Quantification of human pharmaceuticals in water

0 downloads 0 Views 450KB Size Report
Reagents. Methanol (HPLC-grade) and acetone (glass-distilled grade) .... Method validation ... samples were transferred into clean amber glass bottles. Each.
Analytica Chimica Acta 598 (2007) 87–94

Quantification of human pharmaceuticals in water samples by high performance liquid chromatography–tandem mass spectrometry Carolina Nebot ∗ , Stuart W. Gibb, Kenneth G. Boyd Environmental Research Institute (ERI), North Highland College, UHI Millennium Institute, Castle Street, Thurso, Caithness, Scotland KW14 7JD, UK Received 24 April 2007; received in revised form 10 July 2007; accepted 12 July 2007 Available online 14 July 2007

Abstract An improved analytical method for determination of human pharmaceuticals in natural and wastewaters with ng L−1 sensitivity is presented. The method is applicable to pharmaceuticals from a wide range of therapeutic classes including antibiotics, analgesics, anti-inflammatories and anti-cancer compounds. Pharmaceuticals were extracted from waters using solid-phase extraction, and after concentration, analysed by high performance liquid chromatography with tandem mass spectrometric detection (HPLC–MS/MS). Identification of each compound was secured using retention time and by the selected reaction monitoring of two transitions, one of which was additionally used for quantification. Limits of detection ranged from 0.03 to 0.96 ng L−1 and were up to two orders of magnitude lower than those of previously published methods. The method was validated using spiked samples prepared from tap, river and sea water as well as wastewater effluents, collected from the North of Scotland. Analysis of wastewater effluents revealed the presence of mefenamic acid, ibuprofen, erythromycin, diclofenac and trimethoprim. None of the selected pharmaceuticals were detected in river, tap or sea water samples. © 2007 Elsevier B.V. All rights reserved. Keywords: Pharmaceuticals; High performance liquid chromatography–tandem mass spectrometry; Solid-phase extraction; Natural waters

1. Introduction In contemporary society, the use of human pharmaceuticals (pharmaceuticals that are consumed by humans) is both extensive and widespread. Three thousand different active ingredients are licensed for use as pharmaceuticals in the UK [1], and the number of prescriptions increases every year, e.g. in Scotland, the number of prescription dispensed increased 22.5% from 2001 to 2006 [2]. Following administration, human pharmaceuticals may pass through the body and be introduced into the domestic wastewater system via urine and faeces in either metabolised or unmetabolised forms [3]. The direct disposal of unused pharmaceuticals through domestic wastewater is also commonplace. In recent years, a growing number of pharmaceuticals have been detected in final effluents from wastewater treatment plants (WWTPs) indicating incomplete removal or degradation of



Corresponding author. Tel.: +44 1847 889596; fax: +44 1847 890014. E-mail address: [email protected] (C. Nebot).

0003-2670/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2007.07.029

these bio-active compounds. For example, carbamazepine has been detected in final effluents in Germany at 6.3 ng L−1 [4], diclofenac in France at 0.41 ng L−1 [3], salicylic acid in Spain at 570 ng L−1 [5], clofibric acid in Norway at 0.17 ng L−1 [6] and ibuprofen in Canada at 0.48 ng L−1 [7]. Final effluents from WWTPs have been identified as the principal pathway for the introduction of these compounds into the natural aquatic environment [4]. Since by definition, pharmaceutical compounds are ‘designed’ to produce biological effects, their impact on non-target organisms in the aquatic environment is of growing concern [8,9]. In the UK, the Environment Agency of England and Wales have ranked a sub-set of pharmaceuticals according to their potential risk to the aquatic environment (Table 1). Sensitive and reliable analytical techniques are therefore required to provide a robust assessment of the presence and potential risk of human pharmaceuticals in natural waters [9]. The analysis of pharmaceuticals in aqueous media has generally been performed by gas chromatography (GC) with either mass spectrometric (MS) [6,7,10–12] or photodiode array detec-

88

C. Nebot et al. / Analytica Chimica Acta 598 (2007) 87–94

Table 1 Pharmaceuticals selected for this investigation and their application

2.2. Standards

Name

Application [18]

pKa

Paracetamol Trimethoprim Sulfamethoxazole Propanolol Erythromycin Dextropropoxyphene Tamoxifen Lofepramine Diclofenac Mefenamic acid Ibuprofen Clofibric acid

Analgesic Antibiotic Antibiotic Anti-hypertensive Antibiotic Analgesic Anti-cancer Anti-depressant Anti-inflammatory Anti-inflammatory Analgesic Metabolite of lipid regulators

9 7.12 6.1 9.5 8.8 6.3 4 5.5 8.48 4.2 4.4 2.95

tion [3]. In the case of polar pharmaceuticals, such as diclofenac, ibuprofen and clofibric acid derivatization is required prior to GC–MS analysis [7,13,14]. Recent developments in MS have resulted in the introduction of atmospheric pressure ionization (API) and the ability to ionise analytes in the liquid-phase allowing efficient coupling of high performance liquid chromatography (HPLC) and tandem mass spectrometry (MS/MS). As a result the analysis of polar compounds and those with high molecular weight (>900 amu) can now be achieved using HPLC–MS/MS. Consequently, a number of HPLC–MS/MS methods have been proposed for the analysis of human pharmaceuticals in natural and wastewaters [5,15–20]. Here, we present an improved analytical method, which combines solid-phase extraction (SPE) with HPLC–MS/MS to provide a more rapid and sensitive analysis of 11 target human pharmaceuticals in natural and wastewaters. Evaluation of the method on a range of natural and synthetic samples (natural water and wastewater effluent spiked with pharmaceuticals to a known concentration) is described together with (what we understand to be) the first results reporting the analysis of human pharmaceuticals in natural and wastewater samples in Scotland. 2. Experimental 2.1. Reagents Methanol (HPLC-grade) and acetone (glass-distilled grade) were obtained from Rathburn Chemicals Ltd. (Walkerburn, UK); hydrochloric acid (HCl, AR grade), ammonia solution 28–30% (GR for analysis ACR) and acetic acid (GPR grade) from VWR International (Lutterworth, UK), ammonium acetate (GR for analysis ACR) from Merck Pharmaceuticals (West Drayton, UK). Diclofenac (sodium salt), clofibric acid, erythromycin, ibuprofen, mefenamic acid, acetaminophen, (±)-propanolol hydrochloride, sulfamethoxazole, tamoxifen and trimethoprim were obtained from Sigma–Aldrich Co. Ltd. (Dorset, UK), while lofepramine and dextropropoxyphene hydrochloride were obtained from British Pharmacopoeia Commission Laboratory (Teddington, UK). All compounds were of a purity >95%. MilliQ water was used unless otherwise stated.

Standard solutions were prepared in a 50:50 mix of methanol:10 mM ammonium acetate at pH 6 (pH was regulated using 0.1 M HCl). Standards of concentration 2000 ␮g L−1 every 6 months. All standards were stored in the dark at −18 ◦ C. A standard solution containing 20 mg L−1 of the target pharmaceuticals was diluted for the preparation of standards; a 1000 ␮g L−1 solution was used for MS tuning, a 200 ␮g L−1 solution for recovery studies and solutions containing 1, 2, 10, 50 and 250 ␮g L−1 for calibrations. 2.3. Equipment The HPLC–MS/MS system consisted of an Alliance 2695 HPLC from (Waters; Manchester, UK) and a Waters Micromass® Quattro MicroTM detector with electrospray ionization (ESI). Data acquisition and control were carried out using MasslynxTM NT software (Waters; Milford, MA, USA). Analytes were separated on a 250 mm × 10 mm C18 Luna® column (10 ␮m particle size; Phenomenex; Macclesfield, UK) used in conjunction with a 4.0 mm × 2.0 mm C18 guard column (Phenomenex; Macclesfield, UK). Evaporation of extracts was performed with a turboevaporator (Turbo Vap® II from Zyrmark (Hopkinton, MA, USA)) with nitrogen gas (supplied by nitrogen generator; In House Gas, Killearn, UK) model N2 MaxiFlow 60 L, purity 99%). 2.4. Solid-phase extraction (SPE) StrataTM X SPE (reverse phase; polymers; 200 mg; 6 mL; Phenomenex, Macclesfield, UK) was conditioned with methanol (5 mL) and deionised water (5 mL). Samples were transferred through the cartridge at rate of 5 mL min−1 without allowing the cartridge to dry out. After loading, cartridges were dried for 30 min using the vacuum of the SPE manifold. Empty sample vessels were rinsed with acetone (5 mL) and the rinse was added to the SPE column followed by two 5 mL aliquots of methanol. The 15 mL eluate was collected in a glass tube and evaporated to ∼0.1 mL at 55 ◦ C, volume was made up to 0.3 mL with 50:50 mix of methanol:10 mM ammonium acetate at pH 6. Final extracts were transferred directly into amber auto-sampler vials (2 mL, containing 0.5 mL insert vials) and stored at −18 ◦ C prior to analysis by HPLC–MS/MS. 2.5. HPLC–MS/MS for determination of pharmaceuticals Two HPLC methods were used for the analysis of the selected pharmaceuticals: ‘Method A’ was used for the analysis of nine of the pharmaceuticals (paracetamol, trimethoprim, sulfamethoxazole, propanolol, erythromycin, dextropropoxyphene, tamoxifen, diclofenac and mefenamic acid) and ‘Method B’ for ibuprofen and clofibric acid. Both methods used water, methanol, ammonium acetate (10 mM, adjusted to pH 6.0 with 0.1 M HCl) and acetic acid (0.87 M) as

C. Nebot et al. / Analytica Chimica Acta 598 (2007) 87–94 Table 2 Mobile phase compositions for the two HPLC methods used Time (min)

Water (%)

Methanol (%)

Method A (flow 0.17 mL min−1 ) 0 87.0 10.0 3 87.0 10.0 12 30.8 66.0 26 0.0 96.8 31 0.0 96.8 32 86.8 10.0 43 86.8 10.0

10 mM ammonium acetate (%)

0.87 M acetic acid (%)

3.0 3.0 3.0 3.0 3.0 3.0 3.0

0.0 0.0 0.2 0.2 0.2 0.2 0.2

1.0 1.0 1.0 1.0 1.0 1.0

0.2 0.2 0.2 0.2 0.2 0.2

89

fragmentation of the ions, and each pharmaceutical required a specific value of the collision gas energy for its fragmentation (Table 3), the collision energy was changing during data acquisition. Pharmaceuticals were identified by their retention time (Rt ) and selected reaction monitoring (SRM) of two transitions (with the exception of ibuprofen where one SRM transition was used), and quantified using SRM of one transition (Table 3). 2.6. Calibration curves

0.2 mL min−1 )

Method B (flow 0 53.8 3 53.8 6 0.8 11 0.8 17 53.8 20 53.8

45.0 45.0 98.0 98.0 45.0 45.0

mobile phase components, the flow was 0.17 mL min−1 (Method A) and 0.2 mL min−1 (Method B) and the gradients used are presented in Table 2. Auto-sampler tray and column heater temperatures were maintained constant during the analysis at 5 and 25 ◦ C, respectively. Injection volume was 30 ␮L. Positive ionisation mode was used with ‘Method A’ and negative with ‘Method B’. In both ‘Method A’ and ‘Method B’ the following detector parameters were applied: Extractor voltage, 2.20 V; radio frequency lens, 0.20 V; source temperature, 120 ◦ C; low mass (LM) 1 resolution, 12.70; high mass (HM) 1 resolution 1, 12.70; ion energy 1, 0.40; entrance, 0.00; exit, 1.00; LM 2 resolution, 13.20; HM 2 resolution 1, 13.20; ion energy 2, 1.00; multiplier, 650.00 V. In ‘Method A’ capillary voltage was 3.20 kV, desolvation temperature 400 ◦ C, cone gas flow 71.00 L h−1 and desolvation gas flow 420.00 L h−1 . In ‘Method B’ capillary voltage was 1.20 kV, desolvation temperature 200.0 ◦ C, cone gas flow 30.00 L h−1 and desolvation gas flow 256.00 L h−1 . Argon was used for the

Calibration curves for each of the pharmaceuticals were performed daily using standard solutions at concentrations of 1, 2, 10, 50, 250 ␮g L−1 prepared in 50:50 methanol:10 mM ammonium acetate at pH 6. Linear regression coefficients (R2 ) were >0.99 for all compounds with the exception of clofibric acid (R2 ) = 0.96. The MS/MS detector was maintained according to manufacturer’s specifications and cleaned regularly, but when changes in the slopes of the calibration curves were observed more than 50%, the detector received additional cleaning. Quality control standards, standard solutions of mixed pharmaceuticals at 50 ␮g L−1 were analysed after every 12 samples during analysis. 2.7. Optimisation of extraction procedure Recoveries of pharmaceuticals using StrataTM X (Hilton and Thomas protocol [17]) and OasisTM (200 mg; 6 mL; Waters; Milford, MA, USA) cartridges using the protocol of Lee et al. [7] were compared. The procedure reported by Hilton and Thomas [17], was used as the basis for the methodology reported here. However, a number of studies were carried out to optimise recoveries using StrataTM X cartridges including evaluation of SPE eluents, and procedures for washing, drying and storage of the cartridges prior to use.

Table 3 Cone voltage, collision energy, precursor ion and poducts ions used for the MS/MS detection of the selected pharamaceuticals Compounds

Cone voltage (V)

Collision energy (eV)

Precursor ion (m/z)

Product ion 1 (m/z)

Product ion 2 (m/z)

Positive ionization mode (M+H)+ Paracetamol 29 Trimethoprim 33 Sulfamethoxazole 22 Propanolol 28 Erythromycin 31 Dextropropoxyphene 15 Tamoxifen 30 Lofepramine 27 Diclofenac 20 Mefenamic acid 17

20 25 25 23. 36 13 27 21 15 17

152.1 291.1 254.1 260.2 734.2 340.1 372.1 419.0 296.1 242.2

110.1 [M+H-C2 H2 O] 123.1 [M+H-C9 H12 O3 ] 91.5 [M+H-C4 H5 N2 O3 S] 116.1 [M+H-C10 H8 O] 158.3 [M+H-C29 H52 O11 ] 58.0 [M+H-C19 H22 O2 ] 72.0 [C4 H10 N] 224.1 [M+H-C14 H13 N] 250.2 [M+H-CO2 ] 224.3 [M+H-H2 O]

93.0 [M+H-C2 H5 NO] 230.3 [M+H-2OCH3 ] 107.7 [M+H-C4 H5 N2 O2 S] 157.2 [M+H-C5 H13 NO] 116.2 [M+H-C31 H55 NO11 ] 266.3 [M+H-C3 H6 O2 ] 129.2 196.0 [M+H-C16 H16 N] 278.2 [M+H-H2 O] 209.3 [M+H-H2 O-CH3 ]

Negative ionization mode (M−H)− Ibuprofen 15 Clofibric acid 17

8 12

205.00 212.90

161.1 [M−H-CO2 ] 126.6 [M−H-CO2 -(C4 H6 O2 )]

158.9 [M−H-CO-H2 O] 84.6 [M−H-C5 H5 O2 ]

Note: product ion 2 of ibuprofen can only be observed at high concentration (∼100 ␮g L−1 ).

90

C. Nebot et al. / Analytica Chimica Acta 598 (2007) 87–94

2.8. Optimisation of MS detection

2.11. Sample storage experiment

Standard solutions of individual compounds were infused directly into the MS to optimise cone voltage and collision gas energy for each compound (Table 3). The optimum cone voltage was that which generated the most intense signal of a selected precursor ion. The optimum collision gas energy was that which gave the most intense signal of product ions resulting from fragmentation of the selected precursor ion.

Individual tap, river, sea and wastewater effluent samples (1.0 L) were spiked to a final concentration of 10 ␮g L−1 in each of the target pharmaceuticals. Each sample was divided into 10 sub-samples (0.1 L), and stored at 5 ◦ C in the dark. Pharmaceuticals were extracted following 1, 2, 3, 7 and 10 days of storage and recoveries calculated.

2.9. Reproducibility, limits of detection

2.12. Sample collection

A standard mix of pharmaceuticals (25 ␮g L−1 in each pharmaceutical) was injected three consecutive times to calculate reproducibility (R.S.D., %) of Rt s for each pharmaceutical. Reproducibility of response (peak area R.S.D., %) was calculated from the average of three replicate injections from each of three replicate vials of mixed standards containing 0, 1, 5, 25, 150 ␮g L−1 of each compound. Limit of detection (LOD) was determined as the lowest concentration which gave a signal to noise ratio (S/N) ≥ 3 and limit of quantification (LOQ) as the lowest concentration which gave an S/N ≥ 10.

Once the optimum conditions for recovery were established, the method was applied to the determination of pharmaceuticals in 2.5 L samples collected from effluent arising from a wastewater plant, a domestic tap water supply, a river and the sea, all in the North Scotland. All samples were collected in between the 14th and 28th of November 2005.

2.10. Method validation 15 L of samples were collected from effluent arising from a wastewater plant, a domestic tap water supply, a river and the sea. 2.0 and 0.1 L of each type of water was measured and transferred to clean labelled containers. For the preparation of the synthetic samples, each 12 L of sample was spiked with pharmaceuticals to a final concentration of 10 ␮g L−1 of each pharmaceutical. From each 12 L bulk sample, four 2.0 L and three 0.1 L subsamples were transferred into clean amber glass bottles. Each of the 2.0 and 0.1 L samples, where pharmaceuticals were not spiked, were used as blank during the extraction process. The 2.0 and 0.1 L samples were filtered through GF/F glass microfibre filters (Whatman; Middlesex, UK) using either vacuum filtration (0.1 L samples) or positive pressure filtration (2.0 L samples). After filtration, the sample volume was remeasured (for recovery calculations), the pH adjusted to 6 (with 0.1 M HCl or 0.1 M NH3(aq) ) and pharmaceuticals were extracted using StrataTM X cartridges and analysed by HPLC–MS/MS as described above. Filters were stored in amber glass bottles (60 mL) at 5 ◦ C prior to extraction as described below. Residual pharmaceuticals were extracted from the GF/F filters by shaking with 20 mL of dichloromethane for 2 min. The resultant solution was transferred to a graduated glass tube (12 mL; Fisher Scientific Leicestershire, UK). Extracts were concentrated to dryness at 35 ◦ C and the volume made up to 0.2 mL with a 50:50 methanol:10 mM of ammonium acetate at pH 6. Final extracts were transferred directly into autosampler vials (2 mL, amber) and stored at −18 ◦ C prior to analysis by HPLC–MS/MS. Final concentrations were calculated by summing the total mass of analytes quantified in extracts of the SPE cartridges and from the filters in relation to the volume of the sample.

3. Results and discussion 3.1. Optimisation of mass spectrometry Erythromycin, diclofenac, trimethoprim, paracetamol, propanolol, mefenamic acid, sulfamethoxazole, tamoxifen, dextropropoxyphene and lofepramine were detected in positive ionisation mode. Clofibric acid and ibuprofen possess acid groups and lack basic sites making ionisation ineffective in the positive mode (even at cone voltages up to 45 V) and were instead detected in the negative ionisation mode [5,17]. Although it is possible to operate the mass spectrometer in a switching mode, whereby the ionisation alternates between positive and negative modes throughout the analysis, the use of this function results in decreased sensitivity. Furthermore, the rapid switching of polarity can reduce the lifespan of the capillary of the ionisation probe. 3.2. HPLC optimisation To improve chromatographic reproducibility and resolution and to enhance the sensitivity of detection, the following procedures were evaluated using ‘Method A’: 3.2.1. Buffer addition In order to obtain reproducible retention times, the use of buffer was required. However, the addition of buffer has the tendency to lower the signal intensity, due to the suppression of ionisation at MS interface [10]. Ammonium acetate was selected since its signal suppression has been shown to be lower than that of other buffers [6] and has been used previously for pharmaceutical analysis [17,20]. Together with ammonium acetate, 0.87 M acetic acid in water was also introduced to the mobile phase. Acetic acid reduced co-elution of peaks more than other acids tested including formic acid (1%, v/v) and phosphoric acid (5%, v/v).

C. Nebot et al. / Analytica Chimica Acta 598 (2007) 87–94

91

3.2.2. pH optimisation The effect of pH on the Rt s and reproducibility of Rt s of target pharmaceuticals was evaluated through analysis of mixed standards at 100 ␮g L−1 in selected pharmaceuticals in the pH range 2–8 (pH adjusted with 0.1 M NH3(aq) or 0.1 M HCl). The Rt s for paracetamol and tamoxifen showed no dependency on pH. However, for compounds with high pKa values (Table 1) (dextropropoxyphene, erythromycin, propanolol and trimethoprim), Rt s increased with pH (Fig. 1). In contrast, for compounds with low pKa values (diclofenac, mefenamic acid, sulfamethoxazole, ibuprofen and clofibric acid), Rt s decreased with increasing pH. Optimal elution conditions, i.e. those, which resulted in well-resolved peaks within a reasonable run-time, were found to lie in the pH range 5–6. 3.3. Extraction improvement Recoveries of pharmaceuticals using OasisTM cartridges were higher than 50%, where they could be detected but tamoxifen, trimethoprim, propanolol and dextropropoxyphene could not be detected after extraction. Using StrataTM X cartridges, recoveries were significantly different than those published [17], i.e. ∼20% lower of paracetamol, trimethoprim, sulfamethoxazole, ibuprofen and clofibric acid, however, all pharmaceuticals were detected. Steps were therefore undertaken to optimise recovery. Hilton and Thomas [17] suggested the use of methanol and water at pH 3 in pre-conditioning the SPE columns. In this work, a range of organic solvents was evaluated for this step, including acetone, methanol and hexane (results not shown). In agreement with ref. [17], optimum results were obtained using 5 mL methanol followed by 5 mL of deionised water. However, in this work, pH 6 was found to be optimal. Different elution solvents were also tested including acetone only; acetone followed by dichloromethane; 2% acetic acid in methanol; acetone followed by methanol; ethyl acetate only and methanol only (Table 4). Highest recoveries were obtained using acetone followed by methanol for paracetamol, sulfamethoxa-

Fig. 1. The dependency of analyte retention times (Rt s) on extract pH.

zole, dextropropoxyphene, propanolol, diclofenac and clofibric acid. For trimethoprim, recoveries were comparable whether or not acetone was used in conjunction with methanol (Table 4). Washing the SPE columns prior to elution has been shown to remove interfering compounds [6,7,12,13,16]. However,

Table 4 Recoveries (%) of human pharmaceuticals from spiked water samples using different solvents (for elution) and processes on StrataTM X cartridges

Paracetamol Trimethoprim Sufamethoxazole Erythromycin Detropropoxyphene Tamoxifen Propanolol Diclofenac Mefenamic acid Ibuprofen Clofibric acid

Acetonea

Acetone and DCMa

Methanol (acidified)a

Ethylacetatea

Methanolb

Acetone and methanolb

Washed columnc

Dried columna

Column in freezerd

34 78 27 63 30 99 62 56 108 112 52

1.2 87 0.0 35 32 139 51 45 73 105 53

25 52 201 0.0 64 110 66 82 94 165 187

62 41 63 0.0 63 0.0 8.2 61 31 130 82

69 97 63 65 34 64 68 77 93 100 73

105 97 105 31 58 29 73 98 49 79 110

90 89 98 58 55 33 81 91 46 82 86

90 81 92 63 66 39 74 87 44 83 81

96 81 83 48 44 30 66 119 60 71 72

Notes: all data shown are the mean of six replicates. DCM: dichoromethane. a Column not washed, but dried, not stored before elution. b Column not washed or dried, not stored before elution. c Column washed, not dried, not stored before elution. d Columns not washed or dried, columns stored at −80 ◦ C before elution.

92

C. Nebot et al. / Analytica Chimica Acta 598 (2007) 87–94

Table 5 Average R.S.D. (%) for retention time (Rt s) and replicate vials

Paracetamol Trimethoprim Sulfamethoxazole Propanolol Erythromycin Dextropropoxyphene Tamoxifen Lofepramine Diclofenac Mefenamic acid Ibuprofen Clofibric acid a b

Rt s R.S.D. (%)a

Peak area R.S.D. (replicate vials) (%)b

25 (␮g L−1 )

1 (␮g L−1 )

5 (␮g L−1 )

25 (␮g L−1 )

150 (␮g L−1 )

1