Sen et al.: Journal of AOAC International Vol. 98, No. 2, 2015 517 RESIDUES AND TRACE ELEMENTS
Development and Validation of a Simple and Robust Method for Arsenic Speciation in Human Urine Using HPLC/ICP-MS Indranil Sen, Wei Zou, Josephine Alvaran, Linda Nguyen, Ryszard Gajek, and Jianwen She1
California Department of Public Health, Biochemistry Section, Environmental Health Laboratory Branch, Richmond, CA 94804
In order to better distinguish the different toxic inorganic and organic forms of arsenic (As) exposure in individuals, we have developed and validated a simple and robust analytical method for determining the following six As species in human urine: arsenous (III) acid (As-III), As (V) acid, monomethylarsonic acid, dimethylarsinic acid, arsenobetaine (AsB), and arsenocholine. In this method, human urine is diluted using a pH 5.8 buffer, separation is performed using an anion exchange column with isocratic HPLC, and detection is achieved using inductively coupled plasma-MS. The method uses a single mobile phase consisting of low concentrations of both phosphate buffer (5 mM) and ammonium nitrate salt (5 mM) at pH 9.0; this minimizes the column equilibration time and overcomes challenges with separation between AsB and As-III. In addition, As-III oxidation is prevented by degassing the sample preparation buffer at pH 5.8, degassing the mobile phase online at pH 9.0, and by the use of low temperature (–70°C) and flip-cap airtight tubes for long term storage of samples. The method was validated using externally provided reference samples. Results were in agreement with target values at varying concentrations and successfully passed external performance test criteria. Internal QC samples were prepared and repeatedly analyzed to assess the method’s longterm precision, and further analyses were completed on anonymous donor urine to assess the quality of the method’s baseline separation. Results from analyses of external reference samples agreed with target values at varying concentrations, and results from precision studies yielded absolute CV values of 3–14% and recovery from 82 to 115% for the six As species. Analysis of anonymous donor urine confirmed the well-resolved baseline separation capabilities of the method for real participant samples.
A
rsenic (As) has been synonymous with poison for ages. Its compounds are toxic and potentially carcinogenic (1) in humans, depending on the chemical nature and oxidation state at various physiological concentrations (2). Several As compounds or species have been identified in biological or environmental systems with Received May 1, 2014. Accepted by AK December 16, 2014. 1 Corresponding author’s e-mail: [email protected] DOI: 10.5740/jaoacint.14-103
both natural and anthropogenic sources (3–6). Because of its widespread mobility from geological deposits into groundwater, chronic As exposure through drinking water is considered one of the biggest threats to human health (3). Prolonged exposure to low doses of As may result in hyperkeratosis, keratosis of the palms and soles, dermatitis, and skin cancer (7–9). Acute As poisoning symptoms may include diarrhea, blood in the urine, muscle cramping, hair loss, stomach pain, and convulsions. Organs in the body affected by As poisoning are the lungs, skin, kidneys, and liver. As is also related to heart disease (hypertension related cardiovascular disease; 10), cancer (11), stroke (cerebrovascular diseases; 12), chronic lower respiratory diseases (13), and diabetes (7, 8). Treatment of chronic As poisoning includes removal of the toxic source as well as chelation therapy (14, 15). The U.S. Environmental Protection Agency (EPA) has established a drinking water standard for As at 10 ppb to protect users from the effects of long-term, chronic exposure (16). As can be ingested as one or more species, and the species can be metabolized into several different inorganic and organic forms. Inorganic species include arsenous (III) acid (As-III) and As acid (As-V), and organic species include the methylated species such as monomethylarsonic acid (MMA) and dimethylarsinic acid (DMA), as well as arsenobetaine (AsB) and arsenocholine (AsC), which are primarily ingested from seafood and stay unaffected in the body. As inactivates enzyme systems by combining with the sulfhydryl (-SH) groups of the enzymes present in the liver, which are vital for the living system (9, 17). Methylation of As in biological systems, also known as biomethylation, involves a methyl transfer reaction by S-adenosyl methionine to form MMA and DMA (12, 18, 19). As species are excreted in urine, which is routinely used as the matrix to measure the internal dose of As species concentrations. Analyses reflect recent exposure ranging from a few hours to 2 days (20). Chemical structures of six well-known As compounds and their respective LD50 (lethal dose 50%) values (for rats) are listed in Table 1 (21–23). LD50 values provide a measurement of acute toxicity and estimates the dose of a given chemical required to kill 50% of a test population (typically rats). According to the LD50 values for the six As species presented, the most toxic As species are the inorganic As compounds As-III and As-V (24), followed by MMA and DMA, then AsB and AsC (which are least toxic). Since total As concentrations alone cannot reflect the real hazard and exposure risk of an individual (25), determining the concentrations of different As species is critical in distinguishing toxic levels in the body. In order to measure the concentration of each As species and its metabolites, several methods have been reported using separation techniques such as LC and detection methods such
518 Sen et al.: Journal of AOAC International Vol. 98, No. 2, 2015 Table 1. Chemical structures, abbreviations, and LD50 values of As species Name Arsenite
Abbreviation As-III
Structure
LD50, mg/kg 14
Table 2. HPLC operational parameters and settings Instrumental conditions Injector, µL
50
Draw speed, µL/min
200
Injection speed, µL/min
200
Pump flow rate, mL/min
1.0
Pump stop time, min Arsenate
As-V
20
Instrumental settings
Flow pressure (column), bar
15 100–120
Experimental Monomethylarsonate
MMA
700–1800
Dimethylarsonate
DMA
700–2600
Arsenobetaine
AsB
>10000
AsC
>10000
Arsenocholine
as inductively coupled plasma (ICP) spectrometry and atomic fluorescence spectrometry both with and without hydride generation (20, 26–32). A common speciation technique includes the determination of seven As compounds in 16 min by subsequent use of two sets of mobile phases at pH 8.7 and 8.0 in a gradient HPLC pump coupled with ICP-MS with a dynamic reaction cell (28, 33). A similar technique using mobile phase compositions at pH 6.2 and 3.0 has also been published (29, 33). Another known method included the oxidation of As-III to As-V prior to separation using a mobile phase at pH 6.0 (34); however, individual As-III or As-V levels could not be determined by the method. With these methods, analyses are relatively complicated and limited to the number of species that can be determined. The limitation of these methods demonstrates a need to develop a relatively simpler and more robust method capable of measuring most toxic As species. Our objective was to develop a method that would require only an isocratic pump with a single mobile phase composition at a single pH level with the ability to analyze the most prevalent As compounds within a reasonable time. Herein, we report the development and validation of a faster and simpler yet accurate and sensitive analytical method to determine the predominant six As species in human urine samples using an isocratic HPLC pump coupled with an ICP-MS.
Instrumentation The separation of As species was performed with an Agilent (Santa Clara, CA) 1200 series HPLC system consisting of a solvent degassing unit, isocratic pump, autosampler, and column compartment. Typical instrument parameters are reported in Table 2. A Hamilton PRP-X100 column (150 × 4.6 mm with 5 µm particles; Hamilton Cat. No. 79174), Hamilton guard column kit (Hamilton Cat. No. 79383), and Hamilton guard column refill cartridges (Hamilton Cat. No. 79385) were used. Hamilton products were supplied by Fisher Scientific, Pittsburgh, PA. A Hamilton PRP-X100 column is used to separate mixtures of organic and inorganic compounds by using a broad range of mobile phases. It is packed with polystyrene-divinylbenzene resin that is inert with organic solvents as well as aqueous buffers with a wide pH range of 1–13. The column has an increased lifetime and can maintain column-to-column reproducibility. It is very robust, can be cleaned and regenerated easily for persistent chemicals, and is quite suitable for analysis of organic and inorganic compounds in aqueous or environmental samples. Periodic (after every 100 samples) column regeneration [backflush with 2% HNO3 and methanol (CH3OH; Fisher Scientific, Cat. No. A412-1)] and frequent guard column replacements were performed to avoid chloride and other buildup in the column that may contribute to argon chloride (40Ar35Cl) formation causing signal interferences (35). Polyetheretherketone capillary tubing (blue, 1/16 × 0.01 inch id, Agilent Cat. No. 5042-6463) was used to connect the chromatographic outlet to the nebulizer of the ICP-MS instrument. The flow pressure at the HPLC column was monitored at 100–120 bar. Pressure higher than 130 bar in general indicates congestion in the HPLC column or guard column and a need for regeneration and replacement of the guard column. An Agilent 7700 CE ICP-MS instrument equipped with a micromist nebulizer and a helium (He) collision cell (36, 37) was used to detect the As species. A liquid argon tank equipped with an approved gas regulator was used for the ICP-MS instrument. Settings were optimized before each run by using autotune functions. Typical values are reported in Table 3. Mobile Phase A degassed solution of 5 mM ammonium phosphate dibasic [(NH4)2HPO4 (Fisher Cat. No. A686-500)], 5.0 mM ammonium nitrate (NH4NO3; Sigma-Aldrich Cat. No. 256064; St. Louis, MO), 2% CH3OH and 20 ppb germanium (Ge; CPI International
Sen et al.: Journal of AOAC International Vol. 98, No. 2, 2015 519 Table 3. ICP-MS operational parameters and settings Instrumental conditions
Instrumental settings
RF power, W
1550
RF matching, V
1.80
Sample depth, mm
8.0
Carrier gas flow, L/min
1.07
Nebulizer pump, rps
0.3
Spray chamber temperature, ºC
2
Extraction lens 1, V
6.4
Extraction lens 2, V
–135
Omega bias, V
65
Omega lens, V
9.4
Cell entrance, V
–40
Cell exit, V
–56
Deflect, V
13.6
Plate bias, V He flow (for collision cell), mL/min Octapole RF, V
–50 1.9–2.1 160
Octapole bias, V
–8
Energy discrimination, V
5.0
Santa Rosa, CA, Cat. No. S4400-1000201) at pH 9.0 was used as the mobile phase. Diluent A degassed solution of 10 mM potassium phosphate monobasic (KH2PO4; Fluka, Sigma-Aldrich Cat. No. 60216), 2% CH3OH, and 20 ppb Ge at pH 5.8 was used as a diluent. As Standards Individual As species stock standards were obtained from the following commercial sources: AsB [formula weight (FW) 178.06, Fluka, Sigma-Aldrich, Cat. No. 11093]; AsC (FW 245, Argus Chemicals, Vernio, Italy, Cat. No. AR60010); MMA (FW 292, Supelco, Sigma-Aldrich Cat. No. PS-281,); DMA (FW 138, Supelco, Sigma-Aldrich Cat. No. PS-51); 1000 mg/L As-III in 2% HNO3 (SPEX Industries Inc., Edison, NJ, Cat. No. SPEC-AS3); 1000 mg/L As-V in 2% HNO3 (SPEX Industries Inc., Cat. No. SPEC-AS5); and 1000 mg/L Ge as an internal standard (CPI International, Cat. No. S4400-1000201). As species stock standards (1000 ppm) were prepared from commercially available solid compounds and kept in tightly sealed plastic containers at 4°C. Certified 1000 ppm As-III and As-V standard solutions were available commercially (above). In order to prepare multispecies intermediate standard solutions, 100 µL 1000 ppm standard stock solutions of each As species was added into a disposable 15 mL conical tube (Becton-Dickinson, Franklin Lakes, NJ, Cat. No. 2097). The total volume was made up to 10 mL with freshly prepared and degassed diluent solution for a final concentration of 10 ppm for each of the six As species. This multispecies intermediate standard solution remained stable for 1 week when stored at –70°C. Multispecies 100 ppb working standard solutions were prepared fresh on the day of use. To prepare the working standard solution, the 10 ppm intermediate
standard solution was diluted 100x. A 100 µL amount of the multispecies intermediate standard stock solution mixture was added to the freshly prepared and degassed diluent solution for a final volume of 10 mL. This solution was stable at 4°C for 24 h stored in the dark. Multispecies daily calibration standard solutions were prepared fresh on the day of use from the 100 ppb multispecies working standard solution. Total As Method An Agilent 7500 CE ICP-MS instrument equipped with a micromist nebulizer and He collision cell was used for the analysis of total As in human urine. Data acquisition and evaluation were performed with Agilent MassHunter software. Instrumental settings were optimized before each run by using autotune functions, and the parameters were compared with the applicable ranges. Typical values were as follows: radiofrequency (RF) power, 1550 W; RF matching 1.89 V; sample depth 8.0 mm; torch H 0.4 mm; torch V 0 mm; carrier gas flow 0.96 L/min; makeup gas: 0.15 L/min; nebulizer pump 0.1 rps, S/C temperature 2°C; extraction lens 1: 0 V; extraction lens 2: –125 V; omegas bias ce 24 V; omega lens ce 0.2 V; cell entrance –30 V; QP focus: 2 V; cell exit –30 V; deflect 13.6 V; plate bias –50 V; He flow (for collision cell) 4.6 mL/min; H2 gas 3 mL/min; octapole RF 180 V; octapole bias –4 V; and energy discrimination 8.0 V. Human urine was diluted (1:20) using a diluent solution containing the following: 2% HNO3, 0.2% sulfamic acid, 100 µg/L gold (Au) standard solution, 1.5% ethanol, 0.0005% Triton X-100, 50 µg/L Ge (Spec CertiPrep, Cat. No. PLGE92Y) 50 µg/L rhodium (Rh; Spec CertiPrep, Cat. No. PLRH32Y), and 60 µg/L rhenium (Re; Spec CertiPrep, PLRE9-2Y). The samples were prepared by filling a 15 mL conical tube (brand and lots prescreened for contaminants) with 4500 µL diluent, 250 µL 2% HNO3, and 250 µL urine sample. The tubes were inverted to mix. Samples (a) Internal QC.—Internal QC samples were prepared and characterized by the laboratory. Degassed diluent pH 5.8 buffer was used to prepare two levels of surrogate specimen material; a QC-Low level pool with approximately 10 ppb and a QC-High level pool with 100 ppb of each As species. In order to minimize oxygen exposure and As-III oxidation, the resulting pools were aliquoted into 2.0 mL plastic flat-cap graduated tubes (Fisher Scientific, Cat. No. 02-681-258) and stored at –70°C. Initially, QC samples were prepared and stored in slit-cap vials (VWR, Wayne, PA SNP/CRMP PP 0.75 mL, Cat. No. 66065-338) and monitored for a few weeks. Results, however, demonstrated a high absolute CV for As-III when QC samples were stored this way. QC material requires strict preparation and optimal storage conditions in order to maintain the integrity of QC pools. The concentrations of As species in urine samples may change if samples are not prepared and stored properly (38). In order to characterize the QC material, one pair of QC-Low and one pair QC-High samples were analyzed at the beginning and end of analytical runs. Samples were analyzed for 11 days at a rate of one QC pair for each level/day. The resulting 22 sample data points for each QC pool were plotted to establish
520
Sen et al.: Journal of aoaC InternatIonal Vol. 98, no. 2, 2015 (A)
(B)
(C)
(D)
(E)
(F)
For the evaluation, our measured values were in agreement with target values and within the acceptability limits provided by the programs. (c) Anonymous donor urine samples.—In order to test the efficiency of the described As speciation method and apply the method to a real population, urine samples were collected from anonymous donors (n = 26) and analyzed. Prescreened As free plastic specimen containers (Fisher Scientific, SAMCO 120 mL polypropylene urine collection cups, Cat. No. 01 0038) labeled with a unique specimen identification number were used for anonymous donor urine collection. Once collected, urine specimens were stored at –70°C, and freeze-thaw cycles of the specimen were limited to no more than three. Polypropylene HPLC autosampler snap-cap vials (SNP/CRMP PP 0.75ML, VWR, Cat. No. 66065-338) with caps (seal red with pre-slit T/S, VWR Cat. No. 66030-614) were used for sample preparation. For analysis, human urine was diluted 1:10 using a 10.0 mM KH2PO4 buffer with 2% methanol and 20 µg/L Ge at pH 5.8. Results and Discussion Method Detection Limit (MDL), Accuracy, Precision, and Stability
(G)
The MDL for each analyte was determined by analysis of seven replicates of the diluent spiked with 0.5 µg/L of each As species. The SDs of each species were then multiplied by a factor of 3 to establish the MDL values, which ranged from 0.04–0.16 µg/L for the six As species. Precision and accuracy were monitored over a period of a few weeks using internal QC samples. Results are plotted in Figure 1A–H. Data plots of the relatively less toxic As species, AsC and AsB, are not shown. However, the recovery, mean, SD, and CV for each analyte in both QC samples are shown in Table 4. The QC samples were included at the beginning and end of each analytical run and were analyzed in the same manner as urine specimens. The range of the recovery for QC-Low samples was from 72 to 123% and for QC-High samples from 82 to 115%. The SD range for the six analytes for QC-Low samples was from 0.607 to 1.09 and for QC-High samples from 2.4 to 9.0. The range of CV of QC-Low samples of the As species was 5.5–9.9%, except for As-III for which the CV was 14%. The range of CV of QC-High samples was 3.5–9.8%. Accuracy of the method was also examined against a third party SRM, NIST 2669, and third party performance assessment samples, G-EQUAS round 48. Results for four of the toxic species available (39–41) for the G-EQUAS performance assessment
(H)
Figure 1. Data plots of QC samples for (A and B) As-III, (C and D) DMA, (E and F) MMA, and (G and H) As-V.
QC limits. If the CV was ≤15% for each species in a given pool, the QC material was deemed acceptable. (b) External QC.—Analysis of external QC samples was completed to support validation of the method. Our laboratory participated in external proficiency testing (PT) programs including the German External Quality Assessment Scheme (G-EQUAS) and analyzed commercially available standard reference material (SRM) from the National Institute of Standards and Technology (NIST, SRM 2669). The performance assessment samples from PT programs were blind samples, and results were reported to external QC providers for evaluation. Table 4.
The average, recovery, SD, and CV of each As species from low and high internal QC samples (n = 22) QC-Low
As species
Mean, ppb
QC-High
Recovery, %
SD
CV, %
Mean, ppb
Recovery, %
SD
CV, %
AsC
11.0
89–112
0.70
6.39
104
95–109
3.58
3.4
AsB
11.0
89–109
0.67
6.11
104
92–109
3.55
3.4
As-III
9.7
72–123
1.35
13.8
92.5
82–109
9.04
9.8
DMA
10.5
82–114
0.89
8.50
98.5
86–115
6.70
6.8
MMA
10.9
85–109
0.61
5.55
104
94–107
2.92
2.8
As-V
11.0
83–115
1.09
9.93
102
97–107
2.42
2.4
Sen et al.: Journal of AOAC International Vol. 98, No. 2, 2015 521 Table 5. Analytical results and reference ranges of four toxic As species in human urine (A) NIST SRM 2669 Level 1, ppb
As-III
DMA
MMA
Analysis
1.43
4.80
1.87
2.16
1.47 ± 0.10
3.47 ± 0.41
1.87 ± 0.39
2.41 ± 0.30
As-III
DMA
MMA
As-V
Reference range Level 2, ppb Analysis Target range
As-V
5.08
25.72
7.17
5.97
5.03 ± 0.31
25.30 ± 0.7
7.18 ± 0.56
6.16 ± 0.95
(B) G-EQUAS Round 48 Proficiency Test Samples Level 1, ppb
As-III
DMA
MMA
Analysis
5.52
12.62
2.84
7.60
3.4–5.8
7.4–15.8
1.8–4.2
4.9–9.7
As-III
DMA
MMA
As-V
Level 2, ppb Analysis Reference range
10.32
48.11
8.99
22.24
8.4–12.6
38.8–57.4
6.3–12.3
13.9–24.7
samples were reported to the G-EQUAS program for evaluation. Evaluation results indicated our laboratory successfully passed external performance tests based on the program acceptance criteria at all concentration levels provided (Table 5B). This verified that our method is accurate and the results generated by it are comparable to those of other laboratories. The NIST 2669 SRM had two different concentration levels, and results of the toxic species of both levels are shown in Table 5A. In the NIST 2669 Level 1 sample, DMA was measured at 4.8 ppb compared to the certified level of 3.47 ± 0.41 ppb. Though a high recovery of DMA was observed for this sample, it was not observed in other SRMs or internal QC samples. In the NIST 2669 Level 2 sample, trimethyl arsine oxide (TMAO) was included as part of the panel whereas it was not included in the Level 1 sample. TMAO coelutes with AsC and/or AsB, and because our method does not separate TMAO, these analytes were not reported by our laboratory. AsB, however, was included in the Level 1 sample in the absence of TMAO, and we measured 13.59 ppb compared to the certified level of 12.4 ± 1.9 ppb. The sum of all the As species generated from donor urine speciation analyses was compared with the analytical results from As analyses measured by an independent total As method. We discovered that both data sets correlated with each other very well (R = 0.93; Figure 2), suggesting that the concentrations of other species beyond the predominant six As species measured by our method would be very low if they were at all present in the donor urine. Several steps were introduced in the method to achieve optimal reproducibility and accuracy results. Conventionally, a gradient HPLC pump is used with a combination of different mobile phase compositions and pH conditions to optimize the baseline separation of toxic As species and reduce the total elution time. In order to take advantage of the robust and simple instrumentation for isocratic separation, we used an isocratic HPLC pump with a single mobile phase composition at pH 9.0. This allowed us to minimize the column equilibration time as well as optimize the baseline separation of the toxic As species within an elution time of approximately 16 min. The separation between AsC and AsB was reasonable without baseline separation. It is important to select the optimal pH and buffer condition
to keep the prepared As species intact and achieve well-resolved baseline separation between AsB and As-III. AsB may be present in the 100 ppb range along with low concentrations of As-III (1–2 ppb) in real urine samples. The advantage of using phosphate buffer is that it has a strong buffer capacity and can adjust the pH of the injected sample to the pH of the eluent (42). The disadvantage, however, is that phosphate buffers can produce polymeric deposition on the cones and instruments (42). Other studies showed that at a concentration of 15 mM the phosphate buffer could reduce sensitivity by clogging (43) or damage the instrument (38) if not cleaned regularly. In order to avoid this, a low concentration of phosphate buffer is essential. Another study found that ammonium phosphate buffers at low concentrations are superior to sodium phosphate buffers since they produce the least deposits (44). We used a 5 mM ammonium phosphate buffer in the mobile phase with an addition of 2% CH3OH which is known to improve signal response (45). Lastly, we added 5 mM NH4NO3 into the mobile phase to provide the proper ionic strength without any decrease 70
y = 1.173x - 7.523 R² = 0.938
60
50
40
[Total As (ppb)]
Reference range
As-V
30
20
10
[Sum of As species (ppb)]
0 0.00
10.00
20.00
30.00
40.00
50.00
60.00
Figure 2. Correlation of the sum of As species and total As results from donor samples using two independent Figureanonymous 2. Correlation ofurine the sum of As species and total As results from anonymous methods. = Total Astwo concentration (ppb) and X =SPELL sum of Asy AND x urine donor Y samples using independent methods. [AUTHOR: OUT species concentration (ppb). IN EQUATION]
522
Sen et al.: Journal of aoaC InternatIonal Vol. 98, no. 2, 2015
Figure 3. Chromatogram examples of As species in diluent and an anonymous donor’s urine; RT = retention time.
of ionization efficiency for plasma (36); 20 ppb Ge was added as an internal standard. The concentrations of As species in urine samples may change if samples are not prepared and stored properly (38). Hence, preservation of the stability of As-III during sample storage is very important. As-III could oxidize to As-V during storage (46), and the oxidation is more profound in basic solution than in acidic solution (33). In order to prevent the oxidation of As-III, we introduced a few preventative steps in sample preparation. These included using a degassed sample preparation diluent at pH 5.8 and elution using a degassed mobile phase at pH 9.0. Samples were stored at –70°C in flat-cap airtight tubes (Fisher Scientific, Cat. No. 02-681-258) for long-term storage. Method Application to Anonymous Human Urine Samples In order to test the efficiency of the described As speciation method, urine samples were collected from anonymous donors (n = 26) and analyzed. This pilot study demonstrated the method’s capability of achieving very good separation of all six of the As species with baseline resolution for real urine samples. Chromatograms of As species in the diluent and a donor’s urine Table 6. As speciation data from urine samples collected from anonymous donors (n = 26) and geometric means and 95th percentile values of As species from the CDC Fourth National Report on Human Exposure to Environmental Chemicals (survey year 2009–2010, updated tables, March 2013) based on the National Health and Nutrition Examination Survey (NHANES) ppb AsC
AsB
As-III
DMA
MMA
As-V
Maximum
3.9
39
4.1
28.9
2.1
Development and Validation of a Simple and Robust Method for Arsenic Speciation in Human Urine Using HPLC/ICP-MS Indranil Sen, Wei Zou, Josephine Alvaran, Linda Nguyen, Ryszard Gajek, and Jianwen She1
California Department of Public Health, Biochemistry Section, Environmental Health Laboratory Branch, Richmond, CA 94804
In order to better distinguish the different toxic inorganic and organic forms of arsenic (As) exposure in individuals, we have developed and validated a simple and robust analytical method for determining the following six As species in human urine: arsenous (III) acid (As-III), As (V) acid, monomethylarsonic acid, dimethylarsinic acid, arsenobetaine (AsB), and arsenocholine. In this method, human urine is diluted using a pH 5.8 buffer, separation is performed using an anion exchange column with isocratic HPLC, and detection is achieved using inductively coupled plasma-MS. The method uses a single mobile phase consisting of low concentrations of both phosphate buffer (5 mM) and ammonium nitrate salt (5 mM) at pH 9.0; this minimizes the column equilibration time and overcomes challenges with separation between AsB and As-III. In addition, As-III oxidation is prevented by degassing the sample preparation buffer at pH 5.8, degassing the mobile phase online at pH 9.0, and by the use of low temperature (–70°C) and flip-cap airtight tubes for long term storage of samples. The method was validated using externally provided reference samples. Results were in agreement with target values at varying concentrations and successfully passed external performance test criteria. Internal QC samples were prepared and repeatedly analyzed to assess the method’s longterm precision, and further analyses were completed on anonymous donor urine to assess the quality of the method’s baseline separation. Results from analyses of external reference samples agreed with target values at varying concentrations, and results from precision studies yielded absolute CV values of 3–14% and recovery from 82 to 115% for the six As species. Analysis of anonymous donor urine confirmed the well-resolved baseline separation capabilities of the method for real participant samples.
A
rsenic (As) has been synonymous with poison for ages. Its compounds are toxic and potentially carcinogenic (1) in humans, depending on the chemical nature and oxidation state at various physiological concentrations (2). Several As compounds or species have been identified in biological or environmental systems with Received May 1, 2014. Accepted by AK December 16, 2014. 1 Corresponding author’s e-mail: [email protected] DOI: 10.5740/jaoacint.14-103
both natural and anthropogenic sources (3–6). Because of its widespread mobility from geological deposits into groundwater, chronic As exposure through drinking water is considered one of the biggest threats to human health (3). Prolonged exposure to low doses of As may result in hyperkeratosis, keratosis of the palms and soles, dermatitis, and skin cancer (7–9). Acute As poisoning symptoms may include diarrhea, blood in the urine, muscle cramping, hair loss, stomach pain, and convulsions. Organs in the body affected by As poisoning are the lungs, skin, kidneys, and liver. As is also related to heart disease (hypertension related cardiovascular disease; 10), cancer (11), stroke (cerebrovascular diseases; 12), chronic lower respiratory diseases (13), and diabetes (7, 8). Treatment of chronic As poisoning includes removal of the toxic source as well as chelation therapy (14, 15). The U.S. Environmental Protection Agency (EPA) has established a drinking water standard for As at 10 ppb to protect users from the effects of long-term, chronic exposure (16). As can be ingested as one or more species, and the species can be metabolized into several different inorganic and organic forms. Inorganic species include arsenous (III) acid (As-III) and As acid (As-V), and organic species include the methylated species such as monomethylarsonic acid (MMA) and dimethylarsinic acid (DMA), as well as arsenobetaine (AsB) and arsenocholine (AsC), which are primarily ingested from seafood and stay unaffected in the body. As inactivates enzyme systems by combining with the sulfhydryl (-SH) groups of the enzymes present in the liver, which are vital for the living system (9, 17). Methylation of As in biological systems, also known as biomethylation, involves a methyl transfer reaction by S-adenosyl methionine to form MMA and DMA (12, 18, 19). As species are excreted in urine, which is routinely used as the matrix to measure the internal dose of As species concentrations. Analyses reflect recent exposure ranging from a few hours to 2 days (20). Chemical structures of six well-known As compounds and their respective LD50 (lethal dose 50%) values (for rats) are listed in Table 1 (21–23). LD50 values provide a measurement of acute toxicity and estimates the dose of a given chemical required to kill 50% of a test population (typically rats). According to the LD50 values for the six As species presented, the most toxic As species are the inorganic As compounds As-III and As-V (24), followed by MMA and DMA, then AsB and AsC (which are least toxic). Since total As concentrations alone cannot reflect the real hazard and exposure risk of an individual (25), determining the concentrations of different As species is critical in distinguishing toxic levels in the body. In order to measure the concentration of each As species and its metabolites, several methods have been reported using separation techniques such as LC and detection methods such
518 Sen et al.: Journal of AOAC International Vol. 98, No. 2, 2015 Table 1. Chemical structures, abbreviations, and LD50 values of As species Name Arsenite
Abbreviation As-III
Structure
LD50, mg/kg 14
Table 2. HPLC operational parameters and settings Instrumental conditions Injector, µL
50
Draw speed, µL/min
200
Injection speed, µL/min
200
Pump flow rate, mL/min
1.0
Pump stop time, min Arsenate
As-V
20
Instrumental settings
Flow pressure (column), bar
15 100–120
Experimental Monomethylarsonate
MMA
700–1800
Dimethylarsonate
DMA
700–2600
Arsenobetaine
AsB
>10000
AsC
>10000
Arsenocholine
as inductively coupled plasma (ICP) spectrometry and atomic fluorescence spectrometry both with and without hydride generation (20, 26–32). A common speciation technique includes the determination of seven As compounds in 16 min by subsequent use of two sets of mobile phases at pH 8.7 and 8.0 in a gradient HPLC pump coupled with ICP-MS with a dynamic reaction cell (28, 33). A similar technique using mobile phase compositions at pH 6.2 and 3.0 has also been published (29, 33). Another known method included the oxidation of As-III to As-V prior to separation using a mobile phase at pH 6.0 (34); however, individual As-III or As-V levels could not be determined by the method. With these methods, analyses are relatively complicated and limited to the number of species that can be determined. The limitation of these methods demonstrates a need to develop a relatively simpler and more robust method capable of measuring most toxic As species. Our objective was to develop a method that would require only an isocratic pump with a single mobile phase composition at a single pH level with the ability to analyze the most prevalent As compounds within a reasonable time. Herein, we report the development and validation of a faster and simpler yet accurate and sensitive analytical method to determine the predominant six As species in human urine samples using an isocratic HPLC pump coupled with an ICP-MS.
Instrumentation The separation of As species was performed with an Agilent (Santa Clara, CA) 1200 series HPLC system consisting of a solvent degassing unit, isocratic pump, autosampler, and column compartment. Typical instrument parameters are reported in Table 2. A Hamilton PRP-X100 column (150 × 4.6 mm with 5 µm particles; Hamilton Cat. No. 79174), Hamilton guard column kit (Hamilton Cat. No. 79383), and Hamilton guard column refill cartridges (Hamilton Cat. No. 79385) were used. Hamilton products were supplied by Fisher Scientific, Pittsburgh, PA. A Hamilton PRP-X100 column is used to separate mixtures of organic and inorganic compounds by using a broad range of mobile phases. It is packed with polystyrene-divinylbenzene resin that is inert with organic solvents as well as aqueous buffers with a wide pH range of 1–13. The column has an increased lifetime and can maintain column-to-column reproducibility. It is very robust, can be cleaned and regenerated easily for persistent chemicals, and is quite suitable for analysis of organic and inorganic compounds in aqueous or environmental samples. Periodic (after every 100 samples) column regeneration [backflush with 2% HNO3 and methanol (CH3OH; Fisher Scientific, Cat. No. A412-1)] and frequent guard column replacements were performed to avoid chloride and other buildup in the column that may contribute to argon chloride (40Ar35Cl) formation causing signal interferences (35). Polyetheretherketone capillary tubing (blue, 1/16 × 0.01 inch id, Agilent Cat. No. 5042-6463) was used to connect the chromatographic outlet to the nebulizer of the ICP-MS instrument. The flow pressure at the HPLC column was monitored at 100–120 bar. Pressure higher than 130 bar in general indicates congestion in the HPLC column or guard column and a need for regeneration and replacement of the guard column. An Agilent 7700 CE ICP-MS instrument equipped with a micromist nebulizer and a helium (He) collision cell (36, 37) was used to detect the As species. A liquid argon tank equipped with an approved gas regulator was used for the ICP-MS instrument. Settings were optimized before each run by using autotune functions. Typical values are reported in Table 3. Mobile Phase A degassed solution of 5 mM ammonium phosphate dibasic [(NH4)2HPO4 (Fisher Cat. No. A686-500)], 5.0 mM ammonium nitrate (NH4NO3; Sigma-Aldrich Cat. No. 256064; St. Louis, MO), 2% CH3OH and 20 ppb germanium (Ge; CPI International
Sen et al.: Journal of AOAC International Vol. 98, No. 2, 2015 519 Table 3. ICP-MS operational parameters and settings Instrumental conditions
Instrumental settings
RF power, W
1550
RF matching, V
1.80
Sample depth, mm
8.0
Carrier gas flow, L/min
1.07
Nebulizer pump, rps
0.3
Spray chamber temperature, ºC
2
Extraction lens 1, V
6.4
Extraction lens 2, V
–135
Omega bias, V
65
Omega lens, V
9.4
Cell entrance, V
–40
Cell exit, V
–56
Deflect, V
13.6
Plate bias, V He flow (for collision cell), mL/min Octapole RF, V
–50 1.9–2.1 160
Octapole bias, V
–8
Energy discrimination, V
5.0
Santa Rosa, CA, Cat. No. S4400-1000201) at pH 9.0 was used as the mobile phase. Diluent A degassed solution of 10 mM potassium phosphate monobasic (KH2PO4; Fluka, Sigma-Aldrich Cat. No. 60216), 2% CH3OH, and 20 ppb Ge at pH 5.8 was used as a diluent. As Standards Individual As species stock standards were obtained from the following commercial sources: AsB [formula weight (FW) 178.06, Fluka, Sigma-Aldrich, Cat. No. 11093]; AsC (FW 245, Argus Chemicals, Vernio, Italy, Cat. No. AR60010); MMA (FW 292, Supelco, Sigma-Aldrich Cat. No. PS-281,); DMA (FW 138, Supelco, Sigma-Aldrich Cat. No. PS-51); 1000 mg/L As-III in 2% HNO3 (SPEX Industries Inc., Edison, NJ, Cat. No. SPEC-AS3); 1000 mg/L As-V in 2% HNO3 (SPEX Industries Inc., Cat. No. SPEC-AS5); and 1000 mg/L Ge as an internal standard (CPI International, Cat. No. S4400-1000201). As species stock standards (1000 ppm) were prepared from commercially available solid compounds and kept in tightly sealed plastic containers at 4°C. Certified 1000 ppm As-III and As-V standard solutions were available commercially (above). In order to prepare multispecies intermediate standard solutions, 100 µL 1000 ppm standard stock solutions of each As species was added into a disposable 15 mL conical tube (Becton-Dickinson, Franklin Lakes, NJ, Cat. No. 2097). The total volume was made up to 10 mL with freshly prepared and degassed diluent solution for a final concentration of 10 ppm for each of the six As species. This multispecies intermediate standard solution remained stable for 1 week when stored at –70°C. Multispecies 100 ppb working standard solutions were prepared fresh on the day of use. To prepare the working standard solution, the 10 ppm intermediate
standard solution was diluted 100x. A 100 µL amount of the multispecies intermediate standard stock solution mixture was added to the freshly prepared and degassed diluent solution for a final volume of 10 mL. This solution was stable at 4°C for 24 h stored in the dark. Multispecies daily calibration standard solutions were prepared fresh on the day of use from the 100 ppb multispecies working standard solution. Total As Method An Agilent 7500 CE ICP-MS instrument equipped with a micromist nebulizer and He collision cell was used for the analysis of total As in human urine. Data acquisition and evaluation were performed with Agilent MassHunter software. Instrumental settings were optimized before each run by using autotune functions, and the parameters were compared with the applicable ranges. Typical values were as follows: radiofrequency (RF) power, 1550 W; RF matching 1.89 V; sample depth 8.0 mm; torch H 0.4 mm; torch V 0 mm; carrier gas flow 0.96 L/min; makeup gas: 0.15 L/min; nebulizer pump 0.1 rps, S/C temperature 2°C; extraction lens 1: 0 V; extraction lens 2: –125 V; omegas bias ce 24 V; omega lens ce 0.2 V; cell entrance –30 V; QP focus: 2 V; cell exit –30 V; deflect 13.6 V; plate bias –50 V; He flow (for collision cell) 4.6 mL/min; H2 gas 3 mL/min; octapole RF 180 V; octapole bias –4 V; and energy discrimination 8.0 V. Human urine was diluted (1:20) using a diluent solution containing the following: 2% HNO3, 0.2% sulfamic acid, 100 µg/L gold (Au) standard solution, 1.5% ethanol, 0.0005% Triton X-100, 50 µg/L Ge (Spec CertiPrep, Cat. No. PLGE92Y) 50 µg/L rhodium (Rh; Spec CertiPrep, Cat. No. PLRH32Y), and 60 µg/L rhenium (Re; Spec CertiPrep, PLRE9-2Y). The samples were prepared by filling a 15 mL conical tube (brand and lots prescreened for contaminants) with 4500 µL diluent, 250 µL 2% HNO3, and 250 µL urine sample. The tubes were inverted to mix. Samples (a) Internal QC.—Internal QC samples were prepared and characterized by the laboratory. Degassed diluent pH 5.8 buffer was used to prepare two levels of surrogate specimen material; a QC-Low level pool with approximately 10 ppb and a QC-High level pool with 100 ppb of each As species. In order to minimize oxygen exposure and As-III oxidation, the resulting pools were aliquoted into 2.0 mL plastic flat-cap graduated tubes (Fisher Scientific, Cat. No. 02-681-258) and stored at –70°C. Initially, QC samples were prepared and stored in slit-cap vials (VWR, Wayne, PA SNP/CRMP PP 0.75 mL, Cat. No. 66065-338) and monitored for a few weeks. Results, however, demonstrated a high absolute CV for As-III when QC samples were stored this way. QC material requires strict preparation and optimal storage conditions in order to maintain the integrity of QC pools. The concentrations of As species in urine samples may change if samples are not prepared and stored properly (38). In order to characterize the QC material, one pair of QC-Low and one pair QC-High samples were analyzed at the beginning and end of analytical runs. Samples were analyzed for 11 days at a rate of one QC pair for each level/day. The resulting 22 sample data points for each QC pool were plotted to establish
520
Sen et al.: Journal of aoaC InternatIonal Vol. 98, no. 2, 2015 (A)
(B)
(C)
(D)
(E)
(F)
For the evaluation, our measured values were in agreement with target values and within the acceptability limits provided by the programs. (c) Anonymous donor urine samples.—In order to test the efficiency of the described As speciation method and apply the method to a real population, urine samples were collected from anonymous donors (n = 26) and analyzed. Prescreened As free plastic specimen containers (Fisher Scientific, SAMCO 120 mL polypropylene urine collection cups, Cat. No. 01 0038) labeled with a unique specimen identification number were used for anonymous donor urine collection. Once collected, urine specimens were stored at –70°C, and freeze-thaw cycles of the specimen were limited to no more than three. Polypropylene HPLC autosampler snap-cap vials (SNP/CRMP PP 0.75ML, VWR, Cat. No. 66065-338) with caps (seal red with pre-slit T/S, VWR Cat. No. 66030-614) were used for sample preparation. For analysis, human urine was diluted 1:10 using a 10.0 mM KH2PO4 buffer with 2% methanol and 20 µg/L Ge at pH 5.8. Results and Discussion Method Detection Limit (MDL), Accuracy, Precision, and Stability
(G)
The MDL for each analyte was determined by analysis of seven replicates of the diluent spiked with 0.5 µg/L of each As species. The SDs of each species were then multiplied by a factor of 3 to establish the MDL values, which ranged from 0.04–0.16 µg/L for the six As species. Precision and accuracy were monitored over a period of a few weeks using internal QC samples. Results are plotted in Figure 1A–H. Data plots of the relatively less toxic As species, AsC and AsB, are not shown. However, the recovery, mean, SD, and CV for each analyte in both QC samples are shown in Table 4. The QC samples were included at the beginning and end of each analytical run and were analyzed in the same manner as urine specimens. The range of the recovery for QC-Low samples was from 72 to 123% and for QC-High samples from 82 to 115%. The SD range for the six analytes for QC-Low samples was from 0.607 to 1.09 and for QC-High samples from 2.4 to 9.0. The range of CV of QC-Low samples of the As species was 5.5–9.9%, except for As-III for which the CV was 14%. The range of CV of QC-High samples was 3.5–9.8%. Accuracy of the method was also examined against a third party SRM, NIST 2669, and third party performance assessment samples, G-EQUAS round 48. Results for four of the toxic species available (39–41) for the G-EQUAS performance assessment
(H)
Figure 1. Data plots of QC samples for (A and B) As-III, (C and D) DMA, (E and F) MMA, and (G and H) As-V.
QC limits. If the CV was ≤15% for each species in a given pool, the QC material was deemed acceptable. (b) External QC.—Analysis of external QC samples was completed to support validation of the method. Our laboratory participated in external proficiency testing (PT) programs including the German External Quality Assessment Scheme (G-EQUAS) and analyzed commercially available standard reference material (SRM) from the National Institute of Standards and Technology (NIST, SRM 2669). The performance assessment samples from PT programs were blind samples, and results were reported to external QC providers for evaluation. Table 4.
The average, recovery, SD, and CV of each As species from low and high internal QC samples (n = 22) QC-Low
As species
Mean, ppb
QC-High
Recovery, %
SD
CV, %
Mean, ppb
Recovery, %
SD
CV, %
AsC
11.0
89–112
0.70
6.39
104
95–109
3.58
3.4
AsB
11.0
89–109
0.67
6.11
104
92–109
3.55
3.4
As-III
9.7
72–123
1.35
13.8
92.5
82–109
9.04
9.8
DMA
10.5
82–114
0.89
8.50
98.5
86–115
6.70
6.8
MMA
10.9
85–109
0.61
5.55
104
94–107
2.92
2.8
As-V
11.0
83–115
1.09
9.93
102
97–107
2.42
2.4
Sen et al.: Journal of AOAC International Vol. 98, No. 2, 2015 521 Table 5. Analytical results and reference ranges of four toxic As species in human urine (A) NIST SRM 2669 Level 1, ppb
As-III
DMA
MMA
Analysis
1.43
4.80
1.87
2.16
1.47 ± 0.10
3.47 ± 0.41
1.87 ± 0.39
2.41 ± 0.30
As-III
DMA
MMA
As-V
Reference range Level 2, ppb Analysis Target range
As-V
5.08
25.72
7.17
5.97
5.03 ± 0.31
25.30 ± 0.7
7.18 ± 0.56
6.16 ± 0.95
(B) G-EQUAS Round 48 Proficiency Test Samples Level 1, ppb
As-III
DMA
MMA
Analysis
5.52
12.62
2.84
7.60
3.4–5.8
7.4–15.8
1.8–4.2
4.9–9.7
As-III
DMA
MMA
As-V
Level 2, ppb Analysis Reference range
10.32
48.11
8.99
22.24
8.4–12.6
38.8–57.4
6.3–12.3
13.9–24.7
samples were reported to the G-EQUAS program for evaluation. Evaluation results indicated our laboratory successfully passed external performance tests based on the program acceptance criteria at all concentration levels provided (Table 5B). This verified that our method is accurate and the results generated by it are comparable to those of other laboratories. The NIST 2669 SRM had two different concentration levels, and results of the toxic species of both levels are shown in Table 5A. In the NIST 2669 Level 1 sample, DMA was measured at 4.8 ppb compared to the certified level of 3.47 ± 0.41 ppb. Though a high recovery of DMA was observed for this sample, it was not observed in other SRMs or internal QC samples. In the NIST 2669 Level 2 sample, trimethyl arsine oxide (TMAO) was included as part of the panel whereas it was not included in the Level 1 sample. TMAO coelutes with AsC and/or AsB, and because our method does not separate TMAO, these analytes were not reported by our laboratory. AsB, however, was included in the Level 1 sample in the absence of TMAO, and we measured 13.59 ppb compared to the certified level of 12.4 ± 1.9 ppb. The sum of all the As species generated from donor urine speciation analyses was compared with the analytical results from As analyses measured by an independent total As method. We discovered that both data sets correlated with each other very well (R = 0.93; Figure 2), suggesting that the concentrations of other species beyond the predominant six As species measured by our method would be very low if they were at all present in the donor urine. Several steps were introduced in the method to achieve optimal reproducibility and accuracy results. Conventionally, a gradient HPLC pump is used with a combination of different mobile phase compositions and pH conditions to optimize the baseline separation of toxic As species and reduce the total elution time. In order to take advantage of the robust and simple instrumentation for isocratic separation, we used an isocratic HPLC pump with a single mobile phase composition at pH 9.0. This allowed us to minimize the column equilibration time as well as optimize the baseline separation of the toxic As species within an elution time of approximately 16 min. The separation between AsC and AsB was reasonable without baseline separation. It is important to select the optimal pH and buffer condition
to keep the prepared As species intact and achieve well-resolved baseline separation between AsB and As-III. AsB may be present in the 100 ppb range along with low concentrations of As-III (1–2 ppb) in real urine samples. The advantage of using phosphate buffer is that it has a strong buffer capacity and can adjust the pH of the injected sample to the pH of the eluent (42). The disadvantage, however, is that phosphate buffers can produce polymeric deposition on the cones and instruments (42). Other studies showed that at a concentration of 15 mM the phosphate buffer could reduce sensitivity by clogging (43) or damage the instrument (38) if not cleaned regularly. In order to avoid this, a low concentration of phosphate buffer is essential. Another study found that ammonium phosphate buffers at low concentrations are superior to sodium phosphate buffers since they produce the least deposits (44). We used a 5 mM ammonium phosphate buffer in the mobile phase with an addition of 2% CH3OH which is known to improve signal response (45). Lastly, we added 5 mM NH4NO3 into the mobile phase to provide the proper ionic strength without any decrease 70
y = 1.173x - 7.523 R² = 0.938
60
50
40
[Total As (ppb)]
Reference range
As-V
30
20
10
[Sum of As species (ppb)]
0 0.00
10.00
20.00
30.00
40.00
50.00
60.00
Figure 2. Correlation of the sum of As species and total As results from donor samples using two independent Figureanonymous 2. Correlation ofurine the sum of As species and total As results from anonymous methods. = Total Astwo concentration (ppb) and X =SPELL sum of Asy AND x urine donor Y samples using independent methods. [AUTHOR: OUT species concentration (ppb). IN EQUATION]
522
Sen et al.: Journal of aoaC InternatIonal Vol. 98, no. 2, 2015
Figure 3. Chromatogram examples of As species in diluent and an anonymous donor’s urine; RT = retention time.
of ionization efficiency for plasma (36); 20 ppb Ge was added as an internal standard. The concentrations of As species in urine samples may change if samples are not prepared and stored properly (38). Hence, preservation of the stability of As-III during sample storage is very important. As-III could oxidize to As-V during storage (46), and the oxidation is more profound in basic solution than in acidic solution (33). In order to prevent the oxidation of As-III, we introduced a few preventative steps in sample preparation. These included using a degassed sample preparation diluent at pH 5.8 and elution using a degassed mobile phase at pH 9.0. Samples were stored at –70°C in flat-cap airtight tubes (Fisher Scientific, Cat. No. 02-681-258) for long-term storage. Method Application to Anonymous Human Urine Samples In order to test the efficiency of the described As speciation method, urine samples were collected from anonymous donors (n = 26) and analyzed. This pilot study demonstrated the method’s capability of achieving very good separation of all six of the As species with baseline resolution for real urine samples. Chromatograms of As species in the diluent and a donor’s urine Table 6. As speciation data from urine samples collected from anonymous donors (n = 26) and geometric means and 95th percentile values of As species from the CDC Fourth National Report on Human Exposure to Environmental Chemicals (survey year 2009–2010, updated tables, March 2013) based on the National Health and Nutrition Examination Survey (NHANES) ppb AsC
AsB
As-III
DMA
MMA
As-V
Maximum
3.9
39
4.1
28.9
2.1
