Talanta 140 (2015) 115–121
Contents lists available at ScienceDirect
Talanta journal homepage: www.elsevier.com/locate/talanta
Development, validation, and application of an ultra-performance liquid chromatography–sector field inductively coupled plasma mass spectrometry method for simultaneous determination of six organotin compounds in human serum Keith E. Levine a,n, Daniel J. Young a, Scott E. Afton a, James M. Harrington a, Amal S. Essader a, Frank X. Weber a, Reshan A. Fernando a, Kristina Thayer b, Elizabeth E. Hatch c, Veronica G. Robinson b, Suramya Waidyanatha b a
RTI International, Research Triangle Park, NC, United States Division of National Toxicology Program, NIEHS, Research Triangle Park, NC, United States c Department of Epidemiology, Boston University School of Public Health, Boston, MA, United States b
art ic l e i nf o
a b s t r a c t
Article history: Received 10 February 2015 Accepted 15 March 2015 Available online 27 March 2015
Organotin compounds (OTCs) are heavily employed by industry for a wide variety of applications, including the production of plastics and as biocides. Reports of environmental prevalence, differential toxicity between OTCs, and poorly characterized human exposure have fueled the demand for sensitive, selective speciation methods. The objective of this investigation was to develop and validate a rapid, sensitive, and selective analytical method for the simultaneous determination of a suite of organotin compounds, including butyl (mono-, di-, and tri-substituted) and phenyl (mono-, di-, and tri-substituted) species in human serum. The analytical method utilized ultra-performance liquid chromatography (UPLC) coupled with sector field inductively coupled plasma mass spectrometry (SF-ICP-MS). The small (sub-2 mm) particle size of the UPLC column stationary phase and the sensitivity of the SF-ICP-MS enabled separation and sensitive determination of the analyte suite with a runtime of approximately 3 min. Validation activities included demonstration of method linearity over the concentration range of approximately 0.250–13.661 ng mL 1, depending on the species; intraday precision of less than 21%, interday precision of less than 18%, intraday accuracy of 5.3% to 19%, and interday accuracy of 14% to 15% for all species; specificity, and matrix impact. In addition, sensitivity, and analyte stability under different storage scenarios were evaluated. Analyte stability was found to be limited for most species in freezer, refrigerator, and freeze–thaw conditions. The validated method was then applied for the determination of the OTCs in human serum samples from women participating in the Snart-Foraeldre/ MiljØ (Soon-Parents/Environment) Study. The concentration of each OTC ranged from below the experimental limit of quantitation to 10.929 ng tin (Sn) mL 1 serum. Speciation values were confirmed by a total Sn analysis. & 2015 Elsevier B.V. All rights reserved.
Keywords: Organotin compounds Metallomics Speciation Ultra high pressure liquid chromatography Inductively coupled plasma mass spectrometry Serum Biomonitoring
1. Introduction Organotin compounds (OTCs) are a class of organometallic species comprised of a tin (Sn) atom bound covalently to one or more alkyl or aryl groups [1]. The number and structure of the organic substituents bound to Sn can significantly alter its physicochemical characteristics, and as a result, Sn has the greatest variety of organometallic derivatives that are currently in use by industry among any other element [2]. Since the 1940s, the plastics industry has employed mono and n
Corresponding author.
http://dx.doi.org/10.1016/j.talanta.2015.03.022 0039-9140/& 2015 Elsevier B.V. All rights reserved.
dialkyl OTCs as heat and light stabilizers for the production of polyvinyl chloride and other materials [3]. More recently, dialkyl OTCs have been employed in thin film, transparent conductive coatings for liquid crystal display panels [4]. Many OTCs are biocides, with maximum toxicological activity observed for trisubstituted compounds. Tributyltin (TBT) and triphenyltin (TPT) have historically been used in marine antifouling paints to minimize organism growth on ship hulls and as insecticides, miticides, or fungicides for wood preservation or agricultural crop protection [5–7]. The industrial utility of OTCs has led to significant amounts of the chemicals being found in household products and the environment, resulting in widespread potential for human exposure.
116
K.E. Levine et al. / Talanta 140 (2015) 115–121
Several cases of OTC poisoning or exposure were recently reported in workers involved in leather, plastic, or other manufacturing activities [8,9], and OTCs originating from floor wax, diapers, baking paper, clothing, and other consumer products were reported in house dusts [10]. The extent of exposure in the general population is less characterized. The predominant environmental sources of TPT and TBT have been agricultural runoff and the degradation of marine antifouling paints [11,12]; both TPT and TBT are toxic and endocrine disruptors in aquatic organisms, even at sub ng L 1 levels [13]. Numerous reports of gastropod imposex, resulting in permanent female masculinization and species decline [13–15], coupled with OTC bioaccumulation spanning several marine food chain trophic levels [5,12,16–19] have prompted regulatory agencies across the world to restrict use of tin-containing antifouling marine paints to larger ships [20,21]. Even though the use of these paints has been limited, OTCs have persisted in the environment by binding with sediments. When exposed to favorable environmental conditions, aquatic sediments can release sequestered OTCs back into the water column for biological uptake [18]. The differential toxicity, industrial utility, environmental prevalence, and potential for human exposure have fueled the demand for analytical methods capable of determining the concentrations of OTCs in a variety of matrices. These methods must be capable of differentiating target organotin species at very low concentrations in an environmental or biological matrix of interest. Fluorescence spectrometry and bioluminescent assays have recently been reported for detection of organotin compounds [22,23], but the most common analytical approach remains coupling of chromatography with a specific detector. Gas chromatography (GC) coupled to mass spectrometry (MS) [12], tandem mass spectrometry (MS/MS) [13], high resolution MS [24], atomic absorption spectrometry (AAS) [3], flame photometric detection (FPD) [25], pulsed flame photometric detection (PFPD) [20], atomic emission detection (AED) [26], and inductively coupled plasma mass spectrometry (ICP-MS) detection [5] have all been used to quantify organotin species in a variety of matrices. Regardless of the instrumental technique used for detection, OTCs must be extracted and converted to fully alkylated, volatile species to allow analysis by GC. Extraction of ionic OTC species into non-polar solvents has been achieved with complexing agents like tropolone or dithiocarbamate, followed by sample cleanup and concentration. A second analytical approach involves the use of ethylating agents including sodium tetraethylhydroborate and Grignard reagents to convert OTCs to volatile forms that are amenable to analysis by gas chromatography (GC) [20,27]. More recently, headspace solid-phase microextraction (HS-SPME) with in situ derivitization has been employed to simultaneously volatilize, extract, and concentrate OTCs prior to GC separation [28–31]. Although GC has been successfully employed for OTC determinations, challenges associated with the method have prompted a search for alternative methods. Extraction, clean-up, and concentration procedures for OTC measurements by GC can be tedious and time-consuming [32,33]. In addition, GC methods require derivatization of extracted OTCs to more volatile forms, and the extent of derivatization can be dependent on the sample matrix and Sn species present [34,35]. Consolidated extraction and derivatization HSSPME techniques are promising, but are prone to inconsistent matrix effects for OTC measurements [26]. In contrast, liquid chromatography (LC) does not require derivatization prior to analysis and could be capable of separating common OTC environmental contaminants when used with complexing agents. Further, coupling of LC with ICP-MS could provide sensitive detection of OTCs and relatively low limits of detection compared with other detection methods. The objective of this investigation was to develop and validate a rapid, sensitive, and selective ultra-performance liquid
chromatography (UPLC) method coupled with sector field inductively coupled plasma mass spectrometry (SF-ICP-MS) detection for the determination of a suite of OTCs, including monobutyltin (MBT), dibutyltin (DBT), tributyltin (TBT), monophenyltin (MPT), diphenyltin (DPT), and triphenyltin (TPT) (structures presented in Supplementary information, Table S1) in human serum collected from women participating in the Snart-Foraeldre/MiljØ (Soon-Parents/ Environment) Study, a cohort of Danish women who have recently discontinued birth control in order to become pregnant.
2. Experimental 2.1. Reagents The human serum matrix used throughout this investigation to prepare matrix standards and quality control (QC) samples was pooled from six adult female donors and was received and stored frozen (nominal 20 °C) from BioChemed Services (Winchester, VA, USA). Several ampules of a custom organotin standard containing nominal 2000 mg mL 1 concentrations of MBT, DBT, TBT, MPT, DPT, and TPT chlorides in methylene chloride were obtained from Restek (Bellefonte, PA, USA). Semiconductor grade methanol, Ultrex grade acetic acid, and high-purity deionized water ( 18 MΩ, DI H2O) for the mobile phase were obtained from Sigma-Aldrich (St. Louis, MO, USA), J.T. Baker (Center Valley, PA, USA), and Pure Water Solutions (Hillsborough, NC, USA), respectively. High-purity ( Z99%) tropolone and triethylamine were obtained from Santa Cruz Biotechnology (Santa Cruz, CA, USA) and Acros Organics (Geel, Belgium), respectively. Ultrex grade hydrochloric and nitric acids were obtained from J.T. Baker, and a National Institute of Standards and Technology (NIST)-traceable, 10 mg mL 1 Sn standard was purchased from High Purity Standards (Charleston, SC, USA). 2.2. Contamination control With recent advances in instrument technology, the sample preparation method has become the most important source of error in analytical measurements [36]. In order to successfully determine the concentrations of organotin analytes at biologically relevant, sub-ng mL 1 levels, control of the laboratory environment to minimize the potential for contamination and analyte degradation is an absolute necessity. All sample handling activities occurred in a highefficiency particulate air (HEPA)-filtered environment, and work was conducted under UV-free lighting conditions to minimize the potential for photodegradation [37]. In order to minimize Sn background from labware [24], glass UPLC vials (Waters Corporation, Milford, MA, USA) were filled with concentrated hydrochloric acid for at least two hours, rinsed multiple times with high-purity DI water, and were dried and stored under HEPA-filtered air until use. All other labware was soaked in 20% (v/v) hydrochloric acid for a minimum of two hours, rinsed with high-purity DI water, and dried and stored under HEPA-filtered air until use. Multiple lots of each reagent were obtained from several vendors and were screened for total Sn content by SF-ICP-MS. Briefly, reagents were digested in a mixture of nitric and hydrochloric acids prior to analysis against external calibration standards. A total Sn screening procedure was also conducted for blood collection tubes. Tubes were extracted with water and acid for comparison of water-leachable Sn and acid-leachable Sn. Tubes were rinsed three times with high-purity DI water and each rinse was analyzed for total extracted Sn. Tubes were then rinsed with a 2.5% hydrochloric acid/2.5% nitric acid solution and the extract was analyzed for total extracted Sn. The average total water-leachable Sn for six replicates of two lots of commercially-obtained collection tubes was 0.115 and 0.207 ng tube 1. The average sum of water-leachable
K.E. Levine et al. / Talanta 140 (2015) 115–121
and acid-leachable Sn for the same tubes was 5.11 and 10.1 ng tube 1, respectively. It is important to note that the majority of Sn extracted from the collection tubes was detected after acid extraction. The acidic extraction conditions represent a “worst case” scenario for the background Sn and would overestimate the amount of Sn that would leach into samples under biologically relevant conditions. Reagents with the lowest overall Sn background were selected for subsequent use in validation and sample analysis experiments. Particular emphasis was placed on screening tropolone, because high Sn backgrounds in this reagent have posed significant analytical challenges [38]. 2.3. Preparation of solvent and matrix standards Eluent (65% (v/v) methanol, 29% (v/v) DI H2O, 5.5% (v/v) acetic acid, 0.5% (v/v) triethylamine, and 0.075% (w/v) tropolone) and UPLC wash (65% (v/v) methanol, 29.5% (v/v) DI H2O, and 5.5% (v/v) acetic acid) solutions were prepared fresh daily in glass jars. A high-concentration stock solution (approximate species concentrations ranging from 40,000 to 56,000 ng mL 1, as Sn) was prepared fresh daily by adding the custom organotin standard to a glass volumetric flask and bringing to volume with methanol before mixing. A low-concentration stock solution (approximate species concentrations ranging from 2000 to 2700 ng mL 1, as tin) was then prepared fresh daily by adding the high-level stock solution to a glass volumetric flask and diluting to volume with eluent before mixing. Spiking solutions were then prepared fresh daily by transferring a range of volumes of the low-level stock solution to glass volumetric flasks and diluting to volume with eluent before mixing. These spiking solutions were then used to prepare calibration standards and QC samples. Validation experiments were conducted over three days to establish method linearity, intraday and interday precision and accuracy, matrix impact, sensitivity and specificity. Serum matrix standards and QC samples were prepared at eight nominal concentration levels (Table S2) for each species by adding 0.500 mL of serum, 0.050 mL of the appropriate spiking solution, and 1.950 mL of eluent to 10 mL glass vials (CEM Corporation, Matthews, NC, USA). The species concentrations (ng mL 1) were as follows: MPT – 0.319–12.8; MBT – 0.342–13.7; TPT – 0.250–10.0; DPT – 0.280– 11.2; DBT – 0.317–12.7; and TBT – 0.296–11.8. All matrix standards and QC samples were gently mixed by vortex and centrifuged at approximately 3770 rpm on a Beckman (Indianapolis, IN, USA) GS-6 centrifuge for 7 min. Supernatant solutions from this extraction were then transferred to 10 kDa cutoff filter tubes (Sigma-Aldrich, St. Louis, MO, USA) and centrifuged again at approximately 3770 rpm for 20 min as an additional sample clean up step, and the filtrates were transferred directly to UPLC vials for analysis. Matrix standards from the first validation experiment were split and analyzed at both the beginning and at the end of the analytical run to assess instrument drift. Three unfortified serum aliquots were processed as matrix blanks to assess analyte background levels, and an eluent blank was analyzed to determine background in the absence of serum matrix. In order to assess the impact of the extraction matrix on analyte signal, a set of solvent standards was prepared on the first validation experiment day consisting of eluent fortified with the suite of organotin compounds at the same nominal concentration levels as the matrix standards (Table S2). Matrix QC samples were prepared at the same nominal concentration levels to assess intraday and interday precision and accuracy on all three days of validation experiments. An additional six aliquots of the two lowest concentration calibration matrix standards were also prepared and analyzed during the second validation experiment to establish quantitation and detection limits.
117
Stability of OTCs was evaluated under several storage scenarios. Post-extraction stability was evaluated at nominal refrigerator temperature (2–8 °C) storage for up to 14 days with triplicate preparations at three nominal OTC concentrations (1b, 5, and 7 in Table S2). Triplicate serum aliquots fortified at two OTC concentrations (1b and 7 in Table S2) were stored at nominal freezer ( 20 °C) and ultracold freezer ( 80 °C) conditions for up to 60 and 90 days, respectively, to assess long-term storage stability. Fortified serum samples (n ¼3) at three concentrations (1b, 5, and 7 in Table S2) were also stored frozen ( 20 °C) for a minimum of 24 h before being subjecting to three complete freeze–thaw cycles to assess the impact of repeated thawing and refreezing on analyte stability. 2.4. Sample analysis by UPLC–SF-ICP-MS A Thermo (Bremen, Germany) Element-2 SF-ICP-MS equipped with a glass concentric nebulizer and a peltier-cooled cyclonic glass spray chamber was used for all tin measurements. Ultra high purity oxygen from Airgas National Welders (Charlotte, NC, USA) was used to mitigate the impact of the carbon content of the mobile phase on the SF-ICP-MS system. An ACQUITY UPLC system, equipped with a BEH C18 column (2.1 50 mm2 column dimensions, 1.7 mm particle size), was interfaced with the SF-ICP-MS. The system flow rate for analysis was 0.60 mL min 1 and the injection volume used for all standards and samples was 20 mL. Chromatographic data were processed with Thermo XCalibur software version 2.0.6 by integrating peak area and performing an unweighted linear regression of the peak area as a function of species concentration. 2.5. Collection and analysis of serum from Danish women The validated method was applied to 64 serum samples from women participating in the Snart-Foraeldre/MiljØ (Soon-Parents/ Environment) Study, a prospective cohort study of 500 Danish women between the ages of 19 and 40 years who have recently discontinued birth control in order to become pregnant. Miljo is a biospecimen collection sub-study nested within a larger study, Snart-Foraelde (SF), an internet-based prospective cohort study of 10,000 Danish couples who are trying to become pregnant. Data ranges to show analytical method applicability and utility are presented in this report, but a detailed discussion of sample collection and analysis results will be described in a subsequent manuscript. Samples were stored at 20 °C and shipped on dry ice for analysis. All samples were stored at 20 °C until analyzed. Because serum samples were received and analyzed in blinded fashion, without any personally identifiable information, the analytical laboratory received an Institutional Review Board exemption.
3. Results and discussion The analytical method utilized UPLC coupled with SF-ICP-MS. The small (sub-2 mm) particle size of the UPLC column stationary phase and the sensitivity of the SF-ICP-MS enabled separation and sensitive determination of the analyte suite in approximately 3 min per sample (Fig. 1). To the best of our knowledge, the chromatographic run time and detection limits for the suite of analytes are superior to those other literature methods for OTC determinations in biological fluids. The analytical method was validated for linearity, precision, accuracy, specificity, matrix impact, sensitivity, and analyte stability under different storage scenarios through several experiments. The validated method was
118
K.E. Levine et al. / Talanta 140 (2015) 115–121
then successfully applied for the determination of the OTC analyte suite in human serum samples.
Table 1 Intraday precision and accuracy for organotin species. Nominal (OTC)a (ng mL 1)
3.1. Linearity Throughout this investigation, the method was developed and validated to deliver optimal analytical performance for the overall analyte suite and did not focus on optimizing the performance for an individual species at the expense of the other analytes. Calibration curves included six standard concentrations for each analyte. All calibration curves were unweighted and derived using linear least-squares fit of the form y¼mx þb and exhibited correlation coefficients (r) greater than 0.999. A representative calibration curve is presented in Fig. 2. Results for relative error of calibration standards and correlation coefficients for the calibration curves for each day are shown in Table S3. 3.2. Matrix impact: recovery Both matrix and solvent standards were prepared and analyzed to assess matrix impact. For each species and standard concentration level, the ratio of the matrix standard instrument response to the response of solvent standards prepared at the same nominal concentration was calculated and expressed as a percentage. Prior to ratio calculation, both matrix and solvent responses were corrected by subtracting the response from appropriate matrix matched blanks. Across calibration ranges, calculated response ratios for MPT, MBT, TBT, DPT, DBT, and TBT ranged from 150% to 180% (MPT), 140% to 170% (MBT), 110% to 160% (TBT), 120% to 140% (DPT), 140% to 160% (DBT), and 110% to 140% (TBT), respectively, indicating the presence of a serum matrix enhancement. As a result, matrix calibration standards were employed for all subsequent validation experiments. MBT
Intensity (cps)
8 E+05
MPT
MPT 0.319 0.638 0.957 1.28 3.19 6.38 9.57 12.8
4.7 1.7 1.0 3.9 4.1 1.8 12 3.7
MBT 0.342 0.683 1.03 1.37 3.42 6.83 10.2 13.7
16 5.0 3.2 3.6 1.5 1.3 10 5.0
TPT 0.250 0.500 0.750 1.00 2.50 5.00 7.50 10.0
41 34 34 28 24 26 16 20
Precision %RSD
5.8 7.8 3.7 3.1 0.41 5.9 8.7 5.1
17 9.3 4.0 1.8 3.7 7.0 8.9 5.5
21 17 18 5.5 5.7 9.7 16 13
DPT 0.280 0.561 0.841 1.12 2.80 5.61 8.41 11.2
2.1 7.0 1.7 9.4 8.4 6.0 17 12
7.0 7.2 4.7 4.3 1.4 5.1 6.4 3.5
DBT 0.317 0.634 0.952 1.27 3.17 6.34 9.52 12.7
19 10 1.4 1.3 3.4 5.3 6.5 0.72
3.8 7.7 6.2 1.1 2.9 6.0 10 5.3
7 E+05 6 E+05
TBT
TPT
5 E+05
DBT
4 E+05
DPT
3 E+05 2 E+05 1 E+05 0 E+00 0
100
200
Time ( sec) Fig. 1. Representative chromatogram showing phenyltin (MPT, 6.38 ng mL 1), butyltin (MBT, 6.83 ng mL 1), triphenyltin (TPT, 5.00 ng mL 1), diphenyltin (DPT, 5.61 ng mL 1), dibutyltin (DBT, 6.34 ng mL 1), and tributyltin (TBT, 5.92 ng mL 1) in human serum matrix separated in less than 3 min.
0.3 0.25
Area/IS Area
Accuracy mean %RE
a
0.2 0.15
29 21 15 18 15 11 23 15
4.9 7.7 3.5 7.1 3.3 6.2 5.7 3.9
Triplicate preparations at each nominal concentration level.
3.3. Specificity
0.1 0.05 0 0.000
TBT 0.296 0.592 0.888 1.18 2.96 5.92 8.88 11.8
5.000
10.000
15.000
Nominal [Sn] (ng/mL)
Fig. 2. Representative linear regression for the organotin compound, MPT, in human serum obtained during day 1 validation experiment; linear regression equation: y¼0.0220959x þ0.000785919; r¼ 0.9995.
Specificity was demonstrated by comparing the instrument response for each OTC in triplicate matrix blank samples to the response obtained for each species at the experimental limit of quantitation (ELOQ) level. The ratio range of blank instrument response to ELOQ response for all six species ranged from nondetect to 11%.
K.E. Levine et al. / Talanta 140 (2015) 115–121
Table 2 Interday precision and accuracy for organotin species. Nominal (OTC)a (ng mL 1)
Accuracy mean %RE
MPT 0.319 0.638 0.957 1.28 3.19 6.38 9.57 12.8
1.5 3.2 2.5 1.2 0.89 4.7 8.9 11
MBT 0.342 0.683 1.03 1.37 3.42 6.83 10.2 13.7
1.6 4.2 4.0 4.5 0.01 1.8 6.3 11
TPT 0.250 0.500 0.750 1.00 2.50 5.00 7.50 10.0
12 14 9.9 14 11 8.8 3.6 2.6
Precision %RSD
9.1 3.9 4.5 2.5 3.1 4.8 9.0 5.4
0.067 5.2 5.0 5.3 5.5 1.6 12 3.6
27 17 11 9.8 7.4 15 3.3 14
DPT 0.280 0.561 0.841 1.12 2.80 5.61 8.41 11.2
3.0 5.8 9.0 4.3 0.56 1.6 8.3 13
4.5 9.0 5.5 5.1 11 2.7 11 2.5
DBT 0.317 0.634 0.952 1.27 3.17 6.34 9.52 12.7
9.5 6.1 3.4 0.95 4.3 2.7 7.8 6.6
14 11 2.2 2.4 5.5 1.6 6.6 1.5
TBT 0.296 0.5920 0.888 1.18 2.96 5.92 8.88 11.8
8.1 6.1 8.8 2.5 3.0 2.8 15 14
18 13 9.1 7.1 9.3 2.3 10 4.7
a
Triplicate preparations at each nominal concentration level.
119
(730% at the ELOQ level), respectively, for at least four of the six species. This acceptance criterion for intraday precision was met as all species except TPT passed; the intraday accuracy criterion was also achieved as all species except TPT and TBT passed. For DPT, the intraday %RE was marginally outside of 715% for one concentration level at high end of the linear range, but the intraday accuracy was considered fit for purpose because of the high level of accuracy observed at and near the ELOQ level. Interday precision and accuracy criteria were achieved met as all species passed, with the exception of TPT, with marginally failing precision only. Overall, intraday and interday precision and accuracy were considered acceptable for the OTC suite. The failing intraday accuracy for TPT was attributed to degradation during the first day of validation experiments. Determined TPT concentrations were found to gradually decline with each successive QC replicate interspersed throughout the analysis. Although passing, DPT exhibited generally positive %RE, which may indicate that this species is a degradation product of TPT. Interday TPT accuracy passed because analytical run length was shorter for validation experiments on subsequent days. Although TBT failed intraday accuracy with generally positive %RE data, a trend could not be identified on closer investigation of the individual matrix QC replicates. 3.5. Sensitivity Seven matrix QC samples were prepared at the lowest nominal concentration and analyzed to challenge method sensitivity and establish quantitation and detection limits. In order to conservatively be defined as the ELOQ, the acceptance criteria for matrix QC replicates were defined as 7 30% RE and r30% RSD. These criteria were comfortably met for all species (Table 3), providing confidence in accurate, precise measurements at ELOQ levels. The limit of detection (LOD) for each species is also presented and was calculated by multiplying the standard deviation from the determined matrix QC analyte concentrations by 3. 3.6. Stability evaluation Stability was evaluated under several different storage scenarios to determine the optimal application of the study method. Post-extraction refrigerator storage stability was evaluated over 14 days. Although relatively high levels of accuracy and precision were still observed for the butyltin species after 14 days in refrigerated extracts, stability was not observed for the phenyltin species, with apparent degradation of DPT and TPT to MPT (Fig. 3 and Table S4, Supplementary information). Stability data from fortified serum aliquots stored under nominal freezer conditions for up to 60 and 90 days, respectively, are presented in Fig. 4 and Table S5 (Supplementary information). The overall trend for the OTC suite was clear degradation over time. The average relative error of the OTC species stored at ultracold freezer conditions after Table 3 Experimental limits of quantitation and limit of detection. OTC
ELOQa (OTC) (ng mL 1)
Accuracy mean %RE
Precision % RSD
LODa (ng mL 1)
MPT MBT TPT DPT DBT TBT
1.60 1.71 1.25 1.40 1.59 1.48
13 7.5 15 0.69 19 18
6.8 8.9 7.1 5.8 4.1 5.1
0.350 0.500 0.300 0.250 0.250 0.250
3.4. Precision and accuracy Intraday and interday precision and accuracy are presented in Tables 1 and 2, respectively. Intraday and interday precision, expressed as percent relative standard deviation (%RSD), and accuracy, expressed as percent relative error (%RE), of determined concentrations for matrix QC samples were required to be r15% (r 30% at the ELOQ level) and within 715% of the nominal value
a ELOQ and LOD expressed as ng analyte mL 1 of serum; accounts for 5 dilution factor.
K.E. Levine et al. / Talanta 140 (2015) 115–121
20
100 % Relative Error
% Relative Error
120
50 0 -50
0 -20 -40 -60 -80
- 100 0
7 Time (days)
-100
14
50 0 -50
-100 0
% Relative Error
20
40 Time (days)
0
20 Time (days)
60
20 % Relative Error
% Relative Error
100
0
5
10 Time (days)
0 -20 -40 -60 -80 -100 40
60
Fig. 4. Compound stability for OTCs in human serum at freezer storage conditions. Key: A – low concentration stability samples; B – high concentration stability samples. Diamond – MPT; square – MBT; triangle – TPT; X – DPT; þ – DBT; and circle – TBT.
100 50 0 -50 - 100 0
5
10 Time (days)
Fig. 3. Compound stability for OTCs in human serum at refrigerator conditions. Key: A – low concentration stability samples; B – medium concentration stability samples; C – high concentration stability samples. Diamond – MPT; square – MBT; triangle – TPT; X – DPT; þ – DBT; and circle – TBT.
a time period of 90 days was 43% for MBT, 32% for MBT, 68% for TPT, 42% for DPT, 51% for DBT, and 37% for TBT with precision values less than RSD ¼15%. In addition, fortified serum aliquots subjected to three complete freeze–thaw cycles showed similar degradation across all analytes in the suite and all nominal concentration levels. The mean %RE ranged from 48% to 38% (MPT), 47% to 20% (MBT), 91% to 78% (TPT), 41% to 25% (DPT), 52% to 34% (DBT), and 54% to 45% (TBT). Based on these overall stability findings, it is recommended that the sample collection and analytical laboratories closely coordinate activities to minimize species degradation. 3.7. Analysis of serum from Danish women The method was applied to serum from 64 Danish women participating in the Snart-Foraeldre/MiljØ (Soon-Parents/Environment) Study, a prospective cohort study of 500 Danish women between the ages of 19 and 40 years who have recently discontinued birth control in order to become pregnant. The concentrations of OTC species were found to be in the following ranges (ng mL 1): MPT, o LOD – 1.08; MBT, oLOD – 10.929; TPT, oLOD – 0.862; DPT, oLOD – 0.370; DBT, oLOD – 4.495; TPT,
oLOD – 0.513. The samples all exhibited relatively low levels of the OTC suite, suggesting minimal risk of exposure to OTCs in this population. Additional data will be provided in a subsequent manuscript, including confirmatory total tin measurements to establish mass balance.
4. Conclusion A UPLC–SF-ICP-MS method was successfully developed and validated for the determination of a suite of six OTCs (MPT, MBT, TPT, DPT, DBT, and TBT) in human serum. The employed sample preparation and chromatographic procedures allowed for separation of the analytes in less than three minutes without derivatization, resulting in considerably less sample preparation time than that required for many equivalent HPLC or GC separations. This shortened analysis time may offer a significant advantage in that it can facilitate higher sample throughput while minimizing potential instrumental drift and degradation of OTC species, especially TPT. In addition, use of the SF-ICP-MS as the detector enabled sensitive determination of the analyte suite. Overall, sensitivity for the OTCs studied here compares favorably with other literature reports (Table S6) for environmental and biological matrices. Several observations were noted which emphasize the importance of close coordination between the analytical and sample collection laboratories. Careful screening of all lots of employed reagents (most notably, tropolone) prior to use is essential in maintaining low background levels of Sn. Further, the sample collection laboratory could benefit from evaluation of multiple lots of biological fluid collection tubes to minimize the potential for contamination. Care must be taken to thoroughly clean all labware, including chromatography vials and other new, ‘as received’ labware. Ideally, stability considerations will be taken into account when coordinating measurements. To minimize potential degradation, standards and samples should be prepared on
K.E. Levine et al. / Talanta 140 (2015) 115–121
the day of analysis and samples should be analyzed as soon as practical upon receipt. Finally, to the extent possible, study samples should be preserved at the coldest temperature practical and transferred to the analytical laboratory for immediate processing.
Acknowledgements This research was supported by the National Institute of Health/ National Institute of Environmental Health Sciences (NIH/NIEHS; HHSN27320110003C). The authors would like to acknowledge Mr. Bradley J. Collins, Dr. Michelle Hooth, and Dr. Esra Mutlu for their review of this manuscript prior to formal submission.
Appendix A. Supplementary information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.talanta.2015.03.022.
References [1] J.B. Graceli, et al., Reprod. Toxicol. 36 (2013) 40–52. [2] A.O. Sunday, B.A. Alafara, O.G. Oladele, Chem. Speciat. Bioavailab. 24 (4) (2012) 216–226. [3] J. Kuballa, et al., Analyst 120 (1995) 667–673. [4] Y. Suzuki, et al., J. Chromatogr. B 868 (2008) 116–119. [5] M. Furdek, et al., Mar. Pollut. Bull. 64 (2012) 189–199.
121
[6] Y. Liu, Y. Wan, L. Guo, Anal. Lett. 47 (2014) 343–355. [7] M. Hoch, Appl. Geochem. 16 (2001) 719–743. [8] F. Guo, X. Lu, Q. Xu, World J. Emerg. Med. 1 (2) (2010) 122–125. [9] Z. Cui, et al., Talanta 85 (2011) 1028–1033. [10] K. Kannan, et al., Arch. Environ. Contam. Toxicol. 58 (2010) 901–907. [11] J. Gao, et al., Environ. Monit. Assess. 185 (2013) 3831–3837. [12] K.K.Y. Ho, K.M.Y. Leung, Mar. Pollut. Bull. 85 (2014) 634–640. [13] C. Devos, F. David, P. Sandra, J. Chromatogr. A 1261 (2012) 151–157. [14.] G.W. Bryan, et al., J. Mar. Biol. Assoc. U. K. 66 (1986) 611–640. [15] P.E. Gibbs, G.W. Bryan, J. Mar. Biol. Assoc. U. K. 66 (1986) 767–777. [16] Z. Yu, et al., Food Chem. 119 (2010) 364–367. [17] P. Rantakokko, et al., Sci. Total Environ. 408 (2010) 2474–2481. [18] I.Bd Castro, F. Perina, G. Fillmann, Environ. Monit. Assess. 184 (2012) 1781–1799. [19] T. Liu, et al., J. Food Process Eng. 34 (2011) 1125–1143. [20] C.Rd Oliveira, et al., J. Hazard. Mater. 181 (2010) 851–856. [21] Y. Mino, et al., J. Health Sci. 54 (2) (2008) 224–228. [22] M. Bravo, et al., Microchem. J. 106 (2013) 95–101. [23] G. Kabiersch, et al., Anal. Chem. 85 (2013) 5740–5745. [24] M.G. Ikonomou, et al., J. Chromatogr. A 975 (2002) 319–333. [25] T. Horiguchi, et al., Mar. Environ. Res. 76 (2012) 56–62. [26] J. Carpinteiro, I. Rodriguez, R. Cela, Anal. Bioanal. Chem. 380 (2004) 853–857. [27] K. Kannan, K. Senthilkumar, J.P. Giesy, J. Chromatogr. A 33 (1999) 1776–1779. [28.] E. Beceilo-Gonzalez, A. Guimaraes, M.F. Alpendurada, J. Chromatogr. A 1216 (2009) 5563–5569. [29] J.I. Cacho, et al., J. Chromatogr. A 1279 (2013) 1–6. [30] C. Coscolla, et al., Talanta 119 (2014) 544–552. [31] M. Bravo, et al., J. Chromatogr. A 1223 (2012) 9–14. [32] I.R. Pereiro, et al., Anal. Chem. 68 (23) (1996) 4135–4140. [33] T. Sano, et al., J. Chromatogr. A 1217 (2010) 4344–4346. [34] L. Yang, J.W.H. Lam, J. Anal. At. Spectrom. 16 (2001) 724–731. [35] S. Chiron, et al., J. Chromatogr. A 879 (2000) 137–145. [36] C. Dietz, et al., J. Chromatogr. A 1153 (2007) 114–129. [37] K. Dubalska, et al., Crit. Rev. Anal. Chem. 43 (2013) 35–54. [38] S. White, et al., J. Chromatogr. A 794 (1998) 211–218.
Contents lists available at ScienceDirect
Talanta journal homepage: www.elsevier.com/locate/talanta
Development, validation, and application of an ultra-performance liquid chromatography–sector field inductively coupled plasma mass spectrometry method for simultaneous determination of six organotin compounds in human serum Keith E. Levine a,n, Daniel J. Young a, Scott E. Afton a, James M. Harrington a, Amal S. Essader a, Frank X. Weber a, Reshan A. Fernando a, Kristina Thayer b, Elizabeth E. Hatch c, Veronica G. Robinson b, Suramya Waidyanatha b a
RTI International, Research Triangle Park, NC, United States Division of National Toxicology Program, NIEHS, Research Triangle Park, NC, United States c Department of Epidemiology, Boston University School of Public Health, Boston, MA, United States b
art ic l e i nf o
a b s t r a c t
Article history: Received 10 February 2015 Accepted 15 March 2015 Available online 27 March 2015
Organotin compounds (OTCs) are heavily employed by industry for a wide variety of applications, including the production of plastics and as biocides. Reports of environmental prevalence, differential toxicity between OTCs, and poorly characterized human exposure have fueled the demand for sensitive, selective speciation methods. The objective of this investigation was to develop and validate a rapid, sensitive, and selective analytical method for the simultaneous determination of a suite of organotin compounds, including butyl (mono-, di-, and tri-substituted) and phenyl (mono-, di-, and tri-substituted) species in human serum. The analytical method utilized ultra-performance liquid chromatography (UPLC) coupled with sector field inductively coupled plasma mass spectrometry (SF-ICP-MS). The small (sub-2 mm) particle size of the UPLC column stationary phase and the sensitivity of the SF-ICP-MS enabled separation and sensitive determination of the analyte suite with a runtime of approximately 3 min. Validation activities included demonstration of method linearity over the concentration range of approximately 0.250–13.661 ng mL 1, depending on the species; intraday precision of less than 21%, interday precision of less than 18%, intraday accuracy of 5.3% to 19%, and interday accuracy of 14% to 15% for all species; specificity, and matrix impact. In addition, sensitivity, and analyte stability under different storage scenarios were evaluated. Analyte stability was found to be limited for most species in freezer, refrigerator, and freeze–thaw conditions. The validated method was then applied for the determination of the OTCs in human serum samples from women participating in the Snart-Foraeldre/ MiljØ (Soon-Parents/Environment) Study. The concentration of each OTC ranged from below the experimental limit of quantitation to 10.929 ng tin (Sn) mL 1 serum. Speciation values were confirmed by a total Sn analysis. & 2015 Elsevier B.V. All rights reserved.
Keywords: Organotin compounds Metallomics Speciation Ultra high pressure liquid chromatography Inductively coupled plasma mass spectrometry Serum Biomonitoring
1. Introduction Organotin compounds (OTCs) are a class of organometallic species comprised of a tin (Sn) atom bound covalently to one or more alkyl or aryl groups [1]. The number and structure of the organic substituents bound to Sn can significantly alter its physicochemical characteristics, and as a result, Sn has the greatest variety of organometallic derivatives that are currently in use by industry among any other element [2]. Since the 1940s, the plastics industry has employed mono and n
Corresponding author.
http://dx.doi.org/10.1016/j.talanta.2015.03.022 0039-9140/& 2015 Elsevier B.V. All rights reserved.
dialkyl OTCs as heat and light stabilizers for the production of polyvinyl chloride and other materials [3]. More recently, dialkyl OTCs have been employed in thin film, transparent conductive coatings for liquid crystal display panels [4]. Many OTCs are biocides, with maximum toxicological activity observed for trisubstituted compounds. Tributyltin (TBT) and triphenyltin (TPT) have historically been used in marine antifouling paints to minimize organism growth on ship hulls and as insecticides, miticides, or fungicides for wood preservation or agricultural crop protection [5–7]. The industrial utility of OTCs has led to significant amounts of the chemicals being found in household products and the environment, resulting in widespread potential for human exposure.
116
K.E. Levine et al. / Talanta 140 (2015) 115–121
Several cases of OTC poisoning or exposure were recently reported in workers involved in leather, plastic, or other manufacturing activities [8,9], and OTCs originating from floor wax, diapers, baking paper, clothing, and other consumer products were reported in house dusts [10]. The extent of exposure in the general population is less characterized. The predominant environmental sources of TPT and TBT have been agricultural runoff and the degradation of marine antifouling paints [11,12]; both TPT and TBT are toxic and endocrine disruptors in aquatic organisms, even at sub ng L 1 levels [13]. Numerous reports of gastropod imposex, resulting in permanent female masculinization and species decline [13–15], coupled with OTC bioaccumulation spanning several marine food chain trophic levels [5,12,16–19] have prompted regulatory agencies across the world to restrict use of tin-containing antifouling marine paints to larger ships [20,21]. Even though the use of these paints has been limited, OTCs have persisted in the environment by binding with sediments. When exposed to favorable environmental conditions, aquatic sediments can release sequestered OTCs back into the water column for biological uptake [18]. The differential toxicity, industrial utility, environmental prevalence, and potential for human exposure have fueled the demand for analytical methods capable of determining the concentrations of OTCs in a variety of matrices. These methods must be capable of differentiating target organotin species at very low concentrations in an environmental or biological matrix of interest. Fluorescence spectrometry and bioluminescent assays have recently been reported for detection of organotin compounds [22,23], but the most common analytical approach remains coupling of chromatography with a specific detector. Gas chromatography (GC) coupled to mass spectrometry (MS) [12], tandem mass spectrometry (MS/MS) [13], high resolution MS [24], atomic absorption spectrometry (AAS) [3], flame photometric detection (FPD) [25], pulsed flame photometric detection (PFPD) [20], atomic emission detection (AED) [26], and inductively coupled plasma mass spectrometry (ICP-MS) detection [5] have all been used to quantify organotin species in a variety of matrices. Regardless of the instrumental technique used for detection, OTCs must be extracted and converted to fully alkylated, volatile species to allow analysis by GC. Extraction of ionic OTC species into non-polar solvents has been achieved with complexing agents like tropolone or dithiocarbamate, followed by sample cleanup and concentration. A second analytical approach involves the use of ethylating agents including sodium tetraethylhydroborate and Grignard reagents to convert OTCs to volatile forms that are amenable to analysis by gas chromatography (GC) [20,27]. More recently, headspace solid-phase microextraction (HS-SPME) with in situ derivitization has been employed to simultaneously volatilize, extract, and concentrate OTCs prior to GC separation [28–31]. Although GC has been successfully employed for OTC determinations, challenges associated with the method have prompted a search for alternative methods. Extraction, clean-up, and concentration procedures for OTC measurements by GC can be tedious and time-consuming [32,33]. In addition, GC methods require derivatization of extracted OTCs to more volatile forms, and the extent of derivatization can be dependent on the sample matrix and Sn species present [34,35]. Consolidated extraction and derivatization HSSPME techniques are promising, but are prone to inconsistent matrix effects for OTC measurements [26]. In contrast, liquid chromatography (LC) does not require derivatization prior to analysis and could be capable of separating common OTC environmental contaminants when used with complexing agents. Further, coupling of LC with ICP-MS could provide sensitive detection of OTCs and relatively low limits of detection compared with other detection methods. The objective of this investigation was to develop and validate a rapid, sensitive, and selective ultra-performance liquid
chromatography (UPLC) method coupled with sector field inductively coupled plasma mass spectrometry (SF-ICP-MS) detection for the determination of a suite of OTCs, including monobutyltin (MBT), dibutyltin (DBT), tributyltin (TBT), monophenyltin (MPT), diphenyltin (DPT), and triphenyltin (TPT) (structures presented in Supplementary information, Table S1) in human serum collected from women participating in the Snart-Foraeldre/MiljØ (Soon-Parents/ Environment) Study, a cohort of Danish women who have recently discontinued birth control in order to become pregnant.
2. Experimental 2.1. Reagents The human serum matrix used throughout this investigation to prepare matrix standards and quality control (QC) samples was pooled from six adult female donors and was received and stored frozen (nominal 20 °C) from BioChemed Services (Winchester, VA, USA). Several ampules of a custom organotin standard containing nominal 2000 mg mL 1 concentrations of MBT, DBT, TBT, MPT, DPT, and TPT chlorides in methylene chloride were obtained from Restek (Bellefonte, PA, USA). Semiconductor grade methanol, Ultrex grade acetic acid, and high-purity deionized water ( 18 MΩ, DI H2O) for the mobile phase were obtained from Sigma-Aldrich (St. Louis, MO, USA), J.T. Baker (Center Valley, PA, USA), and Pure Water Solutions (Hillsborough, NC, USA), respectively. High-purity ( Z99%) tropolone and triethylamine were obtained from Santa Cruz Biotechnology (Santa Cruz, CA, USA) and Acros Organics (Geel, Belgium), respectively. Ultrex grade hydrochloric and nitric acids were obtained from J.T. Baker, and a National Institute of Standards and Technology (NIST)-traceable, 10 mg mL 1 Sn standard was purchased from High Purity Standards (Charleston, SC, USA). 2.2. Contamination control With recent advances in instrument technology, the sample preparation method has become the most important source of error in analytical measurements [36]. In order to successfully determine the concentrations of organotin analytes at biologically relevant, sub-ng mL 1 levels, control of the laboratory environment to minimize the potential for contamination and analyte degradation is an absolute necessity. All sample handling activities occurred in a highefficiency particulate air (HEPA)-filtered environment, and work was conducted under UV-free lighting conditions to minimize the potential for photodegradation [37]. In order to minimize Sn background from labware [24], glass UPLC vials (Waters Corporation, Milford, MA, USA) were filled with concentrated hydrochloric acid for at least two hours, rinsed multiple times with high-purity DI water, and were dried and stored under HEPA-filtered air until use. All other labware was soaked in 20% (v/v) hydrochloric acid for a minimum of two hours, rinsed with high-purity DI water, and dried and stored under HEPA-filtered air until use. Multiple lots of each reagent were obtained from several vendors and were screened for total Sn content by SF-ICP-MS. Briefly, reagents were digested in a mixture of nitric and hydrochloric acids prior to analysis against external calibration standards. A total Sn screening procedure was also conducted for blood collection tubes. Tubes were extracted with water and acid for comparison of water-leachable Sn and acid-leachable Sn. Tubes were rinsed three times with high-purity DI water and each rinse was analyzed for total extracted Sn. Tubes were then rinsed with a 2.5% hydrochloric acid/2.5% nitric acid solution and the extract was analyzed for total extracted Sn. The average total water-leachable Sn for six replicates of two lots of commercially-obtained collection tubes was 0.115 and 0.207 ng tube 1. The average sum of water-leachable
K.E. Levine et al. / Talanta 140 (2015) 115–121
and acid-leachable Sn for the same tubes was 5.11 and 10.1 ng tube 1, respectively. It is important to note that the majority of Sn extracted from the collection tubes was detected after acid extraction. The acidic extraction conditions represent a “worst case” scenario for the background Sn and would overestimate the amount of Sn that would leach into samples under biologically relevant conditions. Reagents with the lowest overall Sn background were selected for subsequent use in validation and sample analysis experiments. Particular emphasis was placed on screening tropolone, because high Sn backgrounds in this reagent have posed significant analytical challenges [38]. 2.3. Preparation of solvent and matrix standards Eluent (65% (v/v) methanol, 29% (v/v) DI H2O, 5.5% (v/v) acetic acid, 0.5% (v/v) triethylamine, and 0.075% (w/v) tropolone) and UPLC wash (65% (v/v) methanol, 29.5% (v/v) DI H2O, and 5.5% (v/v) acetic acid) solutions were prepared fresh daily in glass jars. A high-concentration stock solution (approximate species concentrations ranging from 40,000 to 56,000 ng mL 1, as Sn) was prepared fresh daily by adding the custom organotin standard to a glass volumetric flask and bringing to volume with methanol before mixing. A low-concentration stock solution (approximate species concentrations ranging from 2000 to 2700 ng mL 1, as tin) was then prepared fresh daily by adding the high-level stock solution to a glass volumetric flask and diluting to volume with eluent before mixing. Spiking solutions were then prepared fresh daily by transferring a range of volumes of the low-level stock solution to glass volumetric flasks and diluting to volume with eluent before mixing. These spiking solutions were then used to prepare calibration standards and QC samples. Validation experiments were conducted over three days to establish method linearity, intraday and interday precision and accuracy, matrix impact, sensitivity and specificity. Serum matrix standards and QC samples were prepared at eight nominal concentration levels (Table S2) for each species by adding 0.500 mL of serum, 0.050 mL of the appropriate spiking solution, and 1.950 mL of eluent to 10 mL glass vials (CEM Corporation, Matthews, NC, USA). The species concentrations (ng mL 1) were as follows: MPT – 0.319–12.8; MBT – 0.342–13.7; TPT – 0.250–10.0; DPT – 0.280– 11.2; DBT – 0.317–12.7; and TBT – 0.296–11.8. All matrix standards and QC samples were gently mixed by vortex and centrifuged at approximately 3770 rpm on a Beckman (Indianapolis, IN, USA) GS-6 centrifuge for 7 min. Supernatant solutions from this extraction were then transferred to 10 kDa cutoff filter tubes (Sigma-Aldrich, St. Louis, MO, USA) and centrifuged again at approximately 3770 rpm for 20 min as an additional sample clean up step, and the filtrates were transferred directly to UPLC vials for analysis. Matrix standards from the first validation experiment were split and analyzed at both the beginning and at the end of the analytical run to assess instrument drift. Three unfortified serum aliquots were processed as matrix blanks to assess analyte background levels, and an eluent blank was analyzed to determine background in the absence of serum matrix. In order to assess the impact of the extraction matrix on analyte signal, a set of solvent standards was prepared on the first validation experiment day consisting of eluent fortified with the suite of organotin compounds at the same nominal concentration levels as the matrix standards (Table S2). Matrix QC samples were prepared at the same nominal concentration levels to assess intraday and interday precision and accuracy on all three days of validation experiments. An additional six aliquots of the two lowest concentration calibration matrix standards were also prepared and analyzed during the second validation experiment to establish quantitation and detection limits.
117
Stability of OTCs was evaluated under several storage scenarios. Post-extraction stability was evaluated at nominal refrigerator temperature (2–8 °C) storage for up to 14 days with triplicate preparations at three nominal OTC concentrations (1b, 5, and 7 in Table S2). Triplicate serum aliquots fortified at two OTC concentrations (1b and 7 in Table S2) were stored at nominal freezer ( 20 °C) and ultracold freezer ( 80 °C) conditions for up to 60 and 90 days, respectively, to assess long-term storage stability. Fortified serum samples (n ¼3) at three concentrations (1b, 5, and 7 in Table S2) were also stored frozen ( 20 °C) for a minimum of 24 h before being subjecting to three complete freeze–thaw cycles to assess the impact of repeated thawing and refreezing on analyte stability. 2.4. Sample analysis by UPLC–SF-ICP-MS A Thermo (Bremen, Germany) Element-2 SF-ICP-MS equipped with a glass concentric nebulizer and a peltier-cooled cyclonic glass spray chamber was used for all tin measurements. Ultra high purity oxygen from Airgas National Welders (Charlotte, NC, USA) was used to mitigate the impact of the carbon content of the mobile phase on the SF-ICP-MS system. An ACQUITY UPLC system, equipped with a BEH C18 column (2.1 50 mm2 column dimensions, 1.7 mm particle size), was interfaced with the SF-ICP-MS. The system flow rate for analysis was 0.60 mL min 1 and the injection volume used for all standards and samples was 20 mL. Chromatographic data were processed with Thermo XCalibur software version 2.0.6 by integrating peak area and performing an unweighted linear regression of the peak area as a function of species concentration. 2.5. Collection and analysis of serum from Danish women The validated method was applied to 64 serum samples from women participating in the Snart-Foraeldre/MiljØ (Soon-Parents/ Environment) Study, a prospective cohort study of 500 Danish women between the ages of 19 and 40 years who have recently discontinued birth control in order to become pregnant. Miljo is a biospecimen collection sub-study nested within a larger study, Snart-Foraelde (SF), an internet-based prospective cohort study of 10,000 Danish couples who are trying to become pregnant. Data ranges to show analytical method applicability and utility are presented in this report, but a detailed discussion of sample collection and analysis results will be described in a subsequent manuscript. Samples were stored at 20 °C and shipped on dry ice for analysis. All samples were stored at 20 °C until analyzed. Because serum samples were received and analyzed in blinded fashion, without any personally identifiable information, the analytical laboratory received an Institutional Review Board exemption.
3. Results and discussion The analytical method utilized UPLC coupled with SF-ICP-MS. The small (sub-2 mm) particle size of the UPLC column stationary phase and the sensitivity of the SF-ICP-MS enabled separation and sensitive determination of the analyte suite in approximately 3 min per sample (Fig. 1). To the best of our knowledge, the chromatographic run time and detection limits for the suite of analytes are superior to those other literature methods for OTC determinations in biological fluids. The analytical method was validated for linearity, precision, accuracy, specificity, matrix impact, sensitivity, and analyte stability under different storage scenarios through several experiments. The validated method was
118
K.E. Levine et al. / Talanta 140 (2015) 115–121
then successfully applied for the determination of the OTC analyte suite in human serum samples.
Table 1 Intraday precision and accuracy for organotin species. Nominal (OTC)a (ng mL 1)
3.1. Linearity Throughout this investigation, the method was developed and validated to deliver optimal analytical performance for the overall analyte suite and did not focus on optimizing the performance for an individual species at the expense of the other analytes. Calibration curves included six standard concentrations for each analyte. All calibration curves were unweighted and derived using linear least-squares fit of the form y¼mx þb and exhibited correlation coefficients (r) greater than 0.999. A representative calibration curve is presented in Fig. 2. Results for relative error of calibration standards and correlation coefficients for the calibration curves for each day are shown in Table S3. 3.2. Matrix impact: recovery Both matrix and solvent standards were prepared and analyzed to assess matrix impact. For each species and standard concentration level, the ratio of the matrix standard instrument response to the response of solvent standards prepared at the same nominal concentration was calculated and expressed as a percentage. Prior to ratio calculation, both matrix and solvent responses were corrected by subtracting the response from appropriate matrix matched blanks. Across calibration ranges, calculated response ratios for MPT, MBT, TBT, DPT, DBT, and TBT ranged from 150% to 180% (MPT), 140% to 170% (MBT), 110% to 160% (TBT), 120% to 140% (DPT), 140% to 160% (DBT), and 110% to 140% (TBT), respectively, indicating the presence of a serum matrix enhancement. As a result, matrix calibration standards were employed for all subsequent validation experiments. MBT
Intensity (cps)
8 E+05
MPT
MPT 0.319 0.638 0.957 1.28 3.19 6.38 9.57 12.8
4.7 1.7 1.0 3.9 4.1 1.8 12 3.7
MBT 0.342 0.683 1.03 1.37 3.42 6.83 10.2 13.7
16 5.0 3.2 3.6 1.5 1.3 10 5.0
TPT 0.250 0.500 0.750 1.00 2.50 5.00 7.50 10.0
41 34 34 28 24 26 16 20
Precision %RSD
5.8 7.8 3.7 3.1 0.41 5.9 8.7 5.1
17 9.3 4.0 1.8 3.7 7.0 8.9 5.5
21 17 18 5.5 5.7 9.7 16 13
DPT 0.280 0.561 0.841 1.12 2.80 5.61 8.41 11.2
2.1 7.0 1.7 9.4 8.4 6.0 17 12
7.0 7.2 4.7 4.3 1.4 5.1 6.4 3.5
DBT 0.317 0.634 0.952 1.27 3.17 6.34 9.52 12.7
19 10 1.4 1.3 3.4 5.3 6.5 0.72
3.8 7.7 6.2 1.1 2.9 6.0 10 5.3
7 E+05 6 E+05
TBT
TPT
5 E+05
DBT
4 E+05
DPT
3 E+05 2 E+05 1 E+05 0 E+00 0
100
200
Time ( sec) Fig. 1. Representative chromatogram showing phenyltin (MPT, 6.38 ng mL 1), butyltin (MBT, 6.83 ng mL 1), triphenyltin (TPT, 5.00 ng mL 1), diphenyltin (DPT, 5.61 ng mL 1), dibutyltin (DBT, 6.34 ng mL 1), and tributyltin (TBT, 5.92 ng mL 1) in human serum matrix separated in less than 3 min.
0.3 0.25
Area/IS Area
Accuracy mean %RE
a
0.2 0.15
29 21 15 18 15 11 23 15
4.9 7.7 3.5 7.1 3.3 6.2 5.7 3.9
Triplicate preparations at each nominal concentration level.
3.3. Specificity
0.1 0.05 0 0.000
TBT 0.296 0.592 0.888 1.18 2.96 5.92 8.88 11.8
5.000
10.000
15.000
Nominal [Sn] (ng/mL)
Fig. 2. Representative linear regression for the organotin compound, MPT, in human serum obtained during day 1 validation experiment; linear regression equation: y¼0.0220959x þ0.000785919; r¼ 0.9995.
Specificity was demonstrated by comparing the instrument response for each OTC in triplicate matrix blank samples to the response obtained for each species at the experimental limit of quantitation (ELOQ) level. The ratio range of blank instrument response to ELOQ response for all six species ranged from nondetect to 11%.
K.E. Levine et al. / Talanta 140 (2015) 115–121
Table 2 Interday precision and accuracy for organotin species. Nominal (OTC)a (ng mL 1)
Accuracy mean %RE
MPT 0.319 0.638 0.957 1.28 3.19 6.38 9.57 12.8
1.5 3.2 2.5 1.2 0.89 4.7 8.9 11
MBT 0.342 0.683 1.03 1.37 3.42 6.83 10.2 13.7
1.6 4.2 4.0 4.5 0.01 1.8 6.3 11
TPT 0.250 0.500 0.750 1.00 2.50 5.00 7.50 10.0
12 14 9.9 14 11 8.8 3.6 2.6
Precision %RSD
9.1 3.9 4.5 2.5 3.1 4.8 9.0 5.4
0.067 5.2 5.0 5.3 5.5 1.6 12 3.6
27 17 11 9.8 7.4 15 3.3 14
DPT 0.280 0.561 0.841 1.12 2.80 5.61 8.41 11.2
3.0 5.8 9.0 4.3 0.56 1.6 8.3 13
4.5 9.0 5.5 5.1 11 2.7 11 2.5
DBT 0.317 0.634 0.952 1.27 3.17 6.34 9.52 12.7
9.5 6.1 3.4 0.95 4.3 2.7 7.8 6.6
14 11 2.2 2.4 5.5 1.6 6.6 1.5
TBT 0.296 0.5920 0.888 1.18 2.96 5.92 8.88 11.8
8.1 6.1 8.8 2.5 3.0 2.8 15 14
18 13 9.1 7.1 9.3 2.3 10 4.7
a
Triplicate preparations at each nominal concentration level.
119
(730% at the ELOQ level), respectively, for at least four of the six species. This acceptance criterion for intraday precision was met as all species except TPT passed; the intraday accuracy criterion was also achieved as all species except TPT and TBT passed. For DPT, the intraday %RE was marginally outside of 715% for one concentration level at high end of the linear range, but the intraday accuracy was considered fit for purpose because of the high level of accuracy observed at and near the ELOQ level. Interday precision and accuracy criteria were achieved met as all species passed, with the exception of TPT, with marginally failing precision only. Overall, intraday and interday precision and accuracy were considered acceptable for the OTC suite. The failing intraday accuracy for TPT was attributed to degradation during the first day of validation experiments. Determined TPT concentrations were found to gradually decline with each successive QC replicate interspersed throughout the analysis. Although passing, DPT exhibited generally positive %RE, which may indicate that this species is a degradation product of TPT. Interday TPT accuracy passed because analytical run length was shorter for validation experiments on subsequent days. Although TBT failed intraday accuracy with generally positive %RE data, a trend could not be identified on closer investigation of the individual matrix QC replicates. 3.5. Sensitivity Seven matrix QC samples were prepared at the lowest nominal concentration and analyzed to challenge method sensitivity and establish quantitation and detection limits. In order to conservatively be defined as the ELOQ, the acceptance criteria for matrix QC replicates were defined as 7 30% RE and r30% RSD. These criteria were comfortably met for all species (Table 3), providing confidence in accurate, precise measurements at ELOQ levels. The limit of detection (LOD) for each species is also presented and was calculated by multiplying the standard deviation from the determined matrix QC analyte concentrations by 3. 3.6. Stability evaluation Stability was evaluated under several different storage scenarios to determine the optimal application of the study method. Post-extraction refrigerator storage stability was evaluated over 14 days. Although relatively high levels of accuracy and precision were still observed for the butyltin species after 14 days in refrigerated extracts, stability was not observed for the phenyltin species, with apparent degradation of DPT and TPT to MPT (Fig. 3 and Table S4, Supplementary information). Stability data from fortified serum aliquots stored under nominal freezer conditions for up to 60 and 90 days, respectively, are presented in Fig. 4 and Table S5 (Supplementary information). The overall trend for the OTC suite was clear degradation over time. The average relative error of the OTC species stored at ultracold freezer conditions after Table 3 Experimental limits of quantitation and limit of detection. OTC
ELOQa (OTC) (ng mL 1)
Accuracy mean %RE
Precision % RSD
LODa (ng mL 1)
MPT MBT TPT DPT DBT TBT
1.60 1.71 1.25 1.40 1.59 1.48
13 7.5 15 0.69 19 18
6.8 8.9 7.1 5.8 4.1 5.1
0.350 0.500 0.300 0.250 0.250 0.250
3.4. Precision and accuracy Intraday and interday precision and accuracy are presented in Tables 1 and 2, respectively. Intraday and interday precision, expressed as percent relative standard deviation (%RSD), and accuracy, expressed as percent relative error (%RE), of determined concentrations for matrix QC samples were required to be r15% (r 30% at the ELOQ level) and within 715% of the nominal value
a ELOQ and LOD expressed as ng analyte mL 1 of serum; accounts for 5 dilution factor.
K.E. Levine et al. / Talanta 140 (2015) 115–121
20
100 % Relative Error
% Relative Error
120
50 0 -50
0 -20 -40 -60 -80
- 100 0
7 Time (days)
-100
14
50 0 -50
-100 0
% Relative Error
20
40 Time (days)
0
20 Time (days)
60
20 % Relative Error
% Relative Error
100
0
5
10 Time (days)
0 -20 -40 -60 -80 -100 40
60
Fig. 4. Compound stability for OTCs in human serum at freezer storage conditions. Key: A – low concentration stability samples; B – high concentration stability samples. Diamond – MPT; square – MBT; triangle – TPT; X – DPT; þ – DBT; and circle – TBT.
100 50 0 -50 - 100 0
5
10 Time (days)
Fig. 3. Compound stability for OTCs in human serum at refrigerator conditions. Key: A – low concentration stability samples; B – medium concentration stability samples; C – high concentration stability samples. Diamond – MPT; square – MBT; triangle – TPT; X – DPT; þ – DBT; and circle – TBT.
a time period of 90 days was 43% for MBT, 32% for MBT, 68% for TPT, 42% for DPT, 51% for DBT, and 37% for TBT with precision values less than RSD ¼15%. In addition, fortified serum aliquots subjected to three complete freeze–thaw cycles showed similar degradation across all analytes in the suite and all nominal concentration levels. The mean %RE ranged from 48% to 38% (MPT), 47% to 20% (MBT), 91% to 78% (TPT), 41% to 25% (DPT), 52% to 34% (DBT), and 54% to 45% (TBT). Based on these overall stability findings, it is recommended that the sample collection and analytical laboratories closely coordinate activities to minimize species degradation. 3.7. Analysis of serum from Danish women The method was applied to serum from 64 Danish women participating in the Snart-Foraeldre/MiljØ (Soon-Parents/Environment) Study, a prospective cohort study of 500 Danish women between the ages of 19 and 40 years who have recently discontinued birth control in order to become pregnant. The concentrations of OTC species were found to be in the following ranges (ng mL 1): MPT, o LOD – 1.08; MBT, oLOD – 10.929; TPT, oLOD – 0.862; DPT, oLOD – 0.370; DBT, oLOD – 4.495; TPT,
oLOD – 0.513. The samples all exhibited relatively low levels of the OTC suite, suggesting minimal risk of exposure to OTCs in this population. Additional data will be provided in a subsequent manuscript, including confirmatory total tin measurements to establish mass balance.
4. Conclusion A UPLC–SF-ICP-MS method was successfully developed and validated for the determination of a suite of six OTCs (MPT, MBT, TPT, DPT, DBT, and TBT) in human serum. The employed sample preparation and chromatographic procedures allowed for separation of the analytes in less than three minutes without derivatization, resulting in considerably less sample preparation time than that required for many equivalent HPLC or GC separations. This shortened analysis time may offer a significant advantage in that it can facilitate higher sample throughput while minimizing potential instrumental drift and degradation of OTC species, especially TPT. In addition, use of the SF-ICP-MS as the detector enabled sensitive determination of the analyte suite. Overall, sensitivity for the OTCs studied here compares favorably with other literature reports (Table S6) for environmental and biological matrices. Several observations were noted which emphasize the importance of close coordination between the analytical and sample collection laboratories. Careful screening of all lots of employed reagents (most notably, tropolone) prior to use is essential in maintaining low background levels of Sn. Further, the sample collection laboratory could benefit from evaluation of multiple lots of biological fluid collection tubes to minimize the potential for contamination. Care must be taken to thoroughly clean all labware, including chromatography vials and other new, ‘as received’ labware. Ideally, stability considerations will be taken into account when coordinating measurements. To minimize potential degradation, standards and samples should be prepared on
K.E. Levine et al. / Talanta 140 (2015) 115–121
the day of analysis and samples should be analyzed as soon as practical upon receipt. Finally, to the extent possible, study samples should be preserved at the coldest temperature practical and transferred to the analytical laboratory for immediate processing.
Acknowledgements This research was supported by the National Institute of Health/ National Institute of Environmental Health Sciences (NIH/NIEHS; HHSN27320110003C). The authors would like to acknowledge Mr. Bradley J. Collins, Dr. Michelle Hooth, and Dr. Esra Mutlu for their review of this manuscript prior to formal submission.
Appendix A. Supplementary information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.talanta.2015.03.022.
References [1] J.B. Graceli, et al., Reprod. Toxicol. 36 (2013) 40–52. [2] A.O. Sunday, B.A. Alafara, O.G. Oladele, Chem. Speciat. Bioavailab. 24 (4) (2012) 216–226. [3] J. Kuballa, et al., Analyst 120 (1995) 667–673. [4] Y. Suzuki, et al., J. Chromatogr. B 868 (2008) 116–119. [5] M. Furdek, et al., Mar. Pollut. Bull. 64 (2012) 189–199.
121
[6] Y. Liu, Y. Wan, L. Guo, Anal. Lett. 47 (2014) 343–355. [7] M. Hoch, Appl. Geochem. 16 (2001) 719–743. [8] F. Guo, X. Lu, Q. Xu, World J. Emerg. Med. 1 (2) (2010) 122–125. [9] Z. Cui, et al., Talanta 85 (2011) 1028–1033. [10] K. Kannan, et al., Arch. Environ. Contam. Toxicol. 58 (2010) 901–907. [11] J. Gao, et al., Environ. Monit. Assess. 185 (2013) 3831–3837. [12] K.K.Y. Ho, K.M.Y. Leung, Mar. Pollut. Bull. 85 (2014) 634–640. [13] C. Devos, F. David, P. Sandra, J. Chromatogr. A 1261 (2012) 151–157. [14.] G.W. Bryan, et al., J. Mar. Biol. Assoc. U. K. 66 (1986) 611–640. [15] P.E. Gibbs, G.W. Bryan, J. Mar. Biol. Assoc. U. K. 66 (1986) 767–777. [16] Z. Yu, et al., Food Chem. 119 (2010) 364–367. [17] P. Rantakokko, et al., Sci. Total Environ. 408 (2010) 2474–2481. [18] I.Bd Castro, F. Perina, G. Fillmann, Environ. Monit. Assess. 184 (2012) 1781–1799. [19] T. Liu, et al., J. Food Process Eng. 34 (2011) 1125–1143. [20] C.Rd Oliveira, et al., J. Hazard. Mater. 181 (2010) 851–856. [21] Y. Mino, et al., J. Health Sci. 54 (2) (2008) 224–228. [22] M. Bravo, et al., Microchem. J. 106 (2013) 95–101. [23] G. Kabiersch, et al., Anal. Chem. 85 (2013) 5740–5745. [24] M.G. Ikonomou, et al., J. Chromatogr. A 975 (2002) 319–333. [25] T. Horiguchi, et al., Mar. Environ. Res. 76 (2012) 56–62. [26] J. Carpinteiro, I. Rodriguez, R. Cela, Anal. Bioanal. Chem. 380 (2004) 853–857. [27] K. Kannan, K. Senthilkumar, J.P. Giesy, J. Chromatogr. A 33 (1999) 1776–1779. [28.] E. Beceilo-Gonzalez, A. Guimaraes, M.F. Alpendurada, J. Chromatogr. A 1216 (2009) 5563–5569. [29] J.I. Cacho, et al., J. Chromatogr. A 1279 (2013) 1–6. [30] C. Coscolla, et al., Talanta 119 (2014) 544–552. [31] M. Bravo, et al., J. Chromatogr. A 1223 (2012) 9–14. [32] I.R. Pereiro, et al., Anal. Chem. 68 (23) (1996) 4135–4140. [33] T. Sano, et al., J. Chromatogr. A 1217 (2010) 4344–4346. [34] L. Yang, J.W.H. Lam, J. Anal. At. Spectrom. 16 (2001) 724–731. [35] S. Chiron, et al., J. Chromatogr. A 879 (2000) 137–145. [36] C. Dietz, et al., J. Chromatogr. A 1153 (2007) 114–129. [37] K. Dubalska, et al., Crit. Rev. Anal. Chem. 43 (2013) 35–54. [38] S. White, et al., J. Chromatogr. A 794 (1998) 211–218.
