From target analysis to suspect and non-target screening of endocrine-disrupting compounds in human urine

Vol.:(0123456789)
1 3
h ps://doi.o g/10.1007/s00216-022-04250-w
RESEARCH PAPER
F om a ge analysis osuspec andnon‑ a ge sc eening
o endoc ine‑dis up ing compounds inhuman u ine
MikelMusa adi1,2 · ClaudiaCaballe o1· Lei eMijangos1,2· Aile eP ie o1,2· Mai aneOli a es1,2· Ola zZuloaga1,2
Recei ed: 30 May 2022 / Re ised: 14 July 2022 / Accep ed: 22 July 2022
© The Au ho (s) 2022
Abs ac
In he p esen wo k, a a ge analysis me hod o simul aneously de e mining 24 di e se endoc ine-dis up ing compounds
(EDCs) in u ine (benzophenones, bisphenols, pa abens, ph hala es and an ibac e ials) was de eloped. The a ge analysis
app oach (including enzyma ic hyd olysis, clean-up by solid-phase ex ac ion and analysis by liquid ch oma og aphy coupled
o andem mass spec ome y (LC–MS/MS)) was op imized, alida ed and applied o olun ee s’ samples, in which 67%
o he a ge EDCs we e quan i ied. Fo ins ance, benzophenone-3 (0.2–13ng g−1), bisphenol A (7.7–13.7ng g−1), me hyl
3,5-dihyd oxybenzoa e (8–254ng g−1), mono bu yl ph hala e (2–17ng g−1) and iclosan (0.3–9ng g−1) we e ound a he
highes concen a ions, bu he p esence o o he analogues was de ec ed as well. The de eloped a ge me hod was u he
ex ended o suspec and non- a ge sc eening (SNTS) by means o LC coupled o high- esolu ion MS/MS. Fi s , well-de ined
wo k lows o SNTS we e alida ed by applying he p e iously de eloped me hod o an ex ended lis o compounds (83),
and hen, o he same eal u ine samples. F om a lis o app oxima ely 4000 suspec s, 33 we e anno a ed a le els om 1 o
3, wi h ood addi i es/ing edien s and pe sonal ca e p oduc s being he mos abundan ones. In he non- a ge app oach, he
sea ch was limi ed o molecules con aining S, Cl and/o B a oms, anno a ing 4 pha maceu icals. The esul s om his s udy
showed ha he combina ion o he lowe limi s o de ec ion o MS/MS and he iden i ica ion powe o high- esolu ion MS/
MS is s ill compulso y o a mo e accu a e de ini ion o human exposome in u ine samples.
Keywo ds Endoc ine-dis up ing compounds· Ta ge analysis· Liquid ch oma og aphy andem mass spec ome y·
Suspec and non- a ge sc eening· High- esolu ion andem mass spec ome y
In oduc ion
The numbe o syn he ic o ganic compounds p oduced and
used nowadays is o e whelming [1]. Despi e hei u ili y,
some o hem pose a se ious h ea no only o he en i on-
men , bu also o humans. In ac , a ound 90% o ch onic
human diseases can be linked o en i onmen al ac o s,
unlike he modes 10% ha can be explained h ough gene -
ics [2, 3]. F om ha en i onmen al con ex , he concep o
exposome eme ged in 2005 [4]. Nowadays, he exposome
engages all kind o exposu es and ac o s ha h ea en
humans h oughou all ou li espans [5], and he ques o
known and unknown chemical compounds is a key ac o
o i s decoding.
In ha amewo k, endoc ine-dis up ing compounds
(EDCs) ha e been in he spo ligh in he las decade since
hey can in e e e wi h he endoc ine sys em leading o, o
ins ance, mu agenic, ca cinogenic o hepa o oxic e ec s [6].
Some examples conce ning EDCs include (i) bisphenols
[7–9], (ii) benzophenones [10, 11], (iii) pa abens [12–14],
(i ) ph hala es [15, 16] and ( ) o ganochlo ides such as i-
closan (TCS) and icloca ban (TCC) [17–19].
Biomoni o ing o hose xenobio ics (i.e., compounds
ha do no occu na u ally in he human o ganism [20])
as well as hei espec i e phase I and phase II me aboli es
[21, 22] is o en pe o med using u ine samples, as i is
shown in se e al epidemiological and clinic s udies [8,
17, 23]. In o de o quan i y hem, mos a ge me hodolo-
gies equi e a deconjuga ion eac ion o ans o m phase
* Mikel Musa adi
[email p o ec ed]
1 Depa men o Analy ical Chemis y, Uni e si y
o  heBasque Coun y (UPV/EHU), Leioa,
BasqueCoun y48940, Spain
2 Resea ch Cen e o Expe imen al Ma ine Biology
andBio echnology (PiE), Uni e si y o  heBasque Coun y
(UPV/EHU), BasqueCoun y, Plen zia48620, Spain
/ Published online: 29 July 2022
Analy ical and Bioanaly ical Chemis y (2022) 414:6855–6869
1 3
II me aboli es, mos ly glucu onides and sulpha e conju-
ga es, in o he unconjuga ed o m [24, 25]. Al hough di ec
quan i ica ion o ee and conjuga ed compounds could
imp o e he in e p e a ion o esul s in clinical esea ch,
a limi ed numbe o conjuga e s anda ds is a ailable [26].
A e wa ds, he unconjuga ed me aboli es a e ex ac ed
om he u ine by a clean-up s ep, which is also used o
emo e in e e ences p esen in he ma ix (endogenous
compounds) and, he e o e, o a oid signal supp ession/
enhancemen (ma ix e ec s) du ing hei analy ical de e -
mina ion. Solid-phase ex ac ion (SPE) is he p e e ed
echnique when dealing wi h se e al EDC amilies [15, 25,
27], specially using e e se-phase (RP) polyme ic so ben s
[28] o mixed-mode SPE ca idges [29].
As o he analysis, liquid ch oma og aphy (LC) cou-
pled o andem mass spec ome y (MS/MS) is he mos
used echnique o analyse o ganic compounds in u ine
[30, 31]. Low- esolu ion LC–MS/MS is used o a ge
analysis and quan i ica ion o EDCs due o i s low de ec-
ion limi s and obus pe o mance [14, 32]. Wi hin ha
a ge ed analysis con ex , mos s udies in he li e a u e a e
ocused on a educed numbe o EDC amilies wi h simi-
la chemical p ope ies, ins ead o de eloping me hods o
simul aneously de e mine a wide a ie y o compounds
wi h di e en chemical beha iou s.
Nowadays, he eme gence o high- esolu ion mass
spec ome y (HRMS) has opened up new oppo uni ies
o look o xenobio ics o me aboli es ha a e a ely ol-
lowed in a ge analysis me hods [33]. Tha way, suspec
o non- a ge sc eening (SNTS) me hods a e p og essi ely
de eloping [34]. Howe e , he men ioned app oaches a e
majo challenges in analy ical chemis y, since he abun-
dan and complex da a ob ained om high- esolu ion
andem mass spec ome y (HRMS/MS) makes he elu-
cida ion o unknowns a ough ask [35, 36]. In ha sense,
ex ensi e LC-HRMS/MS lib a ies a e equi ed [37, 38],
as well as unambiguous c i e ia ega ding quali y con ol
and quali y assu ance (QC/QA) [39] and iden i ica ion
con idence [40]. Mo eo e , sample p epa a ion should
achie e a balance be ween selec i i y, by p ese ing as
many compounds as possible, and sensi i i y, by limi ing
ma ix in e e ences [39].
Despi e he men ioned di icul ies, STNS me hods a e
especially in e es ing o ga he in o ma ion o unde s and-
ing and decoding he human exposome by inding ele an
bioma ke s. The e o e, he objec i es o he p esen wo k
ha e been, on he one hand, (i) o op imize and alida e
a a ge analysis me hod o simul aneously de e mine
se e al EDC amilies (5 benzophenones, 6 bisphenols, 6
pa abens, 5 ph hala es and 2 an ibac e ial) in human u ine
by LC–MS/MS and, on he o he hand, (ii) o ex end he
me hod o SNTS by LC-HRMS/MS.
Ma e ials andme hods
Reagen s andsolu ions
In he a ge analysis me hod, 24 EDCs consis ing o 6 bis-
phenols, 5 benzophenones, 6 pa abens, 5 ph hala e phase
I me aboli es and 2 an ibac e ials we e included based on
he li e a u e, as well as 4 iso opically labelled s anda ds.
Mo eo e , 59 addi ional compounds we e in oduced in
he expe imen s pe o med o ex end he a ge analysis
me hod o SNTS. These compounds consis ed o expo-
some bioma ke s, such as pha maceu icals, indus ial
chemicals, pe luo ina ed alkyl subs ances (PFAS) and
biocides con aining ei he Cl, B o S in hei s uc u e.
All he in o ma ion conce ning he analy es and su oga es
is compiled in TableS1 in he Supplemen a y in o ma ion
(SI). Mo eo e , he model compounds used in he Re en-
ion Time Indices Pla o m (RTI, h p:// i. chem. uoa. g /)
a e also p esen ed in TableS1 in he SI.
As he e a e no eal u ine samples o be used as blanks,
a syn he ic u ine was used o p epa ing blank and qual-
i y con ol (QC) samples o he op imiza ion and alida-
ion o he me hods (see Sec .2.1 in SI) [41]. In o ma ion
abou he es o he eagen s and solu ions used can also
be ound in Sec .2.1 o SI.
De elopmen o  he a ge me hod
byUHPLC‑ESI‑QqQ
Sample ea men
Se e al a iables a ec ing enzyma ic hyd olysis and
clean-up we e e alua ed by spiking syn he ic u ine wi h
he a ge analy es in o de o ge 100ng g−1 in he inal
ex ac . All he esul s we e s a is ically e alua ed using an
analysis o a iance (ANOVA) a a 95% con idence le el.
Rega ding enzyma ic hyd olysis, β-glucu onidase
enzyme uni s (400 and 4000, by adding 20 and 200 µL,
espec i ely) and hyd olysis ime (2 and 12h) we e s ud-
ied o op imize deconjuga ion o glucu onides p esen
in u ine samples. To ha end, 1mL o syn he ic u ine
was hawed o oom empe a u e and spiked a 20ng g−1
wi h bisphenol A glucu onide (BPA-G) alongside he es
o he xenobio ics excep ee bisphenol A (BPA). Two
hund ed mic oli e s o ammonium ace a e (NH4OAc,
1M, pH 5.0) was added o ensu e he p ope media o
β-glucu onidase ac i i y, ollowed by he co esponding
olume o β-glucu onidase solu ion. The deconjuga ion
eac ion was pe o med a 37°C o op imum enzyma ic
ac i i y. All expe imen s we e ca ied ou using h ee ep-
lica es (n = 3), and in all assays, he eac ion was s opped
Musa adi M.e al.
6856
1 3
by adding 2mL o phospha e bu e (0.1M, pH 2.0). The
deconjuga ion e iciency was s udied by ollowing he pe -
cen age o BPA-G le a e he hyd olysis, as well as he
con e sion o ee BPA.
Fo he clean-up, h ee ypes o SPE ca idges we e
es ed acco ding o he li e a u e [15, 17, 18, 25, 29]:
(i) RP SPE, (ii) mixed-mode SPE combining RP wi h
anion exchange and (iii) mixed-mode SPE combining
RP wi h ca ion exchange. Fo RP SPE, polyme ic-based
Oasis HLB (6mL, 200mg, 30µm, Wa e s, Mil o d, MA,
USA) ca idges we e used, while he ollowing elu-
ion sol en s we e e alua ed: (i) ace oni ile (AcN), (ii)
e hyl ace a e (E OAc), (iii) me hanol (MeOH) and se -
e al MeOH combina ions such as (i ) MeOH:ace one, ( )
MeOH:dichlo ome hane (DCM) and ( i) MeOH:E OAc
(all in 50:50, / ). Mo eo e , 4 aliquo s om 3 o
12mL we e eco e ed o s udy he elu ion p o ile using
MeOH:ace one. Unde op imal condi ions, he ca idges
we e ac i a ed and equilib a ed wi h 5mL MeOH:ace one,
5mL Milli-Q wa e and 5mL phospha e bu e (0.1M,
pH 2.0). Then, he u ine samples we e loaded on o he
ca idges and subsequen ly washed wi h 2mL o mic
acid/ o ma e (HCOOH/HCOO−) bu e (1M, pH 2.0)
and 5mL Milli-Q wa e . Finally, he ca idges we e ully
d ied unde acuum and he analy es we e elu ed wi h
3mL MeOH:ace one (50:50, / ).
Besides, Oasis MAX (6mL, 150mg, 30µm, Wa e s)
ca idges we e selec ed o mixed-mode SPE combining RP
and s ong anion exchange, while TELOS neo PCX (3mL,
100mg, 50µm, Kinesis) ca idges we e used o mixed-
mode SPE combining s ong ca ion exchange wi h RP. A
simila ex ac ion p ocedu e explained he eina e was es ed
o bo h cases. Fi s , he ca idges we e ac i a ed and equili-
b a ed wi h 5mL o he ollowing sol en s: (i) E OAc, (ii)
MeOH, (iii) Milli-Q wa e and (i ) phospha e bu e (0.1M,
pH 2.0). A e he u ine samples we e loaded, he ca idges
we e cleaned-up using 10mL Milli-Q wa e and ully d ied
a acuum. Las ly, he analy es we e elu ed wi h 12mL
MeOH ollowed by 12mL E OAc. In he case o he ca i-
onic exchange , an addi ional 5mL o MeOH con aining 1%
ammonia (NH4OH in solu ion) was used. The elu ion sol en
na u e and olume we e no s udied in he mixed-mode SPE.
A e elu ion, 40 µL o dime hyl sul oxide (DMSO) was
added o he elua es as e apo a ion keepe and hey we e
subsequen ly e apo a ed o 40 µL wi h a gen le s eam o
N2 a 35°C using he Tu bo ap LC E apo a o (Zyma k,
Bio age, Uppsala, Sweden). Las ly, he ex ac s we e dilu ed
o 200 µL wi h HPLC wa e (H2O), il e ed h ough polyp o-
pylene il e s (0.22µm, Phenomenex, To ance, CA, USA)
and kep in he eeze a –20°C in ch oma og aphy ials
un il analysis. Reco e y and ma ix e ec we e conside ed
as quan i a i e c i e ia o compa e he e ec i eness o he
es ed ca idges (see Sec .2.2.1 in SI o u he de ails).
UHPLC‑ESI–MS/MS analysis
The sepa a ion and de ec ion o he a ge analy es we e
ca ied ou by a UHPLC sys em (Agilen 1290 In ini y
II) coupled o a iple quad upole (QqQ) mass analyse
(Agilen Technologies 6430 T iple Quad), equipped wi h
a bina y pump, a degasi ying sys em, an au oma ic injec o
and an elec osp ay ioniza ion (ESI) in e ace. The sepa a-
ion o he analy es was pe o med using an ACE Ul a-
Co e 2.5 Supe C18 (2.1mm × 100mm, 2.5µm, A an o ,
Sym a, Mad id, Spain) ch oma og aphic column ha has
a wo king pH ange o 1.5–11, equipped wi h an Ul a-
Co e Supe C18 UHPLC gua d p ecolumn placed in an
ACE UHPLC gua d holde (bo h pu chased om A an o ,
Sym a). The column empe a u e was main ained a 35°C
and 7 µL was selec ed as injec ion olume.
Fo he mobile phases, UHPLC wa e (A line) and
MeOH (B line) we e used a pH 2.5 (0.1% HCOOH) and
10.5 (0.05% NH4OH) o ensu e ioniza ion o all com-
pounds. The low a e was se o 0.3mL min−1, and i was
con inuously in-line il e ed h ough an ACE Ul aCo e 5
Supe C18 (2.1mm × 30mm, 5µm, A an o , Sym a) col-
umn placed be o e he injec o in o de o educe in e e -
ing compounds coming om he LC equipmen . Fu he
de ails on LC-QqQ analysis a e included in Sec .2.2.2 in
SI.
Ta ge me hod alida ion
The a ge analysis me hod by LC-QqQ was alida ed
a h ee concen a ion le els (3ng g−1, 6ng g−1 and
30ng g−1 in u ine) in h ee consecu i e days using 5 epli-
ca es o spiked syn he ic u ine samples (QC samples) [42].
Among he igu es o me i o he alida ion (QC c i e ia),
absolu e and appa en eco e ies ( ueness), epea abil-
i y (in a-day p ecision), ep oducibili y (in e -day p eci-
sion), ins umen al and p ocedu al limi s o quan i ica ion
(iLOQs and pLOQs, espec i ely) and pa ame e s ela ed
o calib a ion cu es’ linea i y (uppe limi s and de e mi-
na ion coe icien s ( 2)) we e de e mined. The de ini ion o
he pa ame e s is included in Sec .2.2.3 in SI.
Rega ding he quali y assu ance (QA) o he analy ical
sequence, all samples we e andomly injec ed, while clean
MeOH was in oduced e e y 5 samples o check o possi-
ble ca yo e . 50 ng g-1calib a ion poin was also injec ed
h oughou he sequence e e y 10 samples as ins umen-
al QC sample o s udy signal in ensi y d i s and e en-
ion ime (RT) shi s. Fo analy e quan i ica ion c i e ia,
bo h m/z ansi ions should be p esen in he sample wi h
an e o o 30% in he a io o hei abundances and RT
should be wi hin ± 0.1min o he pu e s anda d.
F om a ge analysis osuspec andnon‑ a ge sc eening o endoc ine‑dis up ing compounds… 6857
1 3
Suspec andnon‑ a ge sc eening byHRMS
Al hough he sample ea men me hod was de eloped o
he a ge analysis o 24 EDCs, he powe o HRMS was
es ed o see whe he he de eloped p ocedu e could be wid-
ened o SNTS o di e en classes o xenobio ics in u ine
samples. Globally, 83 compounds we e s udied, 24 o which
had been included du ing a ge me hod de elopmen . Ana-
ly es chosen o he SNTS also included a wide ange o
pola i ies wi h log D alues a acidic pHs (loading alue a
he SPE p o ocol) anging om − 0.8 o 6.7 (see TableS1
in SI).
UHPLC‑HESI‑HRMS/MS analysis
A Dionex Ul ima e 3000 UHPLC (The mo Fishe Scien-
i ic, MA, USA) coupled o a high-pe o mance Q Exac i e
Focus O bi ap (qO bi ap, The mo Fishe Scien i ic) mass
analyse wi h a hea ed elec osp ay ioniza ion sou ce (HESI,
The mo Fishe Scien i ic) was used o he analysis o he
EDCs. Same op imized ch oma og aphic condi ions (p ecol-
umn, column, mobile phases, empe a u e, injec ion olume,
g adien and low) men ioned in “UHPLC-ESI–MS/MS
analysis” o LC-QqQ we e used o LC-qO bi ap as well.
The only di e ence was he absence o he column used o
il e ing he mobile phase due o s e ic incon eniences in he
loop. Rega ding qO bi ap ope a ing condi ions, hey we e
se acco ding o he expe ience o he esea ch g oup wi hou
u he op imisa ion [43] and a e de ailed in Sec .2.3.1 in SI.
Da a p ocessing
Th ee di e en da a- ea men app oaches ( a ge , suspec
and non- a ge analyses) we e used o ea he collec ed da a
by LC-qO bi ap. The a ge analysis app oach was used o
calcula e he iLOQs a pH 2.5 and 10.5. To ha end, T ace-
Finde 5.1 (The mo Fishe Scien i ic) so wa e was used,
which con ained he RT, he exac mass and he cha ac e is-
ic agmen ions o he selec ed compounds (see TableS3
in he SI). A 0.1-min window was allowed in he RTs, a
70% i ing in he iso opic pa e ns and a 5-ppm e o in he
monoiso opic mass (MS1) and he mos cha ac e is ic ag-
men s (MS2). Since lowe iLOQs we e achie ed wi h LC-
QqQ, he concen a ion o he EDCs in he olun ee s’ u ine
samples was no de e mined by LC-qO bi ap, and hus, i
was only used o SNTS.
Compound Disco e e 3.2 (The mo Fishe Scien i ic)
was used o SNTS. Wi h ega d o he peak picking c i-
e ia, only ea u es wi h a minimum peak a ea o 106 we e
conside ed when he RSD o he eplica es (n = 5) o each
sample was lowe han 30% and he a io wi h espec o
syn he ic u ine blanks was highe han 10. Speci ic se ings
and pa ame e s o he so wa e ega ding he peak selec ion
ha e been al eady es ablished and desc ibed by he esea ch
g oup [44]. Mo eo e , only ea u es wi h a Lo en zian peak
shape [45] we e manually selec ed o u he anno a ion.
In he case o suspec sc eening (Fig.1), h ee suspec
lis s ob ained om he No man ne wo k we e used: (i)
EUCosme ics [46], which is a combined in en o y o ing e-
dien s employed in cosme ic p oduc s (SCCNFP/0389/00
Final) and e ised in en o y (Decision 2006/257/EC) wi h
a o al o 3334 suspec s, (ii) a collec ion o 52 bisphenols
[47] a ailable a NILU om Table3 o epo 5/17 by KEMI
(Swedish Chemicals Agency) and (iii) 440 exposome bio-
ma ke s [48] om Exposome-Explo e , which is a da abase
dedica ed o bioma ke s o exposu e o en i onmen al isk
ac o s o diseases. The e o e, he inal suspec lis con-
ained a o al o 3826 suspec s, including he molecula o -
mula, exac mass and he s uc u e o each suspec . Besides,
a lis con aining 1311 endogenous u ine me aboli es was
ob ained om he Human Me abolome Da abase (HMDB)
o a oid alse iden i ica ion o endogenous compounds as
xenobio ics [49]. A de ailed desc ip ion o he peak anno a-
ion wo k low is included in Sec .2.3.2 in SI.
Besides suspec sc eening, non- a ge sc eening was ca -
ied ou limi ing i o molecules con aining Cl, B and/o S
due o he speci ic iso opic p o iles o molecules con aining
hose a oms [34]. When acco ding o he iso opic pa e n
(SFi > 50 and pa e n co e age > 90) a ea u e con ained a
leas one a om o Cl, B and/o S, candida es in he Chem-
Spide sea ched by Compound Disco e e 3.2 we e consid-
e ed. When a ailable, mzCloud spec a we e examined and,
when no , in silico agmen a ion was pe o med. Finally,
RT was conside ed in he same e ms as in he suspec
sc eening wo k low, as well as he anno a ion con idence
le els (Fig.1).
Quali y con ol/quali y assu ance
E en hough he concep o me hod alida ion is only used
o a ge analysis me hods, some QC/QA measu es can be
implemen ed o STNS as well [39]. A 10-poin ex e nal
calib a ion was p epa ed in he 0.5–200-ng g−1 ange in
H2O:DMSO (80:20, / ) o calcula ing ins umen al limi s
o iden i ica ion (iLOIs) o he 83 compounds and, he e-
o e, ins umen al alse nega i es. In he case o he iLOIs,
hey we e se as he lowes concen a ion le el ha could be
unequi ocally iden i ied [50], meaning a mass e o lowe
han 5ppm, an iso opic p o ile and agmen a ion spec a i
o a leas 70% and a ± 0.1min e o in he RT.
Fu he mo e, 1-mL syn he ic u ine samples (n = 5) we e
o i ied wi h he 83 compounds a 6ng g−1 (QC samples),
p ocessed ollowing he SPE me hod by Oasis HLB and ana-
lysed by UHPLC-qO bi ap. Then, SNTS wo k lows (see
“Da a p ocessing”) we e used o iden i ying he analy es
and e alua e losses. P ocedu al blanks (n = 5) we e also
Musa adi M.e al.
6858
1 3
p ocessed o a oid iden i ica ion o compounds coming om
elsewhe e. These addi ional expe imen s we e only made
wi h sc eening and anno a ion pu poses, while quan i a-
i e analysis was ou o he scope. Rega ding he analy i-
cal sequence, he same QA c i e ia as in he a ge me hod
alida ion we e ollowed.
Analysis o  eal u ine samples
To es he applicabili y o he de eloped me hods, eal
u ine samples p o ided by 4 olun ee s om he esea ch
g oup we e analysed, while he ob en ion o en i onmen al
o epidemiological esul s was ou o he scope o his wo k.
Those samples we e manipula ed acco ding o he indica-
ions o he E hics Commission o Resea ch and Teaching
o he Uni e si y o he Basque Coun y (CEISH-UPV/EHU,
BOPV 32, 17/2/2014 M10 2021 124 and CEIAB-UPV/EHU,
BOPV 32, 14/2/14, M30 2021 158). Sample handling is u -
he explained in Sec .2.4 in SI.
Resul s anddiscussion
Op imiza ion o  he a ge me hod byLC‑QqQ
Enzyma ic hyd olysis
The pe cen age o BPA-G le in all expe imen s a e
hyd olysis was lowe han 5%, showing ha he enzyme
quan i a i ely deconjuga ed he glucu onide ega dless
o he eac ion ime (2 o 12h) and enzyme uni s (400 o
4000) used. Mo eo e , a e age eco e ies (n = 3) ob ained
o he es o he compounds we e s a is ically compa able
(p > 0.05 o a 95% con idence le el) showing no deg ada-
ion a 37°C independen ly o he eac ion ime (da a no
shown). Speci ically, he eco e y o BPA (all deconju-
ga ed om BPA-G) was 70%, he same as in he expe i-
men s spiking wi h BPA ins ead o BPA-G. The e o e, he
loss obse ed could be a ibu ed o he o he s eps in he
p ocedu e and no o he enzyma ic hyd olysis. Bea ing in
Fig. 1 Scheme o he wo k low used o he anno a ion o sc eened compounds in suspec sc eening
F om a ge analysis osuspec andnon‑ a ge sc eening o endoc ine‑dis up ing compounds… 6859

1 3
mind all o he men ioned, he eac ion was ca ied ou o e -
nigh (12h) du ing alida ion using 400 uni s o he enzyme
(20 µL) o p ac ical pu poses in o de o gua an ee me hod
h oughpu . Ne e heless, conside ing ha he hyd olysis
may di e o o he glucu onide me aboli es in u ine, u -
he in es iga ion is equi ed in ha aspec s udying o he
glucu onides [51]. Addi ionally, he hyd olysis o sulpha e
conjuga es using β-glucu onidase/a ylsul a ase enzymes
should be add essed in u u e wo ks as well.
Op imiza ion o  heclean‑up s ep
As p e iously men ioned, h ee SPE app oaches we e es ed
du ing he clean-up s ep: (i) a RP app oach using Oasis-HLB
ca idges, (ii) a mixed-mode anionic exchange app oach
using Oasis-MAX ca idges and (iii) a mixed-mode ca ionic
exchange using TELOS neo-PCX ca idges. Reco e ies
and ma ix e ec s ob ained wi h each sol en using Oasis
HLB ca idges a e p esen ed in Figu e S1 in he SI. As i
can be obse ed in Figu e S1a, he bes elu ion sol en s o
Oasis HLB in e ms o eco e ies consis ed in hose con ain-
ing MeOH, which could dis up he pola -pola in e ac ion
be ween he a ge compounds and he pola g oups in he
Oasis HLB ca idge by hyd ogen bonding [28]. In ac , ana-
ly es such as BP2, BPS, MDHB o TCC ha con ain –OH
o –NH g oups we e no eco e ed in he absence o MeOH.
Wi h he aim o inc emen ing he eco e ies o ana-
ly es wi h less pola i y, MeOH was combined wi h o he
mo e non-pola sol en s (DCM, ace one and E OAc).
MeOH:DCM mix u e ende ed SPE eco e ies highe han
100%, bu he ex ac s ob ained su e ed om a s onge
signal supp ession a de ec ion (Figu e S1b). Pu e MeOH,
MeOH:ace one and MeOH:E OAc p o ided be e esul s
ega ding ma ix e ec . Finally, he MeOH:ace one mix-
u e was chosen be ween he h ee sol en s since i ende ed
eco e ies close o 100% o less pola analy es such as BP3
o BPZ, while he ma ix e ec was compa able as i can be
seen in Figu e S1b.
E en unde op imum condi ions, mos a ge com-
pounds su e ed om signal supp ession (20–30%) using
Oasis HLB ca idges ha could be due o he high sal
con en o u ine despi e he clean-up s ep. The e o e,
mixed-mode ca ionic and anionic exchange s we e es ed
aiming o dec ease he concen a ion o sal s in he inal
ex ac . In his sense, he objec i e would be he e en ion
o ca ions (u ea, c ea inine, Na+) o anions (Cl−, PO4
3−)
p esen a high concen a ions in u ine using he ionic
exchange mechanism, while he a ge compounds would
be e ained h ough he RP mechanism. The eco e ies
and ma ix e ec s ob ained o spiked syn he ic u ine
a e summa ized in Fig.2. Since he consecu i e elu ion
wi h E OAc in he anionic exchange and wi h E OAc and
MeOH (1% NH4OH) in he ca ionic exchange did no
inc ease he eco e ies, ha da a is no shown. Mo eo e ,
esul s o Oasis HLB using MeOH:ace one 50:50 ( / )
as elu ion sol en a e also included in Fig.2 o a be e
compa ison o he s udied p o ocols.
As can be seen in Fig. 2a, TELOS neo PCX ca -
idges (RP + CX) ende ed he highes eco e ies o he
mixed mode, s a is ically equal (p alue > 0.05, 95% con-
idence le el) o hose ob ained wi h Oasis HLB using
MeOH:ace one as elu ion sol en . In e ms o ma ix e ec
(Fig.2b), mixed-mode app oaches showed a simila sig-
nal supp ession compa able o Oasis HLB. Since he use
o mixed-mode ca idges did no minimize signal sup-
p ession, Oasis HLB ca idges we e chosen conside ing
he ease o un he p ocedu e. Once he SPE ca idge and
sol en na u e we e ixed, an elu ion p o ile was pe o med
by collec ing ou olumes o MeOH:ace one (50:50, / )
om 3 o 12mL. Based on he esul s (Figu e S2 in he SI),
3mL was su icien o a quan i a i e elu ion o he a ge
compounds.
Fig. 2 Boxplo s o he a SPE eco e ies and b ma ix e ec s a de ec-
ion ob ained o he e e se phase (RP) and mixed mode using RP
and ca ionic exchange (CX) and RP and anionic exchange (AX) ca -
idges by LC-QqQ
Musa adi M.e al.
6860
1 3
Ta ge me hod alida ion: igu es o me i
Ins umen al limi s o quan i ica ion andcalib a ion anges
As i can be seen in Fig.3, lowe iLOQs we e ob ained wi h
he LC-QqQ han wi h LC-qO bi ap mass analyse , showing
highe sensi i i y o he DMRM mode han he Full MS-
ddMS2. In ac , mo e han 70% o he analy es p o ided
iLOQ alues lowe han 1ng g−1 in he DMRM ega dless o
he pH o he mobile phase used. The case o he pa abens is
he mos ema kable one in ha sense, whe e a 10- ime di -
e ence is app ecia ed compa ing bo h de ec o s. In LC-QqQ
in gene al, be e alues we e achie ed o he bisphenols in
he basic pH, while pa abens and ph hala es showed lowe
iLOQ in he acidic media. On he con a y, mos iLOQ al-
ues we e be ween 1 and 5ng g−1 o qO bi ap, being mo e
sensi i e basic pH, especially o bisphenols, whose iLOQs
we e no included in he acidic pH o qO bi ap. All iLOQs
a e collec ed in TableS4 in he SI.
All in all, he alues ob ained a e compa able o hose
achie ed in o he wo ks in he li e a u e, which ange
be ween 0.5 and 2ng g−1 [24, 52–54]. Howe e , and as
a endency in analy ical chemis y, in mos o he wo ks
he LOQs a e calcula ed using signal- o-noise (S/N) a ios.
Compa ed o he S/N c i e ia o calcula ing iLOQs, he c i-
e ia applied in he p esen wo k a e s ic e since hey ake
in o accoun no only ha he signal di e s om he blank
bu also he i ing wi h he calib a ion cu e and he p eci-
sion [55].
Besides, uppe limi s o calib a ion anges a e also
included in TableS4 o bo h LC-QqQ and LC-qO bi ap
a bo h pHs. Simila linea i y anges we e obse ed o bo h
de ec o s, showing in a sligh endency o cu e a con-
cen a ions abo e 150–200ng g−1 ha i ed o quad a ic
calib a ion cu es. Howe e , ha endency was less e iden
in he case o LC-QqQ, in which also be e de e mina ion
coe icien s ( 2) we e ob ained. Howe e , since hose high
concen a ions a e no expec ed in u ine, mos cu es we e
limi ed o 100–150ng g−1 in o de o ha e linea cu es and
a oid quad a ic i ing. Taking all in o accoun , LC-QqQ was
used o quan i a i e a ge analysis in eal samples, while
LC-qO bi ap was limi ed o SNTS.
T ueness, p ecision andp ocedu al limi s o quan i ica ion
The indi idual absolu e and appa en eco e ies ob ained a e
included in Tables S5 in SI, o each spiking le el (3, 6 and
30ng·g−1) a he h ee consecu i e days, wi h hei espec-
i e RSDs. The su oga e used o co ec ion o each analy e
is also included a each concen a ion le el. In he cases
whe e an analy e was ionized a bo h pHs, he one p o iding
be e iLOQs was conside ed (see TableS4).
Th ee compounds (BPP, TCC and TCS) could no be
quan i ied a he lowes spiking le el (3ng g−1). Fo he
es o he compounds, absolu e eco e ies be ween 20
and 130% we e ob ained. A medium (6ng g−1) and high
(30ng g−1) spiking le els, sligh ly be e absolu e eco e ies
we e achie ed (30–110%). Acco ding o he ANOVA, he
absolu e eco e ies ob ained we e compa able wi hin he
di e en days a each spiking le el (Fc i ic = 3.15 > Fcalcula ed
= 0.13, Fc i ic = 3.13 > Fcalcula ed = 0.09 and Fc i ic = 3.13 >
Fcalcula ed = 0.31, o low, medium and high le els, espec-
i ely). Mo eo e , due o he p ope su oga e co ec ion,
he appa en eco e ies we e also s a is ically compa able
Fig. 3 Ins umen al limi s o
quan i ica ion ob ained a pH
2.5 and 10.5 o LC-QqQ and
LC-qO bi ap
F om a ge analysis osuspec andnon‑ a ge sc eening o endoc ine‑dis up ing compounds… 6861
1 3
wi hin he days (Fc i ic = 3.22 > Fcalcula ed = 0.85, Fc i ic = 3.1
9 > Fcalcula ed = 0.78 and Fc i ic = 3.21 > Fcalcula ed = 0.98) since
hey anged be ween 70 and 100%. Taking in o conside a-
ion he ANOVA esul s shown, he me hod ep oducibili y
(in e -day p ecision) h oughou he days could be con-
cluded. Tha ou come is c ucial in exposome biomoni o i-
za ion expe imen s, in which la ge sample se s need o be
analysed so he di ision o hem in se e al days is assu ed.
The e o e, he a e age absolu e and appa en eco e ies
ob ained o he 3days a e shown o each le el in Fig.4.
As can be seen in Fig.4, he absolu e and appa en eco e -
ies a e simila a he h ee concen a ion le els, which can be
u he con i med by ANOVA (Fc i ic = 3.14 > Fcalcula ed = 0.
18 and Fc i ic = 3.2 > Fcalcula ed = 0.2 o absolu e and appa en
eco e ies, espec i ely).
Rega ding he in a-day p ecision o he me hod ( epea -
abili y), he eco e ies o he di e en days in each le el
we e combined (n = 15) due o he men ioned me hod
ep oducibili y, and RSD alues o he absolu e eco e ies
we e no s a is ically compa able among spiking le els a
a 95% con idence le el acco ding o ANOVA (Fc i ic = 3.1
3 < Fcalcula ed = 8.45). Tha ou come could be explained by
he highe RSD alues ob ained a he low spiking le el,
al hough only MDHB exceeded he 35% limi . Fo he es
o he analy es, mos alues we e a ound 20–30% a he low
spiking le el, while lowe RSDs (a ound 10–15%) we e
ob ained a medium and high le els. Ne e heless, a e
su oga e co ec ion, he RSDs o he appa en eco e ies
we e compa able (Fc i ic = 3.21 > Fcalcula ed = 3.01) since mos
anged be ween 10 and 15%. The e o e, he me hod was con-
side ed epea able.
In he li e a u e, equi alen ueness and p ecision alues
ha e been ob ained o his ype o EDCs in u ine using
Oasis HLB. Ne e heless, mos wo ks ha e been limi ed
o one o ew amilies, such as ph hala e me aboli es [53],
ph hala es and BPA [56] o BPA and iclosan [57]. Despi e
su oga e co ec ion, i would be in e es ing o assess po en-
ial losses in he il a ion s ep, as well as he e alua ion o
he ma ix e ec a se e al concen a ion le els.
Las ly, he pLOQ alues (see TableS4 o indi idual al-
ues) ob ained we e in he same o de as he concen a ions
expec ed o de ec ed in o he wo ks in u ine [41, 52, 58, 59].
Rega ding he pLOQs o LC-QqQ, mos alues ob ained
we e below 0.5ng g−1, while highe alues we e ob ained
o LC-qO bi ap, anging a ound 1–2ng g−1.
Quali y con ol/quali y assu ance o suspec
andnon‑ a ge sc eening
Ins umen al limi s o iden i ica ion
Fi s , iLOIs we e es ima ed o he 83 analy es using he
wo k lows desc ibed in “Da a p ocessing,” and he esul s
ob ained a e included in TableS3 alongside he MS2 ma ch
using mzCloud o in silico agmen a ion. Due o he s ic e
equi emen s, iLOI alues u ned ou o be highe ha hei
espec i e iLOQ o he a ge analy es, since 90% o he
analy es we e anging be ween 0.5 and 28.1ng g−1. Consid-
e ing all he 83 analy es, be e alues we e achie ed in he
acidic pH in gene al. In ac , he iLOIs a pH 2.5 we e mos ly
be ween 0.1 and 10.8ng g−1, while a pH 10.5, hey anged
be ween 1 and 22ng g−1, due o he poo e agmen a ion
obse ed in he nega i e ioniza ion mode ende ing a lowe
iden i ica ion powe . Rega ding he analy es ha p o ided
high iLOQs, MBuP, MEHHP, MEOHP, ca eine, ace ami-
nophen, MBnP and 2-hyd oxybenzo hiazole had alues
highe han 50ng g−1.
Mos xenobio ics we e sc eened using mzCloud, allow-
ing he calcula ion o hei espec i e iLOI. In he cases o
BnP, MBnP, MEOHP, MEHHP and MDHB, howe e , hey
we e posi i ely iden i ied using in silico agmen a ion. In
he cases o BP3 and BuP, he iden i ica ion was done by
ei he mzCloud o in silico ma ch depending on he ioniza-
ion mode. While BuP was iden i ied by mzCloud ma ch
in he nega i e mode, i was posi i ely anno a ed only by
in silico agmen a ion in he posi i e mode. The opposi e
occu ed o BP3 as i can be obse ed in Figu esS3a and
S3b in SI o posi i e and nega i e modes, espec i ely. This
could be explained because only agmen a ion spec a in he
posi i e and he nega i e mode, espec i ely, a e a ailable
in mzCloud. I could be he e o e highligh ed ha in silico
agmen a ion is a complemen a y ool o he anno a ion o
non- a ge s. Finally, MeP and MBnP could only be iden i ied
in one o he ioniza ion modes, nega i e mode o he i s
Fig. 4 Boxplo s o he a e age (n = 15) absolu e (Abs.) and appa -
en (App.) eco e ies ob ained o syn he ic u ine spiked a low
(3ng g−1), medium (6ng g−1) and high (30ng g−1) concen a ion le -
els
Musa adi M.e al.
6862
1 3
and posi i e o he la e , al hough hey could be quan i ied
in bo h ioniza ion modes in LC-QqQ.
Ne e heless, iLOIs could no be calcula ed o all com-
pounds due o di e en p oblems ha would lead o ins u-
men al alse nega i es. On he one hand, he agmen a-
ion spec a ob ained o some pu e s anda ds could no be
explained ei he using mzCloud lib a y o in silico agmen-
a ion. This was he case o TCS, BPZ and BPP. Rega d-
ing TCS, he spec a in mzCloud con ained he agmen s
160.95664Da and 141.98271Da, while we only ob ained
a single agmen a 91.62964Da (see Figu e S4 in SI). In
he cases o BPZ and BPP, hose wo compounds a e no
included in mzCloud and he in silico agmen a ion was
no able o explain he agmen s ob ained. On he o he
hand, some compounds we e missed due o he il e o
mzCloud ma ch (> 70%), as occu ed o co inine (42%),
chlo olu on (42%) benzo hiazole (41%), mecop op (36%)
and en hion (50%). I should be highligh ed ha excep o
co inine, benzo hiazole and en hion, he es o he com-
pounds men ioned we e ionized in he nega i e mode, which
usually ende s a poo e agmen a ion spec um as s a ed.
The e o e, in-house lib a ies should be implemen ed no o
miss such compounds.
In summa y, in he case o pu e s anda ds, only 9.6%
o he compounds (8 ou o 83) could be conside ed alse
nega i es using he es ablished c i e ia, while 12.5% o he
analy es included in he a ge me hod we e no sc eened (3
ou o 24). As o he calcula ion o iLOIs in human ma i-
ces, i is wo h men ioning ha he e is a huge gap in he
li e a u e in ha a ea [43], since LOIs ha e been mainly
calcula ed in en i onmen al ma ices. The absence o LOIs
in human ma ices con i ms he lack o ha moniza ion in
QC/QA measu emen s in STNS, as i has been p e iously
s a ed by o he expe s in he ield [39].
Sc eening o  a ge s inquali y con ol samples
A e he applica ion o he SNTS wo k lows o pu e s and-
a ds, spiked syn he ic u ine QC samples we e e alua ed o
assess po en ial losses o he SPE p o ocol and ma ix e ec .
In he case o BP8, al hough pu e s anda ds could be posi-
i ely iden i ied bo h in he posi i e and nega i e ioniza ion
modes, in spiked u ine samples, agmen a ion spec a we e
only ob ained in he nega i e mode. In addi ion, no ag-
men a ion spec a we e ob ained o MBnP, EDHB, 4-PBZ,
4,4′-DMA-BP, en hion, TCC, o yzalin, quinme ac, e hion
and PFOSA. Las ly, in he case o BPS, me ibuzin and imi-
daclop id, al hough a agmen a ion spec um was ob ained,
he iden i ica ion ma ch was lowe han 70% compa ed o he
mzCloud lib a y.
As a esul , 13 addi ional analy es could no be sa is ac-
o ily sc eened due o he lack o ex ac ion du ing SPE,
po en ial losses in he e apo a ion o il a ion s ep and/o
ma ix e ec du ing he de ec ion. The e o e, alse nega-
i es inc eased om a 9.6% in pu e s anda ds o a 22.9% in
syn he ic u ine. Conside ing ha 6 o 19 compounds ha
could no be iden i ied using SNTS app oaches had been
p e iously used du ing me hod de elopmen , i could be
highligh ed ha 78.0% o he compounds no conside ed
du ing me hod de elopmen we e sa is ac o ily iden i ied,
and consequen ly, ha he me hod de eloped by Oasis HLB
could be used o SNTS. Ne e heless, i should also be pin-
poin ed ha 6 o 24 a ge s (25.0%) used du ing me hod
de elopmen could no be iden i ied in a SNTS app oach
using LC-qO bi ap in he FullMS-ddMS2 acquisi ion mode
in syn he ic u ine, while hey could be quan i ied e en a
lowe concen a ion wi h a LC-QqQ low- esolu ion mass
analyse in he DMRM mode. The men ioned d awbacks
should be ca e ully assessed in u u e wo ks.
Applica ion o eal samples
Ta ge analysis
The alida ed a ge analysis me hod by LC-QqQ was
applied o he u ine samples o ou olun ee s named A, B,
C and D o es me hod iabili y. The a e age concen a ions
(n = 5, ng g−1 in we weigh ) and he con idence in e als
a 95% (de ined as 2s, whe e s co esponds o he s anda d
de ia ion) measu ed by LC-QqQ a e included in Table1.
Only hose compounds abo e he pLOQ wi h a RSD lowe
han 35% a e included. I should be s a ed ha c ea inine
co ec ion is discou aged o p o iding pollu an s’ concen-
a ions in u ine [60], and he e o e, i was no pe o med.
In gene al, he highes concen a ions o he a ge com-
pounds we e measu ed in sample A. Among benzophe-
nones, BP3 was ound a he highes concen a ions in all
samples ollowed by BP1 and BP2. The me aboli e o ben-
zophenone 4OH-BP was also de ec ed in all he samples.
In he case o bisphenols, al hough BPA was ound a he
highes concen a ion in samples A and B, BPS was quan i-
ied a lowe concen a ions in he ou samples, indica ing
he in oduc ion o al e na i es o BPA in he ma ke [68].
BPAF could also be quan i ied in samples A and C. Among
pa abens, he highes concen a ions we e ound o MDHB
(12–241ng g−1), whe eas EDHB and E P we e ound a
lowe concen a ions, showing he exposu e o chemicals
p esen in pe sonal ca e p oduc s [58]. Rega ding he ph ha-
la es, MBuP was he me aboli e ound by a a he highes
concen a ion (4–12ng g−1), while MBnP, MEHHP, MEHP
and MEOHP we e ound a a simila concen a ion le el,
showing he exposu e o plas icize s [65]. Finally, TCS was
ound a one o de o magni ude highe in sample A han
in he es o he samples, compa able o he es o EDCs
quan i ied.
F om a ge analysis osuspec andnon‑ a ge sc eening o endoc ine‑dis up ing compounds… 6863