Article
Reference
Sociodemographic, behavioral and genetic determinants of allostatic load in a Swiss population-based study
PETROVIC, Dusan, et al.
Abstract
Allostatic load (AL) is a marker of physiological dysregulation which reflects exposure to chronic stress. High AL has been related to poorer health outcomes including mortality. We examine here the association of socioeconomic and lifestyle factors with AL. Additionally, we investigate the extent to which AL is genetically determined. We included 803 participants (52% women, mean age 48±16years) from a population and family-based Swiss study. We computed an AL index aggregating 14 markers from cardiovascular, metabolic, lipidic, oxidative, hypothalamus-pituitary-adrenal and inflammatory homeostatic axes. Education and occupational position were used as indicators of socioeconomic status. Marital status, stress, alcohol intake, smoking, dietary patterns and physical activity were considered as lifestyle factors. Heritability of AL was estimated by maximum likelihood. Women with a low occupational position had higher AL (low vs. high OR=3.99, 95%CI [1.22;13.05]), while the opposite was observed for men (middle vs. high OR=0.48, 95%CI [0.23;0.99]). Education tended to be inversely associated with AL in both sexes(low vs. [...]
PETROVIC, Dusan, et al . Sociodemographic, behavioral and genetic determinants of allostatic load in a Swiss population-based study. Psychoneuroendocrinology , 2016, vol. 67, p. 76-85
DOI : 10.1016/j.psyneuen.2016.02.003 PMID : 26881833
Available at:
http://archive-ouverte.unige.ch/unige:89455
Disclaimer: layout of this document may differ from the published version.
1 / 1
ContentslistsavailableatScienceDirect
Psychoneuroendocrinology
j ou rn a l h om epa ge :w w w . e l s e v i e r . c o m / l o c a t e / p s y n e u e n
Sociodemographic, behavioral and genetic determinants of allostatic load in a Swiss population-based study
Dusan Petrovic
a, Edward Pivin
a, Belen Ponte
a,b, Nasser Dhayat
c, Menno Pruijm
d, Georg Ehret
e, Daniel Ackermann
c, Idris Guessous
a,f, Sandrine Estoppey Younes
a, Antoinette Pechère-Bertschi
b, Bruno Vogt
c, Markus Mohaupt
c, Pierre-Yves Martin
b, Fred Paccaud
a, Michel Burnier
d, Murielle Bochud
a, Silvia Stringhini
a,∗aInstituteofSocialandPreventiveMedicine(IUMSP),LausanneUniversityHospital,Routedelacorniche10,1010Lausanne,Switzerland
bDepartmentofNephrologyandHypertension,UniversityHospitalofGeneva(HUG),RueGabriellePerret-Gentil4,1205Geneva,Switzerland
cUniversityClinicforNephrology,HypertensionandClinicalPharmacology,Inselspital,BernUniversityHospitalandUniversityofBern,Freiburgstrasse15, 3010Bern,Switzerland
dDepartmentofNephrologyandHypertension,LausanneUniversityHospital,RueduBugnon17,1011Lausanne,Switzerland
eDepartmentofCardiology,UniversityHospitalofGeneva(HUG),RueGabriellePerret-Gentil4,1205Geneva,Switzerland
fUnitofPopulationEpidemiology,DivisionofPrimaryCareMedicine,DepartmentofCommunityMedicineandPrimaryCareandEmergencyMedicine, UniversityHospitalofGeneva(HUG),RueGabriellePerret-Gentil4,1205Geneva,Switzerland
a r t i c l e i n f o
Articlehistory:
Received21October2015
Receivedinrevisedform3February2016 Accepted3February2016
Keywords:
Allostaticload socioeconomicstatus physiologicaldysregulation population-based heritability
a b s t r a c t
Allostaticload(AL)isamarkerofphysiologicaldysregulationwhichreflectsexposuretochronicstress.
HighALhasbeenrelatedtopoorerhealthoutcomesincludingmortality.Weexamineheretheassoci- ationofsocioeconomicandlifestylefactorswithAL.Additionally,weinvestigatetheextenttowhich ALisgeneticallydetermined.Weincluded803participants(52%women,meanage48±16years)from apopulationandfamily-basedSwissstudy.WecomputedanALindexaggregating14markersfrom cardiovascular,metabolic,lipidic,oxidative,hypothalamus-pituitary-adrenalandinflammatoryhomeo- staticaxes.Educationandoccupationalpositionwereusedasindicatorsofsocioeconomicstatus.Marital status,stress,alcoholintake,smoking,dietarypatternsandphysicalactivitywereconsideredaslifestyle factors.HeritabilityofALwasestimatedbymaximumlikelihood.Womenwithalowoccupationalposi- tionhadhigherAL(lowvs.highOR=3.99,95%CI[1.22;13.05]),whiletheoppositewasobservedformen (middlevs.highOR=0.48,95%CI[0.23;0.99]).EducationtendedtobeinverselyassociatedwithALin bothsexes(lowvs.highOR=3.54,95%CI[1.69;7.4]/OR=1.59,95%CI[0.88;2.90]inwomen/men).Heavy drinkingmenaswellaswomenabstainingfromalcoholhadhigherALthanmoderatedrinkers.Physical activitywasprotectiveagainstALwhilehighsaltintakewasrelatedtoincreasedALrisk.Theheritability ofALwasestimatedtobe29.5%±7.9%.Ourresultssuggestthatgeneralizedphysiologicaldysregulation, asmeasuredbyAL,isdeterminedbybothenvironmentalandgeneticfactors.Thegeneticcontribution toALremainsmodestwhencomparedtotheenvironmentalcomponent,whichexplainsapproximately 70%ofthephenotypicvariance.
©2016ElsevierLtd.Allrightsreserved.
1. Introduction
Allostaticload(AL)isanindicatorofbiologicaldysregulation representingthecumulativephysiologicaltollexperiencedbyan organismwhenitfailstoadequatelyrespondtochronicpsychoso- cialorphysicalchallengesfromtheenvironment(Dowdetal.,2009;
McEwen,1998).IntroducedintheearlyninetiesbyMcEwenand
∗Correspondingauthor.
E-mailaddress:silvia.stringhini@chuv.ch(S.Stringhini).
Stellar(1993),ALismeasuredthroughasingleindex,resultingfrom acombinationofbiologicalmarkersreflectingthestatesofseveral axesincludingcardiovascular,metabolic,dyslipidemic,neuroen- docrine,hypothalamus-pituitary-adrenal(HPA)andinflammatory (Nicodetal.,2014;Seemanetal.,2010).HighALhasbeenrelated toseveraladversehealthoutcomes,includingphysicalandcogni- tivefunctioning,symptomsofposttraumaticstressdisorder,risk ofcardiovascularevents(Crimminsetal.,2003;Justeretal.,2010;
Seemanetal.,2001),andall-causemortality(Seemanetal.,2004).
TheconceptofALwasoriginallyintroducedtorepresentthe physiologicalconsequencesofchronicstress,itselfinfluencedby
http://dx.doi.org/10.1016/j.psyneuen.2016.02.003 0306-4530/©2016ElsevierLtd.Allrightsreserved.
D.Petrovicetal./Psychoneuroendocrinology67(2016)76–85 77
Fig.1.Simplifiedconceptualframeworkforthedeterminantsofallostaticload.ALisrepresentedwithitssixhomeostaticaxes.Itmaybeinfluenceddirectlybygenetic factors(arrowA)andSES(B),orindirectlythroughbehaviouralandpsychosocialfactors(arrowsCandD).BehaviouralandpsychosocialfactorsmayalsoinfluenceALdirectly (E).
socioeconomicstatus(SES),healthbehaviorsorpsychosocialfac- tors.Inthiscontext,manystudieshavefoundastrongassociation betweenlowSES,asreflectedbyloweducation,adversefinancial conditionsorreceivingsocialtransfers,andhighAL(Gruenewald etal.,2012; Nicodet al.,2014).Therole ofhealth behaviorsin relationtochronicstressandhealthhasalsobeeninvestigatedin previousresearch,whichshowedthatindividuals confrontedto stressfuldailylife(i.e.poverty,crime)arepronetoengagethem- selvesintounhealthybehaviors suchas smokingorovereating, whichmayhelpalleviatesymptomsofpsychologicalstress.How- ever,despitethesepositive,short-termpsychologicaleffects,an unhealthylifestylehasdetrimentalphysiologicalconsequenceson thelongtermandthusresultsinincreasedmorbidityandmortality (Jacksonetal.,2010).
However,theassociations betweenhealth behaviors and AL havenotalwaysbeenconsistentacrossstudies.Galloetal.(2011) showed, for example, that moderate alcohol consumption was associatedwithdecreasedAL,whereasCrimminsetal.(2009)found noassociationbetweenalcoholintakeandAL.Similarly,resultsfor theeffectofsmokingonALwereinconsistent(Crimminsetal., 2009;Huetal.,2007).Finally,AL hasmainlybeenstudiedasa consequenceofchronicenvironmentaldemands,whereasalim- itednumberofstudieshaveexaminedthecontributionofselected geneticdeterminantsusingacandidategeneapproachtothisphe- notype(Brodyet al.,2013; Cicchetti etal., 2011).Thecomplex natureofALsuggeststhatthisphenotypeisinfluencedbymore than one gene (i.e. a polygenic trait). However,previous stud-
ieshavemainlyfocusedontheroleofspecificgeneticmarkers, whichareinvolvedinresponsestocontextualstress,includingSES- associatedrisks,familyorpersonalpressureandtheresponseto physicalabuse(Brodyetal.,2013;Cicchettietal.,2011).Todate, two markers havebeen identified,the SLC6A4serotonintrans- portergene,whoseshortervariantwasassociatedwithhighAL, andCRHR1corticotropinreleasinghormonereceptor1gene,which is involved inHPA axisregulation, andwhose TAT variantwas associated withhighAL. However,to ourknowledge,no study hasyetinvestigatednorassessedheritabilityofAL,whichallows thedeterminetheoverallgeneticcontributiontothisphenotype, irrespectiveofthespecificfunctionofselectedgenes.
In this study,we examine the association of socioeconomic (educationandoccupation)andbehavioralfactors(maritalstatus, smoking,alcoholconsumption,physicalactivity,dietarypatterns, andstress)withALusingdatafromaSwisspopulation-basedstudy.
Further,weinvestigatetheextenttowhichALisgeneticallydeter- minedbyassessingnarrowsenseheritability.Wehypothesizethat ALisinfluencedbybothenvironmental(socioeconomicandbehav- ioral)andgeneticfactors(Fig.1).
2. Methods
2.1. Studypopulationanddesign
DataweredrawnfromtheSKIPOGHstudy(SwissKidneyProject onGenesinHypertension),amulticenterfamily-basedpopulation
studyinitiatedin2009toexplorethegeneticandenvironmental determinantsofbloodpressure(Alwanetal.,2014;Pruijmetal., 2013).
StudyparticipantswererecruitedinthecantonsofBernand GenevaandthecityofLausanne.RecruitmentbeganinDecember 2009andendedinApril2013.Indexcaseswererandomlyselected fromthe population-basedCoLaus study in Lausanne(Firmann etal.,2008),andfromthepopulation-basedBusSantéstudyin Geneva(Guessousetal.,2012).InBern,indexparticipantswere randomlyselectedusingthecantonalphonedirectory.Inclusion criteriawere:(1)writteninformedconsent;(2)minimumageof 18 years;(3) Caucasianorigin; (4) atleast one, and preferably three,first-degreefamilymembersalsowillingtoparticipate.At theendoftherecruitmentperiod,thestudypopulationincluded 1128participants.TheSKIPOGHstudywasapprovedbytheethi- calcommitteesofLausanneUniversityHospital,GenevaUniversity HospitalandtheUniversityHospitalofBern(Ponteetal.,2014).
Participantscamefrom271distinctfamilystructures(pedigrees), most of which included three generations and second degree links(i.e.cousins).Themeanpedigreesize(±SD)was5.05±2.26, withthelargestnuclearfamily(parent–childrenonly)including8 members.Thesepedigreesledto1444parent–offspringpairs,462 siblingpairs,213avuncularpairs,310grandparents–grandchildren pairsand44cousinpairs.
2.2. Clinicalandbiologicaldata
Participantscameforthestudyvisitatoneofthethreemedical centersinthemorning,andfilledinastandardizedquestionnaire athome.Thequestionnairefocusedonavarietyofissuesinclud- inglifestylehabitsaswellasmedicalhistory.Bodyweight(kg), height(cm)and waistand hipcircumferences(cm)weremea- suredaccordingtostandardprocedures.Bodymassindex(BMI) wasdefined asweightinkgdividedbythesquare ofheightin meters.Bloodpressureandheartrateweremeasuredafter10min ofrestinthesittingpositionwithavalidatednon-mercuryaus- cultatorysphygmomanometer(A&DUM-101,A&DCompany,Ltd., ToshimaKu,Tokyo,Japan).Eachparticipant’soffice bloodpres- sureandheartratewerethemeansoffiveconsecutivereadings.
Venousbloodsamplesweredrawnafteranovernightfast.Elec- trolytes,kidneyandliver-functiontests,bloodglucose,cholesterol, triglycerides, insulin, C-reactive protein (CRP), serum uric acid, gamma-glutamyltransferase(GGT)and otherbiological markers weremeasuredinlocaluniversitylaboratoriesusingstandardclin- icallaboratorymethods.Participantswerealsoaskedtocollecta 24-hurinesampleforthemeasurementofurinaryvolume,urinary sodiumandadditionalparameters(Alwanetal.,2014;Ponteetal., 2014;Pruijmetal.,2013).
2.3. Socioeconomicstatus(SES)
Two indicators of SES were used: educational level and occupational position. Highest level of education attained was self-reportedand furtherclassifiedintothree categories:“High”
(University education), “Middle” (Higher secondary education), and“Low”(Lower secondary educationor lower).Occupational position was self-reported and grouped into three categories:
“High”(Managers:liberalprofessions,directors,professors),“Mid- dle” (Lower level executives: teachers, qualified technicians, nurses)and“Low”(Lowqualifiednon-manualsandmanuals:sales assistants,clerks,manualworkers).Participantswhowerenotcur- rently working were assigned their past occupational position.
Participantswho had never worked (studentsand housewives) werenotincludedintheanalysis.
2.4. Lifestylefactors
Lifestyle factors were self reported. Maritalstatus was cat- egorized as “Living alone” or “Living in a couple”. Alcohol consumptionwasassessedusingquestionsonthenumberofalco- holicdrinksusuallyconsumedwithinaweek,thencategorizedas
“Abstainers”(0unit/week;1unit=10gofpurealcohol)“Moder- ate”(1–21/1–14units/weekformen/women)or“Heavydrinking”
(≥21/≥14units/weekformen/women).Smokingstatuswascat- egorizedascurrentandnoncurrentsmoking,thelattercategory includingnever smokers and ex-smokers. Physicalactivity was reported on a scale from 1 to 10, 1 corresponding to a com- pletesedentarylifestyle and 10 correspondingtomanualwork combinedwith sportspractice. Based onthis scale, three cate- goriesweresubsequentlydefined:“Low”(1–4),“Moderate”(5), and“High”(6–10). Dailysalt intakewasassessedthrough24-h urinarysodiumexcretion(mmol/24h)andcategorizedas“Upto 5g/day”,“5–10g/day”and“>10g/day”.Weeklymeatconsumption wascategorized as“Low”(Never–Once/week), “Moderate”(2–4 times/week)and“High”(5–7times/week).Dailyfruitandvegetable consumptionwasclassifiedas“Low”(0–3portions/day),“Moder- ate”(3portions/day)and“High”(>3portions/day).Perceivedstress level(referredtoas“Stress”)wasassessedthroughthequestion:
“Pleaseindicateonascalefrom1to10thepsychologicaltensions andstresstowhichyouareexposedinyoureverydaylife/Veuillez notersuruneéchellede1à10lestensionspsychologiquesetlestress auxquelvousêtesconfrontésactuellementdansvotreviequotidienne [FR]/BittebeurteilenSieanhandeinerSkalavon1–10Ihretägliche AnspannungundStressinIhremAlltag[DE]”withpossibleanswers onascalefrom1to10.Itwasfurthersubdividedinto“Low”(1–3),
“Moderate”(4–6)and“High”(7–10).
2.5. Othercovariates
Use of anti-hypertensive drugs (beta-blockers, angiotensin- converting-enzyme inhibitors, angiotensin receptor blockers, calciumchannelblockers,diuretics),lipid-loweringdrugs(statin, fibrates) and anti-diabetes drugs (oral anti-diabetics, insulin) wasincludedaspotentialconfoundingeffectsintheassociation betweenSESandAL(sensitivityanalyses).
2.6. Allostaticload
WeanalyzedtheconstitutingriskfactorsofALingroupscorre- spondingtosixphysiologicalsystems:cardiovascular,metabolism, hypothalamic-pituitary-adrenalaxis(HPA),lipidicaxis,inflamma- tionand oxidativestress (Galloet al.,2011; Nicodet al.,2014;
Seemanetal.,2010).Comparedtothemarkersusuallyincluded intheassessmentofAL,weomittedautonomicnervoussystem parameters(i.e.adrenaline,noradrenaline)astheywerenotavail- ableinSKIPOGH.Moreover,wedecidedtoincludeanoxidative stressaxisaschronicoxidativestresshasalsobeenlinkedtogener- alizedphysiologicaldysregulation(Devakietal.,2013;Nicodetal., 2014).Further,weseparatedthelipidsfromthemetabolismaxes contrarytopreviousstudies.In total, weassessed14 biological markerswithin6 homeostaticdimensions:mean systolicblood pressure,meandiastolicbloodpressureandheartrate(cardiovas- cularsystem);bloodglucose,bloodinsulin,bmiandwaist-to-hip ratio(metabolism);24hurinecortisol(HPA);high-densitylipopro- teincholesterol, totalcholesterol andtriglycerides(lipidicaxis), serumuricacidandGGT(oxidativestress)andCRP(inflammation).
AllphysiologicalparameterscomposingALwerestratifiedbysex andaresummarizedinSupplementaryTable1(onlineresource1).
Eachbiologicalmarkerwasdichotomizedintohighversuslow- riskvalues(1–0)accordingtoclinicalthresholdsasfoundinthe literature(Ascasoetal.,2003;DowdandGoldman,2006;Dowd
D.Petrovicetal./Psychoneuroendocrinology67(2016)76–85 79
Fig.2. Mixedlinearregressionadjustedmeans(±SE)forallostaticloadbyoccupationalpositionandeducation.Model1(M1)wasadjustedforageandcenterandModel2 (M2)wasadditionallyadjustedformaritalstatus,lifestylebehaviorsandstress.
etal.,2009;Karlamanglaetal.,2012;Krishnamurthy,2013;Perk etal.,2012;Ridker,2003;Zoppinietal.,2012).TheALscorewas computedbysummingthedichotomizedvaluesandthusranged between0and14.TheALscoreandeachofthesixhomeostatic dimensionswerefurtherdichotomizedintohighversuslowriskby usingasacut-offthevalueclosesttothemedian(Supplementary Table2—online resource2).Thiscut-off waschosenasprevious studieshavefoundthatdifferencesinmorbidityormortalityoccur betweengroupswhenALwasdichotomizedatthemedianorat scoresof3–4(Geronimusetal.,2006;Smithetal.,2009).”
2.7. Statisticalanalyses
TheassociationsofSESindicatorsandlifestylefactorswithAL wereanalyzedusingaminimallyadjustedmixedlogisticregres- sionmodelandafullyadjustedmodel,whichincludedadditional adjustmentforallfactors.TheassociationbetweenSESandcon- tinuousAL scorewasfurtherinvestigated throughmixedlinear regressionmodels(M1:minimallyadjustedandM2:fullyadjusted) forwhichadjustedmeanswerecalculated.Astheassociationof occupationalpositionandseverallifestylefactorswithALdiffered bysex(pforinteraction<0.05),allanalyseswerestratifiedbysex.
TheassociationsofSES andlifestyle factorswithAL weresimi- larforthethreecenters(pforinteraction>0.05),sodatafromthe threecenterswerepolledandallanalyseswereadjustedbycenter.
Familialcorrelationsweretakenintoaccountforallanalyses.We assessedheritabilityaspreviouslydescribed(Pruijmetal.,2013).
Heritabilityisameasureoffamilialresemblancethatreliesonthe assumptionthattotalphenotypicvarianceofatraitcanbeparti- tionedintoindependentgeneticandenvironmentalcomponents.
Thegeneticvariancecanfurtherbesubdividedintothreecompo- nents:anadditiveorpolygenicgeneticvariance(1),adominance variance (2) and an epistaticvariance (3). The additivegenetic componentrepresentstheaverageeffectsofindividualalleleson atraitandreflectstransmissibleresemblancebetweenrelatives.
Heritabilityinthenarrowsenseisdefinedastheratiooftheaddi- tivegeneticvariancetothetotalphenotypicvariance.Inthispaper wereferto“heritabilityinthenarrowsense”simplyasheritabil- ity.WeestimatedtheheritabilityofALwithinamodelwhichwas adjustedforage,sexandcenter.Forheritabilitymeasure,totalphe- notypicvariance wassubdividedintorandom(R),polygenic(P) (additivegeneticvariance)andmarital(M)components.Sibship componentwasnotincludedinthemodelasitdidnotsignificantly contributetothetotalphenotypicvariance.For heritability,the SAGEsoftware(StatisticalAnalysisforGeneticEpidemiology)from theASSOCprogramwasused(Ponteetal.,2014).Allotheranalyses wereconductedinSTATA13,(StataCorp.,CollegeStation,Texas, USA).Atwo-sidedp-value<0.05wasusedasasignificantthresh- old.Fig.1wasgeneratedusingMSOfficePowerPointandFig.2was generatedusingMSOfficeExcelandPowerPoint.
3. Results
Of the 1128 participants of the SKIPOGH study, 250 were excludedbecauseofmissingvaluesorincompletedescriptionon oneormorecovariates(N=47foreducationoroccupationalposi- tion,N=171forallostaticloadandN=32forlifestylefactors)and 75 participantswereexcluded becausetheywerenot currently workingandhadnopreviousoccupation(68studentsand7house- wives).Intotal,803individualswereincludedinthepresentstudy, ofwhich388weremen(48%).Excludedparticipantswereyounger (meanage45vs.48years,p<0.05)andtendedtohavealoweredu- cation(13%vs.23%inthehigheducationgroup,p<0.05)thanthose includedinthestudy.
Table 1 summarizesthe main characteristics of thesample.
Womenweremorefrequentlyinalowoccupationalposition,were less frequentlysmokers, weremore frequentlyalcohol abstain- ersandhadhealthierdietarypatterns(lowermeatconsumption anddailysaltintakeandhigherfruitandvegetableconsumption) comparingtomen(allp<0.001).
Table1
Baselinecharacteristicsofparticipantsincludedinthestudy.
Men(N=388) Women(N=415) p-Valuea
Age,mean(±SD,years) 48.28(±16.51) 48.43(±15.65) 0.893
Center 0.040
Lausanne 108(28%) 150(36%)
Geneva 168(43%) 156(38%)
Bern 112(29%) 109(26%)
Educationalattainment 0.379
High 95(24%) 86(21%)
Middle 116(30%) 123(30%)
Low 177(46%) 206(50%)
Occupationalposition <0.001
High 70(18%) 32(8%)
Middle 130(34%) 127(31%)
Low 188(48%) 256(62%)
Maritalstatus 0.241
Livingalone 97(25%) 119(29%)
Livingincouple 291(75%) 296(71%)
Smoking <0.001
No 271(70%) 333(80%)
Yes 117(30%) 82(20%)
Alcoholconsumption <0.001
Moderate 249(64%) 191(46%)
Abstainers 91(23%) 192(46%)
Heavydrinkers 48(12%) 32(8%)
Physicalactivity 0.053
Low 121(31%) 143(34%)
Moderate 92(24%) 119(29%)
High 175(45%) 153(37%)
Dailyfruitandvegetablesconsumption <0.001
Low 168(43%) 90(22%)
Moderate 129(33%) 166(40%)
High 91(23%) 159(38%)
Meatconsumption <0.001
Low 27(7%) 58(14%)
Moderate 204(53%) 250(60%)
High 157(40%) 107(26%)
Saltintake <0.001
Upto5g/day 32(8%) 88(21%)
5–10g/day 194(50%) 253(61%)
>10g/day 162(42%) 74(18%)
Stresslevel 0.752
Low 110(28%) 124(30%)
Moderate 174(45%) 189(46%)
High 104(27%) 102(25%)
DataareN(%)unlessotherwisespecified.
ap-ValuewascomputedaccordingtoChi2testwasbetweeneachvariableandsex.
Table2
Associationbetweensocioeconomicindicatorsandallostaticload.
Men(N=388) Women(N=415)
Adjustedforage andcenter
Adjustedforage, center,lifestyle factors
Adjustedforage andcenter
Adjustedforage, center,lifestyle factors
OR(95%CI) pa OR(95%CI) pa OR(95%CI) pa OR(95%CI) pa
Occupational position
High(Ref.) 1.00 0.600 1.00 0.480 1.00 0.042 1.00 0.190
Middle 0.48[0.23;0.99] 0.43[0.19;0.94] 3.39[1.00;11.48] 3.19[0.88;11.56]
Low 0.69[0.34;1.38] 0.62[0.30;1.31] 3.99[1.22;13.05] 3.23[0.92;11.36]
Education High(Ref.) 1.00 0.107 1.00 0.137 1.00 <0.001 1.00 0.004
Middle 1.07[0.57;2.00] 1.13[0.58;2.20] 2.30[1.08;4.89] 2.50[1.11;5.63]
Low 1.59[0.88;2.90] 1.61[0.84;3.10] 3.54[1.69;7.40] 3.40[1.53;7.57]
OR,oddsratio;CI,confidenceinterval;Ref.,referencelevel.
ap-Valueforlineartrendacross>2categories.
D.Petrovicetal./Psychoneuroendocrinology67(2016)76–8581 Table3
Associationbetweenlifestylefactorsandallostaticload.
Men(N=388) Women(N=415)
Adjustedfor ageandcenter
Adjustedfor age,center, educationand occupation
Adjustedfor ageandcenter
Adjustedfor age,center, educationand occupation
OR(95%CI) pa OR(95%CI) pa OR(95%CI) pa OR(95%CI) pa
Maritalstatus Livingalone(Ref.) 1.00 0.387 1.00 0.389 1.00 0.160 1.00 0.187
Livinginacouple 1.30[0.72;2.37] 1.31
[0.71;2.41)]
0.68[0.40;1.16] 0.70[0.41;1.19]
Smoking No(Ref.) 1.00 0.137 1.00 0.143 1.00 0.617 1.00 0.949
Yes 1.51[0.88;2.59] 1.51[0.87;2.63] 1.17[0.63;2.15] 1.02[0.56;1.85]
Alcoholconsumption Abstainers 0.83[0.47;1.47] 0.195 0.79[0.44;1.40] 0.209 1.90[1.14;3.17] 0.116 1.72[1.03;2.85] 0.166
Moderate(Ref.) 1.00 1.00 1.00 1.00
Heavydrinkers 2.28[0.98;5.27] 2.41[1.02;5.72] 1.09[0.43;2.77] 1.11[0.44;2.83]
Physical activity
Low(Ref.) 1.00 0.011 1.00 0.006 1.00 0.008 1.00 0.003
Moderate 0.41[0.21;0.81] 0.37[0.18;0.74] 0.86[0.49;1.51] 0.79[0.45;1.38]
High 0.44[0.24;0.80] 0.40[0.21;0.74] 0.45[0.25;0.81] 0.43[0.24;0.75]
Fruitsand vegetables
Low(Ref.) 1.00 0.289 1.00 0.408 1.00 0.106 1.00 0.325
Moderate 0.77[0.45;1.33] 0.84[0.48;1.48] 0.77[0.42;1.41] 0.81[0.44;1.47]
High 0.74[0.40;1.36] 0.77[0.41;1.46] 0.60[0.33;1.12] 0.72[0.39;1.36]
Meat Low(Ref.) 1.00 0.346 1.00 0.372 1.00 0.310 1.00 0.375
Moderate 1.92[0.75;4.96] 1.82[0.67;4.91] 1.94[0.91;4.11] 1.80[0.86;3.78]
High 1.98[0.75;5.22] 1.90[0.69;5.21] 1.77[0.76;4.08] 1.64[0.72;3.75]
Saltintake Upto5g(Ref.) 1.00 0.087 1.00 0.184 1.00 0.055 1.00 0.064
5–10g 1.77[0.71;4.42] 1.69[0.67;4.28] 0.95[0.52;1.73] 0.96[0.53;1.76]
>10g 2.26[0.90;5.70] 1.99[0.78;5.08] 2.26[1.03;4.95] 2.20[1.00;4.86]
Stress Low(Ref.) 1.00 0.921 1.00 0.785 1.00 0.803 1.00 0.775
Moderate 0.88[0.49;1.59] 0.83[0.45;1.51] 1.33[0.76;2.31] 1.42[0.82;2.44]
High 0.96[0.50;1.84] 0.90[0.46;1.77] 1.06[0.55;2.05] 1.06[0.55;2.04]
OR,oddsratio;CI,confidenceinterval;Ref.,referencelevel.
ap-Valueforlineartrendacross>2categories.
Table4
Totalvariancepartitionandnarrowsenseheritabilityforallostaticload.
Variancecomponent Estimate(±SD) p-Value
Random(R) 0.35(±0.08) <0.001
Polygenic(P) 0.21(±0.06) <0.001
Marital(M) 0.14(±0.06) 0.01
Total(T) 0.71(±0.03) <0.001
Heritability(P/T)a 29.55%(±7.96%) <0.001
aNarrowsenseheritability(polygenic/total).
Table2showsresultsfortheassociationofSESindicatorswith AL.Menin themiddlevs.highoccupationalpositionhadlower AL(OR=0.48,95%CI[0.23;0.99])whilethosewithalowvs.high educationtendedtohavehigherAL(OR=1.59,95%CI[0.88;2.90]).
Womenin thelowest vs.highest occupational and educational grouphad higher AL (OR=3.99, 95%CI [1.22;13.05]for occupa- tionalpositionandOR=3.54,95%CI[1.69;7.4]foreducation).The associationofoccupationalpositionwithALinwomenwaspartly attenuated after adjusting for lifestyle factors (OR=3.23, 95%CI [0.92;11.36]).Further,we observeda dose-responseassociation betweeneducationandALinmenandwomen(Fig.2allp<0.05).
Resultsfortheassociationoflifestylefactorsand stresswith ALareshowninTable3.Heavydrinkingtendedtobeassociated withincreasedALinmen(OR=2.28,95%CI[0.98;5.27]),whereas abstainingwasassociated withhigherAL in women (OR=1.90, 95%CI[1.14;3.17]).Participantswhowerephysicallyactivehada decreasedriskofhighAL:OR=0.44,95%CI[0.24;0.80]/0.45,95%CI [0.25;0.81]inmen/women.Highsaltintakewasassociatedwith highALinwomen(OR=2.26,95%CI[1.03;4.95]),andtendedtobe associatedinmen(OR=2.26,95%CI[0.90;5.70]).Peopleconsuming ahighamountoffruitandvegetablehadalowerAL,butthedif- ferencewasnotsignificant,whereashighmeatconsumptionand smokingtendedtobeassociatedwithhighAL.
EstimatedheritabilitymeasuresforALarepresentedinTable4.
Heritability for allostatic load was 29.5%±7.9% in the model adjustedforage,sexandcenter.Random,polygenic(additive)and maritalvariancessignificantlycontributedtothetotalphenotypic variance(T).
InSupplementaryTables4–8(onlineresources4–8),wepresent resultsfortheassociationofSESindicators,lifestylefactorsand stresswitheachhomeostaticdimension.Lowereducationtended tobeassociatedwithhighriskofallhomeostaticaxes,exceptfor HPAandlipids.Loweroccupationalpositiontendedtobeassoci- atedwithincreaseddysregulationofthemetabolicandHPAaxes.
Theseassociationsweregenerallystrongerinwomenthaninmen.
Highphysicalactivitywasassociatedwithlowerdysregulationof cardiovascular,metabolic,lipidicaxesandtendedtobeassociated withlowoxidativestress.Heavydrinkingwasassociatedwithdele- teriousdysregulationofcardiovascularandoxidativestressaxes.
Increasedfruitand vegetableconsumption wasassociatedwith lowriskofmetabolicdysregulation.Finally,highsaltintakewas associatedwithhighriskofdysregulationofmetabolicandHPA axes.
4. Sensitivityanalyses
Weconductedasensitivityanalysistoassesswhethermedica- tionintake(anti-hypertensive,lipid-lowering,anti-diabeticdrugs, enteredasseparatedummyvariables)couldconfoundoratten- uatetheassociationbetweenSESand AL.We observedthat an additionaladjustmentforthesecompoundsattenuatedtheasso- ciationbetweenoccupationalpositionandALinwomen(adjusted forage,sex,centerOR=3.9995%CI[1.22;13.05];+lifestylefactors 3.23[0.92;11.36];+medication3.03[0.85;10.78])butnot inmen.
Wealsoobservedanattenuationfortheassociationbetweenoccu-
pationalpositionandtheoxidativestressaxisinwomenbutnotin men.
5. Discussion
Inthismulticentricpopulationandfamily-basedSwissstudy, wefoundastrongassociationbetweenSES,severallifestylefactors, andAL,ameasureofgeneralizedphysiologicaldysregulationand astrongpredictorofmorbidityandall-causemortality(Seeman etal.,2001).Occupationalpositionandeducationwerenegatively relatedwithALinwomen,whiletheassociationsinmentended tobepositive foroccupationalpositionandnegativeforeduca- tion.PhysicalactivitywasnegativelyassociatedwithAL,saltintake waspositivelyassociated,whereastheassociationbetweenalco- holconsumptionandALwasdependentonsex.Finally,ourresults showasignificantgeneticcomponentforAL,asmeasuredbyheri- tability,independentlyofage,center,SESandlifestylefactors.
OccupationalpositionwasnegativelyrelatedtoALinwomen, withwomen witha low occupational position havinga higher riskofhighALthantheirmoreadvantagedcounterparts,inline withresultsfrom otherstudies(Galloet al.,2011;Juster etal., 2013).However,this associationwasreversedin men.A possi- bleexplanationfortheseresultsisthatmeninhighoccupational positionsmayhavehigh-demand jobswithlongworkinghours andconsiderableprofessionalresponsibilities,potentiallyleading tohigherstress,comparingtomenoccupyingnon-manualinter- mediateoccupationscharacterizedbylowerdemandand atthe sametimenotinvolvingphysicalefforts.Ontheotherhand,the adversehealtheffects oflowoccupationalpositionmaybepar- ticularlysevereinwomenastheygenerallyhavetocombinethe physicalandpsychosocialstrainofmanual,lesspaidjobs(Bonjour andGerfin,2001)tothatofhouseholdresponsibilities(Artazcoz etal.,2004).Forexample,inapreviousstudypartlybasedonthe samepopulationofourstudy(Alwanetal.,2014),womenwitha lowoccupationalpositionweremoreaffectedbysleepdeprivation thanmenregardlessofsocialclass.Sexdifferencesintheassoci- ationbetweenoccupationalpositionandALhavebeenreported inapreviousstudyinMontreal(Justeretal.,2013).Aproposed explanationforthisreversedassociationwasthatworkrelatedpsy- chosocialfactors,suchaspsychologicaldemands,decisionlatitude andsocialsupportinteractinsex-specificwayswithAL(Justeretal., 2013).However,wemustpointoutthatthemajorityofstudiesin thisfieldreportanegativeassociationbetweenoccupationalposi- tionandALforbothsexes(Gustafssonetal.,2011).Participants withaloweducationexperiencedhigherphysiologicaldysregula- tionasmeasuredbyAL,inlinewithpreviousresearch(Howard andSparks,2015;Nicodetal.,2014;Seemanetal.,2004)including astudyperformedinaSwisspopulation(Nicodetal.,2014).This mayberelatedtoseveralfactorssuchashealth-relatedknowledge ondetrimentalbehaviors(Kenkel,1991;Noconetal.,2007),use ofhealthpreventiveservicessuchasscreening(Adleretal.,1993), availabilityofpsychosocialresourcessuchassocialsupport,and betterabilitytocopewitheverydayhasslesandstressfulsituations inindividualswithhighvs.loweducation(AdlerandSnibbe,2003;
Seeman,1996).Allthesefactorsmaytranslateintobetterbehav- iorsandlowerexposurestochronicstressinindividualswithhigh education.However,inourstudylifestylefactors(includingmarital statusandstressonly)slightlyattenuatedtheassociationbetween SESindicatorsandAL.Otherfactorssuchaswork-lifebalance,social support,psychosocialfactors,earlylifeconditionsandbettermea- surementsofstressandfinancialstraincanpotentiallycontribute toexplaintheobservedsocialdifferencesinphysiologicaldysreg- ulation(Galloetal.,2011;Hawkleyetal.,2011;Huetal.,2007;
Kubzanskyetal.,1999).Intermsofeducationresultsweremore consistentthanforoccupationalposition(Huetal.,2007;Nicod
D.Petrovicetal./Psychoneuroendocrinology67(2016)76–85 83 etal.,2014;Seemanetal.,2004).Apreviousstudyperformedby
Nicodetal.(2014)alsofoundanegativeassociationbetweenthis indicatorandAL,inaSwisspopulationfromwhichasubsetofthe participantswerealsoincludedinSKIPOGH.
Several lifestyle factors were also related to physiological dysregulationin ourstudy.Moderate alcohol consumptionwas protectiveagainsthighALinbothmenandwomen,inlinewith resultsfromGalloetal.(2011).Thismayberelatedtothebeneficial effectsofmoderatealcoholconsumptiononseveralaxisincluded intheALindexsuchaslipidic,andtothedetrimentaleffectofheavy drinkingonthecardiovascularaxis.Womenabstainingfromalco- holwerealsoatincreasedriskofhighAL.However,individuals mightrestrainfromalcoholconsumption duetomedicalcondi- tionsorotherreasonssuchaspastdrinking(Galloetal.,2011;Hu etal.,2007).Consistentlywithpreviousresearchshowingtheben- eficialeffectsofphysicalactivityonvariousphysiologicalprocesses (Warburtonetal.,2006),wefoundthatmenandwomenreporting highphysicalactivityhadlowerALandlowerdysregulationofsev- eralphysiologicalaxesincludingmetabolism,cardiovascularaxis, lipidicprofileandoxidativestress.Wealsoobservedthathighsalt intakewasassociatedwithincreasedALinwomenandtendedto bepositivelyassociatedinmen.Theseresultsaresupportedbypre- viousresearchshowingthatanincreasedsaltconsumptioncauses anincreaseintheglomerularfiltrationrateinthekidneysthus contributingtohighAL(Berge-LandryandJames,2004).
Surprisingly, we didnot find an associationbetween stress andAL,eventhoughpreviousresearchhasplacedchronicstress as a majordeterminant ofAL (Galloet al., 2011; McEwen and Seeman,1999).Thisisprobablyrelatedtotheroughmeasurement ofstressinourstudy,whereindividualswereaskedtoratetheir perceivedstress ona 10-levelscale, whereas researchfocusing onALgenerallyexaminesstressfulevents,conditionsandexpe- riences(Gruenewaldet al.,2012; McEwen,1998)oruses more elaborateandaccuratetools,suchasthePerceivedStressQues- tionnaire(Levensteinetal.,1993).Moreover,although previous researchhasfoundanassociationbetweensmokingandincreased AL(Crimminsetal.,2009),smokingtendedtobeweaklyandnot significantlyassociatedwithhighALinourstudy.
AL was a significantly heritable trait, after adjustment for age,sexand center.Toourknowledge,this isthefirststudyto demonstrateheritabilityofsucha phenotype.However,despite statistical significance, the genetic contribution to AL remains modestwhencomparedtotheenvironmentalcomponent,which explainsapproximately70%ofthephenotypicvariance.
Itremainshowevertobeclarifiedwhethertheheritabilityof AL comesfrom the individualcontribution of each of its com- ponents(Bartels etal.,2003;Christianetal.,1976;Elbeinetal., 1999;Loombaetal.,2010;McIlhanyetal.,1975;Retterstoletal., 2003), most of which are known to be heritable, or from a
“master-regulator”geneticprocessthataffectsseveralphysiologi- calparameterssimultaneously.
Inadditiontotheresearchthathasinvestigateddeterminants ofALinWesternpopulations,severalstudieshavealsobeencon- ducted in Asian countriessuch asJapan, Taiwanor Nepal, and maythereforeprovideaninterestingethnicorculturalcontrast forthestudyofdeterminantsofAL(Huetal.,2007;Kusanoetal., 2015;WorthmanandPanter-Brick,2008).Kusanoetal.(2015)have shownthatinelderlyJapanese,highalcoholintakewasassociated withhighAL,andincreasedvegetableconsumptionwasassociated withlowerAL,atendencywhichwasalsoobservedinourstudy.
Ontheotherhand,whilewefoundastrongassociationbetween educationoroccupationandALinourstudy,norelationwasfound fortheseSESindicatorsinaJapanesepopulation.Moreover,other Asianstudieshaveshownsimilarordifferenttendenciesforthe determinants ofAL (Huet al.,2007).However,at this point, it remainsdifficulttodeterminewhetherthesimilaritiesordiffer-
encesintheassociationsbetweenSESorlifestylebehaviorsandAL areduetothegeneticbackgroundortoculturalfactors(Gleietal., 2013),andadditionalstudiesfocusingondifferentethnicitiesshall beconductedtoclarifythispoint.
5.1. Strengthsandlimitations
Ourstudyhasseveralstrengths,thefirstbeingtherichnessof thephysiological,geneticandlifestyledataonapopulation-based sampleofaSwisspopulationofEuropeandescent.Thisallowedus tocomputedysregulationindexesforspecificphysiologicalaxes andageneralizedindexofphysiologicaldysregulation(ALscore).
Inaddition,thesignificantheritabilityofALalsosuggestedthatdata qualitywashighwithinourstudy,sincemeasurementerrorstend toreduceheritability(Pruijmetal.,2013).
Ourstudyalsohassomelimitations.First,exceptfordietarysalt intake,theassessmentoflifestylefactorsislikelytobeimprecise andsubjective,astheywereself-reportedbystudyparticipantsand assessedwithbasicquestions.Second,thenotionofALasamarker ofphysiologicaldysregulationstillrequiresfurtherexperimental andclinicalvalidation.Eventhoughthismarkerhasbeenincreas- inglyused(Steptoeetal.,2014),thisconceptremainssomewhat arbitrary,relativelycomplex,andlacksabsoluteconsensusinthe wayitiscomputedthroughoutstudies(Karlamanglaetal.,2012;
Nicodetal.,2014).Itwouldthereforebedesirabletoestablisha widelyacceptedandprecisedefinitionforAL,asitisthecasefor frailty,whichisasimilarnotion(Guessousetal.,2014).Oneofthe issuesregardingALinourstudyisthelackofadditionalmarkersof inflammationandthetotalabsenceofneuroendocrineaxismark- ers,suchasadrenalineornoradrenaline(Karlamanglaetal.,2012).
Thisalsoraisesthequestionofwhetherdifferenthomeostaticaxes andtheircomponentsshallbeweighteddifferentiallywhengen- eratingtheAL score.Further,thelimitedsamplesizemayhave ledtostatisticalpowerissuesforsomeoftheassociations.Finally, ourfindingsareonlyvalidfortheSwisspopulationofEuropean descent,andmaynotbegeneralizedtootherpopulations.
6. Conclusion
Insummary,ourfindingsindicatethatSESactsasastrongdeter- minantofAL,especiallyamongwomen,andthatthisassociation isnotnecessarilyattenuatedbylifestylebehaviors,which affect ALindependently. Moreover,despitethatfactthattheconcepts ofallostasisandAL weremeanttoexpresstheconsequencesof chronicenvironmentaldemands,heritabilityanalysesshowedthat thereisasignificantgeneticpredispositionforAL.
Authorcontributions
SS,MBoandDPdesignedthestudy.BP,ND,MP,GE,DA,IG,SEY, APB,BV,MM,PYM,FP,MBuandMBoactivelycontributedtodata acquisition.DP,SS,EP,MBoanalyzedthedata.SS,DP,EP,BP,ND, MP,DA,IG,FP,MBuandMBocriticallyrevisedthemanuscript.
Conflictsofinterest
Theauthorsdeclarethattheyhavenoconflictofinterest.
Funding
Silvia Stringhini is supported by the Swiss national science foundation(Ambizione Grant no.PZ00P3147998).Thisworkis supportedbytheEuropeancommissionandtheSwissstatesecre- tariatforeducation,researchandinnovation—SERI(Horizon2020 grantno.633666).TheSKIPOGHstudyissupportedbyagrantfrom