Trace level determination of β-blockers in waste waters by highly selective molecularly imprinted polymers extraction followed by liquid chromatography–quadrupole-linear ion trap

Page 1

A

p(mrdaatf2wc©

K

1

ieiAa�s

p

0d

Available online at www.sciencedirect.com

Journal of Chromatography A, 1189 (2008) 374–384

Trace level determination of �-blockers in waste waters by highly selectivemolecularly imprinted polymers extraction followed by liquidchromatography–quadrupole-linear ion trap mass spectrometry

Meritxell Gros a, Tania-Mara Pizzolato b, Mira Petrovic a,c,∗,Maria Jose Lopez de Alda a, Damia Barcelo a

a Department of Environmental Chemistry, IIQAB-CSIC, c/Jordi Girona 18-26, 08034 Barcelona, Spainb Instituto de Quımica-Universidade Federal do Rio Grande do Sul, Av. Bento Goncalves, 9500 Porto Alegre, Brazilc Institucio Catalana de Recerca i Estudis Avancats (ICREA), Passeig Lluis Companys 23, 80010 Barcelona, Spain

Available online 23 October 2007

bstract

This paper describes the development of an analytical methodology to determine eight �-blockers in waste waters using molecularly imprintedolymers (MIPs) as extraction and pre-concentration material, followed by liquid chromatography–quadrupole-linear ion trap mass spectrometryLC–QqLIT MS). The advantages offered by MIPs, in terms of selectivity and specificity, were compared with the most commonly polymericaterials used (the lipophilic–hydrophilic balanced Oasis® HLB cartridges). Even though recoveries achieved with both sorbents were similar,

anging from 50 to 110% for sewage treatment plant (STP) effluent and 40–110 for STP influent, respectively, MIPs provided lower methodetection limits than Oasis® HLB, due to their specificity for target analytes and closely related analogues. Method detection limits (MDL)chieved using MIPs ranged from 0.2 to 6.4 ng/L for STP effluent and from 0.4 to 6.5 ng/L for STP influent. To highlight the advantages of MIPsgainst conventional polymeric cartridges, a detailed matrix effects study as well as cross reactivity tests were performed. For the latter purpose,he extraction efficiency of some pharmaceuticals and pesticides belonging to different therapeutic classes was assessed. LC–QqLIT MS, used

or quantification and confirmation, proved to be a powerful analytical tool, as instrumental detection limits (IDL) achieved ranged from 0.2 to.7 pg injected (in multiple reaction monitoring mode (MRM)). In addition the inclusion of high sensitive MS/MS scans for each compound whenorking in Information Dependent Acquisition mode (IDA) provided extra confirmation for unequivocal identification of target compounds in

omplex environmental matrices. 2007 Elsevier B.V. All rights reserved.

nalysi

roStwcopc

eywords: Molecularly imprinted polymers (MIP); �-Blockers; Waste water a

. Introduction

Several investigations carried out in the last few years, bothn Europe and USA, have revealed the presence of a great vari-ty of drugs in the aquatic environment, which has become anssue of great concern among the scientific community [1,2]mong the most frequently detected pharmaceuticals, which

re also among the top prescribed medications worldwide, are-blockers, used in the treatment of cardiovascular disorders,

uch as hypertension, arrhythmia and heart failure [2–6].

Sewage treatment plants (STP) are major contributors to theirresence in the environment, since sewage waters gather the

∗ Corresponding author. Tel.: +34934006100; fax: +34 932045904.E-mail address: [email protected] (M. Petrovic).

m

ttcht

021-9673/$ – see front matter © 2007 Elsevier B.V. All rights reserved.oi:10.1016/j.chroma.2007.10.052

s; Liquid chromatography–quadrupole-linear ion trap (QqLIT)

esidues present in the excreta, those coming from disposalf unused or expired drugs and pharmaceutical discharges [7].ome of these compounds show low removal during sewage

reatment processes, and therefore, are able to reach surface asell as ground and drinking waters [8,9]. �-Blockers are of con-

ern due to the possible ecotoxicological effects they may havence released into the environment. Some investigations haveointed out that some �-blockers cause acute and chronic toxi-ity to several aquatic organisms at levels close to the maximaeasured in effluents [9].Although several analytical methodologies for the determina-

ion of �-blockers are currently available in the literature, most of

hem focus on their analysis in biological fluids for anti-dopingontrol [10–15] as these substances are included in the list of pro-ibited substances of the International Olympic Committee, dueo their sympathommimetic properties, similar to other central

Page 2

togr.

nbtrm(afmedosoioecdtio

o[lswrtctitloratwsb[npycaZoob[

tpa

oemdtdeaqatoqarts

sebaapohinirAa(fts

2

2

(mwTs

bmida

M. Gros et al. / J. Chroma

ervous system stimulants [16]. Nowadays, an important num-er of methodologies are directed to survey their occurrence inhe aquatic media, due to their significance as emerging envi-onmental contaminants [2,5,6,17,4,18–26]. In these existingethods, polymeric and reversed phase solid phase extraction

SPE) materials, mainly Oasis® hydrophilic–lipophylic bal-nced (HLB) and C18, respectively, are the preferred phasesor pre-concentration and extraction purposes. However, theseaterials, apart from the compounds of interest, are able to

xtract a wide spectrum of substances within a large range ofifferent pKa values [27,28]. Therefore, they are able to extractther organic components present in the sample, leading totrong matrix effects, resulting in suppression or enhancementf the analyte signal [27]. The selectivity of stationary phasess an important parameter to take into account when analysingrganic pollutants at trace levels from complex matrices, such asnvironmental waters, because reducing the level of co-extractedompounds results in better sensitivity and hence, lower limits ofetection. For these reasons, the use of more specific and selec-ive sorbents, such as molecularly imprinted polymers (MIPs),nstead of conventional ones, would allow the accomplishmentf these objectives.

Molecularly imprinting technology is used for the preparationf polymers having specific molecular-recognition properties29–32]. First, the template, which consists on the target ana-yte or a structure related compound, and the monomer form atable complex prior to polymerization. The complex is after-ards polymerized in the presence of a cross-linking agent. The

esulting MIPs are materials possessing microcavities with ahree-dimensional structure complementary in both shape andhemical functionality to that of the template [33,34]. Afterhis process, the template is removed, generating specific bind-ng sites. Then, the polymer can be used to selectively rebindhe template molecule, the analyte or structurally related ana-ogues [35]. The first use of a MIP for solid phase extractionf organic pollutants in water matrices was presented by Fer-er et al. [36] for the determination of the pesticides atrazinend simazine. Nowadays, their application in the environmen-al field as solid phase extraction materials (MIP-SPE) is moreidespread, due to their high potential for single group analy-

is. Several examples showing the good performance achievedy MIPs are presented for the determination of chlorotriazine36], triazine [37,38] and organophosphorous [35] pesticides inatural waters. Other examples dealing with ubiquitous organicollutants have been presented by Caro et al. [39] for the anal-sis of chloro and nitrophenols in water samples, Martin ando-workers [40,41] for the determination of �-agonists and �-ntagonists in plasma and bovine muscle, respectively, and byheng et al. [42] to detect sulphonamide type antibiotics. More-ver, they have been also widely applied for the determinationf pharmaceutical substances in biological fluids, such as clen-uterol [43,44], �-agonists [45]tetracycline and oxytetracycline46] and sulfamethazine [47] antibiotics.

�-Blockers are found at trace levels in natural and wastewa-ers and apart from sensitive and selective extraction materials,owerful instrumental techniques are required for their reli-ble determination. Generally, the method of choice consists

tsns

A 1189 (2008) 374–384 375

n liquid chromatography coupled to tandem mass spectrom-try (LC–MS/MS), where triple quadrupole (QqQ) MS is theost widely employed analyzer. This is attributed to their high

ynamic range and good performance when working in mul-iple reaction monitoring (MRM) mode [28]. However, recentevelopments in hybrid MS instruments have proved to be pow-rful tools to achieve high sensitivity, specificity and selectivitys they combine the main advantages of two analyzers (i.e.uadrupole and time of flight in case of QqTOF or quadrupolend liner ion trap in case of QqLIT). The main advantage ofhe hybrid QqLIT over other LC–tandem MS equipments relyn the fact that the third quadrupole (Q3) can work either as auadrupole or as an IT analyzer, being able to combine their char-cteristic scans in the same experiment and chromatographicun. Hence, a large amount of data for unequivocal identifica-ion and confirmation of target compounds are generated at highensitivity [48].

In this context, the aim of this work was the development of aensitive and selective analytical methodology to determine theight most frequently used �-blockers, using a MIP developedy MIP Technologies (Lund, Sweden) for the pre-concentrationnd extraction of target compounds from both river and wastew-ter samples. Target compounds and their physico-chemicalroperties are shown in Table 1. To prove the potentialf MIPs, a comparison with the lipophilic divinylbenzene-ydrophilic N-virrylpyrrolidone polymers Oasis® HLB, whichs one of the most widely employed materials for the determi-ation of �-blockers in environmental waters [4,5], was alsoncluded in this study. Moreover, matrix effects and cross-eactivity tests were carried out and evaluated in both materials.nalyte identification and confirmation was performed usingLC–QqLIT MS in compliance with the EU regulations

EU Commission Decision 2002/657/EC [49]), which was setor the determination of food residues but was extrapolatedo the analysis of pharmaceutical residues in environmentalamples.

. Experimental

.1. Chemicals and materials

All pharmaceutical standards used were of high purity grade>90%). Atenolol, sotalol hydrochloride, pindolol, timolol,etoprolol, carazolol, propranolol hydrochloride and betaxololere purchased from Sigma-Aldrich (Steinheim, Germany).he isotopically labelled compound atenolol-d7, used as internaltandard, was from CDN Isotopes (Quebec, Canada).

Individual stock standard solutions were prepared on a weightasis in methanol and stored at −20 ◦C. A mixture of all phar-aceutical standards was prepared by appropriate dilution of the

ndividual stock solutions. Further dilutions of this mixture wereone in methanol–water (25:75, v/v) before each analytical runnd were used as working standard solutions. Stock solutions of

he internal standard were also prepared in methanol and weretored at −20 ◦C. A mixture of these standards, used for inter-al standard calibration, was also done by diluting the individualtock solutions in methanol–water (25:75, v/v).

Page 3

376 M. Gros et al. / J. Chromatogr. A 1189 (2008) 374–384

Table 1Structure and physico-chemical properties of the �-blockers analyzed

Compounds Chemical structure Log Kow pKa MW Pv (mmHg)

Atenolol −0.03 9.6 266.34 2.9E−10

Sotalol 0.24 n.d. 272.37 5.3E−9

Pindolol 1.48 9.25 248.32 1.23E−8

Timolol 1.75 9.21 316.42 1.08E−9

Metoprolol 1.69 9.68 267.37 2.88E−7

Carazolol 2.66 n.d. 298.38 1.44E−10

Propranolol 0.74 n.d. 259.80 4.74E−14

Betaxolol 2.98 n.d. 307.43 1.33E−8

(

bTOvmMaNA

2

cwt

www.syrres.com/esc/physprop.htm).

Molecularly imprinted polymers used (MIP4SPETM �-lockers, 25 mg, 10 mL) were kindly supplied by MIPechnologies (Lund, Sweden) and the polymeric cartridgesasis® HLB (60 mg, 3 mL and 200 mg, 6 mL) were kindly pro-ided by Waters Corporation (Milford, MA, USA). HPLC-gradeethanol, acetonitrile and water (LiChrosolv) were supplied by

erck (Darmstadt, Germany). Hydrochloric acid 37%, NH4Ac

nd HAc were purchased from Merck (Darmstadt, Germany).itrogen (99.995% purity) used for drying of extracts was fromir Liquide (Spain).

wu0S

.2. Sample pre-treatment

The method was optimized using influent and effluentollected from a STP located in Rubı (Barcelona, Spain),hich receives urban, domestic and industrial wastewa-

ers. Amber glass bottles pre-rinsed with ultra-pure water

ere used for sample collection. Wastewaters were vac-um filtered through 1 �m glass fiber filters followed by.45 �m nylon membrane filters (Teknokroma, Barcelona,pain).

Page 4

M. Gros et al. / J. Chromatogr. A 1189 (2008) 374–384 377

Table 2MS/MS parameters for the analysis of target analytes by MRM positive ionization mode, using LC–QqLIT

Compounds Rt (min) Precursor ion DP–CE–CXP MRM1 DP–CE–CXP MRM2

Atenolol-d7 (IS) 3.40 274 [M + H]+ 60–20–10 274 > 190 –Atenolol 3.51 267 [M + H]+ 60–35–8 267 > 145 60–35–14 267 > 190Sotalol 3.89 273 [M + H]+ 60–25–8 273 > 213 60–25–6 273 > 255Pindolol 11.84 249 [M + H]+ 60–30–8 249 > 116 60–30–14 249 > 88Timolol 14.52 317 [M + H]+ 60–30–20 317 > 262 60–30–6 317 > 44Metoprolol 14.94 268 [M + H]+ 60–35–10 268 > 121 60–35–8 268 > 133Carazolol 17.42 299 [M + H]+ 60–35–8 299 > 116 60–35–2 299 > 122Propranolol 18.57 260 [M + H]+ 60–30–8 260 > 116 60–30–10 260 > 183Betaxolol 18.93 308 [M + H]+ 60–40–8 308 > 116 60–40–14 308 > 121

D y; IS

2p

ear1Ao2wu2pt(im

2(

oobwbwscgtc1i1aan(

2

(aescc(Peeskt5tfpt

pamui1u

tsstmemm

P, declustering potential; CXP, collision cell exit potential; CE, collision energ

.3. Solid phase extraction using molecularly imprintedolymers (MIP4SPETM)

In the optimised procedure, 25 mL of both effluent and influ-nt wastewaters are loaded at neutral pH onto the MIPs usingBaker vacuum system (J.T. Baker, The Netherlands) at a flow

ate of 1 mL/min. Before extraction, MIPs are conditioned withmL of methanol followed by 1 mL of HPLC grade water.fter sample loading, the cartridges are washed with 2 × 1 mLf HPLC grade water, dried under vacuum for approximatelymin, washed with 1 mL acetonitrile, dried for 10 min, washedith 1 mL dichloromethane and dried for two more minutesnder vacuum. Elution of target analytes is then performed with× 1 mL of methanol containing 10% acetic acid plus 2 × 1 mLure methanol and the eluate obtained is evaporated under a gen-le nitrogen stream and reconstituted with 1 mL methanol–water25:75, v/v) containing 50 �g/L of atenolol-d7. This compounds used for internal standard calibration in positive ionisation

ode (PI) to compensate possible matrix effects.

.4. Solid phase extraction using conventional polymericOasis® HLB) cartridges

The analytical protocol used was adapted from a methodol-gy previously optimized by the authors for the determinationf 28 multiple-class pharmaceuticals, including four of the �-lockers surveyed in the present study, in both surface andastewaters [4]. In order to test the extraction efficiency of all �-lockers investigated, 200 mL of effluent and 100 mL of influentastewater were spiked at an appropriate concentration with a

tandard mixture, containing all target analytes. Cartridges wereonditioned with 5 mL of methanol followed by 5 mL of HPLCrade water at neutral pH. Water samples were then loaded ontohe cartridges at a flow rate of 10 mL/min and thereafter, theartridges were rinsed with 5 mL of HPLC grade water andmL of water containing 5% of methanol, to remove possible

nterferences. Finally, cartridges were dried under vacuum for5–20 min and further eluted with 2 × 4 mL of pure methanol

t 1 mL/min. The eluates were then processed in the same ways those from MIPs, i.e., they were evaporated under a gentleitrogen stream and reconstituted with 1 mL of methanol–water25:75, v/v) containing 50 �g/L of atenolol-d7.

iscw

, internal standard.

.5. LC–QqLIT analysis

LC analysis was performed using an Agilent HP 1100Agilent Technologies, Palo Alto, CA, USA) coupled to

4000QTRAPTM (Applied Biosystems, Foster City, USA)quipped with a Turbospray ESI interface. Chromatographiceparation was achieved with a Purospher Star RP-18 endcappedolumn (125 × 2.0 mm, particle size 5 �m) preceded by a guardolumn with the same packing material, both supplied by MerckDarmstadt, Germany). Target compounds were determined inI mode using a mixture of acetonitrile–methanol (2:1) as elu-nt A and a buffer consisting of NH4Ac 5 mM/HAc at pH 4.7 asluent B, at a flow rate of 0.2 mL/min. The injection volume waset at 20 �L. The elution gradient started with 15% of eluent A,eeping isocratic conditions for 3 min. Then, eluent A increasedo 75% in 20 min and initial conditions were reached again inmin, with a re-equilibration time of 15 min in order to restore

he column. For quantitative analysis, data acquisition was per-ormed in the MRM mode, recording the transitions between therecursor ion and the two most abundant product ions for eacharget analyte.

MRM transitions as well as other compound dependentarameters (declustering, entrance, collision cell exit potentialnd collision energy), were optimized by infusing a standardixture containing all �-blockers at 100 �g/L and optimum val-

es are summarized in Table 2. All transitions were recordedn one single retention time window, with a dwell time of00 ms, and a pause time of 2 ms. Both Q1 and Q3 were set atnit.

In addition, for the purpose of obtaining extra confirma-ion in the identification of the analytes, enhanced product ioncans (EPI) at three different Collision Energies, were recordedimultaneously, in the same chromatographic run, as the MRMransitions. This was performed by operating the system in infor-

ation dependent acquisition (IDA) mode. In a typical IDAxperiment, a MS survey scan (enhanced MS and enchancedulti-charge scan when the third quadrupole operates in ion trapode and MRM and neutral loss when it works as a quadrupole),

s used to generate a peak list of all ions present, which isubjected to a user defined criteria to filter out unwanted pre-ursor ions. The remaining ions are then submitted for MS/MShere Q3 operates in the ion trap mode. This cycle is repeated

Page 5

3 atogr

to

tfdtsatofi

3t5

2

taiwsiltt(ooipww

iwrcwot

ccdwotr

etel

sc

mt0twl(

3

3b

suwwaswaanalytes.

Fig. 1 illustrates the extraction efficiencies achieved for the �-blockers investigated when loading different sample volumes ofHPLC water and wastewater onto the MIP cartridges. For the for-

78 M. Gros et al. / J. Chrom

hrough the duration of the acquisition to generate large amountf informative data.

In this study, the survey scan was an MRM containing nineransitions (the most abundant precursor-to-product transitionor each analyte). Each transition was performed at 100 ms ofwell time and a pause time of 2 ms. The IDA scan intensityhreshold was set to 2000 counts per second (cps). The dependentcan was an EPI scan, which was carried out at three CE (20, 30nd 40 eV), in order to observe progressive fragmentation of allarget compounds in the MS/MS scan, obtaining large amountf structural information. Q1 was set at unit and Q3 set to 20 msll time and scan rate of 1000 amu/s.

Source dependent parameters were as follows: curtain gas,0 V; collision gas (CAD), high; ion spray voltage, 5500 V; theemperature (TEM), 700 ◦C, ion source Gas 1 and 2, respectively,0 V.

.6. Validation of the analytical procedure

The extraction recoveries of the MIP4SPETM optimized pro-ocol, for both STP influent and effluent, were determinedt the three spiking levels (0.1, 1 and 5 �g/L). Recover-es were determined comparing the concentrations obtainedith the initial spiking levels and were determined using the

ame batch of wastewater. Quantification was performed bynternal standard calibration. In each case, samples were ana-yzed by triplicate. Similarly, the extraction efficiencies andhe sensitivity of the method using Oasis® HLB SPE car-ridges for pre-concentration were determined only at one level1 �g/L), since the purpose was to compare the performancef both sorbents, instead of validating an analytical methodol-gy based on SPE with such materials. As both spiked STPnfluent and effluent samples already contained target com-ounds, non-spiked samples were analysed and levels foundere afterwards subtracted from those obtained for the spikedaters.One of the main problems presented by MIPs is the difficulty

n removing the entire template molecule, even after extensiveashing. Then, the leakage of trace amounts of the template

emaining in the MIP is an obstacle in the accurate and pre-ise assay of the target analyte [30]. Therefore, method blanksere carried out to ensure that no leakage of template moleculeccurred, which would lead to an overestimation in the calcula-ion of the recoveries.

Precision using both extraction materials was compared byalculating the relative standard deviation (%RSD) of the tripli-ate spiked samples. For evaluation of the sensitivity, methodetection (MDL), determined in spiked effluent and influentastewaters, were calculated as the minimum detectable amountf analyte with a signal-to-noise ratio of 3. Method quantifica-ion limits (MQL,) was defined and calculated as signal to noiseatio of 10.

The reproducibility and repeatability of the method were

valuated from run-to-run experiments (five successive injec-ions of a 100 �g/L standard solution) and day-to-dayxperiments (five successive days). The instrumental detectionimits (IDL) were estimated from the injection of a standard

Fuc

. A 1189 (2008) 374–384

olution successively diluted until reaching a concentration levelorresponding to a signal-to-noise ratio of 3.

Linear dynamic range was determined by injecting a standardixture of target compound in the concentration range from 0.1

o 200 �g/L. This is equivalent to a concentration range between.004 and 8 �g/L in the water samples, taking into considera-ion a method concentration factor of 25. Calibration curvesere generated using linear regression analysis over the estab-

ished concentration range (0.1–200 �g/L) and gave good fitsr2 > 0.99).

. Results and discussion

.1. Estimation of breakthrough volume in the MIP4SPETM

ased protocol

The determination of the breakthrough volume is of highignificance and a critical parameter in SPE protocols. As MIPssed contained a lower amount of polymeric material comparedith conventional polymeric cartridges, lower sample volumesere used with MIPs to avoid saturation of the sorbent and to

chieve good recoveries of target analytes. For this purpose,everal volumes of HPLC grade water, effluent and influentastewaters were spiked, prior to extraction, with appropri-

te concentrations of a standard mixture containing all target

ig. 1. Recoveries of �-blockers obtained when loading different sample vol-mes of (A) HPLC grade water and (B) influent wastewater spiked at aoncentration of 1 �g/L.

Page 6

togr. A 1189 (2008) 374–384 379

maucwva

3b

tswgedtet1

twawaace

F

aruigfftwsa

dp

TRl

C

ASPTMCPB

ASPTMCPB

M. Gros et al. / J. Chroma

er, higher recoveries for the majority of target compounds werechieved when 100 mL were percolated onto the MIP (from 60p to 100%) and recoveries decreased at higher volumes, espe-ially at 500 mL. However, for more complex matrices, such asastewaters, the best recoveries were obtained using less sampleolumes. For both, STP influent and effluent 25 mL was selecteds the optimum value, yielding recoveries from 40 up to 100%.

.2. Optimization of the washing step in the MIP4SPETM

ased procedure

Another critical step in any MIP-based protocol is the selec-ion of appropriate washing solvents, since they allow the highelectivity to be revealed. Solvents used for the SPE procedureere previously optimized by the manufacturer (MIP Technolo-ies) for the analysis of plasma [55] and on the basis of theirxperiments the selected final conditions were adjusted to thoseescribed in the experimental section for the MIP4SPETM pro-ocol. For this purpose, optimum volumes of HPLC grade water,ffluent and influent wastewaters were spiked, prior to extrac-ion, with appropriate concentrations of a standard mixture (at�g/L) containing all target analytes.

Solvents used for clean-up of interferences, and therefore,o enhance the selectivity of the MIP, were HPLC gradeater, acetonitrile (ACN), a mixture ACN/H2O (60/40, v/v)

nd dichloromethane (DCM). The fist step involved washingith HPLC grade water, which is required to eliminate the salts

nd hydrophilic interferences. After that, organic solvents werepplied. Fig. 2 shows the percentage of elution of bound targetompounds in the following washing steps. The results for efflu-nt wastewaters are depicted, but influent wastewaters followed

icwt

able 3ecoveries obtained for �-blockers in surface and wastewaters in with the MIP4SPETM

imits for each type of water matrix, instrumental detection limits, linearity and preci

ompounds Recoveries at 5 �g/L(%RSD, n = 3)

Recoveries at 1 �g/L(%RSD, n = 3)

Wastewatereffluent

Wastewaterinfluent

Wastewatereffluent

Wastewainfluent

tenolol 110 (±5) 95 (±20) 50 (±7) 43 (±9)otalol 50 (±12) 50 (±5) 65 (±6) 70 (±11indolol 52 (±10) 40 (±8) 50 (±4) 50 (±6)imolol 93 (±4) 60 (±5) 60 (±16) 53 (±10etoprolol 75 (±6) 40 (±4) 64 (±6) 91 (±20arazolol 80 (±15) 62 (±16) 110 (±10) 107 (±5)ropranolol 80 (±5) 60 (±4) 93 (±9) 112 (±12etaxolol 85 (±3) 60 (±8) 87 (±6) 101 (±10

Linearity (r2) (0.1–200 �g/L) IDL (pg injected) MDL (ng/L)

Wastewater e

tenolol 0.9998 2.3 1.5otalol 0.9997 2.7 6.4indolol 0.9994 0.2 0.2imolol 1.000 0.4 0.4etoprolol 0.9999 2.2 0.5arazolol 0.9965 0.6 0.4ropranolol 0.9985 0.7 0.4etaxolol 0.9988 2.0 3.0

ig. 2. Optimization of the washing solvents in the MIP4SPETM protocol.

similar pattern. The clean-up with the mixture of ACN/H2Oesulted in important losses of target compounds, and was notsed in further experiments. Retention of the analytes of interestn the MIP is mainly due to hydrogen bonds between functionalroups in the analyte and specific binding sites in the MIP. There-ore, losses observed using this mixture could be attributed to theact that some of the bonds in the imprinted sites are broken, dueo the polarity of water in addition to the organic solvent (ACN)hich would compete with the compounds to bind the active

ites. As a result, some �-blockers retained would be releasednd lost during the clean-up procedure.

On the other hand, the washing with pure ACN and DCMid not imply significant losses of target analytes, but theyroved to remove other interferences bound in the MIP (from the

mprinted sites and the polymer itself). ACN removes the unspe-ific hydrophobic interferences, as well as other substanceseakly bound in the imprinted sites. DCM is used to prepare

he MIP for the elution of target analytes. Since some of the

procedure, at three different spiking levels, method detection and quantificationsion of the instrument

Recoveries at 0.1 �g/L(%RSD, n = 3)

Repeatability/reproducibility%RSD (n = 5)

ter Wastewatereffluent

Wastewaterinfluent

76 (±7) – 1.1/8.0) 51 (±6) 84 (±2) 3.0/11.0

96 (±20) 70 (±6) 0.7/5.0) 50 (±5) 50 (±1) 2.8/5.0) – 43 (±5) 3.1/4.0

81 (±2) 92 (±9) 7.7/9.0) 75 (±1) 65 (±1) 3.0/4.6) 80 (±13) 62 (±1) 2.0/8.0

MQL (ng/L)

ffluent Wastewater influent Wastewater effluent Wastewater influent

1.1 5.0 4.06.5 22.0 22.00.4 0.6 1.20.4 1.4 1.11.3 1.8 4.30.4 1.3 1.20.5 1.2 1.53.4 10.0 11.3

Page 7

380 M. Gros et al. / J. Chromatogr

Table 4Recoveries obtained with Oasis® HLB cartridges for �-blockers in wastewatersspiked at 1 �g/l level and method detection limits

Compounds Wastewatereffluent

MDL (ng/L) Wastewaterinfluent

MDL (ng/L)

Atenolol 50 (±8) 4 89 (±12) 4Sotalol 81 (±4) 2 60 (±1) 3Pindolol 68 (±3) 1 76 (±3) 8Timolol 50 (±3) 2 54 (±11) 2Metoprolol 80 (±4) 3 42 (±56) 3Carazolol 50 (±7) 1 70 (±3) 4PB

bAa

ubsrt

3p

MemeO

weRoc

HaecSobfrglpi

3

lohps

TR

G

L

P

U

A

T

PT

n

ropranolol 40 (±3) 2 60 (±13) 5etaxolol 60 (±6) 4 70 (±10) 8

indings in the imprinted sites are weakened after the wash withCN, DCM helps the swelling of the MIP, so the imprinted sitesre more open and it is easier to elute the compounds.

Moreover, the order in which solvents are applied, their vol-me and the vacuum drying time between the steps was found toe of high significance. Hence, if 100% ACN was passed withome water still remaining on the MIP, hydrogen bonds thatetain target compounds in the active sites could break, leadingo considerable losses of �-blockers.

.3. Performance of the MIP: Extraction efficiency,recision and sensitivity

Table 3 illustrates the extraction recoveries of theIP4SPETM optimized protocol for both STP influent and efflu-

nt, at the three spiking levels tested (0.1, 1 and 5 �g/L), theethod detection and quantification limits. Table 4 shows the

xtraction efficiencies and the sensitivity of the method usingasis® HLB SPE cartridges for pre-concentration.

irmt

able 5etention of multiple-class pharmaceuticals and pesticides in the MIP: recoveries ob

roup Compounds % Elution ACN fraction

ipid regulators Bezafibrate 7Mevastatin 11

sychiatric drugs Carbamazepine 61Fluoxetine 16Paroxetine 6

lcer healings Lansoprazole 18Loratadine 5Famotidine 2Ranitidine 7

ntibiotics Erythromycin 83Azythromycin 47Sulfamethoxazole n.f.Trimethoprim 26

riazines Atrazine 70Simazine 40Desethylatrazine 15Terbuthylazine 62

henylureas Diuron 70hiocarbamate Molinate 2

.f.: not found.

. A 1189 (2008) 374–384

Results point out that recoveries achieved with both materialsere fairly similar (ranging from 50 to 110% for STP efflu-

nt and 43 to 112 for STP influent, at 1 �g/L spiking level).elative standard deviations for MIPs were quite close to thenes achieved with Oasis® HLB, with the exception of someompounds, such as metoprolol.

However, the main advantages of the MIPs versus Oasis®

LB lay in the fact that good recoveries and sensitivity arechieved even using a pre-concentration factor from four toight times lower (due to less volumes of sample; 25 mL inomparison to 100 and 200 mL, when using Oasis® HLB forTP influent and effluent, respectively). Moreover, the MDLbtained by MIPs were in the same range as the ones achievedy Oasis® HLB, being in some cases lower, such as 0.2 ng/Lor pindolol and 0.4 ng/L for timilol, carazolol and propranolol,espectively. In addition, percolating lower sample volumes is areat advantage against conventional polymeric materials sinceess time is required for the pre-concentration, speeding up sam-le preparation and furthermore, this leads to less matrix effectsn instrumental analysis.

.4. Performance of the MIP: Cross-reactivity tests

Selectivity of MIPs for target compounds and related ana-ogues was evaluated by determining the extraction efficiencyf several pharmaceuticals (lipid regulators, ulcer healings, anti-istamines, antibiotics) and pesticides, which included triazines,henylureas and tiocarbamate pesticides. These pollutants wereelected due to their ubiquity in the aquatic environment, accord-

ng to previous studies performed by the authors’ group ofesearch [4,5,50,51]. Since �-blockers are determined in PIode, only the most widespread pharmaceuticals detected in

he same ionization mode were included. Thus, analgesics and

tained in the washing steps and in the eluate

% Elution DCM fraction % Recovery (%RSD) MIP

n.r. 8 (±4)n.r. n.f.

3 n.f.5 57 (±6)2 67 (±2)

n.f. n.f.n.f. 9 (±5)n.f. 38 (±2)n.f. n.f.

5 21 (±15)3 33 (±1)n.f. n.f.2 38 (±2)

10 n.f.5 n.f.2 n.f.9 n.f.

7 n.f.n.f. n.f.

Page 8

togr. A 1189 (2008) 374–384 381

atnHtSfmTtTesicfaM

rtsattMbctNcc9

3

ti[tpvsmt

Hep

bleuca

Fig. 3. (A) Calibration curves in solvent versus those prepared in spiked wastew-ater influent extracts and (B) comparison between internal standard calibrationcurves in solvent and in spiked wastewater influent extracts, for the ß-blockera

is

ca(oeppatt

n

M. Gros et al. / J. Chroma

nti-inflammatories, in spite of being among the most ubiqui-ous therapeutic groups found in the aquatic environment, wereot considered because they are analyzed by negative ionization.ence, wastewater was spiked at 1 �g/L with two standard mix-

ures containing multiple-class pharmaceuticals and pesticides.ample preparation and recovery calculations were performedollowing the same protocol previously detailed for the deter-ination of �-blockers in MIPs and Oasis® HLB cartridges.o prove that this step is crucial and rules the specificity of

he MIP, aliquots from the washing steps were also collected.able 5 summarized the results obtained. It can be seen that sev-ral compounds were retained, but removed during the washingteps, which proved that this is a process of high significancen any MIP protocol which cannot be neglected. This was thease for the anti-epileptic carbamazepine, the ulcer healingamotidine, and the antibiotics trimethoprim, erythromycin andzythromycin, which would have been highly retained in theIP if no washing with ACN had been performed.Focusing on the recoveries after elution of target compounds,

esults reveal that the majority of pharmaceuticals and all pes-icides are poorly retained in the MIP, because their chemicaltructure do not share any similarity with �-blockers and/orre removed by the washing solvents (see Table 5). However,he antidepressants fluoxetine and paroxetine were an excep-ion to this behaviour, presenting an extraction recovery in the

IP above 50%. Comparing their chemical structure with �-lockers, high retention for fluoxetine could be explained by theommon OCH2CH2CH2NH group. For the rest of compounds,he retention could be due to a combination of hydrophobic andH2 group bindings. In contrast, all pharmaceuticals and pesti-

ides tested presented high extraction efficiencies when using aonventional SPE material that range in general from 60 up to0%.

.5. Performance of the MIP: Matrix effects evaluation

Matrix effects occur because the ESI source is highly suscep-ible to co-extracted components present in the matrix, resultingn a signal suppression or enhancement of the analyte signal52]. The use of highly selective extraction materials and effec-ive sample clean-up procedures can contribute to avoid thisroblem. However, when sorbents used are not able to pro-ide such selectivity, other plausible alternatives are the use ofuitable calibration approaches (i.e. external calibration usingatrix-matched samples, standard addition, internal standard or

he dilution of sample extracts) [53,54].In order to evaluate the selectivity of both MIPs and Oasis®

LB cartridges, external calibration curves prepared with efflu-nt and influent wastewater extracts were compared with thoserepared in solvent (methanol–water (25:75, v/v)).

As polymeric Oasis® HLB cartridges were expected toe less selective for target analytes than MIPs becauseipohilic–hydrophilic composition makes them suitable for the

xtraction of a wide spectrum of substances, the possibility ofsing atenolol-d7 as internal standard (IS) for IS calibration toompensate matrix effects due to co-extracted compounds wasssessed. For this purpose, IS calibration curves were prepared

afa(

tenolol in MIPs and Oasis® HLB cartridges.

n wastewater extracts and compared with those prepared inolvent.

In Fig. 3A the comparison between external calibrationurves for the �-blocker atenolol, prepared using both MIPnd Oasis® HLB influent extracts and solvent (methanol–water25:75, v/v)) is illustrated. If both curves are parallel and totallyverlapped, means that compounds are not liable to matrixffects. As it is depicted, MIPs presented negligible signal sup-ression, because both curves are totally overlapped, whereasolymeric Oasis® HLB showed significant suppression of thenalyte signal. In order to compensate such effect, the use of iso-opically labelled internal standards for calibration is necessaryo solve this problem, as indicated in Fig. 3B.

Fig. 4 compares the extracted ion chromatogram for propra-olol in influent wastewater sample analyzed using Oasis® HLBnd MIP, indicating that the latter is much more selective. Apart

rom providing cleaner extracts, with MIPs higher sensitivity ischieved for the same concentration of �-blockers in the samplenote the different y-scale of intensity).

Page 9

382 M. Gros et al. / J. Chromatogr. A 1189 (2008) 374–384

Fs

hrt1tc

3

paEteiiamtoewoet

ItCbeR1i

Fa

aieoissmpeIwd3

3

fi

ig. 4. Extracted ion chromatogram of propranolol in an influent wastewaterample analyzed by (A) Oasis® HLB cartridges and (B) MIP.

However, this behaviour, apart from being attributed to theigh selectivity that the MIP presents for target analytes andelated analogues, it could also be due to the fact that withhe MIP less sample volume is used (25 mL comparing to the00 mL for influent and 200 mL for effluent wastewaters, respec-ively, for the Oasis® HLB) and consequently, matrix effectsould be reduced.

.6. LC–QqLIT analysis

Working in MRM mode, monitoring two transitions betweenrecursor ions and the two most abundant fragment ionsllowed the accomplishment of the requirements set by theU regulations (EU Commission Decision 2002/657/EC) for

he confirmation and identification of pharmaceuticals innvironmental samples. These regulations determine that thenstrumental technique used should provide a minimum of threedentification points (IP) to ensure a correct detection of targetnalytes. Therefore, when monitoring two transitions in MRMode 4 IP are obtained (1 for the precursor ion and 1.5 for each

ransition product), which is sufficient to confirm the identityf a compound in a sample. Additionally, the inclusion of annhanced product ion scan in the same experiment as the MRM,hen operating in the IDA mode, would increase the numberf IP to 7 (4 for MRM + 1.5 for each product ion), which is anxtra tool for unequivocal confirmation of target compounds inhe sample.

Working in Information Dependent Acquisition Mode inDA was a useful approach to provide unequivocal identifica-ion of target compounds in complex environmental samples.ompounds were identified by monitoring the major transitionetween precursor to product ion in MRM mode and then, for

ach compound, a highly sensitive MS/MS scan is obtained.epresentative chromatograms obtained from the analysis of a00 ng/mL standard mixture, containing all �-blockers and thenternal standard, determined by MRM are illustrated in Fig. 5

Sdos

ig. 5. Chromatograms of a 100 ng/mL standard mixture containing all targetnalytes recorded in the MRM mode.

nd those achieved by IDA in Fig. 6. As it can be observedn Fig. 6 for atenolol, the advantage of performing an IDAxperiment is that, apart from monitoring the MRM transitions,bserving the analyte peak at 3.89 min, structural information forts unequivocal identification is achieved, by performing a highlyelective MS/MS scan. The equipment used demonstrated higherensitivity for the determination of pharmaceuticals in environ-ental samples since IDL (see Table 3). Regarding quantitative

erformance in terms of dynamic range, linear response cov-red three orders of magnitude, providing good fits (r2 > 0.99).n order to ensure correct quantification, precision of the methodas evaluated by analysing five replicates of a 100 �g/L stan-ard, and results indicated a precision from 0.7 to 4.6% and fromto 11% from intra-day and inter-day analysis, respectively.

.7. Analysis of real wastewater samples

To demonstrate the applicability of the developed method,ve influent and effluent grab wastewater samples from a

TP located in Rubı (Barcelona, Spain), which gathers urban,omestic and industrial wastewaters, were analysed. The typef treatment applied in this Wastewater Treatment Plant con-ists on primary settling and secondary treatment by activated

Page 10

M. Gros et al. / J. Chromatogr. A 1189 (2008) 374–384 383

F he det1 M, (B(

ssftptie[

TRti

T

ASPTMCPB

4

bd

ig. 6. Information dependent acquisition (IDA) experiment corresponding to t�g/L in water) wastewater sample. (A) Total ion chromatogram (TIC) of MR

EPI) for atenolol at a CE = 30 eV.

ludge. Results obtained are summarized in Table 6. Atenolol,otalol, propranolol and metoprolol were the �-blockers morerequently detected with concentrations ranging from high ng/Lo low �g/L. Some compounds, such as atenolol, sotalol andropranolol showed poor or no elimination during wastewa-

er treatment process, as similar concentrations were detectedn both influent and effluent wastewaters. In general, lev-ls detected are similar to those reported in previous studies4,5,2,19].

able 6ange of concentrations and average level (expressed in brackets) detected for

arget pharmaceuticals in influent and effluent wastewaters from a STP locatedn Rubı (Barcelona, Spain)

arget compounds Minimum–maximum concentrations (ng/L)

Influent wastewaters Effluent wastewaters

tenolol 1740–2688 (2359) 618–1370 (1060)otalol 243–348 (311) 230–308 (268)indolol bld bldimolol 13–26 (20) bldetoprolol 1203–3047 (2408) 79–547 (375)arazolol bld bldropranolol 74–144 (117) 87–136 (104)etaxolol bld bld

atsc

mvCohe

A

Ui(t

ermination of atenolol in a spiked influent (corresponding to a concentration of) extracted ion chromatogram for atenolol and (C) enhanced product ion scan

. Conclusions

Molecularly imprinted polymers (MIP4SPETM Beta-locker) have proved to be highly selective materials for theetermination of �-blockers in environmental waters. As theyre tailor made materials, synthesized to selectively isolatearget compounds, they show reduced matrix effects and higherensitivity, comparing with the balanced hydrophilic–lipophiliconventional polymeric cartridges Oasis® HLB.

In addition, the application of LC–QqLIT, operating in MRMode, with two transitions monitored for each compound, pro-

ided good sensitivity and selectivity, according to the EUommision Decision 2002/657/EC, enabling the determinationf target compounds at the low ng/L range. The inclusion of aigh sensitive MS/MS scan in the same experiment provided anxtra tool to unequivocally identify the �-blockers studied.

cknowledgements

This work was financially supported by the European

nion EMCO (INCO-CT-2004-509188) and by the Span-

sh Ministerio de Ciencia y Tecnologia project CEMAGUACGL2007-64551). M. Gros acknowledges a grant fromhe Spanish MCyT (project CTM2004-06265-C03-01). Tania

Page 11

3 atogr

MdBa4tart

R

[

[

[[[

[

[[

[

[[

[

[[

[

[

[

[[[[[

[[[[

[

[

[

[

[[

[[

[

[[

[

[[

[

[

[[

153.

84 M. Gros et al. / J. Chrom

ara Pizzolato acknowledges her grant from Coordenacaoe Aperfeicoamento de Pessoal de Nıvel Superior-CAPES-razil. MIP Technologies and Applied Biosystems are gratefullycknowledged for providing MIP cartridges and lending the000QTRAP as well as for their technical assistance, respec-ively. Waters Corporation (USA) and Merck (Germany) are alsocknowledged for providing SPE cartridges and HPLC columns,espectively. R. Chaler and D. Fanjul are also acknowledged forheir excellent assistance in LC–MS troubleshooting.

eferences

[1] C.G. Daughton, T.A. Ternes, Environ. Health Perspect. 107 (1999) 907.[2] S. Castiglioni, R. Bagnati, D. Calamari, R. Fanelli, E. Zuccato, J. Chro-

matogr. A 1092 (2005) 206.[3] B. Gross, J. Montgomery-Brown, A. Naumann, M. Reinhard, Environ.

Toxicol. Chem. 23 (2004) 2074.[4] M. Gros, M. Petrovic, D. Barcelo, Talanta 70 (2006) 678.[5] M.J. Gomez, M. Petrovic, A.R. Fernandez-Alba, D. Barcelo, J. Chromatogr.

A 1114 (2006) 224.[6] D. Bendz, N.A. Paxeus, T.R. Ginn, F.J. Loge, J. Hazard. Mater. 122 (2005)

195.[7] M. Petrovic, M.D. Hernando, M.S. Diaz-Cruz, D. Barcelo, J. Chromatogr.

A 1067 (2005) 1.[8] R. Andreozzi, M. Raffaele, P. Nicklas, Chemosphere 50 (2003) 1319.[9] K. Fent, A.A. Weston, D. Caminada, Aquat. Toxicol. 76 (2006) 122.10] F.T. Delbeke, M. Debackere, N. Desmet, F. Maertens, J. Pharm. Biomed.

Anal. 6 (1988) 827.11] L. Amendola, F. Molaioni, F. Botre, J. Pharmacol. Biomed. Anal. 23 (2000)

211.12] R.D. Johnson, R.J. Lewis, Forensic Sci. Int. 156 (2006) 106.13] M. Thevis, W. Schanzer, J. Chromatogr. Sci. 43 (2005) 22.14] M. Thevis, G. Opfermann, W. Schanzer, Biomed. Chromatogr. 15 (2001)

393.15] M.Y. Liu, D. Zhang, Y.T. Sun, Y.W. Wang, Z.Y. LIu, J.K. Gu, Biomed.

Chromatogr. 21 (2007) 508.16] G.J. Trout, R. Kazlauskas, Chem. Soc. Rev. 33 (2004) 1.17] D. Calamari, E. Zuccato, S. Castiglioni, R. Bagnati, R. Fanelli, Environ.

Sci. Technol. 37 (2003) 1241.18] M.D. Hernando, M. Petrovic, A.R. Fernandez-Alba, D. Barcelo, J. Chro-

matogr. A 1046 (2004) 133.19] N.M. Vieno, T. Tuhkanen, L. Kronberg, J. Chromatogr. A 1134 (2006) 101.20] H.H. Maurer, O. Tenberken, C. Kratzsch, A.A. Weber, F.T. Peters, J. Chro-

matogr. A 1058 (2004) 169.21] M.D. Hernando, M.J. Gomez, A. Aguera, A.R. Fernandez-Alba, TrAC-

Trends Anal. Chem. 26 (2007) 581.

22] H.B. Lee, K. Sarafain, T.E. Peart, J. Chromatogr. A 1148 (2007) 156.23] L.N. Nikolai, E.L. McClure, S.L. McLeod, C.S. Wong, J. Chromatogr. A

1131 (2006) 1.24] M.J. Gomez, A. Aguera, M. Mezcua, J. Hurtado, F. Mocholi, A.R.

Fernandez-Alba, Talanta 73 (2007) 314.

[

[

. A 1189 (2008) 374–384

25] B. Kasprzyk-Hordern, R.M. Dinsdale, A.J. Guwy, J. Chromatogr. A 1161(2007) 132.

26] M.J. Gomez, M.J. Martinez Bueno, S. Lacorte, A.R. Fernandez-Alba, A.Aguera, Chemosphere 66 (2007) 993.

27] I. Ferrer, D. Barcelo, TrAC-Trends Anal. Chem. 18 (1999) 180.28] M. Gros, M. Petrovic, D. Barcelo, Anal. Bioanal. Chem. 386 (2006) 941.29] H.S. Wei, Y.L. Tsai, J.Y. Wu, H. Chen, J. Chromatogr. B 836 (2006) 57.30] F. Qiao, H. Sun, H. Yan, K.H. Row, Chromatographia 64 (2006) 625.31] M.J. Whitcombe, L. Martin, E.N. Vulfson, Chromatographia 47 (1998)

457.32] L.D. Bolisay, J.N. Culver, P. Kofinas, Biomaterials 27 (2006) 4165.33] K.J. Shea, D.Y. Sasaki, J. Am. Chem. Soc. 111 (1989) 3442.34] S. Rimmer, Chromatographia 47 (1998) 470.35] X.L. Zhu, J. Yang, Q.D. Su, J.B. Cai, Y. Gao, J. Chromatogr. A 1092 (2005)

161.36] I. Ferrer, F. Lanza, A. Tolokan, V. Horvath, B. Sellergren, G. Horvai, D.

Barcelo, Anal. Chem. 72 (2000) 3934.37] R. Carabias-Martinez, E. Rodriguez-Gonzalo, E. Herrero-Hernandez, J.

Chromatogr. A 1085 (2005) 199.38] R. Koeber, C. Fleischer, F. Lanza, K.S. Boos, B. Sellergren, D. Barcelo,

Anal. Chem. 73 (2001) 2437.39] E. Caro, R.M. Marce, P.A.G. Cormack, D.C. Sherrington, F. Borrull, J.

Chromatogr. A 995 (2003) 233.40] P. Martin, I.D. Wilson, G.R. Jones, J. Chromatogr. A 889 (2000) 143.41] P.R. Kootstra, C.J.P.F. Kuijpers, K.L. Wubs, D. van Doorn, S.S. Sterk, L.A.

van Ginkel, R.W. Stephany, Anal. Chim. Acta 529 (2005) 75.42] N. Zheng, Y.Z. Li, M.J. Wen, J. Chromatogr. A 1033 (2004) 179.43] G. Brambilla, M. Fiori, B. Rizzo, V. Crescenzi, G. Masci, J. Chromatogr.

B 759 (2001) 27.44] C. Berggren, S. Bayoudh, D. Sherrington, K. Ensing, J. Chromatogr. A 889

(2000) 105.45] K. Farrington, E. Magner, F. Regan, Anal. Chim. Acta 566 (2006) 60.46] E. Caro, R.M. Marce, P.A.G. Cormack, D.C. Sherrington, F. Borrull, Anal.

Chim. Acta 552 (2005) 81.47] A. Guzman-Vazquez De Prada, P. Martinez-Ruiz, A.J. Reviejo, J.M. Pin-

garron, Anal. Chim. Acta 539 (2005) 125.48] G. Hopfgartner, C. Husser, M. Zell, J. Mass Spectrom. 38 (2003) 138.49] Commision Decision (2002/657/EC) of 12 August 2002, Implementing

Council Directive 96/23/EC concerning the performance of analyticalmethods and the interpretation of results, Official Journal of the EuropeanCommunities L221, Brussels, Belgium, pp. 8–36.

50] S. Rodriguez-Mozaz, M.J. Lopez de Alda, D. Barcelo, J. Chromatogr. A1045 (2004) 85.

51] D.A. Azevedo, S. Lacorte, P. Viana, D. Barcelo, Chromatographia 53(2001) 113.

52] M. Stuber, T. Reemtsma, Anal. Bioanal. Chem. 378 (2004) 910.53] A. Kloepfer, J.B. Quintana, T. Reemtsma, J. Chromatogr. A 1067 (2005)

54] T. Benijts, R. Dams, W. Lambert, A. De Leenheer, J. Chromatogr. A 1029(2004) 153.

55] Application Note A1007-MIP Technologies. Extraction of �-blockers fromhuman urine/plasma by LC-MS detection.