Validation of Analytical Method for Determination of Thallium in Rodent Plasma and Tissues by Inductively Coupled Plasma–Mass Spectrometry (ICP-MS)

Abstract

Thallium (Tl) can be released as a byproduct of smelting, mining, and other industries, causing human exposure. There are knowledge gaps on the toxicity of thallium compounds, so the National Toxicology Program is investigating the toxicity of thallium (I) sulfate in rodents. We developed and validated a method to quantitate Tl in rodent plasma and secondary matrices. Primary matrix standards and validation samples were digested with nitric acid and analyzed for Tl by inductively-coupled plasma – mass spectrometry (ICP-MS). Method performance was validated for linearity, accuracy, precision, and other criteria. Calibration was linear from 1.25 to 500 ng Tl/mL plasma; accuracy (RE) was −5.9 to 2.6% for all calibration standards. The lower limit of quantitation (LLOQ) was 1.25 ng Tl/mL plasma, and the limit of detection was 0.0370 ng Tl/mL plasma. Intra- and interday RE and precision (RSD) were −5.6 to −1.7% and ≤0.8% (intraday) and −4.8 to −1.3% and ≤4.3% (interday), respectively, at three sample concentration levels. Standards up to 10.0 × 10³ ng/mL could be analyzed by dilution with digested blank matrix, with −6.4% RE and 5.4% RSD. Method was also evaluated in post-natal day 4 (PND4) Hsd:Sprague Dawley SD (HSD) dam and pup plasma, gestation day 18 (GD 18) HSD rat fetal homogenate, HSD rat urine, female HSD rat brain homogenate, female B6C3F1 mouse plasma. Background Tl was detected in control fetal and brain homogenates and urine at < 30% of LLOQ response. Results demonstrate that the method is suitable for determination of Tl in rodent matrices for toxicology studies.

Introduction

Thallium (Tl) is a post-transition metal that has long been known to be toxic. Although used in some medical applications in the 1800’s, it was also used as a rodent poison and in some instances to deliberately poison humans (Moeschlin 1980). From 1920 – 1965, Tl salts were used as pesticides until several countries banned their application (Arena et al. 1965). Acute LD₅₀ values in mice have been reported as ~20 mg/kg body weight, with toxic effects including alopecia (spot baldness) and neurological effects observed at lower doses (Galván-Arzate and Santamaría 1998; Leung and Ooi 2000). Despite widespread understanding of Tl’s toxic properties, there are still significant data gaps on chronic toxicity (Cvjetko et al. 2010).

Tl is used in several industrial processes, including production of optical lenses, semiconductors, pigments, and fireworks and can be anthropogenically released in wastewater, smelting emissions, or mine drainage (Peter and Viraraghavan 2005). The National Tap Water Database produced by the Environmental Working Group indicates that Tl has been detected in tap water in more than 30 states, representing a significant potential for sub-acute and chronic human exposure (Environmental Working Group 2019). Reported drinking water Tl concentrations have been found as high as 7.2 ng/mL (Twidwell and Williams-Beam 2002). Under the National Primary Drinking Water Regulations, the U.S. Environmental Protection Agency (EPA) has established a maximum contaminant level (MCL) for Tl of 2.0 ng/mL, with a MCL goal of 0.50 ng/mL (Peter and Viraraghavan 2005). However, there have been limited toxicology data to support establishment of a reference concentration for Tl. Due to this data gap, the EPA nominated Tl compounds to the National Toxicology Program (NTP) for testing in rodent models.

Tl levels in plasma and tissues are important to put toxicological findings into context. Inductively coupled plasma – mass spectrometry (ICP-MS) is one of the most broadly-used instrumental approaches for determining trace levels of metals in biological and environmental samples. Sample preparation and analysis methods for the determination of Tl in biological samples have been reported in the literature, but to the best of our knowledge, no ICP-MS methods have been formally validated (Das et al. 2006; Levine et al. 2011). The only method that has been formally validated is a flame atomic absorption spectrometry method, which has significantly higher limits of detection (LOD) and quantitation (lower limit of quantitation, or LLOQ) than analysis by ICP-MS (Anthemidis and Ioannou 2012). In some reports, Tl has been included as part of a broad panel of analytes, which can impact precision and accuracy, detection limit, and other analytical figures of merit (Goullé et al. 2005; Skalny et al. 2016). The purpose of this study was to develop and validate an analytical method for the determination of Tl in biological matrices in support of NTP toxicology studies following perinatal exposure in Hsd:Sprague Dawley SD (HSD) rats and adult exposure in B6C3F1/N mice via drinking water to thallium(I) sulfate (https://ntp.niehs.nih.gov/whatwestudy/testpgm/status/ts-16019.html). This salt was selected as the initial test compound due to the relatively high solubility of the sulfate salt of Tl(I), lower solubility of Tl(III) salts, and the ease of conversion from Tl(III) to Tl(I) in solution (Aldridge 2011). The above factors suggested that it was not practical to perform speciation analysis in these matrices, so the validation efforts focused on measurement of total Tl.

The sample preparation and analysis methods were successfully validated in male Sprague Dawley (SD) rat plasma. Method performance was verified in study sample matrices (secondary matrices): post-natal day 4 (PND4) HSD rat dam plasma, PND4 pup plasma, HSD female rat urine, female B6C3F1 mouse plasma, gestational day 18 (GD18) HSD rat fetus, and HSD female rat brain. The stability of Tl in all matrices was also measured up to 60 days. This study provides a robust analytical method for the determination of Tl in rodent plasma and tissues.

Materials and Methods

Reagents

All standards and samples were prepared with approximately 18 MΩ cm^–1 quality deionized water (DI water; Pure Water Solutions, Hillsborough, NC). Inorganic stock solutions and standards were prepared using National Institute of Standards and Technology (NIST)-traceable standard solutions purchased from High Purity Standards (Charleston, SC). Tl stock solutions were prepared using 10 μg/mL and 1,000 μg/mL stock standards and internal standard (IS) solutions were prepared using 1,000 μg/mL Praseodymium (Pr) stock standard. A 10 μg/mL Tl stock standard from Inorganic Ventures (Christiansburg, VA) was used to prepare matrix standards and validation samples for second-source concentration verification. A 10 μg/mL indium (In) stock standard from High Purity Standards was also used to prepare a tuning solution to optimize system analytical performance and system suitability prior to each analysis. Samples were prepared with concentrated nitric acid (HNO₃; 70%, Trace Metal Grade, Fisher Scientific, Hampton, NH) and 30%, non-stabilized hydrogen peroxide (H₂O₂; Trace Metal Grade, Ricca Chemical Company, Arlington, TX). All biological matrices used for method validation were purchased commercially from BioIVT (Westbury, NY). Both primary matrix (male SD rat plasma) and secondary matrices (PND4 HSD rat dam plasma, PND4 pup plasma, HSD female rat urine, female B6C3F1 mouse plasma, GD18 HSD rat fetus, and HSD female rat brain) were pooled from multiple animals. Plasma was isolated from blood collected using K₃EDTA anticoagulant. Matrices were stored frozen at a nominal temperature of −80 °C until the time of preparation.

Method Validation

Validation experiments were designed to evaluate linearity, precision, accuracy, recovery, selectivity (to address potential analytical interferences on the determination of Tl), sensitivity (defined as the analytical limits of the method), matrix effect, dilution verification, and analysis period stability for Tl in the primary matrix, male Sprague Dawley (SD) rat plasma.

Sample and standards were prepared on each validation day by graphite heating block digestion with nitric acid as described in a later section. An eight-point matrix calibration curve was analyzed on each validation day to evaluate linearity. The accuracy of the determined concentration of calibration standards was expressed as the relative error (RE) by:

$$\left(\text{Measured }\left[\text{Tl}\right]–\text{ Target }\left[\text{Tl}\right]\right)/\text{Target }\left[\text{Tl}\right])\times 100\%$$ (1)

against the nominal standard concentration. Calibration standards were split into two sets for analysis, one at the beginning of the sequence and one at the end, to assess instrument drift over the course of the longest potential analytical period. Three method blank samples consisting of digested control plasma matrix were prepared and run immediately after the highest concentration matrix standard to assess carryover.

Six replicates of the lowest concentration matrix standard were prepared for determination of LOD and LLOQ. The LOD was taken as 3 times the standard deviation of the calculated concentration of the standards and the LLOQ was established as the lowest calibration standard concentration with an average RE for the six replicate standards of ≤ ±20%. Six method blanks containing digestion reagents and IS only were prepared, and measured Tl intensity was compared to the average Tl intensity of the LLOQ standards to assess method selectivity. A solvent blank and a set of solvent standards were prepared at the same concentrations as the matrix calibration standards and analyzed to assess absolute recovery as a ratio of matrix standard response to solvent standard response. The recovery was calculated by:

$$\text{Recovery }=\left({\text{Matrix Standard\hspace{0.17em}}}^{205}\text{Tl Intensity }/{\text{ Solvent Standard\hspace{0.17em}}}^{205}\text{Tl Intensity}\right)\times 100\%$$ (2)

Triplicate matrix quality control (QC) samples were prepared at low (2.00 ng Tl/mL plasma), medium (50.0 ng Tl/mL plasma), and high (400 ng Tl/mL plasma) concentration levels discussed above on three separate days to assess method precision (measured as the relative standard deviation, RSD, of replicate samples) and accuracy (intra- and interday). Intraday reproducibility of the IS signal was calculated as the standard deviation of the method blank replicates. The interday reproducibility of the IS signal was calculated as the standard deviation of the IS signal in the QC matrix samples at all 3 concentration levels on all analytical days. Triplicate matrix standards were also prepared at 10.0 × 10³ ng Tl/mL plasma and diluted post-extraction with method blank to verify the accuracy of sample dilution, or method extension. Triplicate matrix QC samples were digested as described in the following section and then spiked at low, medium, and high concentration levels. Potential matrix effects were assessed by estimating the accuracy and precision of these post-digestion spiked samples.

Matrix Standard Preparation and Extraction

A 1.00 μg/mL Pr IS stock solution in 1% (v/v) HNO₃ was prepared by addition of 1,000 μg/mL Pr stock standard solution and trace metal grade HNO₃ to a polypropylene tube. Tl Stock solutions (0.0100, 0.100, and 1.00 μg/mL Tl) were prepared by pipetting 10 μg/mL Tl stock standard solution into polypropylene tubes in a clean, high-efficiency particulate air (HEPA)-filtered plastic fume hood with concentrated HNO₃ to produce a final concentration of 20% (v/v) HNO₃. Two sets of these stock solutions were prepared using Tl stock standards produced by two different manufacturers (High Purity and Inorganic Ventures) to provide second-source concentration verification.

Matrix calibration and QC standards were prepared by transferring 100 μL aliquots of primary matrix to separate 15-mL polypropylene tubes in a clean hood. Matrix calibration standards were fortified by pipetting Tl stock standards into each tube to produce target concentrations ranging from 1.25 to 500 ng Tl/mL plasma. Matrix QC samples were fortified at target concentrations of 2.00 (low), 50.0 (medium) and 400 (high) ng Tl/mL plasma.

Samples were subjected to complete acid digestion by adding a 1.00 mL aliquot of trace metal grade concentrated HNO₃ to each tube and heating uncapped for 30 minutes at 95 °C in an SCP DigiPrep (Quebec, Montreal, Canada) graphite heating block. Samples were then removed from the block and allowed to cool to room temperature. A 0.500 mL aliquot of 30% H₂O₂ solution was added to each tube and samples were returned to the block for an additional 30 minutes at 95 °C. Samples were then removed and allowed to cool to room temperature, IS was added to a concentration of 5 ng/mL Pr, and all tubes were diluted to the 5.00 mL mark with DI water. Tubes were capped and inverted several times to mix the solution.

A solvent blank and a set of solvent standards were prepared at the same concentrations as the matrix calibration standards and analyzed without digestion following analysis of primary matrix calibration standards.

ICP-MS Analysis and Analyte Quantitation

All standards and samples were analyzed for total Tl content using an X-Series II quadrupole ICP-MS (Thermo Fisher Scientific, Bridgewater, NJ) equipped with a Peltier-cooled spray chamber and an ASX-500 autosampler (Teledyne CETAC Technologies, Omaha, NE). Instrument parameters used on all days of the validation study are shown in Table S1 (Supplemental Information). The ICP-MS was tuned daily by aspiration of a solution containing 10 μg/L indium to maximize signal intensity and stability. The ²⁰⁵Tl signal was measured in all standards and samples as a ratio of the ¹⁴¹Pr IS signal to account for instrument drift. The ratio of Pr signal in each standard and sample to the Pr signal in the initial blank was calculated to adjust the intensity of the Tl signal in each standard and sample, and the adjusted Tl intensity was plotted against standard concentration. Adjusted Tl intensity and concentration were related using a linear least-squares regression to produce the calibration equation. Total Tl in all samples in units of ng Tl/mL extract was calculated against the calibration equation and the final concentration in each sample matrix was calculated using dilution factors arising from the sample digestion process.

Thallium Stability in Matrix and Sample Extracts

Four sets of analysis period stability (APS) samples were prepared in primary matrix, each set containing triplicate matrix standards at low, medium, and high concentrations (2.00, 50.0, and 400 ng Tl/mL plasma, respectively). One set was used for Day 0 reference and was extracted and analyzed on the first analytical day. After analysis, the Day 0 set was left in the autosampler for 5 days to assess autosampler stability. The second APS set was extracted and stored in a refrigerator for five days before analysis to assess refrigerator stability. The third APS set was stored in a freezer (~ −20 °C) for five days and then extracted alongside study samples to assess short-term freezer storage stability. The fourth APS set was stored in the freezer and subjected to three freeze-thaw cycles of at least 12 hours prior to extraction to assess freeze-thaw stability.

Secondary Matrix Evaluation

An evaluation was performed in six matrices to verify the quantification of Tl in several secondary matrices using calibration curves determined in the primary matrix: PND4 HSD rat dam plasma, PND4 pup plasma, HSD adult female rat urine, adult female B6C3F1 mouse plasma, GD18 HSD rat fetus, and HSD adult female rat brain. Selectivity in secondary matrices was determined by preparing six replicate method blanks in each matrix and comparing measured Tl intensity to the LLOQ standard. Precision and accuracy in secondary matrices were assessed by preparation and analysis of six replicate matrix standards at 2.00 ng/mL in each secondary matrix against a calibration curve measured in the primary plasma matrix.

Fetuses and brains analyzed as part of this assessment were homogenized with a polytron (Brinkmann, Kinematica AG, Luzerne, Switzerland) prior to digestion. Five fetuses were selected and allowed to thaw at room temperature. Fetuses were weighed together in an acid-washed polypropylene 50-mL tube, homogenized, and stored at ~ −70 °C if not immediately prepared for analysis. The homogenizer was thoroughly cleaned with DI water between samples. Fetal homogenate was then prepared using the same procedure as the plasma matrix standards described above. Brains were homogenized in a similar manner. Brains were transferred to an acid-washed polypropylene 50-mL tube and weighed before homogenization. Homogenate was then transferred to 15-mL tubes and prepared using the same procedure as the matrix standards described above. Primary matrix calibration standards were prepared and analyzed in the manner described previously.

Long-Term Stability of Thallium in Secondary Matrices

Stability of Tl in secondary matrices was assessed by preparing several sets of matrix standards fortified with Tl at 2 concentrations (2.00 and 250 ng Tl/mL for plasma and urine, ng Tl/g for fetus and brain) and stored in the freezer for up to 521 days. At 15, 30, 63, and 521 days, a set of standards from each matrix and concentration level was analyzed and quantified against a freshly prepared primary matrix calibration curve.

Results and Discussion

Several approaches have been used in previous studies for the determination of Tl in biological samples depending on the matrix as reviewed by (Das et al. 2006). Some studies of urine and whole blood have employed simple dilution approaches with acidic solution for urine and plasma (de Boer et al. 2004; Rodushkin et al. 2004). However, the sample preparation approach used here provides additional versatility to apply the same method to several biomatrices due to the acidic decomposition, which minimizes matrix effects on ionization efficiency during ICP-MS analysis and promotes greater instrument stability and sample throughput. Other studies have employed microwave-assisted sample digestion to achieve a similar effect but transferring the digest between vessels before and after digestion introduces additional opportunities for sample contamination (Bocca et al. 2003; Coni et al. 2000). The graphite digestion block approach employed here mitigates this risk by digesting the samples directly in the vessel used for analysis.

Some studies have included Tl as part of a semi-targeted broad panel approach. Goullé et al analyzed 100 human plasma and urine samples for 27 elements including Tl to determine reference ranges of the element in healthy human adults (Goullé et al. 2005). The reported limits of detection and quantification are lower than the values reported here, but it is unclear how they determined these values. One possible explanation is that the reported limits were calculated using solvent curves or solvent QC standards, which would allow for calculation of analytical limits that could not account for matrix effects. Our approach involves the direct analysis of fortified samples to establish a conservative limit of quantitation that will be consistently achievable during analysis of study samples over an extended time period.

Biological samples were prepared by acid digestion in a graphite heating unit and analyzed by ICP-MS for total thallium content. ICP-MS is a common technique for measuring elements present in samples at or below the ng/g or ng/mL concentration range. The results of the validation indicate that the method provides accurate, sensitive, selective, and reproducible values for total Tl in all the matrices measured here.

Validation in Primary Matrix

Validation results for Tl in male SD rat plasma (primary matrix) are shown in Table 1. Matrix calibration curves were found to be linear (r > 0.999) over the calibration range of 1.25 – 500 ng Tl/mL plasma on each of three validation days. A representative calibration curve from Experimental Day 1 is shown in Figure S1 (Supplemental information) and the least squares linear regression results are shown in Table S2. Accuracy (RE) of the matrix standard concentration ranged from −5.9% to 2.6% in all experiments. Several matrix blanks were analyzed immediately after the calibration curve to assess Tl signal carryover from the highest concentration standard. The measured Tl signal was 28% of the LLOQ in the first blank and < 5% for the remaining blanks, for an average of 9.1% of the lowest standard signal, indicating minimal carryover. Selectivity assesses the potential for analytical interferences in the primary study matrix that can impact method accuracy. In the case of Tl, this is not anticipated to be significant because there are few isobaric or polyatomic interferences that would be endogenous to plasma or any of the secondary matrices. The recovery of Tl estimated by comparing matrix standards with corresponding concentrations of solvent standards ranged from 97.8 – 104%. IS reproducibility was calculated as the relative standard deviation (RSD) of the Pr signal in QC samples, both in six method blank samples measured on the same day, and between analytical runs in QC check standards (n = 27). The IS signal was found to be highly reproducible for both intraday (3.3% RSD) and interday (4.2% RSD) samples.

Table 1.

Method validation data for thallium in male Sprague Dawley rat plasma.

Parameter	Result
Concentration range (ng Tl/mL plasma)	1.25 to 500
Correlation coefficient (r)	> 0.999
Matrix standard accuracy (RE, %)	−5.9 to 2.6
LODa (ng Tl/mL plasma)	0.0370
LLOQb (ng Tl/mL plasma)	1.25
Selectivity (mean matrix blank response relative to LLOQ, %)	8.17
Instrument drift (%)	−5.3 to −1.4
Absolute recovery (%)	97.8 to 104
Internal Standard Reproducibility (RSD, %)	3.3 (Intraday)
	4.2 (Interday)
Matrix Effect (% Difference)	−9.9 to 1.8
Matrix Effect Precision (RSD, %)	0.6 to 1.4
Carryover (mean matrix blank response relative to LLOQ, %)	9.1
Precision and accuracyc
Intraday accuracy (mean RE, %)d	−5.6 to −1.7
Intraday precision (RSD, %)e	0.2 to 0.8
Interday accuracy (mean RE, %)	−4.8 to −1.3
Interday precision (RSD, %)	0.9 to 4.3
Dilution verification (up to 10.0 × 103 ng Tl/mL plasma)
Accuracy (mean RE, %)	−6.4
Precision (RSD, %)	5.4

^aLOD = Limit of detection, calculated as 3 times the standard deviation of 6 replicate preparations of the lowest calibration standard.

^bLLOQ = Lower limit of quantification, or the calibration standard concentration that exhibited −20% to 20% relative error (RE) for 6 replicate analyses.

^cPrecision and accuracy were determined for triplicate fortified QC samples at three levels, 2.00, 50.0, and 400 ng Tl/mL plasma.

^dRE = Percent relative error.

^eRSD = Percent relative standard deviation.

The LLOQ and LOD were experimentally established by analysis of six replicate plasma digests at the lowest calibration concentration, 1.25 ng Tl/mL plasma. The LLOQ value was estimated based on the determined linear range of preliminary method development experiments (data not shown). Analysis of the replicate samples demonstrated an average RE of −1.4% and a 1.0% RSD, establishing this concentration as a suitable LLOQ. The estimated LOD was 0.0370 ng Tl/mL plasma. Analytical limits determined here are comparable to those reported in the literature, which range from 0.03 to 25 ng Tl/mL matrix for non-validated ICP-MS and ETAAS analysis, respectively (Das et al. 2006; Mauras et al. 1993; Yang and Smeyers-Verbeke 1991). Analysis of samples spiked at three concentration levels (2.00, 50.0, and 400 ng Tl/mL plasma) demonstrated intraday accuracy (RE) and precision (RSD) of −5.6 to −1.7 and ≤0.8% at all three concentration levels. Interday accuracy and precision were −4.8 to −1.3% and ≤4.3%, respectively, at all concentration levels. Plasma samples (n = 3) spiked at 10,000 ng Tl/mL plasma diluted into the calibration range with blank matrix had an average RE of −6.4% and RSD of 5.4% demonstrating concentrations above the calibrated range can be successfully diluted into the calibrated range. Matrix effect analysis was done by comparing Tl signal intensity of QC standards spiked before digestion to that in QC standards spiked after digestion. This comparison showed that Tl signal changed between −9.9% and 1.8% in samples spiked after digestion compared to those spiked prior to digestion, with ≤1.4% RSD for triplicate analyses at each concentration level. Taken collectively, these data demonstrated that the assay is suitable to quantitate Tl concentration in plasma.

Secondary Matrix Evaluation

Method performance was evaluated in several secondary matrices to represent study matrices (PND4 HSD rat dam plasma, PND4 pup plasma, HSD adult female rat urine, adult female B6C3F1 mouse plasma, GD18 HSD rat fetus, and HSD adult female rat brain) by verifying selectivity in blanks of each matrix and by measuring precision and accuracy of fortified matrix digests analyzed against a plasma calibration curve. GD18 and PND4 matrices were selected to assess the transfer of Tl during gestation and lactation which provides information on early life exposure. Urine was included as a matrix to provide a comparison of animal data to human exposures in light of some reported human urine biomonitoring data. Brain was selected to assess the potential for Tl to cross the blood-brain barrier.

Demonstrating acceptable method performance for trace metal analysis in biological samples is particularly important due to the possibility of potential endogenous and other background sources of these metals. Previous studies have reported that human and animal matrices contain background Tl concentrations on the order of 0.02–1 ng/g for serum and urine and 1–10 ng/g for tissues (Das et al. 2006; Galván-Arzate and Santamaría 1998; Mulkey and Oehme 1993). These endogenous and other background concentrations usually are a result of consumption of food products containing trace amounts of Tl, as it is present in some soils and enters the food chain through plant uptake (Tremel et al. 1997).

Secondary matrix evaluation results are shown in Table 2. All secondary matrices exhibited Tl signals < 24.3% of the LLOQ standard intensity. Highest background levels were seen in HSD female rat urine, GD18 HSD rat fetal homogenate, and HSD rat brain matrices with IS-corrected Tl intensity values between 16.9% to 24.3% of the LLOQ (approximately 0.211 ng Tl/g fetus homogenate to 0.303 ng Tl/mL urine).

Table 2.

Secondary matrix evaluation data for quantitation of thallium.^a

Parameter	Matrix	Value
Secondary Matrix Selectivity (Mean % of LLOQ Response)	HSDb Maternal Rat Plasma	4.35
	HSD Rat Pup Plasma	2.37
	HSD Female Rat Urine	24.3
	HSD Rat Fetus	16.9
	HSD Rat Brain	22.9
	B6C3F1 Female Mouse Plasma	5.37
Accuracy of secondary matrix samplesc (Mean RE, %d)	HSD Maternal Rat Plasma	−1.1
	HSD Rat Pup Plasma	−0.3
	HSD Female Rat Urine	9.1
	HSD Rat Fetus	9.2
	HSD Rat Brain	12.5
	B6C3F1 Female Mouse Plasma	−1.2
Precision of secondary matrix samplesc (RSD, %e)	HSD Maternal Rat Plasma	0.5
	HSD Rat Pup Plasma	0.9
	HSD Female Rat Urine	0.7
	HSD Rat Fetus	1.4
	HSD Rat Brain	1.5
	B6C3F1 Female Mouse Plasma	1.5

^aTl content in secondary matrices were quantified against a calibration curve generated in primary matrix, male HSD rat plasma.

^bHSD = Harlan Sprague Dawley.

^cPrecision and accuracy were determined for six replicate fortified QC samples at 2.00 ng Tl/mL matrix (ng/g for fetus and brain).

^dRE = Percent relative error.

^eRSD = Percent relative standard deviation.

Quantitation of Tl in additional lots of control brains, fetuses, and urine ranged from 0.057 to 0.517 ng Tl/g matrix (ng/mL for urine), indicating potential background of Tl in these matrices, with significant lot-to-lot variability. All secondary matrices were found to provide analytical accuracy and precision −1.2 – 12.5% RE and ≤ 1.5% RSD demonstrating that the validated method can be successfully used to quantitate Tl concentration in these matrices.

Analysis Period and Free-Thaw Stability of Thallium in Primary Matrix

Results of stability of Tl in extracted plasma samples are shown in Table 3. Ambient autosampler plasma extract storage stability (5 days) demonstrated 98.0 – 101% of Day 0 concentrations with RSD ≤2.6% at all concentration levels. Refrigerated plasma extract storage stability samples (5 days) exhibited 97.2 – 101% of Day 0 concentrations with RSD ≤0.7%. Frozen plasma sample storage stability samples (5 days) demonstrated 96.4 – 101% of Day 0 concentrations with RSD ≤ 0.9%. Fortified plasma samples subjected to 3 freeze-thaw cycles of 16 hours or greater demonstrated 98.4 – 100% of Day 0 concentrations with RSD ≤ 1.6%. Tl-fortified plasma samples were shown to be stable for all storage conditions investigated here.

Table 3.

Stability data for thallium.

Storage condition	RSD, %a	Mean Percentage (%) of Day 0
Extract stabilityb
Ambient extracts (5 days)	≤ 2.6	98.0 – 101
Refrigerated extracts (15 days, 4°C)	≤ 0.7	97.2 – 101
Primary Matrix stability b
Frozen samples (5 days, −20 °C)	≤ 0.9	96.4 – 101
Freeze–thaw (3 cycles over 3 days)	≤ 1.6	98.4–100
Secondary Matrix Stability c
(Up to 521 days, −80 °C)
HSDd Rat Maternal Plasma (PND4)	≤ 2.8	101 – 106
HSD Rat Pup Plasma (PND4)	≤ 2.3	103 – 107
HSD Female Rat Urine	≤ 2.4	106 – 112
HSD Rat Fetus	≤ 1.9	93.4 – 110
HSD Rat Brain	≤ 7.3	99.2 – 107
B6C3F1 Female Mouse Plasma	≤ 6.6	99.1–109

^aRSD = percent relative standard deviation

^bPercent of Day 0 were determined for triplicate fortified QC samples at three levels, 2.00, 50.0, and 400 ng Tl/mL plasma.

^cPercent of Day 0 were determined for triplicate fortified QC samples at two levels, 2.00 and 250 ng Tl/mL matrix or ng Tl/g tissue for brain and fetus.

^dHSD = Harlan Sprague Dawley.

Stability of Thallium in Secondary Matrices

Stability of Tl in secondary matrices when stored frozen at −80 °C are also shown in Table 3. For all secondary matrices, the average Tl concentration up to 521 days was within 93.4 – 112% of Day 0 concentrations with RSD ≤ 7.3% for all replicates. Therefore, Tl can be accurately determined in samples that have been stored in an ultracold storage for up to 521 days. These results are consistent with a recent study on the stability of a panel of trace elements in human blood and plasma, which demonstrated that thallium concentrations were stable for up to 360 days after storage at −80 °C (Tanvir et al. 2021).

Conclusions

Tl has found many industrial uses, making it an economically important element. Significant public health concerns surround its occurrence in drinking water due to well-known acute toxicity, although little data exists around chronic low-level exposure. A versatile bioanalytical method is needed to produce robust, scientifically-defensible analytical results in a variety of tissues and biological fluids to put toxicological findings into context. The method presented here is fit for that purpose, as demonstrated by the results of the analytical validation. ICP-MS analysis allows for sensitive Tl determination in plasma at trace levels with superior performance as demonstrated by the measured accuracy (RE) and precision (RSD) of Tl spiked into plasma. The method boasts analytical limits comparable or superior to previously-reported methods that use earlier analytical techniques like graphite furnace atomic absorption spectrometry. The results reported here demonstrate that the method can reliably generate accurate and precise data for Tl toxicity studies.