexaminetheassociationsbetweentheHFEC282Y mutationandcognitiveperformance.Thevariablesage,level

ofeducation,alcoholconsumption,smoking,BMI,andphys-ical activity were confounders in our study.Although sex, C-reactiveprotein,hemoglobin,andAPOEE4carrierstatus were no actual confounders in our analyses, we included thesevariablesascovariatestoenablecomparisonwithother studies, investigating the associations between body iron levelsandcognitiveperformance(6,22,54–56).Wealso testedforapossibleconfoundingeffectofhomocysteineas allparticipantshadslightlyelevatedplasmatotalhomocysteine levels.However,theresultsweresimilarregardlessifhomo-cysteinewasincludedintheanalyses.Therefore,wedidnot includethisvariableinthefinalstatisticalmodels. Toexaminethepossiblenonlinearrelationshipsbetween iron parameters and cognitive performance, regression models were fitted for each cognitive performance index, withquadratictermsforironparametersastheindependent variables,adjustingforcovariatesandlineartermsforiron parametersinStep1.

Longitudinal analyses.—First,wetestedwhethertreatment withfolicacidwasaneffectmodifierinourdatasetasfolic acidwasfoundtoexertapositiveeffectoncognitiveperfor-mance(37),andpreviousresearchsuggestedthatfolatemight interactwithironmetabolism(57).Wefoundthatfolicacid supplementationwasnotaneffectmodifierinourdataset;the interactiontermsforironparametersandtreatmentcondition were not statistically significant (Supplementary Table 1). Furthermore,theserumironparametersdidnotsignificantly differbetweentheplacebogroupandthetreatmentgroupat theendofthestudy(p=.972forserumironandp=.892for ferritin), indicating that folic acid supplementation did not influencebodyironlevelsinourstudy.Therefore,thelongi-tudinalanalyseswereperformedinthetotalsample.

Usinglinearmixedmodels,weinvestigatedtheassocia-tions between iron parameters and cognitive performance over 3 years of follow-up, adjusting for covariates. This analysismethodtakesintoaccountthecorrelationbetween repeatedmeasurementsandallowstheinclusionofpartici-pantswithmissingobservationsatfollow-up(58).Separate models were fitted for each iron parameter in relation to eachofthefivecognitiveperformanceindices.Anunstruc-tured covariance structure was used. Time (measured in yearssincebaseline)wasincludedtoestimatethechangein cognitiveperformanceover3yearsoffollow-up.Thelongi-tudinaleffectoftheironparameterswasestimatedbythe two-way interaction between time and the specific iron parameter,whichrepresentstherateofchangeincognitive performanceover3yearsasafunctionofthisironparame-ter.ThelongitudinalassociationsbetweentheHFEC282Y mutation and cognitive performance were examined in a similarmanner.

Insecondaryanalyses,westratifiedthestudypopulation bysex,donorstatus,andAPOEE4carrierstatustodeter-mine whether the cross-sectional and longitudinal associa-tionsbetweenironparametersontheonehandandcognitive

functioningontheotherdifferedbetweenmenandwomen, blooddonorsandnon-donors,orcarriersandnoncarriersof theAPOEE4allele.One-samplettestswereusedtodeter- minewhetherserumiron,serumferritin,andcognitivefunc-tioningchangedoverthe3-yearfollow-upperiod.Chi-square testsandindependentsamplesttestswereusedtoinvestigate whether serum iron, total iron-binding capacity, and non– transferrin-boundiron,aswellasthebackgroundvariables age,sex,levelofeducation,alcoholconsumption,smoking, BMI,physicalactivity,C-reactiveprotein,hemoglobin,and APOEE4statusdifferedbetweencarriersandnoncarriersof theHFEC282Ymutation.Becauseofheterogeneityoferror variances, the nonparametric Mann–Whitney test was per-formedtotestfordifferencesinserumferritinandtransferrin saturation between carriers and noncarriers of the HFE C282Ymutation.Hardy–Weinbergequilibriumwasassessed usingachi-squaretest.Normalityofthestandardizedresid-ualsoftheregressionanalyseswasascertainedbymeansof normalP–Pplots. Thestatisticalpowerofthecross-sectionalandlongitudi-nalanalyseswasmorethan.90(smalleffectsize,f2_=.02).

Statistical differences were considered significant at p values<.05.AllanalyseswereperformedusingSPSS16.0 (SPSSInc.,Chicago,IL). Results Participants Table1showsthebaselinecharacteristicsofthepartici-pants.Forty-threeindividuals(5.3%)showeddepletediron stores,asindicatedbyserumferritinconcentrationsbelow thelowernormallimit(ie,15mg/L).Thirty-twoindividuals (3.9%) showed serum hemoglobin concentrations charac-teristicofanemiaaccordingtoWorldHealthOrganization criteria,thatis,lowerthan7.5mmol/Lforwomenandlower than8.1mmol/Lformen.Thecombinationofbothlowfer- ritinandlowhemoglobinlevels,thatis,irondeficiencyane-mia,waspresentinonlysevenparticipants(0.9%).Intotal, 53participants(6.5%)showedserumferritinconcentrations abovetheuppernormallimit(ie,190mg/Lforwomenand 280mg/Lformen).Amongnon-donors,theprevalenceof lowserumferritinwas0.8%andtheprevalenceofhighse- rumferritinwas13.1%,andamongblooddonors,theseper-centageswere11.3%and0.9%,respectively. EightyindividualswereidentifiedascarriersoftheHFE C282Y mutation (10.5%); 2 individuals were homozygous (0.3%)and78personswereheterozygous(10.2%).Theallele frequenciesoftheHFEC282Ypolymorphismdidnotsig-nificantlydifferfromtheexpecteddistributionpredictedby theHardy–Weinbergequilibrium(p=.860)andwerecompa-rablewiththefrequenciesreportedinotherpopulation-based studies(28,29).Serumiron,transferrinsaturation,andnon– transferrin-boundironweresignificantlyincreasedincarriers oftheHFEC282Ymutationascomparedwithnoncarriers, whereastotaliron-bindingcapacitywasdecreasedincarriers oftheHFEC282Ymutationascomparedwithnoncarriers (Table1).Serumferritinandhemoglobinconcentrations,as wellasthebackgroundvariables,didnotsignificantlydiffer betweencarriersandnoncarriersoftheHFEC282Ymutation (Table1).Table2presentstheserumironparametersin Table1. BaselineCharacteristicsoftheStudyPopulationAccordingtoHFEC282YGenotype

TotalSample HFEC282YNoncarriers HFEC282YCarriers p*

n 818 685 80 Age(years) 60.3±5.6 60.3±5.6 59.8±5.6 .508 Femalesex(%) 28.4 27.9 30.0 .690 Levelofeducation,low/middle/high(%) 22.4/38.1/39.5 21.8/38.4/39.9 28.8/31.2/40.0 .282 Alcoholconsumption(g/day)† _{12.5(4.4–23.5) 12.5(4.4–23.4) 14.3(3.7–26.1) .447} Currentsmoker(%) 20.4 20.0 23.8 .431 BMI(kg/m2_{) 26.6±3.6 26.5±3.6 26.6±3.5 .846} Physicalactivity(PASEscore) 152.8±68.6 153.3±69.3 153.1±65.4 .981 Blooddonor,current/former/never(%) 54.2/14.7/31.2 54.5/14.7/30.8 53.8/15.0/31.2 .993 APOEE4alleles,0/1/2(%)‡ _{67.9/29.2/2.8 69.4/28.0/2.6 58.8/38.8/2.5 .134} SerumC-reactiveprotein(mg/dL)†,§ _{1.1(0.6–2.3) 1.1(0.6–2.3) 1.1(0.6–2.1) .956} Serumhemoglobin(mmol/L) 8.9±0.7 8.9±0.7 9.0±0.7 .088 Serumironparameters  Serumiron(mmol/L) 18.5±6.4 18.2±6.2 21.4±7.3 <.001  Totaliron-bindingcapacity(mmol/L) 60.0±8.1 60.4±8.1 56.9±7.8 <.001  Transferrinsaturation(%) 31.4±11.5 30.6±10.8 38.7±15.8 <.001  Ferritin(mg/L)† _{67.0(35.0–131.0) 65.0(34.0–125.0) 86.5(47.8–138.8) .063}  Non–transferrin-boundiron(mmol/L)ǁ‖ _{2.4±0.9 2.4±0.9 2.8±1.0 <.001} Notes:Valuesaremeans±SD.BMI=bodymassindex;PASE=PhysicalActivityScalefortheElderly. *pValuesforthedifferencesbetweencarriersandnoncarriersoftheHFEC282Ymutationusingchi-squaretests,independentsamplesttests,ornonparametric Mann–Whitneytests. †_{Medianvalue(interquartilerange)isgivenbecauseofskeweddatadistribution.} ‡_{n=814inthetotalsample,n=682inHFEC282Ynoncarriers,andn=80inHFEC282Ycarriers.} §_{n=803inthetotalsample,n=678inHFEC282Ynoncarriers,andn=78inHFEC282Ycarriers.} ǁ_{n=808inthetotalsample,n=678inHFEC282Ynoncarriers,andn=78inHFEC282Ycarriers.}

carriersandnoncarriersoftheHFEC282Ymutation,strati-fiedbyAPOEE4carrierstatus.

Cross-Sectional Associations Between Serum Iron Parameters and Cognitive Functioning

The regression analyses including quadratic terms for theironparametersdidnotrevealanycurvilinearrelation-shipsbetweenironparametersandcognitivefunctioning (datanotshown).Cross-sectionallinearregressionanaly- sesindicatedthathigherserumferritinlevelsweresignifi-cantly associated with decreased sensorimotor speed, complexspeed,andinformation-processingspeed,after adjustmentforage,sex,levelofeducation,alcoholcon-sumption, smoking, BMI, physical activity, C-reactive protein,hemoglobin,andAPOEE4carrierstatus(Table3). In addition, higher serum iron was associated with de-creased sensorimotor speed (Table 3). In order to better understandthestrengthoftheassociations,whenthepre-dictive value of serum iron for cognitive performance is comparedwiththerelationshipwithageinthesameregres-sion model, an increase of 10mmol/L serum iron corre-sponds to a lower performance on sensorimotor speed similar to an individual 5.9 years older. Likewise, a 100 mg/Lincreaseinserumferritincorrespondstothesensorimo-torspeedandinformation-processingspeedofsomeone1.2 years older and the complexspeedofanindividual1.3 yearsolder.Incontrasttoserumironandserumferritin,

non–transferrin-bound iron was not significantly associ-atedwithcognitiveperformance.

The observed associations between iron parameters and cognitive performance did not differ between men and women,exceptfortherelationshipbetweenferritinandsen-sorimotorspeed,whichwassignificantinmen(b=−.089, p=.029)butnotinwomen(b=.048,p=0.498).Whenthe analyses were stratified by donor status, only non-donors showedanegativeassociationbetweenserumironandsen-sorimotorspeed(b=−.313,p=.003ascomparedwith b=−.093,p=.409indonors),whichwasstrongerthantheas-sociationfoundinthetotalsample.Statisticalsignificanceof theotherassociationsbetweenironparametersandcognitive performancethatwerefoundinthetotalsamplewaselimi-nated upon stratification by donor status. Stratification by APOE E4carrierstatusshowedthattheobservedrelation-shipbetweenferritinandsensorimotorspeedwassignificant onlyincarriersofoneortwoAPOEE4alleles(n=80)(b= −.150,p=.023ascomparedwithb=−.044,p=.287inthe noncarriergroup);incomparisonwiththetotalsample,the negative relationship between ferritin and sensorimotor speedwasmorepronouncedinAPOEE4carriers.InAPOE E4carriers,thenegativeassociationbetweenserumironand sensorimotorspeedwassimilartotherelationshipobserved inthetotalsample,althoughitwasnotstatisticallysignifi-cant.Overall,thecross-sectionalassociationsbetweeniron parametersandcognitiveperformanceintheanalysesstrati-fiedbysex,donorstatus,orAPOEE4carrierstatuspointed Table2. SerumIronParametersinCarriersandNoncarriersoftheHFEC282YMutationatBaseline,StratifiedbyAPOEE4CarrierStatus SerumIronParameters

HFEC282YNoncarriers HFEC282YCarriers

APOEE4Noncarriers APOEE4Carriers p* APOEE4Noncarriers APOEE4Carriers p n 473 209 47 33 Serumiron(mmol/L) 18.2±6.5 18.1±5.8 .966 21.2±7.7 21.6±6.9 .819 Totaliron-bindingcapacity(mmol/L) 60.5±8.2 60.2±7.7 .691 55.5±6.4 58.8±9.2 .064 Transferrinsaturation(%) 30.6±11.1 30.6±9.9 .958 39.0±16.3 38.4±15.3 .862 Ferritin(mg/L)† _{67.0(35.0–131.0) 63.0(30.0–119.0) .251 91.0(50.0–139.0) 79.0(26.0–147.0) .479} Non–transferrin-boundiron(mmol/L)‡ _{2.4±0.9 2.4±0.8 .369 2.8±1.0 2.9±1.1 .661} Hemoglobin(mmol/L) 8.9±0.7 8.8±0.6 .086 9.1±0.6 9.0±0.7 .728 Notes:Valuesaremeans±SD. *pValuesforindependentsamplesttestsornonparametricMann–Whitneytests. †_{Medianvalue(interquartilerange)isgivenbecauseofskeweddatadistribution.} ‡_{HFEC282Y−/APOEE4–,n=469;HFEC282Y−/APOEE4+,n=206;HFEC282Y+/APOEE4−,n=46;HFEC282Y+/APOEE4+,n=32.} Table3. AssociationsBetweenIronParametersandCognitivePerformanceatBaselineinHealthyOlderAdults* CognitivePerformanceIndices n StandardizedRegressionCoefficient(b)

SerumIron p Ferritin p Non–Transferrin-BoundIron p

Memory 794 −.019 .807 .002 .951 .070 .362 Sensorimotorspeed 789 −.185 .012 −.073 .033 .141 .054 Complexspeed 788 −.081 .296 −.077 .019 .081 .266 Information-processingspeed 791 −.109 .141 −.069 .046 .078 .291 Wordfluency 794 −.050 .519 −.062 .092 .057 .462 Note:*Adjustedforthecovariatesage,sex,levelofeducation,alcoholconsumption,smoking,bodymassindex,physicalactivity,APOEE4carrierstatus,serum C-reactiveprotein,andhemoglobinconcentrationinhierarchicallinearregressionanalyses.

inthesamedirectionastheassociationsobservedinthetotal sample,withfewdifferencesbetweenthegroups,although theytendednottoreachstatisticalsignificance.

In post hoc analyses, we stratified by serum C-reactive proteinconcentration(normal≤3mg/dL,elevated>3mg/dL (59))toascertainwhetherserumferritinwasassociatedwith cognitivefunctioningintheabsenceofinflammation.Higher serum ferritin predicted slower sensorimotor speed (p = .008),complexspeed(p=.007),andinformation-processing speed(p=.018)inindividualswithnormalC-reactivepro-teinconcentrations(n=659)butnotinpersonswithelevated C-reactiveproteinlevelsindicativeofinflammation(n=144) (p=.574,p=.777,andp=.945,respectively).

Longitudinal Associations Between Serum Iron Parameters and Cognitive Functioning

Cognitive performance significantly declined over the 3-year follow-up period on the domains of sensorimotor speed (mean change [95% CI] = −0.08 [−0.11 to −0.05], p=.000),complexspeed(meanchange[95%CI]=−0.06 [−0.10 to −0.02], p = .004), and information-processing speed (mean change [95% CI] = −0.11 [−0.15 to −0.08], p=.000),whereasmemoryimprovedsignificantly(mean change[95%CI]=0.39[0.34–0.44],p=.000)duetothe effectofprocedurallearning.Wordfluencydidnotsignifi-cantlychangeover3years(meanchange[95%CI]=0.03 [−0.03to0.09],p=.320).Ferritinconcentrationsincreased significantlyoverthe3-yearfollow-upperiod(meanchange [95%CI]=16.6mg/L[6.7–26.4],p=.001),whereasserum ironshowedanonsignificantdecrease(meanchange[95% CI] =−0.5mmol/L [−1.0 to 0.01], p = .055). Three-year changesinserumironandferritindidnotdifferbetweenthe twoexperimentalgroups(p=.268forserumironandp= .707forferritin)orbetweenmenandwomen(p=.637for serumironandp=.480forferritin).Inaddition,thelongi- tudinalchangesinserumironandferritindidnotdiffersig-nificantlybetweenblooddonorsandnon-donors(p=.965 for serum iron and p = .088 for ferritin), although blood donors showed a larger 3-year increase in ferritin than non-donors(meanchange[95%CI]=24.4mg/L[8.1–40.7] indonorsand7.1mg/L[−2.4to16.7]innon-donors).

Linearmixedmodelsrevealednosignificantlongitudinal associationsbetweenanyoftheironparametersandcognitive functioning (Table 4), implying that the rate of cognitive changeover3yearsdidnotvaryaccordingtobodyiron concentrations.However,stratifyingouranalysesbydonor status indicated that higher serum iron significantly pre-dictedlessdeclineinsensorimotorspeedover3yearsin non-donors(parameterestimate=.005,p=.010ascom-paredwithparameterestimate=−.001,p=.253inblood donors) and less decline in word fluency over 3 years in blooddonors(parameterestimate=.006,p=.012ascom-pared with parameter estimate = −.004, p = .165 in non-donors), indicating effect modification by donor status. Stratification by sex or APOE E4 carrier status did not revealanydifferencesbetweenmenandwomenorbetween carriersandnoncarriersoftheAPOEE4allele.Overall,the longitudinalassociationsbetweenironparametersandcog-nitiveperformanceintheanalysesstratifiedbysex,donor status,orAPOEE4carrierstatuswerefairlyinconsistentin termsofsizeanddirection.

Associations Between the HFE C282Y Mutation and Cognitive Functioning

To investigate the influence of lifelong exposure to elevatedironlevelsoncognitiveperformanceinlaterlife, weassessedtheassociationsbetweentheHFEC282Ymuta-tionandcognitivefunctioningoneachofthefivedomains. TheHFEC282Ymutationwasnotassociatedwithcognitive performance in both the cross-sectional and longitudinal analyses(SupplementaryTable2).

Discussion

Inthepresentstudy,noneoftheserumironparameters, norHFEC282Ygenotype,wererelatedtocognitiveperfor-manceover3yearsoffollow-up.However,theresultsfrom the cross-sectional analyses suggest that in older individ-ualsserumferritinandserumironmaybenegativelyrelated tothespeedofcognitivefunctioning.Whereashigherferri-tinconcentrationswereassociatedwithdecreasedcognitive functioningacrossthethreedifferentspeedmeasuresat Table4. LongitudinalAssociationsBetweenIronParametersandCognitivePerformanceOver3YearsofFollow-upinHealthyOlderAdults* CognitivePerformanceIndices n ParameterEstimateforLongitudinalEffect

SerumIron p Ferritin p Non–Transferrin-BoundIron† _p

Memory 800 .002 .307 −.001 .292 .000 .965 Sensorimotorspeed 799 .001 .396 .000 .806 .001 .925 Complexspeed 798 −.002 .086 −.001 .221 −.016 .052 Information-processingspeed 799 .001 .501 .000 .578 −.005 .446 Wordfluency 800 .002 .215 .000 .991 .018 .121 Notes:*Adjustedforthecovariatesage,sex,levelofeducation,alcoholconsumption,smoking,bodymassindex,physicalactivity,APOEE4carrierstatus,serum C-reactiveprotein,andhemoglobinconcentrationinlinearmixedmodels. †_{Non–transferrin-boundironwasmeasuredatbaselineonly.Memory,n=794;sensorimotorspeed,n=793;complexspeed,n=792;information-processing} speed,n=793;wordfluency,n=794.

baseline, serum iron appeared to be negatively related to sensorimotorspeedonly.Memoryprocesses,ontheother hand, did not seem to be related to ferritin nor any other serum iron parameter. In addition, non–transferrin-bound ironandtheHFEC282Ymutationwerenotassociatedwith cognitivefunctioninginthecross-sectionalanalyses.

CarriersoftheHFEC282Ymutationtendtoshowhigher bodyironconcentrationsthannoncarriers(28).Asgenetic factors cannot be influenced by cognitive functioning, in-vestigating the associations between HFE genotype and cognitive performance has the benefit of reducing con-foundingandrulingoutthepossibilityofreversecausation (60).Inthepresentstudy,wefoundthattheHFEC282Y mutation was associated with significantly increased con-centrationsofserumironandnon–transferrin-boundiron, as well as a statistically nonsignificant increase in serum ferritin.Contrarytoexpectation,wedidnotfindanyasso- ciationsbetweenHFEC282Ygenotypeandcognitivefunc-tioning. A possible explanation for the lack of such a relationshipistherelativelysmallpercentageofcarriersof theHFEC282Ymutation(10.5%),whichmighthavere-ducedtheprobabilityofdetectingpotentialassociations.In addition,nodatawereavailableconcerninganothercom-mon polymorphism of the HFE gene, H63D, which has been found to interact with the HFE C282Y genotype in determining individual iron levels (28,61). Future studies investigatingtheassociationsbetweenHFEgenotypeand cognitive functioning might increase statistical power by including both polymorphisms of the HFE gene in their analyses.Furthermore,itshouldbenotedthatthegeneraliz-abilityofourstudywaslimitedbythenatureofthestudy population;thepresentstudysamplewasnotrepresentative ofthegeneralpopulation,asparticipantshadslightlyele-vated plasma total homocysteine concentrations and were vitaminB12replete.

Non–transferrin-boundironhasbeenhypothesizedtobe involved in neurodegenerative disorders characterized by irondepositioninthebrain(47,62).Theputativerelationship between non–transferrin-bound iron and cognitive perfor- mance,however,hasnotbeeninvestigatedbefore.Thepres-entresultsdonotoffersupportforanassociationbetween non–transferrin-bound iron and cognitive performance or age-relatedcognitivedecline.Serumnon–transferrin-bound ironconcentrationsaregenerallyverylowinhealthyindivid-uals,whereasindividualsheterozygousfortheHFEC282Y mutationshowslightlyelevatednon–transferrin-boundiron concentrations (63). In hemochromatosis homozygotes, non–transferrin-bound iron is typically present in much largeramounts(63).However,asthepresentstudyincluded onlytwoindividualshomozygousfortheHFEC282Ymuta- tion,therangeofserumnon–transferrin-boundironconcen-trationsmighthavebeentoosmallforanyassociationswith cognitiveperformancetobecomemanifest.Inaddition,non– transferrin-bound iron was only measured at baseline, therebylimitingtheinterpretationofthelongitudinalanalyses

investigatingtheassociationofthisironparameterwithcog-nitivefunctioningover3years.

Although elevated serum ferritin concentrations gener-allyreflectincreasedbodyironstores(45),serumferritin levelsalsotendtobehigherinconditionsofinflammation, asferritinisanacute-phasereactant(23,54).Inthepresent study,wecontrolledforthisconfoundingeffectofinflam-mationbyincludingserumC-reactiveproteinconcentration as a covariate in the statistical models. Post hoc analyses showed that the inverse relationship between ferritin and cognitivespeedwasindependentofC-reactiveproteincon- centration,indicatingthattheobservedcross-sectionalas-sociationbetweenserumferritinandcognitivespeedcould notbeattributedtothepresenceofinflammation. Thepresentfindingthathigherserumferritinlevelswere relatedtoslowersensorimotorspeedincarriers,butnotin noncarriersoftheAPOEE4allele,suggeststhatAPOEE4 carrierstatusmaymodifytheeffectofelevatedbodyiron storesoncognitivefunctioning.AlthoughAPOEE4carrier statushasbeenimplicatedinage-relatedcognitivedecline andneurodegenerativeprocesses(64,65),itisunclearwhich mechanismsmightberesponsiblefortheputativeinterplay between APOE E4 carrier status and serum ferritin in affectingcognitiveperformance.

Todate,onlyafewotherpopulation-basedstudieshave investigated the relationship between iron parameters and cognitive performance in older individuals. These cross-sectionalstudieshaveyieldedconflictingresults.Whereas Gaoandcolleagues(20)didnotfindanyassociationsbe-tweenserumironandcognitiveperformanceinarelatively smallsampleof94menand94women,LaRueandcol-leagues(21)reportedapositivecorrelationbetweenserum transferrinandcognitivefunctioning,thatis,memory,vi-suospatialskills,andabstractreasoning,inasmallsample of 67 men and 70 women. However, this association was not corrected for potential confounders, such as age, sex, BMI,andalcoholconsumption.

A recent cross-sectional study performed by Lam and colleagues(22)inalargepopulation-basedsampleconsist-ingof602menand849womenhasshownaninverselinear associationbetweenserumironandperformanceonmem-orytestsinolderwomen.Interestingly,inoldermen,Lam andcolleaguesfoundaninvertedU-shapedrelationshipbe-tween serum iron and memory performance. Because we hypothesizedthatbothlowandhighbodyironlevelswould beassociatedwithcognitiveimpairment,wealsoexpected tofindaninvertedU-shapedrelationshipbetweenbodyiron levels and cognitive performance. However, although the number of men in our sample (n = 586) was comparable withthenumberofmenincludedinthestudybyLamand colleagues, thereby yielding similar statistical power, no curvilinearassociationsbecameapparentbetweentheiron parametersandcognitivefunctioninginthepresentstudy, neitherinthetotalsamplenorafterstratificationbysex.The lackofcurvilinearassociationsinourstudymaybedueto

therelativelysmallpercentageofindividualswithbodyiron levelsbeloworabovethenormallimits.Therefore,possible nonlinearassociationsbetweenironparametersandcogni- tivefunctioningmayhaveremainedundetected.Indeed,se-rumironconcentrationsinthestudybyLamandcolleagues showedhighermeansandabroaderrangeascomparedwith ourstudy(21.7±7.2mmol/Linmenand19.4±6.2mmol/L inwomenascomparedwith19.0±6.8mmol/Linmenand 17.1±4.8mmol/Linwomeninourstudy).

It is worth noting that the prevalence of depleted iron storesinourstudysamplewasslightlyhigherthanthatre-ported in other population-based studies in older individ-uals (ie, 5.3% as compared with 0.3%–3%) (6,27). In addition, our study showed a lower percentage of ferritin levelsabovetheuppernormallimitthanotherstudiesper-formed in older community-dwelling individuals, which usedsimilarorevenhigherserumferritincutoffvalues(ie, 6.5%ascomparedwith12%–20%)(6,27).Thismaybeex-plainedbythelargenumberofblooddonorsinourstudy sample,asregularwholeblooddonationhasbeenshownto lowerbodyironconcentrations(8).Indeed,whereas0.8% ofnon-donorsshowedferritinlevelsbelowthelowernor-mallimitand13.1%showedferritinlevelsabovetheupper normal limit, the prevalence among blood donors was 11.3%and0.9%,respectively. Uponstratificationofthecross-sectionalanalysesbydo-norstatus,wefoundthathigherserumironwasassociated withdecreasedsensorimotorspeedinnon-donorsbutnotin blooddonors.Whenthelongitudinalanalyseswerestrati-fiedbydonorstatus,wefoundthathigherserumironnot only predicted less decline in sensorimotor speed over 3 yearsinnon-donorsbutalsolessdeclineinwordfluency over3yearsinblooddonors.Althoughthesefindingsare notunequivocal,theydosuggestthatdonorstatusmayin-fluencethenatureoftheassociationsbetweenserumiron parameters and cognitive performance. The underlying mechanismhasyettobeelucidated.

Weobservedasignificantincreaseinserumferritincon- centrationsoverthe3-yearfollow-upperiod,whichiscon-sistentwithearlierreportsindicatingthatbodyironstores increasewithaging(26).Althoughthedifferencewasnot statistically significant, serum ferritin showed a greater 3-year increase in donors as compared with non-donors. Thisdifferencemayberelatedtothepotentialdiscontinua-tionofblooddonationinregularwholeblooddonorsduring thecourseofthestudy.Unfortunately,wewerenotableto verifythisassumptionaswedidnotmonitordonorstatus during the 3-year follow-up period. Serum iron tended to decreaseduringthefollow-upperiod,butthischangewas notstatisticallysignificant.Althoughthefactorsunderlying the3-yearchangeinserumironarenotexactlyclear,itis worth noting that they are not likely to be due to diurnal variation in serum iron levels (66), as we collected the 3-yearbloodsamplesatthesametimeofdayasthebaseline samples.

Inthepresentstudy,weusedserumironparametersas a proxy measure for brain iron levels, even though the exact correlation between central and peripheral iron levels is unclear. Nonetheless, epidemiological studies have found evidence for a positive association between thetwo,byshowingasignificantcorrelationbetweeniron concentrationsinserumandcerebrospinalfluidinolder individuals(67).Ontheotherhand,ithasalsobeensug-gested that brain iron levels may be relatively well iso-latedfromperipheralironlevels,forexample,incarriers oftheHFEC282Ymutation(68).Thus,giventheinabil-itytomeasurecerebrospinalfluidorbrainironlevelsin volunteers, the use of serum iron parameters should be consideredashortcoming.

The main strengths of the present study are the use of longitudinaldata,themeasurementofseveralironparame-ters as well as HFE C282Y genotype, the use of a large community-basedsample,andaverylowattritionratein thelongitudinalphase(2%)(37).Inaddition,thecognitive test battery administered in the present study has been proven a sensitive and robust tool for detecting subtle changesinspecificdomainsofcognitivefunctioning(38– 42).Furthermore,theinclusionofbothmenandwomen,as wellasblooddonorsandnon-donorsinourstudy,enabled ustoinvestigatetherelationshipbetweenironparameters and cognitive functioning within these individual groups. Therelevanceoftakingintoaccountdonorstatusinpopulation-based studies is emphasized by the present finding that regularwholeblooddonationmaymodifytheassociations between iron parameters and cognitive performance. In- deed,thelackofdocumentingdonorstatusmighthavecon-foundedseveralotherpopulation-basedstudiesinvestigating theseassociations,includingthoseperformedbyLamand colleagues(22),Gaoandcolleagues(20),andLaRueand colleagues(21). Inconclusion,thelackofanylongitudinalassociations between iron parameters and cognitive performance, as well as the lack of a relationship between HFE C282Y genotypeandcognitivefunctioning,suggeststhatthereis no clear-cut relationship between serum iron parameters andcognitivefunctioningorage-relatedcognitivedecline in older community-dwelling individuals. However, the presentstudyoffersindicationsforamoreintricaterela-tionship between body iron levels and cognitive perfor-mance,whichmaybepresentonlyinasubsampleofthe community,forexample,inindividualscarryingtheAPOE E4allele.

To our knowledge, our study is the first to address the putative associations between the HFE C282Y mutation andcognitiveperformanceinhealthyolderindividuals,as well as the possible interactions between iron parameters and APOE E4 carrier status or donor status in relation to cognitive functioning. Prospective studies using large population-basedsamplesareneededtofurtherinvestigate the associations between iron parameters and cognitive

performanceandtoestablishwhetherthisassociationmay differ across selected groups, for example, based on sex, donorstatus,APOEE4carrierstatus,orotherriskfactors associatedwithcognitivedecline.Inaddition,moreresearch isnecessarytoidentifytheexactmechanismsbywhichiron parametersmayinfluencecognitivefunctioning. Funding ThisworkwassupportedbytheNetherlandsOrganizationforHealth Research and Development (grant number 200110002); Sanquin Blood Bank (grant number 02-001);Wageningen University; andTop Institute FoodandNutrition. Supplementary material Supplementarymaterialcanbefoundat:http://biomed.gerontologyjournals .org/ Acknowledgment TheauthorswouldliketothankDr.BerryvanTits,Departmentof General Internal Medicine, Radboud University Nijmegen Medical Centre,Nijmegen,TheNetherlands,forthemeasurementofnon–transferrin-boundiron.

References

 1. CarterCS,LeeuwenburghC,DanielsM,FosterTC.Influenceofcalo-rie restriction on measures of age-related cognitive decline: role of increasedphysicalactivity.J Gerontol A Biol Sci Med Sci.2009;64(8): 850–859.

 2. Petursdottir AL, Farr SA, Morley JE, Banks WA, Skuladottir GV. Effectofdietaryn-3polyunsaturatedfattyacidsonbrainlipidfatty acidcomposition,learningability,andmemoryofsenescence-accelerated mouse.J Gerontol A Biol Sci Med Sci.2008;63(11):1153–1160.  3. Stephan BC,Wells JC, Brayne C,Albanese E, Siervo M. Increased

fructoseintakeasariskfactorfordementia.J Gerontol A Biol Sci Med

Sci.2010;65(8):809–814.

 4. YoudimMB.Brainirondeficiencyandexcess;cognitiveimpairment andneurodegenerationwithinvolvementofstriatumandhippocam-pus.Neurotox Res.2008;14(1):45–56.

 5. Brussaard JH, Brants HA, Bouman M, Lowik MR. Iron intake and ironstatusamongadultsintheNetherlands.Eur J Clin Nutr.1997; 51(suppl3):S51–S58.

 6. AhmedF,CoyneT,DobsonA,McClintockC.IronstatusamongAus-tralian adults: findings of a population based study in Queensland, Australia.Asia Pac J Clin Nutr.2008;17(1):40–47.

 7. Lahti-KoskiM,ValstaLM,AlfthanG,TapanainenH,AroA.Ironstatus ofadultsinthecapitalareaofFinland.Eur J Nutr.2003;42(5):287–292.  8. MilmanN,BygKE,OvesenL.IronstatusinDanes1994.II:preva-lenceofirondeficiencyandironoverloadin1319Danishwomenaged 40–70 years. Influence of blood donation, alcohol intake and iron supplementation.Ann Hematol.2000;79(11):612–621.

 9. KillipS,BennettJM,ChambersMD.Irondeficiencyanemia.Am Fam

Physician.2007;75(5):671–678.

10. SandsteadHH.Causesofironandzincdeficienciesandtheireffects onbrain.J Nutr.2000;130(2Ssuppl):347S–349S.

11. Murray RK, Granner DK, Mayes PA, RodwellVW. Harper’s

Bio-chemistry,24thed.Stamford,CT:Appleton&Lange;1996. 12.

ThompsonKJ,ShohamS,ConnorJR.Ironandneurodegenerativedis-orders.Brain Res Bull.2001;55(2):155–164.

13. BeardJL,ConnorJR.Ironstatusandneuralfunctioning.Annu Rev

Nutr.2003;23:41–58.

14. HidalgoC,NúñezMT.Calcium,ironandneuronalfunction.IUBMB

Life.2007;59(4–5):280–285.

15. Murray-KolbLE,BeardJL.Irontreatmentnormalizescognitivefunc-tioninginyoungwomen.Am J Clin Nutr.2007;85(3):778–787.

16. LozoffB,JimenezE,HagenJ,MollenE,WolfAW.Poorerbehavioral and developmental outcome more than 10 years after treatment for irondeficiencyininfancy.Pediatrics.2000;105(4):E51.

17. McCannJC,AmesBN.Anoverviewofevidenceforacausalrelation betweenirondeficiencyduringdevelopmentanddeficitsincognitive orbehavioralfunction.Am J Clin Nutr.2007;85(4):931–945. 18. HaltermanJS,KaczorowskiJM,AligneCA,AuingerP,SzilagyiPG.

Iron deficiency and cognitive achievement among school-aged chil-dren and adolescents in the United States. Pediatrics. 2001;107(6): 1381–1386.

19. BrunerAB,JoffeA,DugganAK,CasellaJF,BrandtJ.Randomised study of cognitive effects of iron supplementation in non-anaemic iron-deficientadolescentgirls.Lancet.1996;348(9033):992–996. 20. GaoS,JinY,UnverzagtFW,etal.Traceelementlevelsandcognitive

function in rural elderly Chinese. J Gerontol A Biol Sci Med Sci. 2008;63(6):635–641.

21. LaRueA,KoehlerKM,WayneSJ,ChiulliSJ,HaalandKY,GarryPJ. Nutritionalstatusandcognitivefunctioninginanormallyagingsam-ple:a6-yreassessment.Am J Clin Nutr.1997;65(1):20–29.

22. Lam PK, Kritz-Silverstein D, Barrett Connor E, et al. Plasma trace elementsandcognitivefunctioninoldermenandwomen:theRancho Bernardostudy.J Nutr Health Aging.2008;12(1):22–27.

23. HandelmanGJ,LevinNW.Ironandanemiainhumanbiology:are-viewofmechanisms.Heart Fail Rev.2008;13(4):393–404.

24. WeissG.Ironmetabolismintheanemiaofchronicdisease.Biochim Biophys Acta.2009;1790(7):682–693. 25. MarshallTA,StumboPJ,WarrenJJ,XieXJ.Inadequatenutrientin-takesarecommonandareassociatedwithlowdietvarietyinrural, community-dwellingelderly.J Nutr.2001;131(8):2192–2196. 26. KoziolJA,HoNJ,FelittiVJ,BeutlerE.Referencecentilesforserum

ferritin and percentage of transferrin saturation, with application to mutationsoftheHFEgene.Clin Chem.2001;47(10):1804–1810. 27. FlemingDJ,JacquesPF,TuckerKL,etal.Ironstatusofthefree-living,

elderly Framingham Heart Study cohort: an iron-replete population with a high prevalence of elevated iron stores. Am J Clin Nutr. 2001;73(3):638–646.

28. NjajouOT,Houwing-DuistermaatJJ,OsborneRH,etal.Apopulation-basedstudyoftheeffectoftheHFEC282YandH63Dmutations onironmetabolism.Eur J Hum Genet.2003;11(3):225–231. 29. WredeCE,HutzlerS,BollheimerLC,etal.Correlationbetweeniron

statusandgenetichemochromatosis(codonC282Y)inalargeGer-manpopulation.Isr Med Assoc J.2004;6(1):30–33.

30. Schmechel DE, Browndyke J, Ghio A. Strategies for dissecting genetic-environmental interactions in neurodegenerative disorders.

Neurotoxicology.2006;27(5):637–657.

31. GuerreiroRJ,BrasJM,SantanaI,etal.AssociationofHFEcommon mutations with Parkinson’s disease, Alzheimer’s disease and mild cognitiveimpairmentinaPortuguesecohort.BMC Neurol.2006;6:24. 32. LehmannDJ,WorwoodM,EllisR,etal.Irongenes,ironloadandrisk

ofAlzheimer’sdisease.J Med Genet.2006;43(10):e52.

33. RobsonKJ,LehmannDJ,WimhurstVL,etal.Synergybetweenthe C2alleleoftransferrinandtheC282Yalleleofthehaemochromatosis gene(HFE)asriskfactorsfordevelopingAlzheimer’sdisease.J Med Genet.2004;41(4):261–265. 34. BerlinD,ChongG,ChertkowH,BergmanH,PhillipsNA,Schipper HM.EvaluationofHFE(hemochromatosis)mutationsasgeneticmod-ifiersinsporadicADandMCI.Neurobiol Aging.2004;25(4):465–474. 35. BiasiottoG,GoldwurmS,FinazziD,etal.HFEgenemutationsina populationofItalianParkinson’sdiseasepatients.Parkinsonism Relat Disord.2008;14(5):426–430.

36. CorreiaAP, Pinto JP, DiasV, Mascarenhas C,Almeida S, Porto G. CAT53andHFEallelesinAlzheimer’sdisease:aputativeprotective role of the C282Y HFE mutation. Neurosci Lett. 2009;457(3): 129–132.

37. DurgaJ,VanBoxtelMPJ,SchoutenEG,etal.Effectof3-yearfolicacid supplementationoncognitivefunctioninolderadultsintheFACITtrial:

a randomised, double blind, controlled trial. Lancet. 2007;369(9557): 208–216.

38. VanderElstW,VanBoxtelMPJ,VanBreukelenGJP,JollesJ.Rey’s verballearningtest:normativedatafor1855healthyparticipantsaged 24–81 years and the influence of age, sex, education, and mode of presentation.J Int Neuropsychol Soc.2005;11(3):290–302.

39. VanderElstW,VanBoxtelMPJ,VanBreukelenGJP,JollesJ.The Stroopcolor-wordtest:influenceofage,sex,andeducation;andnor-mativedataforalargesampleacrosstheadultagerange.Assessment. 2006;13(1):62–79.

40. VanderElstW,VanBoxtelMPJ,VanBreukelenGJP,JollesJ.The Concept Shifting Test: adult normative data. Psychol Assess. 2006;18(4):424–432.

41. VanderElstW,VanBoxtelMPJ,VanBreukelenGJP,JollesJ.The LetterDigitSubstitutionTest:normativedatafor1,858healthypar- ticipantsaged24-81fromtheMaastrichtAgingStudy(MAAS):influ-enceofage,education,andsex.J Clin Exp Neuropsychol.2006;28(6): 998–1009.

42. VanderElstW,VanBoxtelMPJ,VanBreukelenGJP,JollesJ.Norma- tivedatafortheAnimal,ProfessionandLetterMNamingverbalflu-ency tests for Dutch speaking participants and the effects of age, education,andsex.J Int Neuropsychol Soc.2006;12(1):80–90. 43. GambinoR,DesvarieuxE,OrthM,etal.Therelationbetweenchem- icallymeasuredtotaliron-bindingcapacityconcentrationsandimmu-nologicallymeasuredtransferrinconcentrationsinhumanserum.Clin Chem.1997;43(12):2408–2412. 44. VernetM,LeGallJY.Transferrinsaturationandscreeningofgenetic hemochromatosis.Clin Chem.1998;44(2):360–362. 45. CookJD,LipschitzDA,MilesLE,FinchCA.Serumferritinasameasure ofironstoresinnormalsubjects.Am J Clin Nutr.1974;27(7):681–687. 46. Hider RC. Nature of nontransferrin-bound iron. Eur J Clin Invest.

2002;32(suppl1):50–54.

47. BreuerW,HershkoC,CabantchikZI.Theimportanceofnon-transferrin bound iron in disorders of iron metabolism. Transfus Sci. 2000; 23(3):185–192.

48. BreuerW,CabantchikZI.Afluorescence-basedone-stepassayforse-rumnon-transferrin-boundiron.Anal Biochem.2001;299(2):194–202. 49. De Kok JB, Wiegerinck ET, Giesendorf BA, Swinkels DW. Rapid

genotyping of single nucleotide polymorphisms using novel minor groove binding DNA oligonucleotides (MGB probes). Hum Mutat. 2002;19(5):554–559.

50. JeffreyGP,ChakrabartiS,HegeleRA,AdamsPC.Polymorphismin intron 4 of HFE may cause overestimation of C282Y homozygote prevalenceinhaemochromatosis.Nat Genet.1999;22(4):325–326. 51. LynasC.Acheaperandmorerapidpolymerasechainreaction-restriction

fragment length polymorphism method for the detection of the HLA-H gene mutations occurring in hereditary hemochromatosis.

Blood.1997;90(10):4235–4236.

52. De Bie SE. Standaardvragen 1987: Voorstellen voor uniformering

van vraagstellingen naar achtergrondkenmerken en interviews (Dutch) [Standard Questions 1987: Proposal for Uniformization of

Questions Regarding Background Variables and Interviews].2nded. Leiden,TheNetherlands:LeidenUniversityPress;1987. 53. WashburnRA,SmithKW,JetteAM,JanneyCA.ThePhysicalActiv-ityScalefortheElderly(PASE):developmentandevaluation.J Clin Epidemiol.1993;46(2):153–162. 54. CuniettiE,ChiariMM,MontiM,etal.Distortionofironstatusindi-cesbyacuteinflammationinolderhospitalizedpatients.Arch Geron-tol Geriatr.2004;39(1):35–42. 55. PulliamJF,JenningsCD,KryscioRJ,etal.AssociationofHFEmuta- tionswithneurodegenerationandoxidativestressinAlzheimer’sdis-easeandcorrelationwithAPOE.Am J Med Genet B Neuropsychiatr

Genet.2003;119B(1):48–53.

56. ShahRC,WilsonRS,TangY,DongX,MurrayA,BennettDA.Rela-tion of hemoglobin to level of cognitive funcShahRC,WilsonRS,TangY,DongX,MurrayA,BennettDA.Rela-tion in older persons.

Neuroepidemiology.2009;32(1):40–46.

57. Mitsopoulos E, Zanos S, Ginikopoulou E, et al. Do serum and red blood cell folate levels indicate iron response in hemodialysis pa-tients?ASAIO J.2006;52(2):163–168.

58. Verbeke G, Molenberghs G. Linear Mixed Models for Longitudinal

Data.NewYork:Springer;2000.

59. RidkerPM.ClinicalapplicationofC-reactiveproteinforcardiovascu-lar disease detection and prevention. Circulation. 2003;107(3):363– 369.

60. TobinMD,MinelliC,BurtonPR,ThompsonJR.Commentary:devel- opmentofMendelianrandomization:fromhypothesistestto‘Mende-liandeconfounding’.Int J Epidemiol.2004;33(1):26–29.

61. Gómez-Llorente C, Miranda-León MT, Blanco S, Gandia-Pla S, Gómez-CapillaJA,Fárez-VidalME.Frequencyandclinicalexpression ofHFEgenemutationsinaSpanishpopulationofsubjectswithab-normalironmetabolism.Ann Hematol.2005;84(10):650–655. 62. KartikasariAE,GeorgiouNA,VisserenFL,VanKats-RenaudH,Van

AsbeckBS,MarxJJ.Endothelialactivationandinductionofmono-cyte adhesion by nontransferrin-bound iron present in human sera.

FASEB J.2006;20(2):353–355.

63. DeValkB,AddicksMA,GosriwatanaI,LuS,HiderRC,MarxJJ. Non-transferrin-boundironispresentinserumofhereditaryhaemo-chromatosisheterozygotes.Eur J Clin Invest.2000;30(3):248–251. 64. Deary IJ,Whiteman MC, PattieA, et al. Cognitive change and the

APOEepsilon4allele.Nature.2002;418(6901):932. 65. SmithJD.Apolipoproteinsandaging:emergingmechanisms.Ageing Res Rev.2002;1(3):345–365. 66. DaleJC,BurrittMF,ZinsmeisterAR.Diurnalvariationofserumiron, iron-bindingcapacity,transferrinsaturation,andferritinlevels.Am J Clin Pathol.2002;117(5):802–808. 67. MizunoS,MiharaT,MiyaokaT,InagakiT,HoriguchiJ.CSFiron, ferritinandtransferrinlevelsinrestlesslegssyndrome.J Sleep Res. 2005;14(1):43–47.

68. Haba-Rubio J, Staner L, Petiau C, Erb G, Schunck T, Macher JP. Restless legs syndrome and low brain iron levels in patients with haemochromatosis. J Neurol Neurosurg Psychiatry. 2005;76(7): 1009–1010.