Chemicals, standards and reference materials
The following commercial chemicals were used: ortho phosphoric acid (p.a., or TraceSELECT Ultra) from Fluka (Buchs, Switzerland); pyridine from Merck (Merck, Darmstadt, Germany); and hydrogen peroxide 30 % (p.a.), aqueous ammonia 25 % (suprapure), 65 % nitric acid (p.a.), and formic acid (p.a.) from Roth (Carl Roth, Karlsruhe, Germany). Chemicals were used without further purification except for the nitric acid which was distilled in a quartz sub-boiling distillation unit. Water used throughout was from a Milli-Q Academic water purification system (Millipore GmbH, Vienna, Austria) with a specific resistivity of 18.2 MΩ*cm.
Individual standard solutions (1000 ± 3 μg L−1
in 2 % nitric acid) for total element determinations of As, Cd, Mo, Pb, Sb, Se, U, W, and Zn (in the urine samples) and Ge, In, and Lu (internal standards) were obtained from CPI International (Santa Rosa, CA, US). For arsenic speciation, stock solutions containing 1000 mg As L−1
of each of the following species were prepared in water: arsenite (As(III) and arsenate (As(V)) prepared from NaAsO2
O, respectively, purchased from Merck (Darmstadt, Germany); dimethylarsinate (DMA) prepared from sodium dimethylarsinate purchased from Fluka (Buchs, Switzerland); methylarsonate (MA) prepared in-house from sodium arsenite and methyl iodide (Meyer reaction); and arsenobetaine (AB), as the bromide salt, prepared in-house following the method of Cannon et al.11
The purity of the synthesized standards (MA and AB) was established by NMR and HPLC/mass spectrometry. Other arsenic standards (trimethylarsine oxide, arsenocholine, tetramethylarsonium ion, oxo and thio-dimethylarsinylethanol and oxo- and thio-dimethylarsinylacetic acid) were prepared as previously reported;12,13
these standards were used to check the identity of minor peaks which occasionally appeared in the chromatograms.
The certified reference materials for total element measurements were NIST 1643e, trace elements in water (National Institute of Standards & Technology, Gaithersburg, Maryland, US) certified for As, Cd, Mo, Pb, Sb, Se, & Zn; and NIES No. 18, human urine (National Institute for Environmental Studies, Tsukuba, Japan) certified for As, Se & Zn. In addition, Seronorm™ control urine (Sero AS, Billingstad, Norway) and an in-house urine sample served as non-certified reference materials. The certified reference material for determining arsenic species was NIES No 18, human urine, certified for AB and DMA. Our in-house reference urine was used as a control for iAs, MA, DMA, and AB.
Total element determinations and arsenic speciation analyses were performed with an Agilent 7700x ICPMS (Agilent Technologies, Waldbronn, Germany) equipped with a Micro Mist nebulizer (Glass Expansion, Melbourne, Australia) and a Scott double pass spray chamber. For total element analysis, the integrated auto sampler G3160B from Agilent Technologies was used. The sample cups (1.5 mL capacity, made from polystyrene) were purchased from Sarstedt AG & Co (Sarstedt, Nurnbrecht, Germany). Data acquisition and evaluation was performed with ICPMS Masshunter B.01.01 software.
For arsenic speciation analyses, an Agilent 1100 Series HPLC system consisting of a solvent degassing unit, a quaternary pump, an autosampler and a thermostated column compartment was used as the chromatographic system. The outlet of the HPLC column was connected via PEEK capillary tubing (0.125 mm i.d.) to the nebulizer of the ICPMS. The ion intensity at m/z 75 (75As) was monitored using the ICPMS Masshunter B.01.01 software. Additionally, the ion intensities at m/z 53 (40Ar13C/40Ar12C1H) and 77 (40Ar37Cl) were monitored. Instrumental settings used throughout this work were optimized before each run using the software autotune function. Typical values were: forward power 1600 W; carrier gas flow 1.00 L min−1; nebulizer pump 0.1 rps (1.02 mm inner diameter tubing); extraction lens 1 0 V, extraction lens 2 −195 V; omega bias −80 V; omega lens 10 V; cell entrance −30 V; cell exit −50 V; deflect 17 V; plate bias −50 V; octopole RF 170 V octopole bias −8 V; quadrupole bias −3 V. With these settings the performance of the instrument was 7Li >5*104 cps/μg L−1, 89Y >1.3*105 cps/μg L−1; 205Tl >1.3*105 cps/μg L−1 with a 156CeO/140Ce ratio < 0.01.
Cd (m/z=111), Sb (m/z=121), W (m/z=182), Pb (m/z=208), U (m/z=238) and the internal standards Ge (m/z=74), In (m/z=115) and Lu (m/z=175) were measured in the normal mode. One channel was measured per isotope. In the collision gas mode, 4.0 mL min−1 He was used and the instrument settings were changed to cell entrance −40 V; cell exit −60 V; deflect 3 V; plate bias −60 V; octopole RF 190 V octopole bias −18 V; quadrupole bias −15 V. Under these conditions, the ratio of 156CeO/140Ce was typically < 0.006. Zn (m/z=66), As (m/z=75), Se (m/z=78), Mo (m/z=95), and the internal standards Ge (m/z=74) and In (m/z=115) were measured in the helium collision gas mode. One channel was measured per isotope. From each sample, five replicate measurements were recorded.
Sample collection and storage
Urine was collected from the Strong Heart Study participants on three occasions from 1989 to 1999 (1989–1991, 1993–1995, and 1998–1999). Following a physical examination conducted in the morning, participants were asked to void urine into a plastic cup, and approximately 8 mL was transferred to a 14 mL polypropylene screw-cap tube. These samples were frozen within 1–2 hours of collection and shipped on dry ice to the Medstar Health Research Institute, Hyattsville, MD, USA where they were stored at −80 °C. In the subsequent years, the samples had been thawed briefly and sub-samples removed for analysis of albumin and creatinine. During 2009 and 2010, portions (0.5 – 1.0 mL) of the samples were shipped on dry ice to Graz University, Austria where they were stored at −80 °C until analysis.
Total element measurements
A portion (170 μL) of the thawed urine sample was transferred to a Plastibrand® microtube (Brand GmbH + Co KG, Wertheim, Germany) and diluted to 1.76 g with 10 % v/v nitric acid containing the internal standards Ge, In, Lu at a concentration of 44 μg L−1. The mixture was centrifuged for ten minutes at 1.3 × 104 g. The resultant supernatant was then analysed for As, Cd, Mo, Pb, Sb, Se, U (only for some samples), W, and Zn by ICPMS.
Determination of arsenic species
The remainder of the thawed urine sample was filtered through a 0.2 μm Nylon filter (Whatman GmbH, Dassel, Germany) into a 250 μL polypropylene crimp vial (Agilent Technologies). This filtered sample was analysed directly by anion-exchange HPLC/ICPMS. Additionally, a portion (90 μL) of the filtered sample was removed from the HPLC vial and 10 μL of H2O2 were added, to convert any trivalent- and thio-arsenicals to their pentavalent and/or oxygenated forms, and the mixture was allowed to stand for at least two hours at a temperature > 23°C before analysis by anion-exchange HPLC/ICPMS.
The anion-exchange HPLC conditions (identical for both non-oxidised and oxidised urine samples) were: PRP-X100 column (4.6 mm × 150 mm, 5 μm particles; Hamilton Company, Reno USA) at 40°C with a mobile phase of 20 mM aqueous phosphoric acid adjusted with aqueous ammonia to pH 6 at a flow rate of 1 mL min−1
. Injection volume was 20 μL. A carbon source (1% CO2
in argon) was introduced directly to the plasma, as previously described for selenium,14
to provide a 4-5-fold increase in sensitivity. The CO2
was introduced via the T-piece of the high matrix sample introduction kit and the optional gas was set to 0.17 L min−1
. Under these chromatographic conditions, As(III) elutes near the void volume, very close to AB and most other cationic arsenic species. This void-volume peak was assigned as AB + As(III) in the non-oxidised sample, and as AB in the oxidised sample (), based on the premise that AB is the only arsenic cation found in significant quantities in urine (see below).15
The total iAs content [As(III) + As(V)] was obtained from the As(V) peak in the oxidised sample. For all HPLC runs, peaks were quantified against the respective standard. Calibration was usually performed in the range 0.10 to 20.0 μg As L−1
(six-point calibration curve); limit of detection was 0.1 μg As L−1
for iAs [As(V) peak], MA, DMA and AB, and the intra-assay coefficient of variation was better than 5 % for all species.
Figure 1 Anion-exchange HPLC/ICPMS chromatograms of a urine sample (grey line) and the same sample after treatment with H2O2 (black line). The expanded regions show the two areas where changes occurred. HPLC conditions were: PRP-X100 column (4.6 mm × 150 (more ...)
The premise that AB was essentially the only cationic arsenic species in the urine samples was tested by performing cation-exchange HPLC/ICPMS on 188 samples that had shown a significant peak at the void volume during anion-exchange HPLC/ICPMS of the oxidized samples. A Zorbax 300-SCX column (4.6 mm × 150 mm, 5 μm particles; Agilent Technologies) at 30°C was used with a mobile phase of 10 mM pyridine at pH 2.3 (adjusted with formic acid) at a flow rate of 1.5 mL min−1. The injection volume was 10 μL. ICPMS was used as a detector with the settings described above for anion-exchange HPLC/ICPMS.
Quality assurance/quality control
Before dispatch to the analytical laboratory in Graz, each sample was assigned a unique six digit code. A total of 5,095 urines samples were analysed over a two-year period in batches of 79 samples. In each batch, samples were run together with calibration blanks and checks (every ten samples), sample preparation blanks and spikes, Reference Materials, and “in-house” reference urine under the QA/QC regime described below. A typical batch sample sequence is summarised in .
Analysis sequence for trace element and arsenic species analyses (typical batch of 79 samples)
Calibration blanks and calibration standards were used to construct the primary calibration curves for quantification of total element content in the samples. They were prepared in the same acid matrix as the urine samples. The standard curve was constructed from blank + five concentrations of the element spanning the range of sample element concentrations. Analysis of samples proceeded only when the correlation coefficients of the standard curves were >0.999. Calibration checks included the re-analysis of selected calibration standards every ten samples throughout each sample analytical run. They served as a check of the calibration curve to account for changes in instrument sensitivity during the analysis run.
Sample preparation blanks were portions of Milli-Q water which underwent the same sample handling procedure applied to the urine samples. They were used to monitor and correct for procedural and analytical contamination resulting from reagents, glassware/plastic-ware, handling, etc. Sample preparation spikes were sample preparation blanks that were spiked with known quantities of arsenic compounds. They were used to evaluate sample preparation and analysis performance (recovery, reproducibility, etc.).
Reference Materials – namely NIST 1643e, NIES No 18 and the control material Seronorm™ for total element measurements, and NIES No 18 and the in-house urine reference sample for both total element and arsenic species measurements – were analysed with every batch of urine samples.
Unbeknown to the analytical laboratory, 47 samples were supplied in duplicate which provided an unbiased check on the repeatability of the methods. Additionally, the reference samples were independently measured for total arsenic and arsenic species by an external laboratory in Norway at the National Institute of Nutrition and Seafood Research (NIFES). A further check was made by our participation in a round robin exercise organized by Recipe Chemicals + Instruments GmbH (Munich, Germany).