Development and validation of a semi-quantitative ultra-high performance liquid chromatography-tandem mass spectrometry method for screening of selective androgen receptor

A semi-quantitative method was developed to monitor the misuse of 15 SARM compounds belonging to nine different families, in urine matrices from a range of species (equine, canine, human, bovine and murine).

jou rn al h om ep a g e : w w w e l s e v i e r c o m / l o c a t e / c h r o m a

a Institute for Global Food Security, School of Biological Sciences, Queen’s University Belfast, BT9 5AG, United Kingdom

b Laboratory, Irish Greyhound Board, Limerick Greyhound Stadium, Ireland

c Applied Science Department, Limerick Institute of Technology, Moylish, Limerick, Ireland

d Laboratorio Antidoping, Federazione Medico Sportiva Italiana, Italy

e RIKILT Wageningen University & Research, European Union Reference Laboratory, Wageningen, the Netherlands

f Irish Diagnostic Laboratory Services Ltd., Johnstown, Co Kildare, Ireland

a r t i c l e i n f o

Article history:

Received 25 January 2019

Received in revised form 16 April 2019

Accepted 17 April 2019

Available online 22 April 2019

Keywords:

Selective androgen receptor modulators

UHPLC-MS/MS

Urine

Doping control

Residue and food safety

a b s t r a c t

Investigation of alternative pharmacophores to

anabolic-androgenicsteroids(AAS)whichcanseparateanaboliceffectson

muscleandbonefromandrogenicactivityinothertissuessuch

astheprostateandseminalvesicles[1 hasledtotheemergence

ofselectiveandrogenreceptormodulators(SARMs),aclassof

non-steroidalagentswithaffinityfortheandrogenreceptor(AR)similar

tothatofdihydrotestosterone(DHT)[2 Asaheterogeneousgroup

ofmoleculesincorporatingarangeofpharmacophoresthatlackthe

∗ Corresponding authors.

E-mail addresses: eventura01@qub.ac.uk, emiliano.ventura@outlook.it

(E Ventura), a.gadaj@qub.ac.uk, agadaj@gmail.com (A Gadaj).

steroidnucleusoftestosteroneanddihydrotestosterone[2 SARMs behaveaspartialARagonistsinandrogenictissues(prostateand seminalvesicle)butactmainlyasfullARagonistinanabolic tis-sue(muscleandbone)[1,3 Thestructuralmodificationofknown

AR antagonists, suchas thenonsteroidal antiandrogens bicalu-tamide,flutamide,hydroxyflutamideandnilutamide[2,4 resulted

intheinitialgenerationofnovelnonsteroidalARagonistswithan arylpropionamide-nucleus,namelySARMS-1andandarine(S-4), forpotentialuseastherapeuticsinbenignprostatichyperplasia (BPH)andandrogen-deficiencyrelateddisorders[5–7].Sincethen, severalclasses of chemical scaffoldswith SARM-likeproperties havebeendevelopedexhibitingstronganabolicactivityandhigh tissueselectivity,elevatedabsorptionratesviaoraladministration, andreducedundesirableandrogenicside-effects[8–11].Potential pharmacologicapplicationsofSARMshavebeenfocusedtowards https://doi.org/10.1016/j.chroma.2019.04.050

0021-9673/© 2019 The Authors Published by Elsevier B.V This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.

cancerandotherchronicdiseases,aswellasinhypogonadism,

hor-monereplacementtherapy,malecontraception,benignprostatic

hyperplasia,breastandprostatecancer[8,9

Easeofavailability,simplicityofuse,advantageousbiological

effects[12] andshortdetectionwindows[13,14,23–25,15–22]

arekey featuresincreasing thepotentialfor SARMmisuse, and

consequentlythey are widely recognised as drugs of abuse in

both human and animal (e.g equine and canine) sports, and

asemerging candidatesforillicit useinfood-producing species

[19].AlthoughmanySARMcompoundsarecurrentlyundergoing

evaluationinvariousstudies,asyetnoneareapprovedfor

pharma-ceuticaluse[8 thereiswidespreadSARMavailabilityvia

black-andgrey-marketsources.Recently,variousSARMs(e.g.S-4,S-22

andLGD-4033) havebeenidentified withinblack-market

prod-ucts[26–29],onlinevendors[30–34],andconfiscatedgoods[35]

SARMshave gained particular popularityin professional sports

andarebannedbytheWorldAnti-DopingAgency(WADA)[36],

theInternationalAgreementonBreeding,Racing andWagering

(IABRW)[37]andFédérationEquestreInternationale(International

Equestrian Federation, FEI)[38], withmany reports of positive

findingsfromroutinetesting[39–42].Morerecently,65adverse

analytical findings (AAFs) for a range of SARMs (e.g.andarine,

ostarine,LGD-4033andRAD140)werereportedinhumansportin

2017alone[43].ThepotentialforSARMstobefurtheradoptedfor

useinfood-producinganimals(e.g.incattlelivestock)toincrease

musclegrowthandreducefatmassalsoremainsadistinctthreat

[44]

Advancedand reliablescreening andconfirmatory analytical

assaysare requiredtodetect SARMusefor dopingpractices in

sportandmonitorforpotentialmisuseinstockfarming.Anumber

ofSARMcompoundshavebeensuccessfullyincludedintohuman

anti-dopingcontrol[45–52]withsomeassaysdevelopedforequine

racinganimals[14,17,18,53].However,todateonlyalimited

num-ber of analytical procedures covering solely arylpropionamides

havebeenestablishedforfoodsafetyanalysis[15,16,54,55]

LC-MSandoccasionallyGC-MS-basedapproacheshavebeenapplied

toelucidatethemetabolicpathwaysofsomeemergingSARMsin

variousspeciestosupportthedevelopmentofdetectionassaysfor

thesecompounds[14,19,22,56].Moreover,thedetectionofSARMs

andassociatedmetabolitesincanine[57,58],rodents[57–62],as

wellas humanspecimens [63–67] were conducted to support

SARMclinicalstudies.Whilsturineandbloodarecommon

matri-cesofchoice,faeceshavebeenproposedasanalternativematrix

fortheanalysisofarylpropionamide-derivedcompoundsinbovine

[15,16],canine[57]andrats[57,62].However,thereported

screen-ingand/orconfirmatoryassaysaretypicallycapableofanalysisof

eitherasingleSARMcompoundoralimitednumberofSARMsand

relatedmetabolitesinasinglespecimen(Table1)

In the present study, an innovative fast, simple and

cost-effectivesemi-quantitativemulti-residueUPLC-MS/MSscreening

assay wasdeveloped for a group of 15 key SARM compounds

with different physicochemical properties, chosen based upon

theirreporteduseinhumanandanimalsportsandavailabilityas

certifiedanalyticalstandards.TargetSARMcompoundsincluded

AC-262536,andarine(S-4),bicalutamide,BMS-564929,GLPG0492,

LGD-2226,LGD-4033,Ly2452473, ostarine(S-22),PF-06260414,

RAD140,S-1,S-6,S-9andS-23(Fig.1).Thedevelopedmethodhas

beenvalidatedinurinematricesfromarangeofspecies(equine,

canine,human,bovineandmurine)inaccordancewiththeEU

Com-missionDecision2002/657/ECcriteria[68]and EuropeanUnion

ReferenceLaboratoriesforResidues(EU-RLs)guidelines[69].The

assaywasemployedtoscreenforSARMresiduepresenceinurine

sourcedfromracinganimals(equineandcanine),amateurandelite

athletes,aswellasfarm(bovine)andexperimentallytreated

ani-mals

Sample volume

bicalutamide, ostarine

Linearity 0.0–5.0

1 RSDr

Linearity 2.5–250

bicalutamide, hydroxyflutamide, ostarine

Sensitivity 0.25

Linearity 0.25–30

Precision 0.6–17.6

volume (mL)

Sample preparation Detection limits Method performance Reference

buffer (pH 5, 0.2 M) eLOQ0.1ngmL−1 N/A

[ 15 ] 3.0 Enzymatic hydrolysis

followed by SPE (as above)

Andarine (S-4), bicalutamide, hydroxyflutamide, ostarine (S-22)

followed by SPE (Oasis HLB)

LOD 0.015–0.142 ng mL−1

Linearity 0.25–25 ng mL−1

[ 55 ]

USFC-Q-IM-ToF (mode: MS E )

LOD 0.0018–0.0406 ng mL −1

Andarine (S-4), ostarine (S-22)

UHPLC-HRMS (modes: MS, DDA)

followed by on-line SPE (Oasis HLB)

eLOD 1.25 ng mL −1

(S-22), 5 ng mL−1 (S-4)

Inter-day precision 9.4–11.7 % Recovery 11–15 %

[ 53 ]

Andarine (S-4), ostarine (S-22), ostarine glucuronide, S-23, S-24

HPLC-HRMS Human 0.09 “Dilute-and-shoot” LLOD <0.1 ng mL −1 Intra-day precision

3.2–7.7 % Inter-day precision 4.4–14.5 %

[ 45 ]

Andarine (S-4), bicalutamide, ostarine (S-22)

followed by LLE (TBME,

K 2 CO 3 /NaHCO 3 buffer)

CC␣ 0.025 ng mL−1 CC␤

0.025-0.05 ng mL -1

Linearity 0.0–2.0 ng mL−1 Accuracy 89-105 % RSD r 2.6–10.4 RSD RL 2.9–12.2 %

[ 19 ]

Andarine (S-4), ostarine (S-22)

followed by SPE (Bond-Elut Plexa PCX), 2% aq HCOOH

Andarine (S-4), M5 metabolite of S-4, ostarine (S-22)

followed by alkaline LLE (pentane and diethyl ether)

LOD 0.1 ng mL −1

(S-4, S-22)

Andarine (S-4), metabolite of S-4

HPLC-MS/MS Human 0.09 Dilute-and-shoot¨¨ LOD1.0ngmL−1 Intra-day precision

5.5–10.3 % Inter-day precision 0.38-4.7 %

[ 47 ]

Andarine (S-4), S-1, S-9, S-24

7.6–11.6 % Inter-day precision 9.9–14.4 % Recovery 85-105 %

[ 48 ]

M1 metabolite of S-1

10–10,000 ng mL−1 Relative recovery 89%

[ 62 ]

Arylpropionamide,

pyrrolidinyl-benzonitrile

Andarine (S-4), ostarine (S-22), S-1, LGD-4033, metabolites:

O-dephenyl andarine, O-dephenyl ostarine

GC-EI-Q-ToF (modes: MS and MS/MS by continuous switching)

followed by LLE (TBME NaHCO 3 /K 2 CO 3 buffer (pH 9.5)) and derivatisation (MSTFA/ethanethiol/NH 4 I)

LLOD 0.2–10 ng mL −1

Bicyclic hydantoin,

quinolinone

BMS-564929, LGD-2226

followed by LLE and derivatisation

LLOD 0.2 ng mL −1

(LGD-2226) LLOD

10 ng mL −1

(BMS-564929)

Intra-day precision 6.8–16.6 % Inter-day precision 12.7–17.7 % Recovery 83–85 %

[ 80 ]

Table 1 (Continued)

volume (mL)

Sample preparation Detection limits Method performance Reference

Bicyclic hydantoin,

benzimidazole

BMS-564929, 5,6-dichloro-benzimidazole derivatives (n = 4)

20 ng mL−1 (BMS-564929)

Intra-day precision 2.4–13.2 % Inter-day precision 6.5–24.2 % Recovery 89–106 %

[ 79 ]

HLOQ 50 ng mL −1

Ly2452473 UHPLC-HRMS

(modes: MS, MS/MS)

HLOQ 10 ng mL −1 Precision ≤ 6.5 %

Accuracy ≤ 9.8 %

[ 65 ] Quinolinone US 6,462,038,

LG-121071

followed by LLE (TBME,

pH 9.6, carbonate buffer (0.1 M)) and

derivatisation (MSTFA/NH 4 I/dithiothreitol)

LOD 1.0 ng mL −1 Linearity

5.0–500 ng mL −1

Intra-day precision 8.1–14.8 % Inter-day precision 9.5-16.2 % Recovery 97-101%

[ 76 ]

LGD-2226, 6-alkylamino-2-quinolinones (n = 2)

followed by SPE (Oasis HLB) and derivatisation (MSTFA)

LOD 0.01–1.0 ng mL −1

LOQ 0.03–3.0 ng mL −1

Linearity LOQ-100 ng mL −1

Intra-day repeatability 5–9 % Recovery 92–111

%

[ 77 ]

LGD-2226, 6-alkylamino-2-quinolinones (n = 3)

followed by LLE (TBME,

pH 9.6, K 2 CO 3 /NaHCO 3

buffer)

LLOD 0.01-0.2 ng mL −1 Intra-day precision

3.2-8.5 % Inter-day precision 6.3-16.6 % Recovery 81-98 %

[ 51 ]

Pyrrolidinyl-benzonitrile

LGD-4033 UHPLC-Q-ToF

(mode: MS E )

followed by SPE (Oasis HLB)

LOD 2.6 ng mL −1

[M-H] − , 0.5 ng mL−1 [M+HCOOH-H]−

Tetrahydroquinolinone LG121071 HPLC-MS/MS Human 1.0 Enzymatic hydrolysis

followed by LLE (TBME,

pH 9.6, carbonate buffer (20%))

LLOD 0.5 ng mL −1 Linearity

0.5–5.0 ng mL −1 , 1–200 ng mL −1

Intra-day precision 2.3–8.5 % Inter-day precision 7.2–11.7 % Recovery 40 %

[ 49 ]

Tricyclic

tetrahy-droquinoline

Tricyclic tetrahy-droquinoline derivatives (n = 3)

followed by LLE (TBME,

pH 9.6, K 2 CO 3 /NaHCO 3

buffer, Na 2 SO 4 )

LLOD 0.2–0.6 ng mL −1

Intra-day precision 6.4–15.1 % Inter-day precision 11.3–21.8 % Recovery 92–97 %

[ 50 ]

9 pharmacophores 15 analytes UHPLC-MS/MS Equine, bovine,

canine, human, murine

0.2 LLE (TBME, NH 4 OH aq.

(50 mM, pH 10.5))

CC␤ 1 ng mL−1,

2 ng mL−1(S-4),

5 ng mL −1

(BMS-564929) eLOD 0.01–0.75 ng mL −1

Precision 9.8–33.6 % (equine) Sensitivity 95–100 % Recovery 74–94 %

Actual method

Fig 1.Chemical structures of SARMs included in the actual UHPLC-MS/MS method.

2.1 Reagentsandapparatus

Ultra-purewater(18.2MOhm)wasgeneratedinhouseusing

a Milliporewater purification system(Millipore, Cork, Ireland)

Methanol (MeOH)and acetonitrile (MeCN), both ChromasolvTM

LC–MSgrade,aswellasammoniumhydroxidesolution,≥25%in

H2Oandaceticacid,botheluentadditivesforLC–MS,weresourced

fromHoneywell(VWRInternational,Dublin,Ireland).LiChrosolv®

LC grade tert-butyl methyl ether (TBME), ethanol (puriss p.a.,

ACSreagent,absolutealcohol,withoutadditive,≥99.8%),dimethyl

sulfoxide(ACSreagent,≥99.9%)andacetonitrile-D,99.5%

(MeCN-D)weresourced from Sigma-Aldrich(Dublin, Ireland).SafeSeal

polypropylene micro tubes (2mL) were obtainedfromSarstedt

(Nümbrecht,Germany) ADVX-2500 multi-tube vortexer(VWR

International,Dublin, Ireland), a Hettich Micro 200R centrifuge

fromDavidson&Hardy(Belfast,UK)andaTurbovapLV

evapo-ratorfromCaliperLifeSciences(MountainView,USA)wereused

duringsamplepreparation.Inthisstudy,thedensityofurinewas

measuredthroughspecificgravity(SG)oftheurinesamplesusing

apocketrefractometerPAL-USG(CAT)fromAtago(Tokyo,Japan)

AC-262536 (P/N 96443-25MG), andarine (S-4, P/N

78986-25MG),bicalutamide(P/NPHR-1678-1G),LGD-2226(P/N

07682-25MG), Ly2452473 (P/N CDS025139-50MG), PF-06260414 (P/N

PZ0343-5MG), S-1 (P/N 68114-25MG), S-6 (P/N 79260-25MG)

andS-23(P/N55939-25MG)werepurchasedfromSigma-Aldrich

(Dublin, Ireland) LGD-4033 (P/N CAY9002046-50mg), ostarine

(S-22, P/N MK-2866) and RAD140 (P/N CAY18773-1mg) were

purchasedfromCambridgeBioscienceLtd.(Cambridge,UK)

BMS-564929(10mMsolutioninDMSO,P/NHV-12111)andGLPG0492

(10mMsolutioninDMSO,P/NHY-18102)werepurchasedfrom

MedChem Express (Sollentuna, Sweden) S-9 (P/N D289535),

Bicalutamide-D4(P/NB382002)andS-1-D4 (P/ND289532)were

purchased from Toronto Research Chemicals (TRC; Toronto,

Canada).Allstandardsandinternalstandardsstocksolutionswere

preparedataconcentrationof1mgmL−1 inMeCN,DMSO,EtOH

andMeCN-D,respectively.Intermediatemixedstandardsolutions werepreparedatthefollowingconcentrations:20/40/100,1/2/5 and0.1/0.2/0.5␮gmL−1inMeCNbyserialdilutions.Working qual-itycontrolstandardsolutionataconcentrationof10/20/50ngmL−1 waspreparedinMeCN.Intermediateinternalstandardmix solu-tions were prepared at 20 and 1␮g mL−1, respectively, using MeCN-Dasthediluent.Aworkinginternalstandardmixsolution waspreparedat50ngmL−1 inMeCN-D.Allstandardsand inter-nalstandardsstocksolutionswerefoundtobestableforatleast oneyearwhenstoredat−20◦Cduring‘in-house’stabilitystudies.

Workingqualitycontrolstandardandworkinginternalstandard mixsolutionswerefoundtobestableforatleast3monthswhen storedat−20◦C.

2.2 Preparationofextractedmatrixscreenpositiveandrecovery controlchecks

Negativequalitycontrol(QC)sampleswereobtainedby pool-ingaliquots(n=5–10)ofnegativeurinesamples.Extractedmatrix screenpositivecontrolswerepreparedbyfortifyingthreenegative

QCsamples(200␮L)priortoextractionwith20␮Lofquality con-trolstandardsolutiontogiveascreeningtargetconcentrationof

1ngmL−1 inurineforallanalytesexcludingandarineand

BMS-564929giving a concentrationof 2and 5ngmL−1,respectively Additionally,twoblankQCsampleswerespikedafterextraction withqualitycontrolstandardsolution(20␮L)tomonitorforloss

ofanalytesduringextraction

2.3 Samplepreparation All sampling and analysis were performed under the guid-anceandapprovaloflocalethicalregulations.Urinesampleswere storedat−80◦Cpriortoanalysis.Urinesampleswerecentrifuged

at4500×gfor10minat4◦C,andfollowingcheckingofpHand specificgravity(SG),aliquoted(200␮L)into2mLmicrotubes Sam-pleswerefortifiedwith20␮Lofa 50ngmL−1 internalstandard mixsolutionandlefttostandfor15min,anda200␮Lvolumeof

con-tentswerevortexedfor60sand1.2mLofTBMEwassubsequently

added.Followingvortexingfor15min,sampleswerecentrifugedat

15,000rpm(21,380×g)for10minat4◦C,andsupernatants

trans-ferredintocleanempty2mLmicrotubesandevaporatedtodryness

underflowofnitrogen(≤5bar) at40◦C onaTurbovap LV

sys-tem.SampleswerereconstitutedinH2O:MeCN(4:1,v/v;100␮L)

byvortexing(5min)and9␮Lofextractswereinjectedontothe

UHPLC-MS/MSsystem

2.4 UHPLC-MS/MSconditions

SeparationswereperformedusingaWaters(Milford,MA,USA)

AcquityI-ClassUPLC®systemcomprisingofastainlesssteelLuna®

OmegaPolarC18analyticalcolumn(100×2.1mm,100Å,1.6␮m)

(Phenomenex,P/N00D-4748-AN)equippedwithKrudKatcherTM

UltraHPLCin-linefilter(Phenomenex,P/NAF0-8497)maintained

ata temperatureof45◦Candthepumpwasoperatedata flow

rateof0.40mLmin−1.Abinarygradientsystemwasusedto

sep-arateanalytescomprisingofmobilephaseA,0.1%(v/v)aceticacid

inwaterandmobilephaseB,0.1%(v/v)aceticacidinMeOH.The

gradientprofilewasasfollows:(1)0.0min20%B,(2)0.5min20%

B,(3)9.0min99%B,(4) 10.0min99%B,(5)10.1min20%B, (6)

12.0min20%B.Theinjectionvolumewas9␮L.Aftereach

injec-tiontheneedlewaswashedandpurgedwithH2O:MeOH(1:1,v/v)

andH2O:MeOH(4:1,v/v)solutions,respectively.Adivertvalvewas

usedtoreducesourcecontamination(8.50–11.50minaflowwas

divertedtowaste)

SARMresidues weredetectedusing a Waters Xevo® TQ-MS

triplequadrupolemassanalyser(Manchester,UK)operatingboth

inpositiveandnegativeelectrosprayionisationmodes(ESI±).The

UHPLC-MS/MSsystem wascontrolled by MassLynxTM software

anddatawasprocessedusingTargetLynxTMsoftware(bothfrom

Waters).Theelectrosprayvoltagewassetat2.5kV(ESI+)and1.0kV

(ESI−),respectively.Thedesolvationandsourcetemperatureswere

setat550and120◦C,respectively.Nitrogenwasemployedasthe

desolvationandconegases,whichweresetat900Lh−1and50L

h−1,respectively.Argonwasemployedasthecollisiongas,ataflow

rateof0.15mLmin−1,whichtypicallygavepressuresof2.5×10-3

mbar.TheMSconditionswereoptimisedusingIntelliStartby

infu-sionof1␮g mL−1 standard solutionsand 50%mobilephases A

andBatflowratesof5␮Lmin−1and0.2mLmin−1,respectively

Theconevoltagewasoptimisedforeach precursorion andtwo

tofourmostabundantproductfragmentionswereselected.The

selectedreactionmonitoring(SRM)windowsweretime-sectored,

anddwell timeand inter-channeldelays weresettoget

maxi-mumresponsefortheinstrument.Theseconditionsareoutlined

inTable2.Interscan delaywassetto5msbetweensuccessive

SRMwindows,inter-channeldelaywassetto5msandpolarity

switching20ms.Dwelltimesrangedfrom0.005to0.300s(Table2)

Availablestableisotope-labelledanaloguesofbicalutamideand

S-1(bicalutamide-D4 andS-1-D4)wereusedasinternalstandards

forarylpropionamideresidues(Table2).Theresponsefactorwas

obtainedforarylpropionamidesasaratiobetweenanalytepeak

areaandinternalstandardpeakarea,whileinthecaseoftheother

SARMresidues,peakareawasusedastheresponse

2.5 Methodvalidation

Themethod wasvalidatedaccording tothe EUCommission

Decision2002/657/ECcriteriaandEuropeanUnionReference

Lab-oratoriesforResidues(EU-RLs)20/1/2010guidelinesforscreening

assays The following performance studies were carried out to

prove the suitability of the method in achieving the goal for

which it wasdeveloped:selectivity, specificity, detection

capa-bility(CC␤), sensitivity, precision,limit of detection(LOD) and

absoluterecoveryaswellasapplicability,ruggednessandmatrix effects.Validationwascarriedoutatthescreeningtarget concen-tration (Cval)of 1ngmL−1 excludingandarineand BMS-564929 validatedat2and5ngmL−1,respectively.Thedetection capabil-ity(CC␤), definedin2002/657/EC,wascalculatedinaccordance withtheEU-RLs20/1/2010guidelines,byassessingthresholdvalue (T)and cut-off factor (Fm).TodeterminetheT-value, 61blank equineurinesamplesofdifferentoriginswereanalysedusingthe methoddescribedaboveonanumberofoccasionsbytwodifferent analyststoobtaintotalof61 datapoints.TheT-valuewas esti-matedforat leasttwo transitionsfor each analyteasa sumof

ameanresponseand1.64timesthestandarddeviationofnoise levelsacquiredfor61blanksamples.Todeterminethecut-off fac-tor(Fm),61blankequineurinesamplesofdifferentoriginswere fortifiedatthescreeningtargetconcentration(Cval)onnumerous occasionsandthesampleswereanalysedbytwodifferentanalysts This gave a total of 61 independent data points for each ana-lyteatthetargetedconcentrationof1ngmL−1excludingandarine (2ngmL−1)andBMS-564929 (5ngmL−1), respectively The cut-offfactor(Fm)wasestimatedforatleasttwotransitionsforeach analyteasameanresponsedecreasedby1.64timesthestandard deviationofresponseacquiredfor61fortifiedsamples.According

totheEuropeanUnionReferenceLaboratoriesforResidues (EU-RLs)20/1/2010guidelines,thedetectioncapability(CC␤)ofthe screeningmethodisvalidatedwhenthecut-offfactorisgreater thanthethresholdvalue(Fm>T).ItcanthenbededucedthatCC␤

istrulybelowthevalidationlevel.Sincetheveryfirstrequirement expectedfromascreeningmethodistoavoidfalsenegative(also called“falsecompliant”) results, thedetectioncapability ofthe methodwasestimatedastheconcentrationlevelwhere≤5%of false-negativeresultsremain

Thesensitivityofthemethodwasexpressedasthepercentage basedontheratioofsamplesdetectedaspositiveintruepositive samplesi.e.followingthefortification[70].Asensitivity≥95%at thescreeningtargetconcentration(Cval)meansthatthenumberof false-negativesamplesistruly≤5%.Despitebeingarequired per-formancecharacteristictobedeterminedsolelyforquantitative methods[68],precisionwascalculatedasthecoefficientof vari-ation(CV)oftheresponsefollowingfortificationatthescreening targetconcentration(Cval).Limitofdetection(LOD)wasestimated

atasignal-to-noiseratio(S/N)atleastthreemeasuredpeaktopeak Followinginitialdeterminationofthedetectioncapability(CC␤) forequineurine,thedevelopedmethodwasappliedtothesame matrixtypefromfourdifferentspecies-bovine,canine,human andmurineurine, respectively.Murineurinewasincludedasa matrix within the validation process in recognitionthat many SARMmetabolisminvivostudiesutiliseexperimentalrodent mod-elsandassuchthedevelopedmethodmayfindapplicationinsuch studies.Theapplicabilityofthescreeningmethodwasevaluatedby analysing20blankurinesamples(n=5perspecies)andthesame

20blankurinesamples(n=5perspecies)fortifiedatthescreening targetconcentration(Cval)usedpreviouslyforequineurine Ani-malspecieswereincludedintheruggednessstudyasfactorsthat couldinfluencetheresults.Moreover,toinvestigatethe rugged-nessofthedevelopedassay,15differentblankurinesamples(n=5 perspecies)andthesame15blankurinesamples(n=5perspecies) fortifiedatthescreeningtargetconcentrationwereanalysedata differentdayandbyadifferentoperatorthatexecutedthe appli-cabilitystudy[69].Toevaluatematrixeffects inequine,bovine, canine,humanandmurineurine,25blanksamplesfrom differ-entsourcesofeachmatrix(n=5)werepost-extractionspikedat theconcentrationequaltotwotimesthescreeningtarget concen-tration(2×Cval),namely2ngmL−1excludingandarine(4ngmL−1) andBMS-564929(10ngmL−1),respectively.Matrixeffectsforeach analytewerecalculatedaspercentagedifferencesbetweenthe sig-nals obtainedwhen matrix extracts were injected and when a

UHPLC-MS/MS conditions for urine samples.

Bicalutamide-D 4 Arylpropionamide C 18 H 10 D 4 F 4 N 2 O 4 S 5.88 433.2 > 255.1 0.007 26 14 13 – N/A

Andarine Arylpropionamide S-4, GTX-007 C 19 H 18 F 3 N 3 O 6 5.83 440.2 > 150.0 d 0.005 30 30 15 – Bicalutamide-D 4

Bicalutamide Arylpropionamide C 18 H 14 F 4 N 2 O 4 S 5.90 429.2 > 255.0 d 0.007 24 16 13 – Bicalutamide-D 4

LGD-4033 Pyrrolidinyl-benzonitrile VK5211 C 14 H 12 F 6 N 2 O 6.70 337.1 > 267.2 d 0.005 28 10 8 – N/A

Ly2452473 Indole

CDS025139, TT-701 C22H22N4O2 6.51 375.2>272.1

Ostarine Arylpropionamide

S-22, EnoboSarm, GTx-024, MK-2866

C 19 H 14 F 3 N 3 O 3 6.20 388.1 > 118.0 d 0.009 30 20 9 – Bicalutamide-D 4

S-9 Arylpropionamide

4-Desacetamido-4-chloro andarine

C 17 H 14 ClF 3 N 2 O 5 7.26 417.2 > 127.0 d 0.009 30 28 12 – Bicalutamide-D 4

a TR, retention time.

b CE, collision energy.

c SRM 1 (6.45–7.05 min); SRM 2 (4.50–5.10 min); SRM 3 (3.60–4.50 min); SRM 4 (6.20–6.80 min); SRM 5 (5.90–6.50 min); SRM 6 (6.55–7.15 min); SRM 7 (5.75–6.35 min); SRM 8 (6.40–7.00 min); SRM 9 (5.90–6.50 min); SRM

10 (6.60–7.20 min); SRM 11 (6.90–7.50 min); SRM 12 (7.00–7.60 min); SRM 13 (5.60–6.20 min); SRM 14 (7.10–7.70 min); SRM 15 (5.55–6.15 min).

d

Fig 2.Overlay of UHPLC-MS/MS traces of equine urine fortified with 15 SARMs at 1/2/5 ng mL −1

standardsolutionofequivalentconcentrationwasinjected,divided

bythesignalofthelatter[71]

2.6 Applicationofthemethod

Themethoddevelopedinthisstudyhasbeenappliedtoroutine

screeningforthepresenceofSARMresiduesinbovineurine

sam-ples(n=51)fromabattoirsacrossIreland,equineurinesamples

(n=61)donatedbytheIrishEquineCentre(IEC),canineurine

sam-ples(n=109)providedbytheIrishGreyhoundBoardandhuman

urinesamplesdonatedbynon-professionalvolunteerathletes(n=

22)aswellasurinesamplesfromathletes(n=20)suppliedbythe

WADAaccreditedAnti-DopingLaboratoryofRome(Italy),selected

amongthosealreadyreportedasnegative,andafter

anonymiza-tion

3.1 Methoddevelopment

3.1.1 UHPLC-MS/MSconditions

In this study,SARM residues wereanalysed by electrospray

ionisationmassspectrometry(ESI-MS)usingbothpositiveand

neg-ativeionisationmodes.DataacquiredinSRMmodebymonitoring

protonated[M+H]+anddeprotonated[M−H]−molecules,

respec-tively Diagnostic ions obtainedwere in agreement with those

reportedintheliterature.Atleasttwomostabundantproduct

frag-mentionsweremonitoredforeachSARMcompoundyieldingat

leastfouridentificationpoints[68].Theelectrosprayvoltage,

des-olvationandsourcetemperatures,desolvation,coneandcollision

gasflowrateswereoptimisedtogetmaximumresponseforthe

instrument.SRMwindowsweretimesectoredandadequate

con-ditionswereestablishedthrougheffectiveset-upofdwelltimes,

inter-scanandinter-channeldelayaswellaspolarityswitching.A

totalof12–15datapointsweretypicallyobtainedacrossapeakto

attainreproducibleintegrationandthusachievehighlyrepeatable

analysis

A number of different mobile phases and additives

includ-ingvolatilebuffer(ammoniumformate)andacid(formic,acetic)

wereassessedwitharangeofUHPLCcolumnchemistries,namely

AcquityUPLC®:HSST3andCSHC18,Cortecs®:C18andT3(allfrom

Waters),Kinetex:F5,EVOC ,andLunaOmegaPolarC (allfrom

Phenomenex).Comparisonofcolumntypeandmobilephase per-formanceweremadebasedonpeakshape(Supplementarydata

-Fig.1)andrelativeabundanceofanalytes(Supplementarydata -Fig.2and3).OptimalLCconditionswereidentifiedasthatbasedon mobilephasescomprisedofwaterandmethanolbothcontaining 0.1%(v/v)aceticacidemployingaLunaOmegaPolarC18column Gradientconditionsandflowratewereadjustedinordertoachieve mostfavourablechromatographicseparation,andaspresentedin Fig.2,allanalyteswereseparatedwithinthefirst7.70minofthe chromatographicrun

3.1.2 Samplepreparation Oneofthemaingoalsofthisstudywastodeveloparapid, sim-pleandcost-effectivesamplepreparationprocedurethatwouldbe suitableforallthe15SARMsofinterestinurinematrixfromfive dif-ferentanimalspecies:equine,bovine,canine,humanandmurine, respectively.Liquid-liquidextraction(LLE)procedureshavebeen successfullyemployedinbothhumanandequinesportdrug test-ing,aswellasinfoodcontrolapplyingarangeoforganicsolvents e.g.tert-butyl-methylether(TBME),ethylacetate(EA)anddiethyl ether[25,42,51].Nevertheless,tothebestoftheauthors’ knowl-edge,nomulti-residueanalyticalmethodbasedonaLLEhasbeen proposedthatcoversallthe15SARMcompoundsincludedinthe current study.Thisresearchinvestigated theimpactof a range

ofextractionparameters,suchasvolumeofequineurinesample (0.2–2.0mL)andorganicsolvent(ratio1:3,1:6,v/v),pH(3.0,5.0, 9.0and 11.0),saltaddition(sodium andammoniumsulphates), andconcentrationfactor(2,4,13.3)inordertoachieve satisfac-toryrecoveryofallthe15analytes.ThepHhadasignificantimpact bothontheextractionofalltheanalytesandmatrixcoextractive interferences.Overall,apHof5.0workedadequatelyforallthe analytesprovidingwithhigherabsoluterecoveryvalues(78–108

%)inequineurine,butontheotherhanditledtothe unaccept-ablesignalsuppressionforsomeoftheSARMs(e.g.BMS-564929, GLPG0492and RAD140)in comparisontoaLLEatpH9.0 Con-sequently,theoptimumresultswereachievedbytheadditionof

200␮Lofabuffersolution(50mMaqueousNH4OH,pH10.5)to

200␮Lofequineurine,settingthepHvaluearound9.0priortoa liquid-liquidextractionwith1.2mLofTBME

Moreover,supported liquid extraction(SLE) in equine urine was tested employing the Isolute SLE+cartridges (1 and 2mL) and96-wellplate(400␮L).Arangeofparameterswereevaluated,

Fig 3.Average absolute recoveries (and standard deviations, shown by error bars) obtained applying SLE and LLE in equine urine fortified at 1 ng mL −1 (n = 2).

includingurinesample volume(0.2–1mL),pHvalue,aswellas

organiceluent(TBME,EAandDCMasrecommendedbythe

man-ufacturer).AmongtheSLEprotocols,recoveryandprecisionwere

thebestworkingwith200␮Lurineand400␮L96-wellplateunder

alkalineconditions(200␮L50mMNH4OHpH10.5)withTBME

Nevertheless,SLEwasnotdeterminedtobeaprocedureofchoice

duetoabsoluterecoverieslower(42–95%)thanthoseobtainedfor

theabove-mentionedLLEwithTBME(69–91%)asoutlinedinFig.3

Followingextraction(LLEandSLE),theorganicsolvent(TBME)

wasevaporatedat40◦Ctodrynessunderastreamofnitrogen.It

wasfoundthatevaporationofsolventtodrynessdidnotleadtoany

significantlosesofanalytesandconsequentlytheuseofdimethyl

sulfoxide(DMSO)asa“keeper”wasavoided.Moreover,arangeof

differentreconstitutionsolventswasinvestigated,andH2O:MeCN

(4:1,v/v),wasfoundtoprovidesatisfactorysensitivitywith

accept-ablepeakshapesofalltheanalytes.Finally,theoptimumconditions

describedinSection2.3providedwithaverageabsolute

recover-ies,calculatedatthescreeningtargetconcentration,intherange

of74–94%forallSARMsof interestinalltestedurine matrices

(Table4)

3.2 Methodvalidation

3.2.1 Selectivity,specificity,andmatrixeffectstudies

Thespecificityofthemethodwasinvestigatedthrough

monitor-ingforinterferencesintheUHPLC-MS/MStracesfortheanalytes

andinternalstandards.Theabsenceofcrosstalkwasverifiedby

injecting analytesand internal standards singly The selectivity

of themethodwasestablishedthrough testing 263urine

sam-ples from different sources coming from five different species

(bovine,canine,equineand murineanimalsaswellashumans)

withoutobservedinterferences.Carry-overwasassessedduring

thevalidation study byinjecting blanksolvent (MeOH)

follow-ingthesamplefortifiedattheconcentrationequaltofivetimes

thescreeningtargetconcentration(5×Cval)anditwasalso

moni-toredduringaroutineanalysisbyinjectingblanksolvent(MeOH)

followingthesamplefortifiedatthescreeningtarget

concentra-tion(screenpositivecontrol).Noanalytesignalappearedinblank

solvent(MeOH) Matrix effects evaluation(Table4)highlighted

bothsuppressionandenhancementeffectsinfivematrices,namely

equine,bovine,canine,humanandmurinespecies,respectively

ThegreatestamountofsuppressionwasobservedforBMS-564929

inequine (72%)and human(47%)urine, both BMS-564929and

RAD140inbovine(50%)andcanine(57%)urine,andRAD140in murine(81%)urinematrix,respectively.On theotherhand,the greatestamountofenhancementwasobservedforbicalutamidein equine(29%)andmurine(29%)urinematrix,respectively Alterna-tively,intheeventthatotherisotope-labelledanaloguesrelatedto SARMcompoundsofinterestaredevelopedand/orbecomemore affordable,theycanbeimplementedasinternalstandardsintothe methodtocompensateforsignallossresultingfrommatrixeffects

soastoimproveaccuracyandprecision

3.2.2 Detectioncapability(CCˇ) SincearecommendedconcentrationforSARMsinurinehasnot beenestablished[38,72],thescreeningtargetconcentrationwas basedontheiranabolicpropertiesandsetatlevelsofexogenous anabolicandrogenicsteroidsandotheranabolicagents[72,73] Val-idationwasperformedatthescreeningtargetconcentration(Cval) setat1ngmL−1excludingandarine(2ngmL−1)andBMS-564929 (5ngmL−1),respectively,andasingleMS/MStransitionwas suffi-cienttoensurethescreeningoftheanalyteaccordingtothecurrent legislation[69].However,thecut-offfactors(Fm)wereabove T-valuesfor atleasttwo transitionsforallSARMsofinterest.The determinedCC␤valueswerebeloworequalthevalidationlevels foratleasttwotransitionsforallanalytes(Table3,Table4and Sup-plementarydataTable1).ThesensitivityashighlightedinTable3 (andSupplementarydata-Table1)was≥95%foratleasttwo tran-sitionsforallSARMs.Moreover,thedeterminedionratioswere within±30%tolerancerangeforalltransitionsofinterest[74].To conclude,allSARMsofinterestcanbedetectedinequineurineby applyingthisscreeningassaywithariskofafalse-negativerate≤5%

asrequiredbythecurrentlegislation[68,69].Inaccordancewith theEUCommissionDecision2002/657/EC,precisionexpressedas

CV,inthecaseofaquantitativemethod,shouldbeaslowas pos-sible(analyteconcentrationbelow100ngmL−1).Theprecisionof thecurrentscreening assaywasobserved tobein therangeof 9.8–30.9%inequineurine(Table3 whereasinthecaseofallother specieswasfoundtorangefrom6.4to48.2%(Supplementarydata –Table2)

Relativecut-off factor(RFm)wascalculatedfor eachanalyte (Table3)(andSupplementarydata–Table1)asthepercentage basedontheratioofthecut-offfactorandthemeanresponseof fortifiedsamples,andwasappliedtoscreenpositivecontrols(QC samples)duringroutineapplicationofthisscreeningtest

Table 3

Validation results for fortified equine urine samples (n = 61).

a Values calculated response-based.

b Estimated LOD (S/N≥3).

c Screening target concentration.

d Calculated as the percentage based on the ratio of the cut-off factor and the mean response of fortified samples.

e Calculated as coefficient of variation (CV) of the response following fortification.

f Expressed as percentage based on the ratio of samples detected as positive in true positive samples, following fortification.

Table 4

Recovery and matrix effect data.

Bicalutamide 94 11.5 −28.9 ± 27.8 −9.5 ± 12.5 −1.4 ± 8.1 −2.0 ± 11.0 −28.9 ± 18.7

PF-06260414 87 10.1 52 ± 13.6 27.4 ± 9.4 30.4 ± 10.1 30.0 ± 4.8 51 ± 12.0

a Recovery was determined by comparing results from fortified samples to those of negative samples spiked post-extraction at the screening target concentration (C val ,

n = 2) Recovery is based on data collected from routine application of the method in five species of interest over 15 month period (n = 25 analytical runs).

b Ion suppression results for urine matrices are based on the analysis of 25 samples (n = 5 per species) from different sources Values calculated as described in Section 2.5 Negative values indicate matrix enhancement.

Resultsfromon-goingQCsamples(negativecontrols(pooled

blankurine)andscreenpositivecontrolsfortifiedatthescreening

target concentration) are being recorded continuously and the

datautilisedtoverifythatthescreeningassayperformsreliably

androbust

3.2.3 Extensionofvalidation:applicationtobovine,canine,

humanandraturine

Followinginitialvalidation ofthedevelopedassayin equine

urine,anextensionofvalidationwasperformedonthesamematrix

typefromfourdifferentspecies-bovine,canine,humanandmurine

urine,respectively.Thevalidationstudywascarriedoutontwo

consecutivedaysonaseriesof20blankurinesamples(n=5per

species)andthesame20blankurinesamples(n=5perspecies)

fortifiedatthescreeningtargetconcentration(Cval),thesameasfor

equineurine,1ngmL−1excludingandarine(2ngmL−1)and

BMS-564929(5ngmL−1),respectively.Thesensitivityashighlightedin

Supplementarydata(Tables2–3),was≥95%foratleasttwo

tran-sitionsfor allSARMs in all matrices of interest and there was

maximumoneresultbelowthecut-offfactorestablishedinitially

forequineurine.Insuchacaseitcanbeconcludedthatthe

devel-oped screening assay is applicable tothe new species, namely bovine,canine,humanandmurineurine,withthesamedetection capability(CC␤)valuesforalltargetanalytesastheoriginalmatrix Theruggednessstudyofthedevelopedassayresultedin cor-rect classificationof all tested samples.In detail, respective 15 blanksamples(n=5perspecies)wereall“screennegative”whereas thecorresponding fortified oneswere all “screenpositive”(i.e exceededthecut-offfactor)

3.2.4 ApplicationofmethodtoanalysisofurinefromSARM exposedanimals

Due to the unavailability of a suitable proficiency test, an inter-laboratory studywasperformed in conjunctionwith RIK-ILT(Wageningen,theNetherlands).Threebovineurinesamples providedbyRIKILT,collectedwithintheframeofanostarine (S-22)administrationstudyinasteercalf[19],weretestedblindly Allsampleswereidentifiedcorrectlyasfollows:onesamplewas screenednegative (collectedbeforethetreatment) andanother twowerescreenedpositive(collected2hand3days,respectively, followinganoraladministrationofostarine)-Fig.4