Gonads and strife : sex hormones vary according to sexual orientation for women and stress indices for both sexes

Gonads and strife: Sex hormones vary according to sexual orientation for women and stress indices for both sexes

Robert-Paul Juster

, Daniel Almeida

, Christopher Cardoso

, Catherine Raymond

, Philip Jai Johnson

, James G. Pfaus

, Adrianna Mendrek

, Annie Duchesne

,

Jens C. Pruessner

, Sonia J. Lupien

aDepartmentofPsychiatry,ColumbiaUniversity,NewYork,NY,UnitedStates

bDepartmentofNeurologyandNeurosurgery,McGillUniversity,Montreal,Quebec,Canada

cDepartmentofPsychology,ConcordiaUniversity,Montreal,Quebec,Canada

dDépartmentdeNeuroscience,UniversitédeMontréal,Montreal,Quebec,Canada

eDépartmentdePsychiatrie,UniversitédeMontréal,Montreal,Quebec,Canada

fDepartmentofPsychology,Bishop’sUniversity,Sherbrooke,Quebec,Canada

gDepartmentofPsychiatry,McGillUniversity,Montreal,Quebec,Canada

hDepartmentofPsychology,McGillUniversity,Montreal,Quebec,Canada

Keywords:

Sexualorientation Testosterone Estradiol/progesterone

Dehydroepiandrosterone-sulphate Cortisol

Allostaticload

a b s t r a c t

Thisstudyassessedsexualorientationandpsychobiologicalstressindicesinrelationtosalivarysexhor- monesaspartofawell-validatedlaboratory-basedstressparadigm.Participantsincluded87healthy adultsthatwereonaverage25yearsoldwhoself-identifiedaslesbian/bisexualwomen(n=20),het- erosexualwomen(n=21),gay/bisexualmen(n=26),andheterosexualmen(n=20).Twosalivasamples werecollectedfifteenminutesbeforeandfifteenminutesafterexposuretoamodifiedTrierSocialStress Testtodeterminetestosterone,estradiol,andprogesteroneconcentrationsviaenzyme-immuneassay- ing.Meansexhormoneswerefurthertestedinassociationtostressindicesrelatedtocortisolsystemic output(areaunderthecurvewithrespecttoground)basedontenmeasuresthroughoutthetwo-hour visit,allostaticloadindexedusing21biomarkers,andperceivedstressassessedusingawell-validated questionnaire.Resultsrevealedthatlesbian/bisexualwomenhadhigheroveralltestosteroneandproges- teroneconcentrationsthanheterosexualwomen,whilenodifferenceswerefoundamonggay/bisexual menincomparisontoheterosexualmen.Lesbian/bisexualwomenandheterosexualmenshowedposi- tiveassociationsbetweenmeanestradiolconcentrationsandallostaticload,whilegay/bisexualmenand heterosexualwomenshowedpositiveassociationsbetweenmeantestosteroneandcortisolsystemicout- put.Insummary,sexhormonevariationsappeartovaryaccordingtosexualorientationamongwomen, butalsoasafunctionofcortisolsystemicoutput,allostaticload,andperceivedstressforbothsexes.

1. Introduction

Explorationintotheneurobiologicalcorrelatesofsexualori- entationhashadacontroversialhistory.Earlyresearchreflected homosexuality’s classification asa mental illness until 1973,at whichtimetheAmericanPsychiatricAssociationremoveditfrom itsdiagnostic manual(Friedmanand Downey, 1994).Biological

∗Correspondingauthorat:ColumbiaUniversity,Dept.ofPsychiatry,Divisionof BehavioralMedicineandDivisionofGender,Sexuality,andHealth,ColumbiaUni- versityMedicalCenter,622West168thSt.,PH1540G,NewYork,NY10032,United States.

E-mailaddress:rjuster2108@cumc.columbia.edu(R.-P.Juster).

explanations ofnon-heterosexual behaviorhaveoften hypothe- sized adysregulationofsex-specifichormoneprofiles,resulting in anomalies in the organizational and activational effects of thesehormonesontheneurodevelopmentofcircuitryunderlying species-specificsexualbehavior.Animalmodelsinvolvingprena- tal androgen deficits, for example,were first believed tocause malehomosexuality(Phoenixetal.,1959),whileprenatalandro- genoverabundancepresumablyresultedinfemalehomosexuality.

Ourstudyendeavourstoshowthatsuchsexhormonedifferences assumedtobeattributabletosexualorientationarealsomodulated byunexploredstressphenomena.

Advancesinbehavioralneurosciencehasledtotheintroduction ofsophisticatedgeneticmodelsinvolvingnon-functionalandrogen receptors,furtherallowingresearcherstocharacterizetherelation-

Konstanzer Online-Publikations-System (KOPS) URL: http://nbn-resolving.de/urn:nbn:de:bsz:352-2-1f6t7wu8qvgcl6 Erschienen in: Psychoneuroendocrinology ; 72 (2016). - S. 119-130

https://dx.doi.org/10.1016/j.psyneuen.2016.06.011

shipbetweenandrogens,themasculinizationoftherodentbrain, andsexualbehavior(Zuloagaetal.,2008).Guidedbythisliterature, perturbationsinthehumanprenatalenvironmentduringacrit- icalperiod(between10to22weeksgestation)mightexposethe humanfetustoandrogenimbalancesthathavethepotentialtoalter structureandfunctionofkeyneuroanatomicalregionsimplicated inhumansexuality(EllisandAmes,1987).Whilethisearlyexpo- sureofandrogenstothedevelopingbrainisunderstoodtoaffect post-natalsex/gender,sexualorientation,andrelatedbehaviors, theliteraturehasbeeninconclusiveandthusreflectsthemultifac- torialnatureofhumansexuality.

Several human studies during the 1970s using analytes extractedfromeitherurine,serum,orplasmarevealedthatmen belongingtoasexualminorityshowedhighertestosterone(Brodie etal.,1974;Doerretal.,1976;TourneyandHatfield,1973),lower testosterone(Brodieetal.,1974;Kolodnyetal.,1972;Kolodnyetal., 1971;Loraineetal.,1971;Loraineetal.,1970;Pillardetal.,1974;

Rohdeetal.,1977;Stahletal.,1976),nodifferencesintestosterone (Barlow et al.,1974;Birk etal., 1973;Doerret al.,1973;Jaffee etal.,1980),higherestrogen(Doerretal.,1973;Doerretal.,1976), orlowerestrogen(Evans,1972)whencomparedtoheterosexual controls. Likewise,amongsexualminoritywomen comparedto age-matchedheterosexualcontrols,studieshavereportedlower estrogen(Loraineetal.,1971;Loraineetal.,1970),highertestos- terone(Loraineetal.,1971;Loraineetal.,1970),nodifferences intestosterone(Downeyetal.,1987),andnodifferencesinestro- genor progesterone (Griffiths et al., 1974; Seyler et al., 1978).

Similarlyfor gonadotropins,studiesreporting elevatedluteiniz- inghormoneconcentrationsamongsexualminoritymen(Kolodny etal.,1972)andwomen(Loraineetal.,1971)havebeenmatched byanabundanceofresearchfailingtoshowdifferencesinluteiniz- inghormone, folliclestimulatinghormone, as wellasprolactin (FriedmanandFrantz,1977;Jaffeeetal.,1980;Kolodnyetal.,1971;

Parksetal.,1974).

Inacriticalreviewoftheliterature,Meyer-Bahlburgconcluded thatfindingswereoverallinconsistentamongsexualminoritymen (Meyer-Bahlburg,1977).Bycontrast,aboutone-thirdofparticipat- ingsexualminoritywomenmanifestedelevatedandrogenlevels while otherwise showing no endocrine abnormalities (Meyer- Bahlburg,1979).Importantly,methodologicaldifferencesbetween studiesrenderedcomparisonsandanyfinalconclusionsdifficult.

Thisbodyofhumanresearchdidnotsupporttheneurohormonal hypothesisofsexualorientation(BanksandGartrell,1995).Meyer- Bahlbergacknowledgedthatsomeearlyresearcherswererightfully cautiousintheirconclusions(Meyer-Bahlburg,1977).Inparticu- lar,theinconsistenciesinobservedHPG-axispatternsmightnot betheprimarycauseof sexualminorityorientation,but rather asecondaryconsequencerelatedtounmeasuredfactorssuchas psychosocialstress(Kolodnyetal.,1972;Meyer-Bahlburg,1979).

Stressresearchersduringthe1970sbeganshowingthatpsy- chological factors could modulate the HPG-axis. For example, a longitudinal study among military men undergoing stressful trainingrevealedthatplasmatestosteronelevelswerelowestdur- ingtheearliernovicephasecompared tothelaterseniorphase (Kreuzetal.,1972).ThisparalleledthepioneeringworkofMason who systematically studied stressful situations (e.g., parachute jumping,air-trafficcontrolling) andidentifiedkeypsychological determinants(e.g.,novelty,uncontrollability)thatactivatedstress responses(Mason,1968).Unfortunatelythisknowledgewasnot appliedtounderstandingthemixedHPG-axisliteratureonsex- ualorientation,despitespeculationthatpsychosocialstressmight beinvolvedinstudyinconsistencies.Becausesexualminoritiesare atanincreasedriskforstress-relatedpathologies duetostigma anddiscrimination(IOM,2011;Meyer,2003),itishighlyproba- blethatuniquepsychosocialcontextsinfluenceHPG-axisprofiles

andmightconfoundtheliteraturedescribingbiologicaldifferences focusingpurelyonsexualorientation,identity,&behavior.

Advancesinpsychosocialandbiologicalapproachestostudy- ing stress propelled an entire field of psychoneuroendocrine research aimedat identifying mechanisms of disease suscepti- bility.In particular,thedevelopmentof laboratory-basedstress inductionparadigms havesubstantiatedthatthesexes differin theirstressresponsepatternsofthestresshormonecortisoleas- ilycollectedviasaliva.StudiesusingthepopularTrierSocialStress TestorTSST(Kirschbaumetal.,1993)consistentlyshowthatmen mountagreatercortisolresponsethanwomenofreproductiveage (Kirschbaumetal.,1992).Inturn,womenshowfurtherattenuation whenusingoralcontraceptives(Kirschbaumetal.,1995)ordur- ingthehighestrogen(follicular)phaseoftheirmenstrualcyclesas opposedtoduringthelutealphase(Kirschbaumetal.,1999).

Beyond sex differences in stress reactivity, research apply- ing stress biomarkers are beginning to be used to understand how stigmaaffects the health and wellbeing of sexualminori- ties(Hatzenbuehleretal.,2013).Ourgrouphasrecentlyprovided novelevidencethatsexualorientationmodulatescortisolreactiv- ity.Specifically,lesbian/bisexualwomenshowhigherpost-stressor cortisolconcentrationscomparedtoheterosexualwomen,while gay/menshowoveralllowercortisolconcentrationscomparedto heterosexualmenaftercontrollingforbasalsexhormoneconcen- trations(Juster etal., 2015).In additionto this, we foundthat sexualminorityparticipantswhohaddisclosedtheirsexualorien- tationtofamilyandfriendsevidencedlowermorningcortisollevels andlesspsychiatricsymptomsthanthosewhohadnotcompletely disclosedirrespectiveofsex(Justeretal.,2013b).Thesereportssug- gestthatimportantpsychosocialandbehavioralfactorsmayresult indistinctbiologicalsignatures.

Itremainsunknownhowever,howcirculatingsexhormones inthecontextoftheTSSTparadigmvaryasafunctionofone’s sexualorientation,andhowtheseassociationsarefurthermod- ulatedby stressphenomena.In thecurrent study,weexplored whethersexualminoritiesdifferfromsame-sexheterosexualcon- trolsintermsofsalivarytestosterone,estradiol,andprogesterone concentrationsbeforeandafterexposuretotheTSST.Guidedbyour previousreportsshowingthatbiopsychosocialstressisuniquely experiencedbetweenandwithinsexualorientations(Justeretal., 2015;Justeretal.,2016a;Justeretal.,2013b),wefurtherassessed whetherchangesinsexhormoneswereassociatedwithcortisol systemicoutputsummarizedusing10measurementsthroughout theTSST,allostaticloadindexedusing21stress-relatedbiomarkers, andfinallyperceivedstress.

GiventhemixedfindingsintheHPG-axisliteratureonsexual orientationandthelackofstudieslinkedtostressindices,wedid nothypothesizedirectionalityofassociations.We did,however, hypothesizethatpsychobiologicalstressindiceswouldcorrelate withmeansexhormoneconcentrationsbeyondthoseassociations attributabletosexualorientation.

2. Methods 2.1. Participants

Eighty-sevenparticipantsages18–45(M=24.61±0.61SE)iden- tifying as lesbianor gay (8 women and 20 men), bisexual(13 womenand6men),and heterosexual(20womenand 21men) wererecruitedfromMontrealaspartofabroaderresearchpro- gram(Justeretal.,2015;Justeretal.,2016a;Justeretal.,2013b).

Toequalizegroupsduetofewerlesbiansandbisexualmen, we combinedlesbian/gaywithbisexualindividuals (20women and 26men)andcontrastedthemtoheterosexuals(20womenand21 men).

Table1

Sampledescriptivestatisticsaccordingtosexandsexualorientation.

Information Sample Lesbian/Bisexual♀ Heterosexual♀ Gay/Bisexual♂ Heterosexual♂ ^p

N 87 20 20 26 21 –

Demographic:

Age,M(SE) 24.61(0.61) 24.10(1.34) 25.45(1.13) 23.77(0.98) 25.33(1.47) 0.685

Race/ethnicity:

White,% 70.1 75.0 55.0 73.1 76.2 0.147

Black,% 5.7 15.0 10.0 0.0 0.0 0.147

Asian,% 12.6 0.0 10.0 23.1 14.3 0.147

Hispanic,% 6.9 0.0 20.0 3.8 4.8 0.147

Arab,% 4.6 10.0 5.0 0.0 4.8 0.147

Occupation:

Workers,% 34.5 40.0 35.0 23.1 42.9 0.490

Students,% 65.5 60.0 65.0 76.9 57.1 0.490

Working/studyinghours/week,M(SE) 28.02(1.82) 28.40(3.63) 28.5(4.38) 29.96(3.24) 24.91(3.54) 0.789

Sexualorientation^A:

Sexualattractions,M(SE) 3.51(0.26) 4.75(0.33) 1.40(0.15) 5.96(0.30) 1.29(0.10) <0.001

Sexualbehaviors,M(SE) 3.34(0.26 4.25(0.44) 1.30(0.11) 6.00(0.29) 1.114(0.8) <0.001

Sexualfantasies,M(SE) 3.61(0.26) 5.15(0.25) 1.65(0.25) 5.81(0.32) 1.29(0.12) <0.001

Lifestylepreferences,M(SE) 3.37(0.27) 5.15(0.44) 1.115(0.08) 5.54(0.33) 1.10(0.07) <0.001

Sexualidentity,M(SE) 3.52(0.27) 5.05(0.43) 1.15(0.08) 6.04(0.26) 1.19(0.09) <0.001

Socio-economic:

Post-secondaryeducation,% 95.3 95.0 90.0 100.0 95.0 0.575

Personalannualincome,$CAD,M(SE) 16,000(0.17) 14,500(0.34) 19,000(0.53) 14,000(0.19) 16,800(0.33) 0.737 Householdannualincome,$CAD,M(SE) 32,100(0.32) 37,000(0.68) 25,000(0.54) 27,100(0.52) 39,000(0.75) 0.311 Healthandwell-being:

Medicationuse,% 18.4 25.0 15.0 15.4 14.3 0.105

Oralcontraceptiveuse,% 16.1 20.0 50.0 − − 0.017

Minorphysicalcondition,% 34.5 50.0 40.0 23.1 28.6 0.238

Psychiatrichistory:

None,% 28.7 15.0 40.0 26.9 33.3 0.522

Pasthistory,% 8.0 15.0 5.0 11.5 33.3 0.522

Familyhistory,% 35.6 35.0 45.0 30.8 33.3 0.522

Bothpastandfamilyhistory,% 27.6 35.0 10.0 30.8 33.3 0.522

Subjectivedimensions^B:

Self-ratedhealth,M(SE) 3.72(0.09) 3.50(0.21) 3.95(0.14) 3.69(0.17) 3.75(0.20) 0.419

Self-ratedphysique,M(SE) 3.37(0.10) 3.15(0.22) 3.70(0.18) 3.27(0.14) 3.40(0.28) 0.284

Self-rateddiet,M(SE) 3.30(0.11) 3.25(0.19) 3.40(0.25) 3.34(0.19) 3.15(0.25) 0.832

Behavioral:

Tobaccosmoking:

Smokers,% 11.5 5.0 10.0 11.5 19.0 0.376

Socialsmokers,% 14.9 10.0 5.0 19.2 23.8 0.376

Non-smokers,% 73.6 85.0 85.0 69.2 57.1 0.376

Alcoholconsumption(weekly):

0orinfrequently,% 25.3 25.0 50.0 18.1 4.8 0.134

1to5,% 40.2 45.0 40.0 26.9 52.4 0.134

6to10,% 26.4 25.0 10.0 38.5 28.6 0.134

11ormore,% 8.0 5.0 0.0 11.5 14.3 0.134

Elicitdruguse:

None,% 66.7 60.0 85.0 61.5 61.9 0.334

Occasional(monthlyorannually),% 24.1 35.0 5.0 30.8 23.8 0.334

Regular(dailyorweekly),% 9.2 5.0 10.0 7.7 14.3 0.334

Interpersonal:

Single,% 72.1 60.0 80.0 80.8 65.0 0.315

Children,% 4.7 10.0 5.0 0.0 5.0 0.463

Siblings,% 86.0 95.0 85.0 80.8 85.0 0.578

Parentsalive,% 94.2 95.0 90.0 92.3 100.0 0.559

Infrequentfamilygatherings,% 36.1 30.0 30.0 57.7 20.0 0.169

Non-religious/spiritual,% 78.8 76.5 84.2 84.0 68.4 0.569

During screening, women provided information concerning theirmenstrualcycles(e.g.,dateoflastmenses,averagecycledays) andoralcontraceptiveuse.Onewomanwhodidnotprovideinfor- mationonreproductivefunctioningandwasthereforeremoved frommainanalyses.Themainexclusionarycriteriainthisstudy weremajorhealthproblems,severementalillness,oruseofsyn- theticglucocorticoids.Transgenderindividualswerenotsolicited duetohormonaltreatmentsthatdirectlyaffecttheHPG-axis;how- ever,thisrepresentsagroupthatoughttobeincludedinfuture researchgivencriticalhealthinequalities(IOM,2011)Thesample’s information on demographics, socio-economics, health, wellbe- ing,lifestylebehaviors,andinterpersonalfeaturesarereportedin Table1.

2.2. Generalprotocol

Thisstudywasapprovedbytheresearchethicsboardofthe InstitutuniversitaireensantémentaledeMontréal.Upona15-min studyexplanationandscreeninginterviewviatelephone,eligible participantswerescheduledforafirstvisitattheCentreforStud- iesonHumanStress.Allparticipantsprovidedinformedconsent uponarrivaltoourlaboratory.TestingwasconductedbetweenJuly 2010toNovember2010,constraininganyseasonalvariationthat confoundsexhormones(vanAndersetal.,2006).

Amorningvisitwasscheduledbetween8:00AMand11:00AM andlastedapproximately40min.Thisvisitinvolvedablooddraw fromacertifiednurse,aContinentalbreakfasttobreaka12hfast, completionofquestionnaires,andfinally,detailedinstructionsfor

take-homebiomarkercollectionprotocolsfirstreportedelsewhere (Justeretal.,2013b).Betweenvisitsspacedonaverageoneweek apart,participantscompletedtake-homematerialsbeforereturn- ingtothelaboratory.

Anafternoon/eveningvisitwasscheduledbetween12:00PM and7:00PMandlasted120min.Thecurrentstudyfocusesonthis afternoon/eveningvisitinwhichparticipantswereexposedtoa modifiedversionoftheTSSTwhileprovidingsalivasamples.Upon completion of study requirements, participantswere explained theirbloodresultsindetailandcompensatedwith$50CADupon debriefing.

2.3. Sexualorientation

Sexualorientationwasascertainedand cross-validatedusing three combined methods: (1) response to one of three sepa- rateadvertisementssearchingforeitherlesbian/gay,bisexual,or heterosexualparticipants;(2)askingparticipantstheiridentified sexualorientation in an open-ended mannerduring telephone screening;and(3)administrationofamodified5-itemKleinSexual OrientationScale(Kleinetal.,1990).Thisinstrumentusesa7-point Likertscalefrom1(othersexonly)to7(samesexonly)toassess sexualattractions,sexualbehaviors,sexualfantasies,lifestylepref- erence,andsexualidentity.Theentiresample’sresponseswere internallyconsistent(␣=0.982).Basedoncorrespondenceamong these three methods, sexual orientation was coded as “sexual minority”(n=46)or“heterosexual”(n=41).

2.4. Perceiveddistress

The14-itemPerceivedStressScale(Cohenetal.,1983)measures perceivedstressusinga5-pointLikertscalefrom0(never)to4(very often).Originalpsychometricpropertiesrevealedstronginternal consistency(mean␣=0.85),test–retestreliability(meanr=0.85), andevidenceofconcurrentvaliditywithdepression(meanr=0.71) andphysicalcomplaints(meanr=0.59)amongyoungstudents.In thecurrentsample,perceivedstressshowedstronginternalcon- sistency(␣=0.88).

2.5. Stressreactivityparadigm

Duringthetwo-hourafternoonvisittoourlaboratory,partici- pantswereexposedtoamodifiedversion(Andrewsetal.,2007;

Wadiwalla et al., 2010)of the Trier Social Stress Test or TSST (Kirschbaumetal.,1993).Uponaten-minuteanticipationphase, participantswereledtoaseparateroomwheretheywereaskedto deliverafive-minutemockjobinterviewfollowedbyfiveminutes of mental arithmetic in front of an unseen, ostensible behav- ioralexpertseatedbehindaone-waymirror.Theparticipantand the“behavioralexpert”communicatedviaanintercommunication deviceandtheparticipant’sperformancewasrecordedbyavideo camera.Aseminalmeta-analysis(DickersonandKemeny,2004) posits that laboratory-based stressors eliciting social-evaluative threatincludeevaluativeaudiences,negativesocialcomparison(s), and/orrecordedperformancethatmaximizeHPA-axisreactivity.

Previous studiesby ourgroup demonstratethat placingthe evaluativeaudiencebehindaone-waymirror(‘panel-out’)further maximizes between-sexdifferences in cortisolstress reactivity.

Specifically,menexposedtothistype oftheTSSTshownosig- nificant differences in HPA-axis reactivity when compared to thestandard performance in front of theaudience (‘panel-in’);

(Andrewsetal.,2007).Incontrast,heterosexualwomenexposedto the‘panel-out’conditionshowdecreasedcortisolstressreactivity incomparisontoheterosexualwomeninthe‘panel-in’condition (Wadiwallaetal.,2010).Ourrecentstudyemployingheterosex- ualwomeninthe‘panel-out’conditionalsoreports comparable

decreased cortisolreactivity(Marin etal., 2012); however,it is unknownhowthisisrelatedtosexhormonevariations.

2.6. Visitorder

Participantsvisitedourlaboratorytwice:(1)morningvisitfor ablooddrawtoassessallostaticloadand(2)afternoonvisitfor theTSST.Visitorderwascounterbalancedrandomlytomanipulate experiencednoveltyofthetestingenvironment.Inthefirstgroup (morning/afternoon;n=49),participantsreceivedablooddrawin themorningduringtheirfirstvisitandwereexposedtotheTSSTin theafternoonduringtheirsecondvisitaboutaweeklater;thiswas reversedforthesecondgroup(afternoon/morning;n=37).Because thesecondgroupwasarrivingforthefirsttimetoourlaboratory whenexposedtotheTSST,weexpectedthattheywouldbemore distressedthanthefirstgroupwhohadalreadyfamiliarizedthem- selveswiththesetting.Thisacclimationisbelievedtohelpdiminish participants’distressexperiencedinnoveltestingenvironments (Sindietal.,2013)

Whilethisdidnotsignificantlymodulatecortisolconcentrations inapreviousreport(Justeretal.,2015),preliminaryanalyseswere conductedinthecurrentstudytoassesswhetherthismanipula- tionmodulatedsexhormonedynamics.TheHPG-axisisbelievedto beinvolvedinanticipatorystatesandexpectationofpreoperative stress(Gerraetal.,2000).Additionallynoteworthyintermsofday- timevariability,ithasbeensuggestedthattestosteronecollection shouldbeconductedintheafternoonandeveningtoprovidethe bestpossiblerepresentationofanindividual’sgeneralresponsive- nesstosocialinteractionsandnotthosebasedonnaturaldiurnal variation(Grayetal.,2004).

2.7. Salivacollectionandendocrineassays

Twosalivasamples(approximately2mLpersample)werecol- lected−15minpre-TSSTand+15minpost-TSSTtoassesssalivary testosterone,estradiol,andprogesteroneateachtime-point.Col- lectionwasachievedusingthepassivedroolmethodguidedwith sterilizedstraws.Sampleswerefrozenimmediatelyuponcollec- tionat−20^◦Cand assayswereperformedwithinthreemonths of testingto prevent degradationknown toinfluenceHPG-axis biomarkers (Granger et al., 2004). Because collection occurred between12:00PMand7:00PM(M=2:34PM,SE=0.11),weassessed forcircadianvariationinsexhormonevaluesfortheentiresam- ple:nosignificantcorrelationswitharrivaltimewerefound.Thirty minutespriortosalivasampling,participantswereinstructedto avoidmajormeals,cigarettes,caffeinated/sugarybeverages,and dairyproducts.Theywerefurthermoreaskedtorefrainfromoral hygieneandstrenuousphysicalactivitytwohoursbeforesampling.

Tensaliva samplesdestined for cortisoldetermination were collected in ten-minute intervals at the following time-points:

−40min, −30min, −20min, −10min (anticipation phase), and immediatelybeforetheTSST,aswellas+10min,+20min,+30min, +40min,and+50minthereafter.Attheendofthetestingsessions, thesesalivasampleswerefrozenat−20^◦ untilassaying.Apre- viousreportusingthissampleshowedthatlesbian/bisexualhad highercortisolconcentrationsthanheterosexualwomen+40min post-TSST,while gay/bisexualmen showedlower overallcorti- solconcentrationsthanheterosexualmenthroughoutthesession (Justeretal.,2015).Inthecurrentanalysis,wesummarizeall10 cortisoltime-pointstoassessassociationswithmeansexhormone concentrations.

All saliva samples were analyzed at the Centre for Studies onHumanStress(www.humanstress.ca).Priortoassayingeach biomarker,frozensampleswerebroughttoroomtemperatureto becentrifugedat1500xg(3000rpm)for15min.Fortestosterone determination,weusedanexpandedrangeenzymeimmuneassay

kit(Salimetrics^®,StateCollege,PA,CatalogueNo.1-2402)where the lower limit of sensitivity is 1pg/ml. For estradiol, or more specifically17␤-estradiol,determination, weused ahighsensi- tivityenzymeimmuneassaykit(Salimetrics^®,StateCollege,PA, CatalogueNo.1-3702)wheretherangeofdetectionis1–32pg/ml.

Forprogesteronedetermination,weusedahighsensitivityenzyme immuneassaykit(Salimetrics^®,StateCollege,PA,CatalogueNo.1- 1502)wherethelowerlimitofsensitivityis5pg/ml.Inter-assay andintra-assay coefficientsofvariancewererespectivelybelow 12.80% and6.34%for testosterone,3.7%and6.78% forestradiol, and8.08%and6.78%forprogesterone.Forcortisoldetermination, weusedahighsensitivityenzymeimmuneassaykit(Salimetrics^® StateCollege,PA,CatalogueNo.1-3102)wheretherangeofdetec- tionisbetween0.012–3ug/dL.Assayswereruninduplicatesand averageswereusedinstatisticalanalyses.

2.8. Allostaticloadbiomarkers

Allostaticloadisdefinedasthemulti-systemic‘wearandtear’

thatchronicstressandunhealthybehaviorsexactonthebodyand brain(McEwenandStellar,1993).Twenty-onebiomarkerswere usedtocalculateacount-basedallostaticloadindexconstructed accordingtothesample’sdistributionforhigh-riskpercentilesas previouslydone(Seemanetal.,1997).Systemicbiomarkersrepre- sentneuroendocrine(salivarycortisolwastransformedintoboth AMcortisol and PMcortisolslopes tocapturedynamic diurnal HPA-axisfluctuations;serumdehydroepiandrosterone-sulphate;

12h overnighturinary adrenalin, noradrenalin, and dopamine), immune/inflammatory(plasmainterleukin-6andtumour-necrosis factor-alpha;serumc-reactiveprotein;andserumfibrinogenpre- servedwithsodiumcitrate),metabolic(serumalbumin,creatinine, insulin, glycosylated haemoglobin (%), total cholesterol, high- densitylipoprotein,andtriglycerides;waist-to-hipratiomeasured withgraduatedtapeandcalculatedbydividingrespectiveinches;

bodymassindexcalculatedasmass(kg)dividedbyheight(m²);

and cardiovascular (mean of three seated systolic and diastolic bloodpressurerecordingsmeasuredwithanelectronicsphygmo- manometer;A&DMedical©:ModelUA-631V)functioning.Table2 listsallbiomarkersusedwiththeirrespectivecut-offsandgroup characterizations.

Previous allostaticload findings in this sample showedthat gay/bisexualmenhadlowerallostaticloadlevelsthanheterosex- ualmenandthatthiswasdrivenbylowertriglycerides,bodymass indices,andtrendingtumour-necrosisfactor-alphaconcentrations (Justeretal.,2013a).Amongsexualminorities,thosethatenacted avoidancecopingstrategiesduringsexualidentityformationand disclosureevidencedelevatedallostaticloadandpsychosocialdis- tress(Justeretal.,2016a).Inthecurrentanalysis,ourrationalefor usingallostaticloadwastoassesspotentialconvergenceofassocia- tionswithsalivarysexhormonesandtoassessdifferencesbetween women for individualbiomarkers which wedidnot previously assesssincewomendidnotshowallostaticloaddifferences.

2.9. Statisticalanalysis

AllstatisticalanalyseswererunusingtheStatisticalPackage for the Social Science Version 22 for Macintosh. Group differ- encesindescriptiveinformationarereportedinTable1.Groups only differedaccordingtosexuality asexpected(all ps<0.001), while heterosexualwomen usedoral contraceptivesmore than lesbian/bisexualwomen(p=0.017).

Due tothe vast inter-individual differences in sex hormone concentrations and intra-individual variations as a function of contextual stimuli and reproductive considerations, we care-

fullyaccounted for potentialconfounders in ourmain analyses Table2 Allostaticloadbiomarkerinformation. ABiomarker(Unit)ReferenceCut-OffFreq.SampleM(SE)Lesbian/Bisexual♀M(SE)Heterosexual♀M(SE)Gay/Bisexual♂M(SE)Heterosexual♂M(SE) CortisolAM(Slope)N/A−0.03292or0.19069200.06161(0.01599)0.04172(0.02739)0.07145(0.02727)0.07445(0.03772)0.05396(0.03018) CortisolPM(Slope).N/A−0.06902or0.0029418−0.03202(0.00431)−0.03481(0.00959)−0.02995(0.00635)−0.02843(0.00964)−0.03613(0.00808) Dehydroepiandrosterone-sulphate(mol/L)1.2–10.4<3.5218.49(0.40)7.34(0.64)5.58(0.37)10.19(0.73)10.20(0.83) Adrenalin(nmol/L)0–110>82.52337.18(6.17)55.89(26.71)29.40(3.84)33.77(3.95)31.60(2.97) Noradrenalin(nmol/L)70–475>373.7522168.14(10.57)181.37(35.19)171.10(17.69)161.77(14.36)160.90(17.41) Dopamine(nmol/L)420–2600>2055.0211681.53(76.58)1647.05(114.94)1685.30(150.15)1826.62(162.18)1521.90(167 Tumour-necrosisfactor-␣(pg/mL)>8>6.0233.45(0.91)2.20(0.34)2.14(0.49)2.02(0.14)7.68(3.65) Interleukin-6(pg/mL)>2.5>1.875220.93(0.16)0.63(0.10)1.09(0.45)0.86(0.29)1.16(0.38) C-reactiveprotein(mg/L)0–5>3.75202.74(0.93)3.13(1.85)2.16(1.07)1.41(0.83)4.48(3.15) Fibrinogen(g/L)2.28–5.4>4.62202.98(0.08)3.23(1.85)3.14(0.08)2.74(0.14)2.99(0.17) Insulin(pmol/L)19–154>120.252148.02(2.81)50.98(6.42)44.80(5.08)44.61(4.61)52.64(6.68) Glycosylatedhaemoglobin(%)0.042–0.062>0.057260.515(0.00022)0.052(0.00043)0.052(0.00037)0.051(0.00044)0.051(0.00042) Creatinine(mol/L)54–110>96.02368.28(1.19)62.00(2.25)62.10(2.84)72.81(1.54)74.24(1.69) Albumin(g/L)35–50<38.752644.22(0.60)41.84(0.76)42.20(0.78)46.77(1.56)45.14(0.74) Triglycerides(mmol/L)0.5–1.7>1.4210.85(0.06)0.88(0.18)0.68(0.06)0.76(0.07)1.08(0.14) Totalcholesterol(mmol/L)2–5.2>4.4224.27(0.09)4.22(0.19)4.63(0.18)4.05(0.14)4.24(0.19) Highdensitylipoproteincholesterol(mmol/L)1.05–1.55<1.175231.44(0.05)1.63(0.10)1.68(0.13)1.23(0.05)1.22(0.07) Systolicbloodpressure(mmHg)90–140>127.522113.05(1.22)111.42(1.95)107.65(2.20)116.67(2.42)115.37(2.63) Diastolicbloodpressure(mmHg)60–90>82.52269.93(0.87)71.58(1.69)68.53(1.67)69.53(1.48)70.22(2.21) Waist-hipratio(ratio)0.8–1>0.95220.83(0.12)0.824(0.015)0.081(0.010)0.843(0.009)0.822(0.043) 2Bodymassindex(kg/m)18.5–25>23.3752222.95(0.46)23.53(1.08)21.43(1.09)22.41(0.51)24.52(1.00) ANote:Clinicalreferencerangesaccompanyingbloodresultswereusedtodeterminehigh-riskcut-offsandarereportedelsewhere(Justeretal.,2016a,b;Justeretal.,2013a,b).

of repeated measures. Specifically for each sex hormone in sequence,potentialconfoundingeffectswerefirstassessedusing repeated-measures ANOVAs as a function of (1) visit order (morning/afternoon: n=49; afternoon/morning group: n=37), (2) menstrual cycle divided equally according to follicular les- bian/bisexual women (n=11) and heterosexual women (n=11) aswellasluteallesbian/bisexualwomen(n=9)andheterosexual women(n=9),andfinally(3)oralcontraceptiveuse(users:n=14;

non-users:n=26) thatwasmore commonamong heterosexual women (users: n=10, non-users: n=10) than lesbian/bisexual women (users: n=4;non-users:n=16). Using information that contrastedtheTSSTvisitdateandthedateoflastmenstruation forwomen(M=18.25,SE=2.01),preliminaryone-wayANOVAsof studyoutcomesasafunctionofmenstrualcyclestatusrevealed nosignificantdifferencesbetweenwomeninthefollicular(days 1–14)andluteal(days15–28)phases.Notethatmenstrualstatus wasunknownforoneparticipant,sofinalsampleN=86.

Mixed-design repeated-measures analysis of covariance (ANCOVA) was run with sexual orientation entered as the between-subjects factor and sex hormones (two measures of testosterone, estradiol, and progesterone each 15min before and 15min after TSSTexposure) entered asthe within-subject factorsinsex-specificanalyseswhilecontrollingforageandvisit order as well as menstrual cycle status and oral contraceptive useamongwomen.Post-hocanalysesemployedone-wayANOVA or paired-samplet-test as required. In all such analyses, effect sizesare reportedthroughout andcanbeinterpretedaccording tothefollowingconventions:²P∼=0.01representsasmalleffect, ²P∼=0.06amediumeffect,and²P∼=0.14constitutealargeeffect (Fritzetal.,2012).

Next, we assessed associations among mean sex hormones in relationtocortisolsystemic output,allostatic load,and per- ceivedstressusing bivariatecorrelationssplit accordingtosex.

Time-dependentstressreactivecortisolconcentrationsweretrans- formedintosummaryscoresbasedontheareaunderthecurve formulaebasedonthetrapezoidformula(Pruessneretal.,2003).

Theareaunderthecurvewithrespecttoground(AUCg)wascalcu- latedtorepresentsystemicoutputthroughouttheTSSTvisit.

Insupplementalanalyses,wefollowedwithanexplorationof 21individualbiomarkerdifferencesusingANCOVAscontrollingfor ageandbivariatecorrelationswithotherstudyvariables.

3. Results 3.1. Testosterone

Inpreliminaryanalysesofvisitorder,theafternoon/morning groupshowedhighersalivarytestosteroneconcentrationsamong both women (F_(1,38)=7.003, p=0.012, ²P=0.156) and men (F_(1,44)=7.645,p=0.008,²P=0.148)incomparisontothemorn- ing/afternoon group.A time effect was alsofoundamong men (F_(1,44)=12.241,p<0.001,²P=0.218),revealingthattestosterone concentrationssignificantlydecreasedfollowingtheTSST.Among women,menstrualstatuswasnotrelatedtodifferencesinconcen- trationsoftestosterone(p=0.26).Bycontrast,oralcontraceptive usewasrelatedtolowertestosteroneconcentrationsincomparison tonon-use(F_(1.38)=6.267,p=0.017,²P=0.142).

For women (Fig. 1A), lesbian/bisexual women showed higher testosterone concentrations than heterosexual women (F_(1,34)=8.648, p=0.006, ²P=0.203). Covariation effects were foundfor visitorder (F_(1,34)=4.182,p=0.049,²P=0.110) while trending for oral contraceptive use (F_(1,34)=3.591, p=0.067, ²P=0.096) and time X menstrual cycle phase (F_(1,34)=3.044, p=0.090,²P=0.082).

For men (Fig. 1B), no significant within-subjects effects or interactionsweredetected(ps>0.364).Between-subjectsresults revealednogroupdifferences(ps>0.15)otherthanthecovarying effectofvisitorder(F_(1,42)=4.892,p=0.032,²P=0.104).

3.2. Estradiol

Preliminary analyses demonstrated that visit order was not significantly related to salivaryestradiol concentrations among women(p=0.120)andmen(p=0.121)despiteatrendtowardsa timebygroupinteractionamongwomen(F_(1,38)=2.991,p=0.092, ²P=0.073)andatrendtowardsatimeeffectinmen(F_(1,44)=3.411, p=0.071, ²P=0.072) marked in both cases by non-significant decreasesinresponsetotheTSST.Estradiolconcentrationsdidnot differaccordingtomenstrualcycle(p=0.508)oraccordingtooral contraceptiveuse(p=0.233)amongwomen.

Forwomen(Fig.1C),nobetween-subjectseffects(ps>0.41)or within-subjectseffectsorinteractions(ps>0.18)weredetectedfor estradiolconcentrationswiththeexceptionofavisitordercovari- ationeffect(F_(1,34)=4.881,p=0.034,²P=0.126).

Formen(Fig.1D), nobetween-subjectseffects(ps>0.183)or within-subjectseffectsorinteractions(ps>0.319)weredetected forestradiolconcentrations.

3.3. Progesterone

Inpreliminaryanalyses,theafternoon/morninggroupshowed higher salivary progesterone concentrations than the morn- ing/afternoon group among women (F_(1,38)=4.083, p=0.050, ²P=0.097)butnotmen(p=0.272).Forwomen,nogroupdiffer- encesinprogesteroneconcentrationsweredetectedasafunction ofmenstrualstatus(p=0.776)ororalcontraceptivesuse(p=0.282).

For women (Fig. 1E), lesbian/bisexual women displayed higher progesterone concentrations than heterosexual women (F_(1,34)=4.084,p=0.05,²P=0.107).BeyondatimeXmenstrualsta- tuscovariationinteraction(F_(1,34)=3.772,p=0.06,²P=0.10).No otherbetween-subjectseffects(ps>0.17)orwithin-subjectseffects orinteractions(ps>0.13)weredetected.

For men (Fig. 1F), nosignificant findings were detected for between-subjecteffects (ps>0.224) or within-subject effects or interactions(ps>0.388).

3.4. Sex-specificassociationsamongmeansexhormonesand stressindices

Table3reportsalldescriptivestatisticsandcorrelativestatis- ticsfor meansexhormones,cortisol systemicoutput,allostatic load,andperceivedstressstratifiedaccordingtosex.First,testos- teronewaspositivelyassociatedwithcortisolsystemicoutputfor bothsexesandwithperceivedstressformen.Second,estradiolwas positivelyassociatedwithallostaticloadandperceivedstressonly amongmen. Third,progesteronewaspositivelyassociatedwith cortisolsystemicoutputforbothsexes.

Re-analysessplitfurther bysexand sexualorientationwere nextperformed.Forlesbian/bisexualwomen,allostaticloadwas positivelyassociatedwithestadiol(r=0.458,p=0.049).Forhetero- sexualwomen,cortisolsystemicoutputwaspositivelyassociated withtestosterone(r=0.629,p=0.001)andprogesterone(r=0.483, p=0.012). For gay/bisexual men, cortisol systemic output was positivelyassociated with testosterone (r=0.501,p=0.024). For heterosexualmen,allostaticloadwaspositivelyassociatedwith estradiol(r=0.508,p=0.022)andperceivedstresswaspositively associated with testosterone (r=0.676, p=0.001) and estradiol (r=0.457,p=0.043).

-15 +15 40

60 80 100 120 140

Lesbian/Bisexual Women Heterosexual Women

**

Time

Testosterone (pg/ml)

-15 +15

2 3 4 5 6

Lesbian/Bisexual Women Heterosexual Women

Time

Estradiol (pg/ml)

-15 +15

0 50 100 150 200

Lesbian/Bisexual Women Heterosexual Women

*

Time

Progesterone (pg/ml)

-15 +15

40 60 80 100 120 140

Gay/Bisexual Men Heterosexual Men

Time

Testosterone (pg/ml)

-15 +15

2 3 4 5 6

Gay/Bisexual Men Heterosexual Men

Time

Estradiol (pg/ml)

-15 +15

0 50 100 150 200

Gay/Bisexual Men Heterosexual Men

Time

Progesterone (pg/ml)

A B

C D

E F

Fig.1.Sex-specificestimatedmean(±SE)salivarytestosterone(AandB),estradiol(CandD),andprogesterone(EandF)—15minbeforeand+15minafterexposuretothe TrierSocialStressTest.Valuesareadjustedforageandvisitorderforbothsexesaswellasmenstrualstatusandoralcontraceptionforwomen.Note:**=p<0.01;*=p<0.05.

Table3 Descriptivestatisticsandcorrelationcoefficientsstratifiedaccordingtosex. WomenMen 1.2.3.4.5.6.7.1.2.3.4.5.6.7. 1.SexualOrientationA–– 2.Testosterone(M).586***–−0.66– 3.Estradiol(M)0.060.314*–0.1000.213– 4.Progesterone(M).434**.425**.429**–0.063.349*.664***– 5.CortisolSystemicOutput(AUCg)0.245.428**0.198.382*–−0.018.496***0.004.382**– 6.AllostaticLoad0.1620.0150.0340.1550.114–−0.278†0.140.333*0.187−0.044– 7.PerceivedStress0.0540.064−0.0600.0700.0480.179–0.033.455**.404**0.2390.0840.183– M–84.236.58144.2311.252.3325.0–168.555.99104.4912.302.7624.85 SD–30.712.35111.434.881.496.71–40.611.958.87.01.881.88 A=sexualorientationwascodedas0=heterosexualand1=homosexual/bisexual;AUCg=areaunderthecurvewithrespecttoground;M=mean;SD=standarddeviation. †<0.10. *p<0.05. **p<0.01. ***p<0.001.

3.5. Supplementalresultsforindividualsbiomarkersamong women

Individualbiomarkerscomprisingtheallostaticloadindexwere assessed toexplore convergence of salivarysex hormone find- ingsreportedabovewithotherbiomarkersextractedfromblood, urine,etc.AnalysisemployedseparateANCOVAscontrollingfor ageamong women only. Results revealedthat lesbian/bisexual women had higher levels of dehydroepiandrosterone-sulphate (F_(1,36)=5.31, p=0.027,²P=0.129; Fig.2A)and lowerlevels of LDL-cholesterol(F_(1,36)=4.155,p=0.049,²P=0.103;Fig.2B)than heterosexualwomen.

4. Discussion

Thisstudy exploredwhetherunbound salivarytestosterone, estradiol,andprogesteroneconcentrationsvaryasafunctionof sexualorientationaswellasbyacomprehensivearrayofstress indices. We foundthat lesbian/bisexualwomen showed higher overalltestosteroneandprogesteroneconcentrationsasagroup in comparison to heterosexual women, while menshowed no differencesinHPG-axisfunctioning.Insupplementalanalyses,les- bian/bisexualwomen alsoshowedhigherconcentrations ofthe androgenprecursorDHEA-SandlowerLDL-cholesterolthanhet- erosexualwomen.Besidesthesesexualorientationeffectsamong women vis-à-vis basal biomarkers, we also found that mean sexhormoneswereassociatedtostressindices;namely,cortisol systemicoutput,allostaticloadcomprising21biomarkers,andper- ceivedstress.Furthermore,ourfindingssuggestthatrecalibrations inHPG-axisphysiologymaybeexplainedbyintertwinedpsychoso- cialprocesses(e.g.,perceivedstress),concomitantbiobehavioral activities(e.g.,cortisoldynamics), and cumulative physiological dysregulations(e.g.,allostaticload)ratherthanstrictlyaccording tosexualorientation.

Independentofsexandsexualorientation,meansexhormones variedaccordingtostressindices.First,cortisolsystemicoutput representing10repeatedmeasurementsthroughouttheTSSTvisit was positively associated with mean testosterone and proges- teronelevelsforwomenandmen.Second,wefindnovelevidence that allostaticloadindexed with21 biomarkerswas correlated withestradiolonlyamongmen.Third,perceivedstresswaspos- itivelyassociatedwithtestosteroneandestradiolonlyamongmen.

Whenbrokendownaccordingtosexandsexualorientationgroups, lesbian/bisexualwomen and heterosexual men showedsimilar positive associations betweentestosterone and cortisol (AUCg), whilegay/bisexualmenandheterosexualwomenshowedposi- tiveassociationsbetweenestradiolandAL.Thisintriguing“gender reversal” in directionalassociations is consistent withour cor- tisolreactivityfindingsinthis samplewherebylesbian/bisexual womenhyper-reactedlikeheterosexualmenwhilegay/bisexual menhypo-reactedakintoheterosexualwomen(Justeretal.,2015).

Our findings among women contribute to a mixed liter- ature. Specifically, higher basal testosterone concentrations in lesbian/bisexualwomenis inaccordwithsomeexisting reports (Loraineetal.,1971; Loraineet al.,1970), butalsoindisaccord withstudiesshowingnodifferences(Downeyetal.,1987;Griffiths etal.,1974).Lesbian/bisexualwomen alsodemonstratedhigher basalprogesteroneconcentrationsincomparisontoheterosexual women.Whilethisresultisnovelforwomenbelongingtoasex- ualminoritygroup,astudyof92heterosexualwomen(Fleischman etal.,2015)foundthathomoeroticmotivationwassignificantly andpositivelyassociated withsalivaryprogesteronelevels.It is important tonote,however, that thevariation in progesterone concentrations varied considerably. In addition, we also found forthefirsttimethat lesbian/bisexualwomen showedelevated

0 2 4 6 8 10

Sexual Orientati on

*

Lesbian/Bisexual Women Heterosexual Women

0 1 2 3

Sexual Orientati on

LDL- Choles ter ol (mmol/l )

Lesbian/Bisexual Women Heterosexual Women

A B

Fig.2. Women-specificestimatedmean(±SE)(A)dehydroepiandrosterone-sulphate(DHEA-S)and(B)low-density-lipoprotein(LDL)-cholesterolbetweenlesbian/bisexual womenandheterosexualwomen.

Note:*=p<0.05.

concentrationsof serumDHEA-Sandreduced concentrationsof LDL-cholesterolaspartofblooddrawonaseparatedayfromthe TSSTsession. Thisprovidessomeconvergenceof evidencewith salivarysteroidhormone datathatstrengthensourconclusions.

Inaddition,DHEA-Swaspositivelycorrelatedwithtestosterone, whichisanexpecteddirectionofassociationgiventhebiochemical synthesisoftestosteronefromDHEA-Sthroughandrostenedione.

Alongstandingproposalinneuroendocrinologyhasbeenthat elevated androgens among sexual minority women might be explainedby environmentalstress(Meyer-Bahlburg,1979), and this may also be true of DHEA-S. Similar to cortisol, serum concentrationsofDHEA-Sareelevatedinresponsetoacutepsy- chosocialstressinducedbytheTSST(Lennartssonetal.,2012b).

Giventhatwealsofoundthathigherpost-stressorcortisolreac- tivityin lesbian/bisexualwomen incomparisontoheterosexual women (Juster et al., 2015), as well as the antagonizing rela- tionship between DHEA-S and cortisol, it is possible that the uniquepsychosocialcontextsdrivingthisendocrinephenomenon maysubsequentlyraisebasalDHEA-Sinacompensatoryfashion.

Finally,thesynthesisofsteroidhormonesdependsinpartonLDL- cholesterol, which explainswhylevels werelow among sexual minoritywomen.

Our findings among lesbian/bisexual women converge with literature suggesting a higher exposure to elevated prenatal androgensotherwisenotpresentamongmen.Forexample,pre- natal androgen exposure acts to reduce otoacoustic emissions (OAE;echo-likewaveformsemittedbythecochlea)(McFadden, 2002)andtodecreasethe2D:4Dfingerratio(Manning,2002).A meta-analysisshowedthat heterosexualwomen havea greater (morefeminine)2D:4Dratiothanlesbianwomen,butnodiffer- ences emerge for men(Grimbos et al., 2010).Lesbian/bisexual womenalsoshowdecreasedOAEs(McFaddenandPasanen,1998).

Furthermore, prenatal exposure to exogenous estrogens (e.g., diethylstilbestrol)inwomenisassociatedwithnon-heterosexual proclivities(Ehrhardtetal.,1985;Meyer-Bahlburgetal.,1995),but inconsistentlyamongmen(Hines,2011).Inthecontextofprena- talstress,oneproposalwhythesesexhormoneassociationsare notpresentinmenmaybeduetotesticularandrogenproduction thatcompensatesforstressalterationsinandrogensecretionthat isabsentamongwomen(Hines,2011).Whileprenatalconditions alone likely cannotdetermine sexual orientation(Ehrhardtand Meyer-Bahlburg,1981),thehigherandrogensamonglesbiansin ourstudy(Fig.1A)conformwiththisnotionthatdifferencesinpre-

natalandrogenexposuremightaccountforsomesexualminorities amongwomen,butnotamongmen(Fig.1B).

Regardlessofsexualorientation,menshowedunprecedented decreasesintestosteronefollowingtheTSST.Studiescontrasting timechangesintestosteronefollowingtheTSSTreportnodiffer- encesat+10minor+25min(SchoofsandWolf,2011),butthen increasedtestosteronewasdetectedat+30min(Gerraetal.,2000), at+40min(Lennartssonetal.,2012a),andfinallythroughoutper- sonalizedpeaklevelsrangingfrom+25min,+47min,and+77min (Bedgoodetal.,2014).Ourfindingofdecreasedtestosteronefrom

−15minpre-TSSTto+15minpost-TSSTissurprisingly fast.One possibilityisrapidautonomicregulationand/oranticipatorymod- ulation(e.g.,visitorder)ofstressresponsesignallingthatdampens testosterone concentrations for men. Despite noclearevidence thatluteinizinghormonediffersamongLGBindividuals,luteinizing releasinghormonesecretionisregulatedbyhypothalamicneurons thataremodulatedbyfast-actingcatecholamines(Goorenetal., 1990).Alternatively,themainmethodologicaldifferencebetween previousHPG-axisTSSTstudiesandourstudyisthatweusedthe

‘panelout’TSSTvariation wheretheevaluativejudge ishidden behindaone-waymirror.Althoughpurelyspeculative,thelackof visualsocial-evaluativethreatmightdecreasetestosteroneamong menandcouldbeexploredfurtherinthestressreactivityliterature.

Given that sex hormones are characterized by situational- dependent release, and are synergized by other biobehavioral functions(NewmanandJosephs,2009;Turanetal.,2014),consti- tutionalfactorsandsocialbehaviorsmightmodulatesexhormone physiology.Forinstance,testosteroneandcortisolinteracttopre- dictaggressiveanddominantbehaviors(CarréandMehta,2011).

Highlevelsofsexualminoritystresscoupledwithinadequatecop- ingresourcesamongstlesbianandbisexualwomen(IOM,2011) mayrecalibrateHPGaxisequilibriatofavouranendocrinemilieu underlyingalternativecopingstrategies.Futurestudiesmustassess thesepsychosocialconstructsinbetterdetailthanthecurrentstudy allowed.Bycontrast,theabsoluteabsenceofdifferencesinsexhor- monesamongmenisremarkablyconsistentwithexistingliterature (Meyer-Bahlburg,1977).

While males who identify with a sexual minority iden- tity/orientationfacesimilarpressuresandstigma,thereisempirical andtheoreticaljustificationstoconsidertheircopingmechanisms separatefromsexualminoritywomen(Lewisetal.,2012).Con- sideringthisframework,alesbian/bisexualspecificup-regulation oftestosteroneasaresultofauniquecombinationofpsychoso-