Td

MUa

b

a

ARRAA

KCGSDL

1

impgltwmmip

aF

h0

Journal of Chromatography A, 1440 (2016) 105–111

Contents lists available at ScienceDirect

Journal of Chromatography A

j o ur na l ho me page: www.elsev ier .com/ locate /chroma

racing gadolinium-based contrast agents from surface water torinking water by means of speciation analysis

arvin Birkaa, Christoph A. Wehea, Oliver Hachmöllera, Michael Sperlinga,b,we Karsta,∗

University of Münster, Institute of Inorganic and Analytical Chemistry, Corrensstraße 28/30, 48149 Münster, GermanyEuropean Virtual Institute for Speciation Analysis (EVISA), Mendelstraße 11, 48149 Münster, Germany

r t i c l e i n f o

rticle history:eceived 4 August 2015eceived in revised form 8 February 2016ccepted 14 February 2016vailable online 17 February 2016

eywords:ontrast agentsadoliniumpeciation analysisrinking waterC-ICP-MS

a b s t r a c t

In recent decades, a significant amount of anthropogenic gadolinium has been released into the envi-ronment as a result of the broad application of contrast agents for magnetic resonance imaging (MRI).Since this anthropogenic gadolinium anomaly has also been detected in drinking water, it has becomenecessary to investigate the possible effect of drinking water purification on these highly polar microcon-taminats. Therefore, a novel highly sensitive method for speciation analysis of gadolinium is presented.For that purpose, the hyphenation of hydrophilic interaction liquid chromatography (HILIC) and induc-tively coupled plasma-mass spectrometry (ICP-MS) was employed. In order to enhance the detectionpower, sample introduction was carried out by ultrasonic nebulization. In combination with a novelHILIC method using a diol-based stationary phase, it was possible to achieve superior limits of detectionfor frequently applied gadolinium-based contrast agents below 20 pmol/L. With this method, the con-trast agents Gd-DTPA, Gd-DOTA and Gd-BT-DO3A were determined in concentrations up to 159 pmol/L

in samples from several waterworks in a densely populated region of Germany alongside the river Ruhras well as from a waterworks near a catchment lake. Thereby, the direct impact of anthropogenic gadolin-ium species being present in the surface water on the amount of anthropogenic gadolinium in drinkingwater was shown. There was no evidence for the degradation of contrast agents, the release of Gd3+ orthe presence of further Gd species.

© 2016 Elsevier B.V. All rights reserved.

. Introduction

Contrast agents based on the rare earth element (REE) gadolin-um (Gd) are frequently applied prior to medical examinations with

agnetic resonance imaging (MRI) because of the paramagneticroperties of Gd3+ [1]. Those properties cause an increased lon-itudinal relaxation rate T1 of water protons during MRI scans,eading to an improved contrast of the resulting images. Due tohe toxicity of Gd3+, it is delivered to the patients as chelateith polyaminocarboxylates, resulting in a fast and unmetabolized,ostly renal, excretion [2]. The first of these compounds to be com-

ercially available was Gd-DTPA (gadopentetate), which has been

ntroduced into the market in 1988. Since then, several other com-ounds from different pharmaceutical companies with different

∗ Corresponding author at: University of Münster, Institute of Inorganicnd Analytical Chemistry, Corrensstraße 30, 48149 Münster, Germany.ax: +49 251 8336013.

E-mail address: uk@uni-muenster.de (U. Karst).

ttp://dx.doi.org/10.1016/j.chroma.2016.02.050021-9673/© 2016 Elsevier B.V. All rights reserved.

chelating agents were developed and commercialized. Fig. 1 showsthe chemical structures of widely applied contrast agents, whichwere subject of this study [3]. These complexes can be separatedin two groups: on the one hand those with linear ligands such asthe above mentioned Gd-DTPA and Gd-BOPTA (gadobenate), andon the other hand those with macrocyclic ligands like Gd-DOTA(gadoterate) and Gd-BT-DO3A (gadobutrol) [1]. The dosage perinfusion of commercially available contrast agents is 0.05 mmol upto 0.3 mmol per kg body weight, meaning that about 1 g of gadolin-ium is being applied for one contrast agent enhanced MRI scan[3].

As a result of these high dosages and the unmetabolized excre-tion of the compounds, large amounts of anthropogenic gadoliniumare released into the wastewater. In 1996, Bau and Dulski describedan enrichment of gadolinium relatively to the other REEs in watersamples from rivers and lakes as a result of this input [4]. This

anthropogenic gadolinium anomaly has been investigated sincethen in a series of studies. It was shown that this phenomenoncan be observed in rivers and lakes in highly populated regionswith developed health care around the world [5–12]. Furthermore,

1 atogr.

s[ttgsii

wwlotd[hfb

a(ec[oliarstbctottttpd

06 M. Birka et al. / J. Chrom

ignificant gadolinium anomalies were found even in the North Sea13]. The emitted amount of anthropogenic gadolinium indicateshat the contrast agents are not being removed during wastewaterreatment. In 2010, investigations by Verplanck et al. revealed largeadolinium anomalies in treated wastewater in contrast to sewageludge [14]. These results were confirmed by Telgmann et al., show-ng only a minor removal of approximately 10% of total gadoliniumn batch experiments with a simulated aeration tank [15].

Since the supply with drinking water also partly relies on surfaceater, it is possible to detect anthropogenic gadolinium in the tapater of urban agglomerations. Kulaksiz and Bau reported in 2011

arge anthropogenic gadolinium anomalies in the western districtsf Berlin (Germany) and only minor amounts in the eastern dis-ricts. This was a result of the drinking water supply in the westernistricts by means of bank filtration and thereby from surface water16]. Samples from processing and purification of drinking waterave not been analyzed so far. Furthermore, the formation of trans-

ormation products for example through ozonation or disinfectiony UV irradiation has to be considered.

Therefore, it is necessary to develop highly sensitive speci-tion analysis methods. Methods based on ion chromatographyIC), high performance liquid chromatography (HPLC), and sizexclusion chromatography (SEC) have been developed for thehromatographic separation of gadolinium-based contrast agents17–19]. In 2008, Künnemeyer et al. demonstrated the separationf commonly applied contrast agents using hydrophilic interactioniquid chromatography (HILIC) hyphenated to electrospray ion-zation mass spectrometry (ESI-MS) [20]. Since then, HILIC waspplied in several studies for the analysis of biological and envi-onmental samples using zwitterionic and unboned silica-basedtationary phases. Especially the hyphenation of HILIC with induc-ively coupled plasma-mass spectrometry (ICP-MS) has proven toe a powerful method for the determination of gadolinium basedontrast agents, mostly due to the uniquely high and element selec-ive sensitivity of ICP-MS [21–24]. Nevertheless, the direct analysisf drinking water samples remains a difficult task with regardo the low concentration of the individual gadolinium species athe pmol/L level [25]. In general, the sample introduction into

he plasma is a crucial parameter for sensitivity and therefore forhe limit of detection using ICP-MS. Compared to conventionalneumatic nebulization, ultrasonic nebulization with subsequentesolvation offers a ten times increased efficiency of aerosol gener-

Fig. 1. Structures of the frequently applied gadolinium-based contrast agents for MR

A 1440 (2016) 105–111

ation and an improved transport efficiency into the plasma [26]. Theapplication of ultrasonic nebulization for HPLC-ICP-MS has beenaccomplished for the determination of several species of metalsand metalloids such as mercury, platinum, arsenic and selenium[27–31].

This paper describes the development of a HILIC-ICP-MS methodusing ultrasonic nebulization for sample introduction. A new HILICseparation with a diol-based stationary phase was accomplishedfor the most commonly applied contrast agents. This method wasthen applied for the analysis of samples from different steps ofdrinking water purification in a series of waterworks. Additionally,total concentrations of gadolinium were determined by means ofICP-MS.

2. Materials and methods

2.1. Sampling and sample preparation

Water samples from six waterworks in Germany were analyzedduring this study, covering different methods of drinking waterpurification from surface water. Those methods involve ground fil-tration, filtration through activated carbon as well as ozonation.Disinfection of the drinking water is carried out in those water-works by addition of chlorine dioxide or UV irradiation. The locationof the waterworks is shown in Fig. 2. One of those waterworks (A)is located near a large catchment lake, from which the raw wateris obtained. The other five waterworks (B–F) are located in one ofthe most densely populated region of Germany alongside the riverRuhr. Therefore, it was assumed that significant amounts of anthro-pogenic gadolinium could be detected in the drinking water. Intotal, 21 Samples from different steps of the drinking water purifi-cation were taken with vessels of polypropylene (PP) at a singleday in October 2014. The detailed sampling points in the respec-tive waterworks are shown in the results section. Since the contrastagents Gd-DTPA and Gd-BOPTA show strong adsorption on glasssurfaces, the use of glass vessels had to be avoided during samplingas well as during sample preparation [21]. The collected samples

were cooled immediately below 10 ◦C for transport and subsequentanalysis.

For determination of total gadolinium concentrations withICP-MS, the samples were filtered through syringe filters with

I examinations with respective trademarks, which were subject of this study.

M. Birka et al. / J. Chromatogr. A 1440 (2016) 105–111 107

F ar a c

paofitsbwm

2

((SIFstba

Fc

ig. 2. Location of the waterworks being part of his study alongside the Ruhr and ne

olytetrafluoroethylene (PTFE) membrane. 100 �L nitric acid (65%)nd 1 mL rhodium solution as internal standard were added to 8 mLf the filtrate in a PP vessel and diluted to a volume of 10 mL. Thenal concentration of the internal standard was 1 �g/L. For specia-ion analysis with HILIC-ICP-MS, about 1.5 mL of the surface wateramples were also filtered through syringe filters with PTFE mem-rane and filled into PP vials. The procedure of sample preparationas repeated three times for ICP-MS as well as for HILIC-ICP-MSeasurements.

.2. Chemicals and consumables

Nitric acid (65%, Suprapur), gadolinium standard for ICP-MS1000 mg/L), and acetonitrile were purchased from Merck KGaADarmstadt, Germany). Rhodium solution (1000 mg/L) from SCPcience (Baie D’Urfé, Canada) was used as internal standard forCP-MS. Formic acid and ammonium formate were obtained fromluka Chemie GmbH (Buchs, Switzerland). The contrast agent

tandards were purchased as infusion solutions from the respec-ive pharmaceutical companies: Dotarem (Gd-DOTA, 0.5 mol/L)y Guerbet (Sulzbach, Germany), Magnevist (Gd-DTPA, 0.5 mol/L)nd Gadovist (Gd-BT-DO3A, 1.0 mol/L) by Bayer-Schering Pharma

ig. 3. HILIC-ICP-MS chromatograms of standard solutions (Gd-BOPTA, Gd-BT-DO3A, Gd-hromatograms show the sum of the intensities of the isotopes 156Gd, 157Gd, 158Gd and 16

atchment lake. The drinking water is produced by purification of the surface water.

AG (Berlin, Germany), and Multihance (Gd-BOPTA, 0.5 mol/L) byNycomed (Konstanz, Germany). All chemicals were used in thehighest quality available. Water was purified by an AquatronWater Still purification system model A4000D (Barloworld Scien-tific, Nemours Cedex, France). Syringe filters with PTFE membranes(0.2 �m pore size) were purchased from VWR International (Darm-stadt, Germany).

2.3. Stock and standard solutions

The determination of total gadolinium concentrations with ICP-MS was carried out by external calibration in a range from 5 to500 ng/L. Rhodium solution was added as internal standard priorto final dilution of the sample solutions as well as to the calibrationsolutions with a final concentration of 1 �g/L.

Solutions of the contrast agents Gd-BOPTA, Gd-BT-DO3A, Gd-DOTA and Gd-DTPA were prepared in a concentration of 1 nmol/L

each by dilution of the infusion solutions for determination ofthe retention times. For quantification of single contrast agents bymeans of HILIC-ICP-MS, a mixed stock solution with a concentrationof 50 nmol/L for Gd-BOPTA, Gd-BT-DO3A, Gd-DOTA and Gd-DTPA

DOTA and Gd-DTPA, c = 10–500 pmol/L) and a blank solution (deionized water). The0Gd.

108 M. Birka et al. / J. Chromatogr.

Fig. 4. HILIC-ICP-MS chromatograms of a standard solution (Gd-BOPTA, Gd-BT-DO3A, Gd-DOTA and Gd-DTPA, c = 500 pmol/L) and the samples from waterworksF. The chromatograms show the sum of the intensities of the isotopes 156Gd, 157Gd,158Gd and 160Gd.

Table 1Comparison of LODs from different studies involving the determination of Gd-basedcontrast agents in environmental samples and drinking water by means of HILIC-ICP-MS.

Reference Year Gd-based contrast agents LOD (pmol/L)

Künnemeyer et al. [21] 2009 Gd-DTPA, Gd-BOPTA, 1000Gd-DTPA-BMA, Gd-DOTA,Gd-BT-DO3A

Raju et al. [22] 2010 Gd-DTPA, Gd-BOPTA, 150Gd-DTPA-BMA, Gd-DOTA,Gd-BT-DO3A

Telgmann et al. [15] 2012 Gd-DTPA, Gd-BOPTA, 820Gd-DTPA-BMA, Gd-DOTA,Gd-BT-DO3A

Birka et al. [23] 2013 Gd-DTPA, Gd-BOPTA, 80–100Gd-DOTA, Gd-BT-DO3A

Lindner et al. [24] 2014 Gd-DTPA, Gd-BOPTA, 330–970Gd-DTPA-BMA, Gd-DOTA,Gd-BT-DO3A

Lindner et al. [25] 2015 Gd-DTPA, Gd-BOPTA, 10–22

w1

2

2

dFQmSbdts(tpaio

Gd-DTPA-BMA, Gd-DOTA,Gd-BT-DO3A

as prepared. External calibration was carried out in a range from0 to 500 pmol/L.

.4. Instrumentation and experimental parameters

.4.1. ICP-MSThe total gadolinium concentration in the water samples was

etermined using a quadrupole-based iCAP Qc ICP-MS (Thermoisher Scientific, Bremen, Germany) that was controlled by thetegra software version 2.4 (Thermo Fisher Scientific). The instru-ent was equipped with a PFA MicroFlow nebulizer (Elemental

cientific, Omaha, NE, USA), a Peltier-cooled cyclonic spray cham-er (Thermo Fisher Scientific), a quartz injector pipe with an inneriameter of 3.5 mm and a SC-4-S autosampler (Elemental Scien-ific). The interface consisted of a nickel sampler and a nickelkimmer. Measurements were carried out in the standard modeSTD) without pressurizing the optional collision/reaction cell ofhe ICP-MS in order to achieve highest sensitivity. The following

arameters were used: RF power 1550 W, cool gas flow 13.8 L/min,uxiliary gas flow 0.8 L/min and nebulizer gas flow 1.0 L/min. Thesotopes 103Rh, 158Gd and 160Gd were monitored with a dwell timef 0.1 s each.

A 1440 (2016) 105–111

2.4.2. HILIC-ICP-MSThe chromatographic separation for speciation analysis was car-

ried out with an HPLC system (Shimadzu, Duisburg, Germany)consisting of two LC-10ADVP pumps, a SIL-10A autosampler, aDGC-14A degasser, a CTO-10ASVP column oven and a SCL-10AVPsystem controller. The system was controlled with the LCSolutionsoftware version 1.2.2 (Shimadzu). An YMC-Triart diol HILIC col-umn (YMC, Dinslaken, Germany) was used (150 mm length, 4.6 mminner diameter, 3 �m particle size) at a temperature of 30 ◦C, whilethe mobile phase consisted of 50 mmol/L aqueous ammonium for-mate adjusted to pH 3.7 as eluent A and acetonitrile as eluent B.The HILIC separation was carried out at a flow rate of 800 �L/minwithin 15 min in isocratic mode with 25% A and 75% B. The injectionvolume was 5 �L.

The HPLC was coupled to an ElementXR double focusing sectorfield ICP-MS (Thermo Fisher Scientific), which was controlled bythe Element software version 3.1.0.231 (Thermo Fisher Scientific).An U6000AT+ ultrasonic nebulizer (Teledyne CETAC Technologies,Omaha, NE, USA) was employed for sample introduction using aheater temperature of 120 ◦C and a cooler temperature of −5 ◦Cfor desolvation without the additional membrane desolvator. TheICP-MS instrument was equipped with a quartz injector pipe withan inner diameter of 1 mm and an interface consisting of a plat-inum sampler and a nickel skimmer. In order to achieve highestsensitivity due to maximum ion transmission the measurementswere carried out in E-scan with low resolution using the followingparameters: RF power 1500 W, cool gas flow 16 L/min, auxiliarygas flow 0.8 L/min, nebulizer gas flow 0.6 L/min and oxygen flow0.04 L/min. The oxygen flow was added after the ultrasonic neb-ulizer into the sample aerosol in order to prevent carbon buildupin the interface. The isotopes 156Gd, 157Gd, 158Gd and 160Gd weremonitored with a sample time of 0.01 s each. 100 samples perpeak were measured in a mass window of 10% for each iso-tope. The software Origin version 9.1.0 (OriginLab Corporation,Northampton, USA) was employed for data analysis. The HILIC-ICP-MS chromatograms based on the sum of intensities of the isotopesmentioned above were smoothed with the Savitzky–Golay algo-rithm using a second order polynomial regression over 90 datapoints.

3. Results and discussion

3.1. Method development

The determination of gadolinium species in drinking water is achallenging task especially considering the low concentrations tobe expected. Therefore, a novel HILIC-ICP-MS method was devel-oped combining a diol HILIC separation and improved ICP-MSdetection. The latter was achieved by ultrasonic nebulization andsubsequent desolvation for sample introduction. Since the nebu-lization efficiency directly affects the transport efficiency of theanalytes into the plasma, this has to be considered to be a cru-cial parameter regarding the sensitivity of a method. The efficiencyof ultrasonic nebulization and the resulting signal stability arestrongly depending on the flow rate of the introduced solution. Itcould be observed that it was possible to achieve a sufficiently sta-ble signal under HILIC conditions with a flow rate of 800 �L/min.Additionally, the employed double focusing sector field ICP-MSgenerally provides a high sensitivity if used in low resolution,because of the maximized ion transmission. The relatively highaccelerating voltage also improves sensitivity compared to most

quadrupole-based systems.

The employed diol HILIC method combines this high flow ratewith an isocratic separation of the most commonly applied contrastagents, which is shown in Fig. 3. As can be seen, it was possible

atogr.

tD5ssp

3

taLBGGAtiIbTfaGa1

3

tte

TSw

M. Birka et al. / J. Chrom

o separate and to detect the contrast agents Gd-BOPTA, Gd-BT-O3A, Gd-DOTA and Gd-DTPA in a concentration range from 10 to00 pmol/L. The relative standard deviation (RSD) of the baselineignal from a blank analysis was 5.4%. Thus, the sensitivity and theignal to noise ratio of this HILIC-ICP-MS method were sufficient toerform speciation analysis of gadolinium in drinking water.

.2. Figures of merit

The LODs were determined by means of signal to noise ratio ofhree (S/N = 3), whereas the LOQs were determined by means of

signal to noise ratio of ten (S/N = 10). For total gadolinium theOD was 0.6 pmol/L and the LOQ was 2.1 pmol/L. The LOD for Gd-T-DO3A was as low as 8 pmol/L, for Gd-DOTA 11 pmol/L and ford-DTPA 14 pmol/L. Thus, the respective LOQ was 26 pmol/L ford-BT-DO3A, 37 pmol/L for Gd-DOTA and 46 pmol/L for Gd-DTPA.s shown in Table 1, this is a significant improvement compared

o earlier studies involving the determination of contrast agentsn environmental samples and drinking water by means of HILIC-CP-MS. Linearity in the relevant concentration range was verifiedased on the triplicate measurement of the calibration solutions.he resulting regression coefficients were 0.999 for total Gd, 0.999or Gd-BT-DO3A, 0.999 for Gd-DOTA and 0.997 for Gd-DTPA. Theverage RSD of retention times was 0.2% for Gd-BT-DO3A, 0.6% ford-DOTA and 0.6% for Gd-DTPA. The column recovery for speciationnalysis was found to be 93% for Gd-BT-DO3A, 95% for Gd-DOTA and01% for Gd-DTPA.

.3. Analysis of water samples

The above described HILIC-ICP-MS method was employed forhe analysis of samples from different waterworks. Additionally,otal gadolinium concentrations were determined with ICP-MS forvaluation of the mass balance. The HILIC-ICP-MS chromatograms

able 2ampling points in the respective waterworks and quantification results. Gadolinium-baseas determined by means of ICP-MS.

Sample Gd-based contrast agents

Gd-BT-DO3A (pmol/L) Gd

Waterworks ASurface water <LOQ 55After filtration <LOQ 49Drinking water – <L

Waterworks BSurface water <LOQ <LGround filtrate <LOQ <LDrinking water after disinfection (UV irradiation) – –

Waterworks CSurface water 40 (±3) <LGround filtrate 30 (±4) <LDrinking water after disinfection (chlorine dioxide) <LOQ –

Waterworks DSurface water 30 (±5) 56Ground filtrate <LOQ 48After ozonation <LOQ <LAfter filtration <LOQ 40Drinking water after disinfection (UV irradiation) <LOQ 45

Waterworks ESurface water 49 (±14) 61Ground filtrate 26 (±6) 81Drinking water after disinfection (chlorine dioxide) <LOQ 74

Waterworks FSurface water 38 (±20) 85After ozonation <LOQ 46Ground filtrate <LOQ 37Drinking water after disinfection (chlorine dioxide) <LOQ 44

A 1440 (2016) 105–111 109

of the analyzed samples from waterworks F and a 500 pmol/L stan-dard solution are displayed in Fig. 4.

The chromatograms show that the contrast agents Gd-BT-DO3A,Gd-DOTA and Gd-DTPA could be detected in samples from differ-ent steps of the drinking water purification, while Gd-BOPTA wasnot detectable in any sample. The direct comparison of the sampleswith the standard indicates that the concentration of the individ-ual species could be expected to be found in the lower pmol/Lrange. The signal intensity of Gd-BT-DO3A and Gd-DOTA is slightlydecreased after ozonation, while it is constant for Gd-DTPA. There-fore, the HILIC-ICP-MS measurements demonstrate that the abovementioned contrast agents in the drinking water can be traced backthrough the whole purification process.

The results of the quantification of individual species by HILIC-ICP-MS and total gadolinium by ICP-MS are shown in Table 2,Figs. 5 and 6. It can be seen that individual contrast agents couldbe detected and quantified in the surface water, after all purifica-tion steps and in the disinfected drinking water. The comparisonof summed up concentrations of individual contrast agents and thetotal Gd concentrations in the surface water and drinking watersamples shows a gap in the mass balance. This gap is increasedin the drinking water mostly due to the extremely low concen-trations of Gd-BT-DO3A, which is found below the LOQ in mostsamples. In case of the surface water sample from the catchmentlake at waterworks A, the contrast agents represent 71% of the totalGd concentration. In the respective drinking water sample, thisamount was decreased to 63%. The contrast agents in the samplesfrom the Ruhr represent 87% up to 108% of the total gadolinium.There is also a larger gap in the mass balance of the drinking watersamples with an amount of 66–81%. Nevertheless, this cannot be

considered as evidence for the release of Gd3+ or a significant degra-dation of individual species, because Gd-BT-DO3A was quantifiedslightly above the LOQ or was found below the LOQ in most samples.

d contrast agents were determined by means of HILIC-ICP-MS and total gadolinium

Total Gd (pmol/L)

-DOTA (pmol/L) Gd-DTPA (pmol/L) � (pmol/L)

(±20) 161 (±1) 216 307 (±4) (±15) 167 (±8) 216 273 (±5)OQ 159 (±8) 159 254 (±3)

OQ 88 (±1) 88 82 (±1)OQ 89 (±5) 89 107 (±1)

82 (±22) 82 100 (±2)

OQ 93 (±6) 133 133 (±1)OQ 121 (±20) 151 130 (±1)

94 (±13) 94 123 (±2)

(±20) 116 (±4) 202 231 (±5) (±9) 128 (±22) 176 227 (±4)OQ 116 (±5) 116 206 (±8)

(±9) 109 (±13) 149 219 (±2) (±4) 110 (±27) 155 188 (±3)

(±2) 100 (±9) 210 242 (±6) (±5) 137 (±42) 244 274 (±7) (±24) 122 (±10) 196 298 (±1)

(±12) 139 (±8) 262 265 (±9) (±3) 140 (±43) 186 225 (±5) (±18) 124 (±25) 161 218 (±4) (±3) 103 (±20) 147 216 (±5)

110 M. Birka et al. / J. Chromatogr. A 1440 (2016) 105–111

Fig. 5. Quantification results of samples from waterworks A–C. Gadolinium-based contrast agents were determined by means of HILIC-ICP-MS and total gadolinium wasdetermined by means of ICP-MS.

F ts wem

oGGtbis

fawwwfiBs

ptoat

ig. 6. Quantification results of samples from D–F. Gadolinium-based contrast ageneans of ICP-MS.

As mentioned above, the raw water from waterworks A isbtained from a catchment lake. In this case, the contrast agentsd-DOTA and Gd-DTPA were quantified in the surface water, whiled-BT-DO3A was found to be below the LOQ. In the drinking water,

he concentration of Gd-DTPA was 167 pmol/L and Gd-DOTA waselow the LOQ. Total gadolinium concentrations were determined

n a range of 254 pmol/L in the drinking water to 307 pmol/L in theurface water.

In contrast to waterworks A, the further samples were collectedrom waterworks alongside the river Ruhr. Only the linear contrastgent Gd-DTPA could be quantified in the samples from water-orks B with a concentration of 82 pmol/L in the drinking water,hich was disinfected by UV irradiation. The results of water-orks C, some kilometers downstream the river Ruhr, confirm thisnding. In addition to Gd-DTPA, the macrocyclic contrast agent Gd-T-DO3A was quantified with a concentration of 40 pmol/L in theurface water and 30 pmol/L in the ground filtrate.

Waterworks D is located in more densely populated region com-ared to waterworks B and C. As a result of that, there is an increasedotal gadolinium concentration as well as increased concentrations

f individual species in the surface water, exposing a significantnthropogenic input. Again, this amount of contrast agents andhereby the anthropogenic input was also found in the drinking

re determined by means of HILIC-ICP-MS and total gadolinium was determined by

water. In the surface water, the concentration was 30 pmol/L forGd-BT-DO3A, 56 pmol/L for Gd-DOTA and 116 pmol/L for Gd-DTPA.Nearly similar concentrations were found in the drinking waterafter disinfection by means of UV irradiation for Gd-DOTA and Gd-DTPA, while Gd-BT-DO3A was below the LOQ. The total gadoliniumconcentration was 188 pmol/L in the drinking water.

The analysis of the samples from waterworks E also shows highconcentrations of total gadolinium and contrast agents comparedto the waterworks B and C upstream the Ruhr. A total gadoliniumconcentration of 242 pmol/L was determined in the surface water,whereas it was 274 pmol/L in the ground filtrate and 298 pmol/L inthe drinking water after addition of chlorine dioxide. The concen-tration in the surface water was 100 pmol/L for Gd-DTPA, 61 pmol/Lfor Gd-DOTA and 49 pmol/L for Gd-BTDO3A. In the drinking waterafter disinfection by chlorine dioxide, the concentration of Gd-BT-DO3A was below the LOQ, while Gd-DTPA and Gd-DOTA werequantified in a concentration of 110 pmol/L and 45 pmol/L, respec-tively.

Since Waterworks F is located in close vicinity of waterworksE, the initial amount of anthropogenic gadolinium in the surface

water was also increased compared to the waterworks upstream.The resulting concentration of the most abundant contrast agentGd-DTPA was 103 pmol/L in the drinking water. Therefore, the

atogr.

aricp

4

sHBabwwTpgwc

popfwaTsdespa

A

WGWp

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

M. Birka et al. / J. Chrom

nalysis of samples from the waterworks E and F substantiates theesults from waterworks D, indicating an increased anthropogenicnput of gadolinium downstream the Ruhr. The concentrations ofontrast agents were also clearly increased in the drinking waterroduced from the related waterworks.

. Conclusions

In this work, a new highly sensitive HILIC-ICP-MS method forpeciation analysis of gadolinium was presented, involving a diolILIC column and sample introduction by ultrasonic nebulization.ecause of the resulting superior LODs and LOQs, the frequentlypplied contrast agents Gd-DTPA, Gd-DOTA and Gd-BT-DO3A coulde detected and quantified in samples from the process of drinkingater purification. Furthermore, the linear contrast agent Gd-DTPAas by far the most abundant species to be detected in all samples.

here was no evidence for additional Gd species or transformationroducts being present in the analyzed samples. Since only sin-le samples were taken and considering the standard deviations asell, an interpretation of the data regarding a possible removal of

ontrast agents is not possible.It was possible to demonstrate the direct impact of anthro-

ogenic gadolinium species in the surface water on the amountf anthropogenic gadolinium in the drinking water in a denselyopulated region of Germany at the river Ruhr. In the waterworksurther downstream the river, the analysis of the produced drinkingater revealed higher concentrations of individual contrast agents

s well as total gadolinium compared to the waterworks upstream.his finding directly correlates with the respective surface wateramples. In case of the samples from a catchment lake and therinking water from this lake, there was a similar correlation. Nev-rtheless, it is necessary to conduct further investigations on a lagercale, especially regarding the time dependency of sampling. Theresented methods provide the capability to perform such studiesnd serve as a basis for further method development.

cknowledgements

The authors would like to thank Dr. Claus Schlett from theestfälische Wasser- und Umweltanalytik GmbH as well as the

ELSENWASSER AG, the Wasserwerke Westfalen GmbH and theassergewinnung Essen GmbH for collecting and providing sam-

les.

eferences

[1] J.M. Idée, M. Port, I. Raynal, M. Schaefer, S. Le Greneur, C. Corot, Clinical andbiological consequences of transmetallation induced by contrast agents formagnetic resonance imaging: a review, Fundam. Clin. Pharmacol. 20 (2006)563–576.

[2] P. Caravan, J.J. Ellison, T.J. McMurry, R.B. Lauffer, Gadolinium(III) chelates asMRI contrast agents: structure, dynamics, and applications, Chem. Rev. 99(1999) 2293–2352.

[3] S. Haneder, W. Kucharczyk, S.O. Schoenberg, H.J. Michaely, Safety of magneticresonance contrast media: a review with special focus on nephrogenicsystemic fibrosis, Top. Magn. Reson. Imaging 24 (2015) 57–65.

[4] M. Bau, P. Dulski, Anthropogenic origin of positive gadolinium anomalies inriver waters, Earth Planet, Sci. Lett. 143 (1996) 245–255.

[5] M. Bau, A. Knappe, P. Dulski, Anthropogenic gadolinium as a micropollutant inriver waters in Pennsylvania, and in Lake Erie, northeastern United States,Chem. Erde—Geochem. 66 (2006) 143–152.

[6] A. Knappe, P. Möller, P. Dulski, A. Pekdeger, Positive gadolinium anomaly insurface water and ground water of the urban area Berlin, Germany, Chem.Erde—Geochem. 65 (2005) 167–189.

[7] F. Elbaz-Poulichet, J.L. Seidel, C. Othoniel, Occurrence of an anthropogenicgadolinium anomaly in river and coastal waters of Southern France, Water

Res. 36 (2002) 1102–1105.

[8] M. Rabiet, F. Brissaud, J.L. Seidel, S. Pistre, F. Elbaz-Poulichet, Positivegadolinium anomalies in wastewater treatment plant effluents and aquaticenvironment in the Hérault watershed (South France), Chemosphere 75(2009) 1057–1064.

A 1440 (2016) 105–111 111

[9] Y. Nozaki, D. Lerche, D.S. Alibo, M. Tsutsumi, Dissolved indium and rare earthelements in three Japanese rivers and Tokyo Bay: evidence for anthropogenicGd and In, Geochim, Cosmochim. Acta 64 (2000) 3975–3982.

10] Y.B. Zhu, R. Hattori, D. Rahmi, S. Okuda, A. Itoh, E. Fujimori, T. Umemura, H.Haraguchi, Fractional distributions of trace metals in surface water of LakeBiwa as studied by ultrafiltration and ICP-MS, Bull. Chem. Soc. Jpn. 78 (2005)1970–1976.

11] P.L. Verplanck, H.E. Taylor, D.K. Nordstrom, L.B. Barber, Aqueous stability ofgadolinium in surface waters receiving sewage treatment plant effluent,Boulder Creek, Colorado, Environ. Sci. Technol. 39 (2005) 6923–6929.

12] M.G. Lawrence, S.D. Jupiter, B.S. Kamber, Aquatic geochemistry of the rareearth elements and yttrium in the Pioneer River catchment, Australia, Mar.Freshwater Res. 57 (2006) 725–736.

13] S. Kulaksiz, M. Bau, Contrasting behaviour of anthropogenic gadolinium andnatural rare earth elements in estuaries and the gadolinium input into theNorth Sea, Earth Planet. Sci. Lett. 260 (2007) 361–371.

14] P.L. Verplanck, E.T. Furlong, J.L. Gray, P.J. Phillips, R.E. Wolf, K. Esposito,Evaluating the behavior of gadolinium and other rare earth elements throughlarge metropolitan sewage treatment plants, Environ. Sci. Technol. 44 (2010)3876–3882.

15] L. Telgmann, C.A. Wehe, M. Birka, J. Künnemeyer, S. Nowak, M. Sperling, U.Karst, Speciation and isotope dilution analysis of gadolinium-based contrastagents in wastewater, Environ. Sci. Technol. 46 (2012) 11929–11936.

16] S. Kulaksiz, M. Bau, Anthropogenic gadolinium as a microcontaminant in tapwater used as drinking water in urban areas and megacities, Appl. Geochem.26 (2011) 1877–1885.

17] A. Mazzucotelli, V. Bavastello, E. Magi, P. Rivaro, C. Tomba, Analysis ofgadolinium polyaminopolycarboxylic complexes by HPLC-ultrasonicnebulizer-ICP-AES hyphenated technique, Anal. Proc. 32 (1995) 165–167.

18] V. Loreti, J. Bettmer, Determination of the MRI contrast agent Gd-DTPA bySEC-ICP-MS, Anal. Bioanal. Chem. 379 (2004) 1050–1054.

19] P. Pfundstein, C. Martin, W. Schulz, K.M. Ruth, A. Wille, T. Moritz, A. Steinbach,D. Flottmann, IC-ICP-MS analysis of gadolinium-based MRI contrast agents,LC-GC N. Am. (2011) 27–29.

20] J. Künnemeyer, L. Terborg, S. Nowak, A. Scheffer, L. Telgmann, F. Tokmak, A.Günsel, G. Wiesmüller, S. Reichelt, U. Karst, Speciation analysis ofgadolinium-based MRI contrast agents in blood plasma by hydrophilicinteraction chromatography/electrospray mass spectrometry, Anal. Chem. 80(2008) 8163–8170.

21] J. Künnemeyer, L. Terborg, B. Meermann, C. Brauckmann, I. Möller, A. Scheffer,U. Karst, Speciation analysis of gadolinium chelates in hospital effluents andwastewater treatment plant sewage by a novel HILIC/ICP-MS method,Environ. Sci. Technol. 43 (2009) 2884–2890.

22] C.S.K. Raju, A. Cossmer, H. Scharf, U. Panne, D. Lück, Speciation of gadoliniumbased MRI contrast agents in environmental water samples using hydrophilicinteraction chromatography hyphenated with inductively coupled plasmamass spectrometry, J. Anal. At. Spectrom. 25 (2010) 55–61.

23] M. Birka, C.A. Wehe, L. Telgmann, M. Sperling, U. Karst, Sensitivequantification of gadolinium-based magnetic resonance imaging contrastagents in surface waters using hydrophilic interaction liquid chromatographyinductively coupled plasma sector field mass spectrometry, J. Chromatogr. A1308 (2013) 125–131.

24] U. Lindner, J. Lingott, S. Richter, N. Jakubowski, U. Panne, Speciation ofgadolinium in surface water samples and plants by hydrophilic interactionchromatography hyphenated with inductively coupled plasma massspectrometry, Anal. Bioanal. Chem. 405 (2013) 1865–1873.

25] U. Lindner, J. Lingott, S. Richter, W. Jiang, N. Jakubowski, U. Panne, Analysis ofGadolinium-based contrast agents in tap water with a new hydrophilicinteraction chromatography (ZIC-cHILIC) hyphenated with inductivelycoupled plasma mass spectrometry, Anal. Bioanal. Chem. 407 (2014)2415–2422.

26] J.E. Carr, K. Kwok, G.K. Webster, J.W. Carnahan, Effects of liquidchromatography mobile phases and buffer salts on phosphorus inductivelycoupled plasma atomic emission and mass spectrometries utilizing ultrasonicnebulization and membrane desolvation, J. Pharm. Biomed. Anal. 40 (2006)42–50.

27] C.W. Huang, S.J. Jiang, Speciation of mercury by reversed-phaseliquid-chromatography with inductively-coupled plasma-mass spectrometricdetection, J. Anal. At. Spectrom. 8 (1993) 681–686.

28] R.D. Wilken, R. Falter, Determination of methylmercury by thespecies-specific isotope addition method using a newly developed HPLC-ICPMS coupling technique with ultrasonic nebulization, Appl. Organomet. Chem.12 (1998) 551–557.

29] B. Gammelgaard, O. Jons, Comparison of an ultrasonic nebulizer with across-flow nebulizer for selenium speciation by ion-chromatography andinductively coupled plasma mass spectrometry, J. Anal. At. Spectrom. 15(2000) 499–505.

30] K. Falk, H. Emons, Speciation of arsenic compounds by ion-exchangeHPLC-ICP-MS with different nebulizers, J. Anal. At. Spectrom. 15 (2000)643–649.

31] C.J. Smith, I.D. Wilson, F. Abou-Shakra, R. Payne, T.C. Parry, P. Sinclair, D.W.

Roberts, A comparison of the quantitative methods for the analysis of theplatinum-containing anticancer drug{cis-[amminedichloro(2-methylpyridine)]platinum(II)} (ZD0473) by HPLCcoupled to either a triple quadrupole mass spectrometer or an inductivelycoupled plasma mass spectrometer, Anal. Chem. 75 (2003) 1463–1469.