Gene-environment interaction and Mendelian randomisation

Environmental Neurology

Gene-environment interaction and Mendelian randomisation

P. Kolber

, R. Kru¨ger

*

aLuxembourgCentreforSystemsBiomedicine,ClinicalandExperimentalNeuroscience,UniversityofLuxembourg, 4362Belval,Esch-sur-Alzette,Luxembourg

bNeurology,CentreHospitalierdeLuxembourg,Luxembourg,Luxembourg

cLuxembourgInstituteofHealth,Luxembourg,Luxembourg

1. Introduction

Parkinson’s disease (PD) is the most common movement disorder and the second most frequent neurodegenerative diseaseafterAlzheimer’sdisease(AD)[1,2].Firstdescribedtwo centuries ago, the underlying cause of PD still remains unknownforthe majorityofthecases, however,thereare recognizedfactorswhichincreasetheriskforPD,including aging,environmentalexposure(pesticidesorheavymetals) and genetic factors (genetic risk variants and monogenic Mendeliantraits)–currentlyupto30%ofallPDcasescanbe referredtoageneticcontribution[3].

Fundamentally,themotorsymptomsofPD–asaprototype forneurodegenerativedisorders–aremainlycharacterizedbya progressive loss of dopaminergic neurons linked with an accumulation ofaggregation-prone alpha-synuclein protein.

These intracellular alpha-synuclein inclusions are a more generalized process affecting different types of neurons in the central and peripheral nervous system contributing to the spectrum of non-motor symptoms in PD and forming the pathognomonic Lewy bodies in affected brain regions (reviewed in[4]).Therefore,specific clearancepathways are involved inPD, thatinclude onone handimpaired protein degradation viatheubiquitin-proteasomesystem[5]andon anotherhandimpairedlysosomal clearanceofproteinsand info article

Articlehistory:

Received14April2019 Accepted20April2019

Availableonline19September2019

Keywords:

Parkinson’sdisease Mendelianrandomisation Gene-environmentinteraction Genome-wideassociationstudies (GWAS)

abstract

GeneticfactorsonlyaccountforuptoathirdofthecasesofParkinson’sdisease(PD),while theremainingcasesareofunknownaetiology.Environmentalexposures(suchaspesticides orheavymetals)andtheinteractionwithgeneticsusceptibilityfactors(summarizedin theconceptofimpairedxenobioticmetabolism)arebelievedtoplayamajorroleinthe mechanismsofneurodegeneration.Besideoftheclassicalassociationstudies(e.g.genome- wideassociationstudies),anovelapproachtoinvestigateenvironmentalriskfactorsare Mendelianrandomisationstudies.Thisreviewexploresthegene-environmentinteraction andthegainofMendelianrandomisationstudiesinassessingcausalitiesofmodifiablerisk factorsforPD.

#2019TheAuthors.PublishedbyElsevierMassonSAS.Thisisanopenaccessarticleunder theCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/).

*Correspondingauthorat:LuxembourgCentreforSystemsBiomedicine(LCBS),ClinicalandExperimentalNeuroscience,Universityof Luxembourg,6avenueduSwing,4362Belval,Esch-sur-Alzette,Luxembourg.

E-mailaddress:rejko.krueger@uni.lu(R.Kru¨ger).

Availableonlineat

ScienceDirect

www.sciencedirect.com

https://doi.org/10.1016/j.neurol.2019.04.010

0035-3787/#2019TheAuthors.PublishedbyElsevierMassonSAS.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://

creativecommons.org/licenses/by-nc-nd/4.0/).

organellesviaautophagy[6,7].Thesubsequentaggregationof misfoldedalpha-synucleinanditsprion-likespreadwithinthe centralandautonomousnervoussystemissupposedtobea critical step in disease progression [4,8,9]. Whether these aggregates or the preceding (proto-)fibrillary structures of alpha-synucleininterferewiththeneuron’svitalfunctionsis currentlysubjectofdebate[9].However,thesequestrationof otherproteinsandtranscriptionfactorsintoaggregates,aswell as impaired mitochondrial function also increase the cells’

vulnerabilitytoexcitotoxicityandoxidativestress,leadingto energydepletion.Inaddition,duetodeficiencyoftheubiquitin- proteasomesystem,theautophagylysosomepathwayplaysan importantrolefordegradationofmisfoldedproteinsanddefect mitochondria [10]. These mechanisms lead in fine to the activationoftheapoptotic cascadeandcelldeathrelatedto neurodegeneration.Someneuronalpopulationsseemtomore vulnerable to these pathogenic mechanisms than others within the nervous system, thus determining the clinical phenotypeofthediseaseaccordingtothepreferentialsitesof neuronaldysfunctionandsubsequentdegeneration[11,12].

Inthelasttwodecadestheinvestigationofrarefamilial forms of PD with monogenetic cases showing classical Mendelianinheritanceprovidedthefirstmajorinsightsinto the underlying molecular mechanisms involved in PD, by analysing in vitro and invivo the resultingdysfunction of the proteinsencoded by thesegenes, e.g. alpha-synuclein.

This allowed to dissect pathophysiological mechanisms of impairedproteindegradation,oxidativestressandmitochon- drialdysfunction,andthereforeraremonogenicformsofPD servedasanentry pointto understandthemorecommon sporadic form of PD [13]. Indeed, some of the mutations identified inmonogenicformsofPDwerealsoobservedin patientswithout apositive familyhistory or sometimes in unaffectedindividuals,soapparentlynotsufficienttocause thedisease.Thisindicatesreducedpenetranceforsomeof themutations.CarriersoftheG2019SmutationintheLRRK2 geneareagoodexample,wherepatientswerediagnosedwith sporadic PD mirroring the reduced penetrance of certain geneticriskfactors[14–20].Unbiasedgenome-wideassocia- tion studies (GWAS) have demonstrated a role of more commongenetic variantsin thepathogenesis ofidiopathic

PD [20–25]. The ‘‘graded risk’’ concept (Fig. 1) includes Mendelian mutations, low frequency genetic variants and common polymorphisms and introduces the concept of a continuum from more common genetic variants to rare disease-causing mutationswith anassociatedmoreor less strong impact on the expressivity and contribution to the developmentofthedisease[3,26].

Besidestheparticipation ofgeneticsusceptibilityfactors inthepathogenesisofneurodegeneration,thesefindingson reduced penetranceand potentialpolygenic contributionto disease risk indicate animportant role ofadditional, most probably environmental, risk factors and an important interplay between both genetic and environmental factors inmodulatingtheclinicalexpressionofthedisease[27].

2. Gene-environment interactions

Aging, at the cellular level, reflects an accumulation of changes affecting the physiological functioning of the cell andthewholeorganism,leadingtoanincreasedvulnerability to death. ‘‘Chronologic age’’ does not necessarily reflect necessarily‘‘biologicalage’’, asadditional variablessuchas individualgeneticbackground,lifestyleanddiseaseprocesses influence cellular aging. Hallmarks of aging are genetic alterations including telomere attrition, increased genomic instability and epigenetic alterations involving changes in DNA methylationpatterns,histonemodificationsandchro- matinremodeling[28].

Differentbiomarkerspreviouslyproposedfor definingthe biologicalageofaspecificcellortissue,suchastelomerelength [29]orage-dependentdeletionsofmitochondrialDNA[30],were insufficientintheirprecision(e.g.sensitivity,specificity)and practicality(e.g.accessibility,cost-effectiveness)[31].Epigenetic changes, especially the DNA methylation status provides a morereliablebiomarkerinthisregard[32–34].Especiallysome CpGsites(5’-cytosine-phosphate-guanine-3’sites)exhibitmore linearDNAmethylationchangesthroughoutagingandarethus valuablebiomarkersforageprediction[35–37].

Fig.1–Thegraded-riskconcept(adaptedfromManolioetal.,2009;[26]).

Interestingly, ‘‘epigenetic drift’’ – the divergence of the epigenomeduetorandomchangesinmethylation(orerrorsin the maintenance of the epigenome) over a lifespan – is influencedsubstantiallybyenvironmentalfactors,asshown intwinstudies[36,37].Thisisbelievedtocausetheindividual susceptibilities,prognoses,treatmentoutcomesandinalarger sensetheindividualclinicalphenotypesofcommondiseases.

Overtime, the accumulation ofepigenetic changes –or epimutations–isbelievedtoinduceinstabilityofthegenome, tofacilitate geneticmutations and finallypathologicalpro- cessesdefiningadisease.ThesechangesinDNAmethylation andtheireffectongeneexpressiongiveusnewinsightinage- relateddiseasessuchasPD[38,39].

2.2. Gene-environmentinteraction

Withinthelasttwodecades,geneticcausesforPDhavemore andmorecometotheforefrontinPDresearch[40].However, mostcasescannotbeexplainedyetbyamonogeneticformof PD,norbyriskvariants.Inversely,eventhoughenvironmental factorsarerecognizedtocontributetoPD,noteveryonewithPD hasbeenexposedtooneofthecurrentlyknownenvironmental riskfactorsandviceversanoteveryoneexposedtoanenviron- mental risk factor is developing the disease. Commonly, pesticides are thought to alter mitochondrial function and increase oxidativestress indopaminergicneurons,but also to accelerate the formation of alpha-synuclein fibrils and interferewiththefunctionoftheubiquitin-proteasomesystem [41–43]. Professional or occupational pesticide usage may increase expressionofthedopaminetransporter(DAT)and leadtoanaccumulationoftoxicantsindopaminergicneurons, affectingdopaminergicneurotransmissioninsubjectscarrying DATsusceptibilityalleles[44,45].Exposuretoorganophospha- tes of subjects carrying polymorphisms in the ACHE/PON1 locusmightincreasetheriskforinsecticide-inducedPDviaa neurotransmissionimbalanceduetoimpairedacetylcholines- terase (AChE) and paraoxonase (PON1) activity(and in fine impairedorganophosphatedegradations)[46,47].Thecombi- nation and interaction of geneticsusceptibility factors and environmental exposure is summarized in the concept of impairedxenobioticmetabolismandshedsfirstlightintothe mechanisms of neurodegeneration. These initial insights may translate into new therapiesfor PDpatients that take intoaccounttheindividualgeneticbackgroundsreflectedin enzymaticactivities[48–50].

EvenonceassociationsofPDwithenvironmentalfactors wereestablished,itwasyetnotclearwhetherthereis(inverse) causality between PD and these factors, e.g. the reverse association between PD and smoking [51]. One way to investigateenvironmentalfactorsandtheircausalityforPD mightbebyusingso calledMendelianrandomisation(MR) methods.

3. Mendelian randomisation

Association studies are a common tool to investigate aetiologies of pathologies, however it is now known that

associationandcausalityarenotequivalentbynomeans,as an association betweena disease and a riskfactor canbe causedbyunknownfactors(confounders)orthediseasecan cause the increased presence of a risk factor (reverse causation), as represented in Fig. 2A. In order to avoid confounding factors (selection bias, cofounders, reverse causation...), randomisedcontrol trials(RCT), byrandomly assigning study participants in two or more groups, are considered as gold standard for investigating causality in prospective clinical trials. Besides the ethical concerns of exposingsubjectsdeliberatelytotheinvestigatedriskfactor, thesestudieshavethedownsidethattheyareexpensiveand time consuming as a longitudinal observation period is needed.Additionally,RCTsarelimitedinthenumberofrisk factors investigated and they only reflect the effect of an exposureduringacertainperiodoftimeinthelifeofthestudy participant.

TheaimofaMRstudyistoassesswhetheranon-genetic/

modifiable environmental exposure is associated with the investigatedpathologybyintroducingarandomisationmethod intoanobservationalstudyandavoidingtheabove-mentioned downsidesofRCTs[52].Asshowninadirectedacyclicgraph (Fig. 2A), MR is able to test the null hypothesis that the pathology(outcome)isnotcausedbyanexposure,byassuming that genetic variants used as instrumental variables (IV) are robustlyassociated withthe investigated exposure, are independent of confounders and are associated with the investigated outcome solely (and linearly, unaffected by statisticalinteractions)viatheexposure[53].Therandomisa- tionmethodinMRstudiesisbasedonMendelianprinciples of inheritance [54]. MR studies use genetic variants, such assinglenucleotidepolymorphisms(SNPs)asIVstomeasure theeffectofannon-geneticexposureonanoutcome(Fig.2A) [55,56]. This randomisation method of using germline geneticvariants(whicharesupposed toberandomlydistri- butedduringconceptioninthegeneralpopulation)asvaria- bles,isusedforprovingcausalassociationsandtoaccurately estimatetheeffectofalifelongexposuretoanenvironmental/

lifestyleriskfactoronanoutcome[57],bybeinglesssusceptible tothebiasesofobservationalstudies[58].

3.2. MendelianrandomisationinParkinson’sdisease

Inthepastdecades,classicalepidemiologicalstudies(includ- ing potentialrecall biases) could determinesome environ- mentalriskfactorsforPD,forexamplepesticideusageorhead trauma[59–62].Butalso,protectivefactorshavinganegative association with the PD risk, such as smoking, caffeine drinkingorelevateduratelevelsinserumwereidentifiedin suchstudies.Thesenegativeassociations(orpotentialprotec- tivefactors)arewidelydiscussedandMRstudiesprovidea waytoanalysetheseassociationsfortheircausalitiesandto shedlightonpotentialprotectivefactors.

Potentiallyduetoitsantioxidativeeffect,elevatedplasma uratelevelswereshowntohaveanegativeassociationwith PDrisk[63].AJapaneseclinicalstudyhadalreadyinvestigated whetherinosine(aurateprecursor)wouldincreasetheplasma uratelevelofPDpatients,withoutprovingcausality[64].Two independent Danish studies recently used MR methodsto assesscausalityofPDincidenceandmetabolicdatasuchas

plasmauratelevels[65,66]. Theepidemiological association betweenelevatedplasmaurateandareducedriskforPDwas shown,butMRanalysisremainednegativeandprovidedno evidenceforacausalrelationshipbetweenthismodifiablerisk factor and the outcome.The initially demonstrated epide- miologicalassociationmightthereforebeattributedtoother unknownfactorsthantheelevateduratelevels,makingthe rationaleofclinicalinosinestudiespotentiallylessrelevant [67].

ThenegativeassociationofPDriskandcoffeedrinkingor nicotinesmoking waspreviously shownin differentepide- miologicalstudies[68],raisingthecontroversyoverwhether, forexample,smoking cigarettesreducestheriskforPD,or PD patientsare less likelyto becomeaddicted tosmoking becauseoftheirunderlying‘‘primarypersonality’’(Fig.2B).

Serumcaffeineanditsmetaboliteswereeveninvestigatedas potentialdiagnosticbiomarkersforearlyPD,consistentwith itssupposedneuroprotectiveeffect[69].However,theunder- lyingcausalityisnotknownyet(neuroprotectivesubstances absorbedbythelungsordigestivetractarespeculated)andan RCTexposingintentionallysubjectstocigarettesmokingto investigatetheriskforPDwouldbehighlyunethical.ManyMR studieshavebeenconductedasanewapproachtoinvestigate potential protective environmental factors, such as coffee consumption,andtheireffectonhealth[70].Alsohighserum ironlevelsandahighbodymassindex(BMI)wereshownin MRstudiestohaveanegativecausaleffectonPDrisk[71,72].

LowLDLcholesterollevelsontheotherhand,mighthavea

causaleffectonloweringtheriskforAD,withoutaneffecton PDrisk[73].

ArecentMRstudyondailycoffeeconsumptionandPDrisk couldnotdefineacausaleffectbetweenthesetwo,butdidfind a positive causal association between being a ‘‘morning person’’andincreasedPDrisk[74].Thecircadianrhythmis highlydeterminedbygeneticvariantsandthepresumptionis thattheresultingphenotype‘‘morningperson’’ismoreactive inthemorninganddoesnotneedtodrinkasmuchcoffeeto startitsday.However,the‘‘nightowl’’(havingalowerriskfor PD)wouldneedmorecoffeeinthemorningtocopewiththe

‘‘8 a.m. to 5p.m.’’ work schedule dictatedby the society, explaining the reduced risk for PD associated with coffee consumptionasanepiphenomenon.Thiscausalassociation of the sleep-wake rhythm and PD (or neurodegenerative diseasesingeneral)makessenseinlightofstudiesshowing amyloid-betaaccumulationinsleepdeprivedbrains[75].

3.3. LimitationsofMendelianrandomisation

Complementingtheclassicalobservationalstudiesbyavoiding theirbiases,MRisnotwithoutitslimitations.Inordertobe usedtoassessthe effectofanon-genetic riskfactoronan outcome,MRreliesontheavailabilityofrobustgeneticdata.

ManyGWASstudieshaveenlightenedinthelastdecadesnot onlyassociationswithdiseases,butalsocomplexmeasurable parameterssuchasbodymassindex(BMI;[76]),serumlevelof metabolites (such as serum uric acid or cholesterol levels;

Fig.2–TheprinciplesofMendelianrandomisationrepresentedinadirectedacyclicgraph(A).TheexampleofParkinson’s diseaseisrepresentedinB.Geneticvariantsofalpha-synuclein(orother)canpotentiallycausethePDphenotypeorevena

‘‘primarypersonality’’foratendencytoexposetonon-geneticfactors(suchascigarettesmoking).Ifcigarettesmokingis

‘‘protectivefactor’’forPD(causality)orduetothis‘‘primarypersonality’’(reversecausality)isdebated.

[77,78]),cognitivefeaturesormanyotherbehaviourpatterns likedietsandhabits[79–81].Forexample,Yengoandcolleagues analysedGWASdataofaboutsevenhundredsubjectsusingMR and identified hundreds ofexposure-associated SNPs (data availablebecause oftheincreasingnumberofgeneticasso- ciationstudies and data banks) [76]. The discovery of new genetictraitsandSNPsandtheassessmentofcausalitiesare moreaccurateifalargesampleofgeneticdataisavailable.

Therefore,MRstudies cannotbeusedifnoSNPshavebeen identifiedtobeassociatedfortheexposure,ortheydon’thave sufficientpowertodetectcausalityiftheSNPsdon’texplainthe integralexposurevariation[57].

4. Conclusion

Mendelian randomisation studies are an efficient method, complementary to classical association studies, to assess causalitiesofmodifiableenvironmentalriskfactors ifsuffi- cientgeneticdataisavailable.Therefore,cohortsassessinga hugearrayofdifferentclinicaldataandprovidingadetailed geneticcharacterizationareneededinthefuturetoenhance thepoweroftheseMRstudies[82].Avoidingconfounderand reverse causality biases of association studies, MR thus enablestoinvestigategene-environmental interactionsand determinepotentialprotectivefactorstobeimplementedin future preventive medicine campaigns.They mightoffer a goodtoolforguidingfutureclinicalRCTS,byavoidingerrancy andwasteofresearchresources.

5. Disclosure of interest

The work of R.K. and P.K. is supported by Grants of the Luxembourg National Research Fund (FNR; NCER-PD and PEARL [FNR/P13/6682797/Kru¨ger]) and by funding from the European Union’s Horizon 2020 research and innovation programmeunderGrant agreementNo692320(TWINNING;

Centre-PD)toRK.

references

[1] AscherioA,SchwarzschildMA.Theepidemiologyof Parkinson’sdisease:riskfactorsandprevention.Lancet Neurol2016;15:1257–72.http://dx.doi.org/10.1016/S1474- 4422(16)30230-7.

[2] PoeweW,SeppiK,TannerCM,HallidayGM,BrundinP, VolkmannJ,etal.Parkinsondisease.NatRevDisPrim 2017;3:17013.

[3] LarsenSB,HanssZ,Kru¨gerR.Thegeneticarchitectureof mitochondrialdysfunctioninParkinson’sdisease.Cell TissueRes2018;373:21–37.http://dx.doi.org/10.1007/s00441- 017-2768-8.

[4] GoedertM,ClavagueraF,TolnayM.Thepropagationof prion-likeproteininclusionsinneurodegenerative diseases.TrendsNeurosci2010;33:317–25.http://dx.doi.org/

10.1016/j.tins.2010.04.003.

[5] AtkinG,PaulsonH.Ubiquitinpathwaysin

neurodegenerativedisease.FrontMolNeurosci2014;7:63.

http://dx.doi.org/10.3389/fnmol.2014.00063.

[6] CuervoAM,WongESP,Martinez-VicenteM.Protein degradation,aggregation,andmisfolding.MovDisord 2010;25:S49–54.http://dx.doi.org/10.1002/mds.22718.

[7] PickrellAM,YouleRJ.TheRolesofPINK1.Parkin,and MitochondrialFidelityinParkinson’sDisease.Neuron 2015;85:257–73.http://dx.doi.org/10.1016/

j.neuron.2014.12.007.

[8] RecasensA,DehayB.Alpha-synucleinspreadingin Parkinson’sdisease.FrontNeuroanat2014;8:159.http://

dx.doi.org/10.3389/fnana.2014.00159.

[9] GoedertM.Alzheimer’sandParkinson’sdiseases:The prionconceptinrelationtoassembledAb,tau,anda- synuclein.Science2015;80:349.http://dx.doi.org/10.1126/

science.1255555.

[10] PanT,KondoS,LeW,JankovicJ.Theroleofautophagy- lysosomepathwayinneurodegenerationassociatedwith Parkinson’sdisease.Brain2008;131:1969–78.http://

dx.doi.org/10.1093/brain/awm318.

[11] BraakH,DelTrediciK,Ru¨bU,DeVosRAI,JansenSteur ENH,BraakE.Stagingofbrainpathologyrelatedtosporadic Parkinson’sdisease.NeurobiolAging2003;24:197–211.

http://dx.doi.org/10.1016/S0197-4580(02)00065-9.

[12] BraakH,Ru¨bU,GaiWP,DelTrediciK.Idiopathic Parkinson’sdisease:possibleroutesbywhichvulnerable neuronaltypesmaybesubjecttoneuroinvasionbyan unknownpathogen.JNeuralTransm2003;110:517–36.

http://dx.doi.org/10.1007/s00702-002-0808-2.

[13] SchieslingC,KieperN,SeidelK,Kru¨gerR.Review:Familial Parkinson’sdisease–genetics,clinicalphenotypeand neuropathologyinrelationtothecommonsporadicformof thedisease.NeuropatholApplNeurobiol2008;34:255–71.

http://dx.doi.org/10.1111/j.1365-2990.2008.00952.x.

[14] MaraganoreDM,DeAndradeM,ElbazA,FarrerMJ, IoannidisJP,Kru¨gerR,etal.Collaborativeanalysisof a-synucleingenepromotervariabilityandParkinson disease.JAmMedAssoc2006;296:661–70.http://dx.doi.org/

10.1001/jama.296.6.661.

[15] Abou-SleimanPM,MuqitMMK,McDonaldNQ,YangYX, GandhiS,HealyDG,etal.Aheterozygouseffectfor PINK1mutationsinParkinson’sdisease?AnnNeurol 2006;60:414–9.http://dx.doi.org/10.1002/ana.20960.

[16] GoldwurmS,ZiniM,MarianiL,TeseiS,MiceliR,SironiF, etal.EvaluationofLRRK2G2019Spenetrance.Neurology 2007;68:1141–3.http://dx.doi.org/10.1212/

01.wnl.0000254483.19854.ef.

[17] DiFonzoA,Rohe´ CF,FerreiraJ,ChienHF,VaccaL,StocchiF, etal.AfrequentLRRK2genemutationassociatedwith autosomaldominantParkinson’sdisease.Lancet 2005;365:412–5.http://dx.doi.org/10.1016/S0140- 6736(05)17829-5.

[18] LatourelleJC,SunM,LewMF,SuchowerskyO,KleinC, GolbeLI,etal.TheGly2019SermutationinLRRK2isnot fullypenetrantinfamilialParkinson’sdisease:theGenePD study.BMCMed2008;6:32.http://dx.doi.org/10.1186/1741- 7015-6-32.

[19] MuellerJC,FuchsJ,HoferA,ZimprichA,LichtnerP,IlligT, etal.Multipleregionsofa-synucleinareassociatedwith Parkinson’sdisease.AnnNeurol2005;57:535–5341.http://

dx.doi.org/10.1002/ana.20438.

[20] ForoudT,UniackeSK,LiuL,PankratzN,RudolphA,Halter C,etal.Heterozygosityforamutationintheparkingene leadstolateronsetParkinsondisease.Neurology 2003;60:796–801.http://dx.doi.org/10.1212/

01.WNL.0000049470.00180.07.

[21] Simo´n-Sa´nchezJ,SchulteC,BrasJM,SharmaM,GibbsJR, BergD,etal.Genome-wideassociationstudyreveals geneticriskunderlyingParkinson’sdisease.NatGenet 2009;41:1308.

[22] InternationalParkinsonDiseaseGenomicsConsortium, NallsMA,PlagnolV,HernandezDG,SharmaM,Sheerin UM,etal.Imputationofsequencevariantsfor

identificationofgeneticrisksforParkinson’sdisease:a meta-analysisofgenome-wideassociation,studies.Lancet 2011;377:641–9.http://dx.doi.org/10.1016/S0140-

6736(10)62345-8.

[23] BeechamAH,PatsopoulosNA,XifaraDK,DavisMF, KemppinenA,etal.Analysisofimmune-relatedloci identifies48newsusceptibilityvariantsformultiple sclerosis.NatGenet2013;45:1353.

[24] LambertJ-C,Ibrahim-VerbaasCA,HaroldD,NajAC,SimsR, BellenguezC,etal.Meta-analysisof74,046individuals identifies11newsusceptibilitylociforAlzheimer’sdisease.

NatGenet2013;45:1452.

[25] SaadM,LesageS,Saint-PierreA,CorvolJC,ZelenikaD, LambertJC,etal.Genome-wideassociationstudyconfirms BST1andsuggestsalocuson12q24astherisklocifor Parkinson’sdiseaseintheEuropeanpopulation.HumMol Genet2011;20:615–27.http://dx.doi.org/10.1093/hmg/

ddq497.

[26] ManolioTA,CollinsFS,CoxNJ,GoldsteinDB,HindorffLA, HunterDJ,etal.Findingthemissingheritabilityofcomplex diseases.Nature2009;461:747.

[27] RobakLA,JansenIE,vanRooijJ,UitterlindenAG,KraaijR, JankovicJ,etal.Excessiveburdenoflysosomalstorage disordergenevariantsinParkinson’sdisease.Brain 2017;140:3191–203.http://dx.doi.org/10.1093/brain/awx285.

[28] Lo´pez-Otı´nC,BlascoMA,PartridgeL,SerranoM,Kroemer G.Thehallmarksofaging.Cell2013;153:1194–217.http://

dx.doi.org/10.1016/j.cell.2013.05.039.

[29] BlascoMA.Telomeresandhumandisease:ageing,cancer andbeyond.NatRevGenet2005;6:611–22.http://dx.doi.org/

10.1038/nrg1656.

[30] CortopassiGA,ShibataD,SoongNW,ArnheimN.Apattern ofaccumulationofasomaticdeletionofmitochondrial DNAinaginghumantissues.ProcNatlAcadSci 1992;89:7370–4.http://dx.doi.org/10.1073/pnas.89.16.7370.

[31] MeissnerC,Ritz-TimmeS.Molecularpathologyandage estimation.ForensicSciInt2010;203:34–43.http://

dx.doi.org/10.1016/j.forsciint.2010.07.010.

[32] FragaMF,EstellerM.Epigeneticsandaging:thetargetsand themarks.TrendsGenet2007;23:413–8.http://dx.doi.org/

10.1016/j.tig.2007.05.008.

[33] WeidnerCI,LinQ,KochCM,EiseleL,BeierF,ZieglerP,etal.

AgingofbloodcanbetrackedbyDNAmethylationchanges atjustthreeCpGsites.GenomeBiol2014;15:R24.http://

dx.doi.org/10.1186/gb-2014-15-2-r24.

[34] HorvathS.DNAmethylationageofhumantissuesandcell types.GenomeBiol2013;14:3156.http://dx.doi.org/10.1186/

gb-2013-14-10-r115.

[35] HannumG,GuinneyJ,ZhaoL,ZhangL,HughesG,SaddaS, etal.Genome-widemethylationprofilesreveal

quantitativeviewsofhumanagingrates.MolCell 2013;49:359–67.http://dx.doi.org/10.1016/

j.molcel.2012.10.016.

[36] BocklandtS,LinW,SehlME,Sa´nchezFJ,SinsheimerJS, HorvathS,etal.EpigeneticPredictorofAge.PLoSOne 2011;6:1–6.http://dx.doi.org/10.1371/journal.pone.0014821.

[37] KochCM,WagnerW.Epigenetic-aging-signatureto determineageindifferenttissues.Aging(AlbanyNY) 2011;3:1018–27.http://dx.doi.org/10.18632/aging.100395.

[38] TanQ,HeijmansBT,HjelmborgJVB,SoerensenM, ChristensenK,ChristiansenL.Epigeneticdriftintheaging genome:aten-yearfollow-upinanelderlytwincohort.Int JEpidemiol2016;45:1146–58.http://dx.doi.org/10.1093/ije/

dyw132.

[39] RenaniPG,TaheriF,RostamiD,FarahaniN,AbdolkarimiH, AbdollahiE,etal.Involvementofaberrantregulationof

epigeneticmechanismsinthepathogenesisofParkinson’s diseaseandepigenetic-basedtherapies.JCellPhysioln.d.;0.

https://doi.org/10.1002/jcp.28622.

[40] SingletonAB,HardyJA,GasserT.Thebirthofthemodern eraofParkinson’sDiseaseGenetics.JParkinsonsDis 2017;7:S87–93.http://dx.doi.org/10.3233/jpd-179009.

[41] AbdollahiM,RanjbarA,ShadniaS,NikfarS,RezaieeA.

Pesticidesandoxidativestress:areview.MedSciMonit 2004;10:RA141–7.

[42] BetarbetR,Canet-AvilesRM,ShererTB,Mastroberardino PG,McLendonC.KimJ-H,etal.Intersectingpathwaysto neurodegenerationinParkinson’sdisease:Effectsofthe pesticiderotenoneonDJ-1,(-synuclein,andtheubiquitin– proteasomesystem).NeurobiolDis2006;22:404–20.http://

dx.doi.org/10.1016/j.nbd.2005.12.003.

[43] UverskyVN,LiJ,FinkAL.Pesticidesdirectlyacceleratethe rateofalpha-synucleinfibrilformation:apossiblefactorin Parkinson’sdisease.FEBSLett2001;500:105–8.

[44] KeladaSNP,CheckowayH,KardiaSLR,CarlsonCS,Costa- MallenP,EatonDL,etal.5⁰and3⁰regionvariabilityinthe dopaminetransportergene(SLC6A3),pesticideexposure andParkinson’sdiseaserisk:AHypothesis-Generating Study.HumMolGenet2006;15:3055–62.http://dx.doi.org/

10.1093/hmg/ddl247.

[45] RitzBR,ManthripragadaAD,CostelloC,LincolnSJ,Farrer LJ,CockburnM,etal.DopamineTransporterGenetic VariantsandPesticidesinParkinson’sDisease.Environ HealthPerspect2009;117:964–9.http://dx.doi.org/10.1289/

ehp.0800277.

[46] Benmoyal-SegalL,VanderT,ShifmanS,BrykB,EbsteinRP, MarcusE-L,etal.Acetylcholinesterase/paraoxonase interactionsincreasetheriskofinsecticide-induced Parkinson’sdisease.FASEBJ2005;19:452–4.http://

dx.doi.org/10.1096/fj.04-2106fje.

[47] GattoNM,CockburnM,BronsteinJ,ManthripragadaAD, RitzB.Well-WaterConsumptionandParkinson’s DiseaseinRuralCalifornia.EnvironHealthPerspect 2009;117:1912–8.http://dx.doi.org/10.1289/ehp.0900852.

[48] LeCouteurDG,McLeanAJ,TaylorMC,WoodhamBL,Board PG.PesticidesandParkinson’sdisease.Biomed

Pharmacother1999;53:122–30.http://dx.doi.org/10.1016/

S0753-3322(99)80077-8.

[49] BjorklundG,StejskalV,UrbinaMA,DadarM,ChirumboloS, MutterJ.MetalsandParkinson’sDisease:Mechanisms andBiochemicalProcesses.CurrMedChem2018;25:2198–

214.http://dx.doi.org/10.2174/

0929867325666171129124616.

[50] HamzaTH,ChenH,Hill-BurnsEM,RhodesSL,Montimurro J,KayDM,etal.Genome-WideGene-EnvironmentStudy IdentifiesGlutamateReceptorGeneGRIN2Aasa Parkinson’sDiseaseModifierGeneviaInteractionwith Coffee.PLOSGenet2011;7:1–15.http://dx.doi.org/10.1371/

journal.pgen.1002237.

[51] ElbazA,CarcaillonL,KabS,MoisanF.Epidemiologyof Parkinson’sdisease.RevNeurol(Paris)2016;172:14–26.

http://dx.doi.org/10.1016/j.neurol.2015.09.012.

[52] TaylorAE,DaviesNM,WareJJ,VanderWeeleT,SmithGD, Munafo` MR.Mendelianrandomizationinhealthresearch:

Usingappropriategeneticvariantsandavoidingbiased estimates.EconHumBiol2014;13:99–106.http://dx.doi.org/

10.1016/j.ehb.2013.12.002.

[53] LawlorDA,HarbordRM,SterneJAC,TimpsonN,Davey SmithG.Mendelianrandomization:Usinggenesas instrumentsformakingcausalinferencesinepidemiology.

StatMed2008;27:1133–63.http://dx.doi.org/10.1002/

sim.3034.

[54] CornishAJ,TomlinsonIPM,HoulstonRS.Mendelian randomisation:Apowerfulandinexpensivemethodfor identifyingandexcludingnon-geneticriskfactorsfor

colorectalcancer.MolAspectsMed2019;69:41–7.http://

dx.doi.org/10.1016/j.mam.2019.01.002.

[55] SmithGD,EbrahimS.‘‘Mendelianrandomization’’:can geneticepidemiologycontributetounderstanding environmentaldeterminantsofdisease?IntJEpidemiol 2003;32:1–22.http://dx.doi.org/10.1093/ije/dyg070.

[56] SheehanNA,DidelezV,BurtonPR,TobinMD.Mendelian RandomisationandCausalInferenceinObservational Epidemiology.PLOSMed2008;5:1–6.http://dx.doi.org/

10.1371/journal.pmed.0050177.

[57] SmithGD,EbrahimS.Mendelianrandomization:prospects, potentials,andlimitations.IntJEpidemiol2004.http://

dx.doi.org/10.1093/ije/dyh132.

[58] SmithGD,HemaniG.Mendelianrandomization:

geneticanchorsforcausalinferenceinepidemiological studies.HumMolGenet2014;23:R89–98.http://dx.doi.org/

10.1093/hmg/ddu328.

[59] VlaarT,KabS,SchwaabY,Fre´ryN,ElbazA,MoisanF.

AssociationofParkinson’sdiseasewithindustrysectors:a Frenchnationwideincidencestudy.EurJEpidemiol 2018;33:1101–11.http://dx.doi.org/10.1007/s10654-018-0399-3.

[60] DelamarreA,MeissnerWG.Epidemiology,environmental riskfactorsandgeneticsofParkinson’sdisease.PresseMed 2017;46:175–81.http://dx.doi.org/10.1016/j.lpm.2017.01.001.

[61] NicolettiA,VastaR,MostileG,NicolettiG,ArabiaG,Iliceto G,etal.HeadtraumaandParkinson’sdisease:resultsfrom anItaliancase-controlstudy.NeurolSci2017;38:1835–9.

http://dx.doi.org/10.1007/s10072-017-3076-5.

[62] FactorSA,WeinerWJ.Priorhistoryofheadtraumain Parkinson’sdisease.MovDisord1991;6:225–9.http://

dx.doi.org/10.1002/mds.870060306.

[63] WeisskopfMG,O’ReillyE,ChenH,SchwarzschildMA, AscherioA.PlasmaurateandriskofParkinson’sdisease.

AmJEpidemiol2007;166:561–7.http://dx.doi.org/10.1093/

aje/kwm127.

[64] IwakiH,AndoR,MiyaueN,TadaS,TsujiiT,YabeH,etal.

Oneyearsafetyandefficacyofinosinetoincreasethe serumuratelevelforpatientswithParkinson’sdiseasein Japan.JNeurolSci2017;383:75–8.http://dx.doi.org/10.1016/

j.jns.2017.10.030.

[65] KobyleckiCJ,NordestgaardBG,AfzalS.Plasmaurateand riskofParkinson’sdisease:Amendelianrandomization study.AnnNeurol2018;84:178–90.http://dx.doi.org/

10.1002/ana.25292.

[66] KiaDA,NoyceAJ,WhiteJ,SpeedD,NicolasA,collaborators I,etal.Mendelianrandomizationstudyshowsnocausal relationshipbetweencirculatinguratelevelsand Parkinson’sdisease.AnnNeurol2018;84:191–9.http://

dx.doi.org/10.1002/ana.25294.

[67] SchwarzschildMA,AscherioA,BealMF,CudkowiczME, CurhanGC,HareJM,etal.Inosinetoincreaseserumand cerebrospinalfluidurateinparkinsondiseasea

randomizedclinicaltrial.JAMANeurol2014;71:141–50.

http://dx.doi.org/10.1001/jamaneurol.2013.5528.

[68] Herna´nMA,TakkoucheB,Caaman˜o-IsornaF,Gestal-Otero JJ.Ameta-analysisofcoffeedrinking,cigarettesmoking, andtheriskofParkinson’sdisease.AnnNeurol 2002;52:276–84.http://dx.doi.org/10.1002/ana.10277.

[69] FujimakiM,SaikiS,LiY,KagaN,TakaH,HatanoT,etal.

Serumcaffeineandmetabolitesarereliablebiomarkersof earlyParkinsondisease.Neurology2018;90:e404–11.http://

dx.doi.org/10.1212/WNL.0000000000004888.

[70] CornelisMC,MunafoMR.Mendelianrandomization studiesofcoffeeandcaffeineconsumption.Nutrients 2018;10:1343.http://dx.doi.org/10.3390/nu10101343.

[71] PichlerI,DelGrecoMF,Go¨geleM,LillCM,BertramL,DoCB, etal.SerumIronLevelsandtheRiskofParkinson Disease:AMendelianRandomizationStudy.PLOSMed 2013;10:1–13.http://dx.doi.org/10.1371/

journal.pmed.1001462.

[72] NoyceAJ,KiaDA,HemaniG,NicolasA,PriceTR,DePablo- FernandezE,etal.Estimatingthecausalinfluenceofbody massindexonriskofParkinsondisease:AMendelian randomisationstudy.PLOSMed2017;14:1–19.http://

dx.doi.org/10.1371/journal.pmed.1002314.

[73] BennM,NordestgaardBG,Frikke-SchmidtR,Tybjærg- HansenA,LowLDL.cholesterol,PCSK9andHMGCRgenetic variation,andriskofAlzheimer’sdiseaseandParkinson’s disease:MendelianRandomisationStudy.BMJ

2017;357:j1648.http://dx.doi.org/10.1136/bmj.j1648.

[74] NoyceAJ,KiaD,HeilbronK,JepsonJ,HemaniG, ConsortiumIPDG,etal.Tendencytowardsbeinga

‘‘Morningperson’’increasesriskofParkinson’sdisease:

evidencefromMendelianrandomisation.BioRxiv 2018;288241.http://dx.doi.org/10.1101/288241.

[75] Shokri-KojoriE,WangG-J,WiersCE,DemiralSB,GuoM, KimSW,etal.b-Amyloidaccumulationinthehumanbrain afteronenightofsleepdeprivation.ProcNatlAcadSci 2018;115:4483–8.http://dx.doi.org/10.1073/

pnas.1721694115.

[76] YengoL,SidorenkoJ,KemperKE,ZhengZ,WoodAR, WeedonMN,etal.Meta-analysisofgenome-wide associationstudiesforheightandbodymassindex in700,000individualsofEuropeanancestry.HumMol Genet2018;27:3641–9.http://dx.doi.org/10.1093/hmg/

ddy271.

[77] DehghanA,Ko¨ttgenA,YangQ,HwangSJ,KaoWL, RivadeneiraF,etal.Associationofthreegeneticlociwith uricacidconcentrationandriskofgout:agenome-wide associationstudy.Lancet2008;372:1953–61.http://

dx.doi.org/10.1016/S0140-6736(08)61343-4.

[78] KathiresanS,MelanderO,GuiducciC,SurtiA,BurttNP, RiederMJ,etal.Sixnewlociassociatedwithbloodlow- densitylipoproteincholesterol,high-densitylipoprotein cholesterolortriglyceridesinhumans.NatGenet 2008;40:189–97.http://dx.doi.org/10.1038/ng.75.

[79] FigueiredoJC,HsuL,HutterCM,LinY,CampbellPT,Baron JA,etal.Genome-WideDiet-GeneInteractionAnalysesfor RiskofColorectalCancer.PLoSGenet2014;10:e1004228.

http://dx.doi.org/10.1371/journal.pgen.1004228.

[80] TrampushJW,YangMLZ,YuJ,KnowlesE,DaviesG, LiewaldDC,etal.GWASmeta-analysisrevealsnovelloci andgeneticcorrelatesforgeneralcognitivefunction:A reportfromtheCOGENTconsortium.MolPsychiatry 2017;22:1651–2.http://dx.doi.org/10.1038/mp.2016.244.

[81] DeMoorMHM,CostaPT,TerraccianoA,KruegerRF,De GeusEJC,ToshikoT,etal.Meta-analysisofgenome-wide associationstudiesforpersonality.MolPsychiatry 2012;17:337–49.http://dx.doi.org/10.1038/mp.2010.128.

[82] HippG,VaillantM,DiederichNJ,RoompK,SatagopamVP, BandaP,etal.TheLuxembourgParkinson’sStudy:a ComprehensiveApproachforStratificationandEarly Diagnosis.FrontAgingNeurosci2018;10:326.http://

dx.doi.org/10.3389/FNAGI.2018.00326.