Evaluation of Multiple Blood Matrices for Assessment of Human Exposure to Nerve Agents

Abstract

Biomedical samples may be used to determine human exposure to nerve agents through the analysis of specific biomarkers. Samples received may include serum, plasma, whole blood, lysed blood and, due to the toxicity of these compounds, postmortem blood. To quantitate metabolites resulting from exposure to sarin (GB), soman (GD), cyclosarin (GF), VX and VR, these blood matrices were evaluated individually for precision, accuracy, sensitivity and specificity. Accuracies for these metabolites ranged from 100 to 113% with coefficients of variation ranging from 2.31 to 13.5% across a reportable range of 1–100 ng/mL meeting FDA recommended guidelines for bioanalytical methods in all five matrices. Limits of detection were calculated to be 0.09–0.043 ng/mL, and no interferences were detected in unexposed matrix samples. The use of serum calibrators was also determined to be a suitable alternative to matrix-matched calibrators. Finally, to provide a comparative value between whole blood and plasma, the ratio of the five nerve agent metabolites measured in whole blood versus plasma was determined. Analysis of individual whole blood samples (n = 40), fortified with nerve agent metabolites across the reportable range, resulted in average nerve agent metabolite blood to plasma ratios ranging from 0.53 to 0.56. This study demonstrates the accurate and precise quantitation of nerve agent metabolites in serum, plasma, whole blood, lysed blood and postmortem blood. It also provides a comparative value between whole blood and plasma samples, which can assist epidemiologists and physicians with interpretation of test results from blood specimens obtained under variable conditions.

Introduction

Organophosphorus nerve agents (OPNAs) are a class of highly toxic, synthetic cholinesterase inhibitors (1) that have been stockpiled for warfare use since the early 1900s. Even with the ratification of the Chemical Weapons Convention in 1997 (2), concerns remain that nerve agents will be implemented for terrorist activities. Uses in Syria (3), Matsumoto (4), Tokyo (5, 6) and Iran (7) support these concerns and highlight the continued need to identify human exposure to these compounds. To determine the causative agent, information beyond visible cholinesterase poisoning symptomology is needed. Measurement of unique biomarkers such as metabolites and protein adducts in clinical samples (8–10) can confirm exposure and identify the specific nerve agent.

The most prevalent nerve agent biomarker is the alkyl phosphonic acid metabolite, which accounts for up to 90% of the dose detected from exposure to these compounds (11). Although the majority of these metabolites are excreted in urine, they have also been detected in blood following human and animal exposures to VX, sarin (GB) and cyclosarin (GF) (10–13). Nerve agent metabolite concentrations of 2–135 ng/mL were measured in human serum samples collected between 1.5 and 2.5 h postexposure to VX (10). Similarly, a direct dermal exposure to VX resulted in serum metabolite concentrations of 1,250 ng/mL (12). Other reports confirmed that alkyl phosphonic acids have remained in blood for days postexposure (9, 14). Although detection of these metabolites may be limited by dose, sample collection time following exposure and analytical sensitivity, it can be concluded that alkyl phosphonic acids are unique, specific biomarkers indicative of OPNA exposure.

To quantify alkyl phosphonic acids in serum or plasma, analytical techniques such as liquid chromatography tandem mass spectrometry (LC–MS-MS) and gas chromatography mass spectrometry (GC–MS) have been used (11, 12, 15–17). Various chromatographic separations were used to isolate these compounds from biological matrices, including reversed phase and hydrophilic interaction chromatography (11, 15, 18). Furthermore, when LC–MS-MS or GC–MS analyses were coupled with additional sample preparation such as dilution, liquid–liquid extraction or solid-phase extraction (SPE), detection limits of these metabolites were documented as low as 0.5 ng/mL (14, 15). Sample preparation using anion exchange, reversed phase and aqueous normal phase SPE as well as combinations of these chemistries (15, 19, 20) have all been successfully used to develop nerve agent exposure analysis methods for clinical matrices.

Following an exposure event, a variety of biomedical samples may be submitted for analysis. In addition to serum and plasma, nonideal samples such as hemolyzed, clotted or postmortem blood samples may be received (21, 22). There is also the possibility that whole blood may be received with insufficient volume to partition the serum or plasma thus requiring the direct analysis of whole blood. Analysis of received samples is crucial regardless of quantity or quality obtained, particularly due to the high potential consequences of OPNA exposure events. To ensure accurate quantitation, assay performance should be evaluated for each of these potential matrices (22–24), which to the authors' knowledge has not been reported for whole blood, lysed blood or postmortem blood.

Prior results used to establish OPNA metabolite concentrations following exposures have been measured in serum or plasma samples (11, 12, 14). However, when whole blood is the only available sample to analyze, a means to translate the concentrations between whole blood and plasma is needed. A reliable comparison of whole blood measurements with plasma measurements has been achieved for other pharmaceutical compounds and corresponding metabolites using a blood-to-plasma (B:P) ratio (25, 26). A correlation between the hematocrit and the B:P ratio was observed for some pharmaceutical compounds and the corresponding metabolites (25, 26). Once established, the B:P ratio can be used to estimate equivalent metabolite concentrations in plasma from quantitative values obtained from whole blood, which would allow the comparison of exposure results regardless of the matrix received during an emergency.

In this study, serum, plasma, whole blood, lysed blood and postmortem blood were fortified with five OPNA metabolites corresponding to human exposure to GB, soman (GD), GF, VX and VR. These fortified matrices were evaluated for precision and accuracy using both matrix-matched and serum calibrators. Sensitivity and specificity were also investigated to identify the reportable range for all five metabolites in each of the matrices and ensure minimal false-positive results. Although postmortem blood was used as a matrix for sample analysis, this study did not address postmortem concentration changes resulting from redistribution of these analytes. Furthermore, to permit for future comparison of OPNA exposure data between plasma and whole blood, a B:P was established using individually fortified blood samples.

Experimental

Materials and chemicals

The following analytes were evaluated for this method: ethyl methylphosphonic acid (EMPA), metabolite of VX [O-ethyl S-(2-diisopropylaminoethyl) methylphosphonothioate; isopropyl methylphosphonic acid (IMPA), metabolite of GB [isopropyl methylphosphonofluoridate]; 2-(methyl)propyl methylphosphonic acid (MMPA), metabolite of VR [O-2-(methyl)propyl S-2-(diethylaminoethyl) methylphosphonothioate]; pinacolyl methylphosphonic acid (PMPA), metabolite of GD [pinacolyl methylphosphonofluoridate] and cyclohexyl methylphosphonic acid (CMPA), metabolite of GF [cyclohexyl methylphosphonofluoridate]. Individual concentrates of EMPA, PMPA, MMPA and CMPA were purchased from Cerilliant Corporation (Round Rock, TX) at 1,000 µg/mL in methanol; IMPA solid (99.40% purity) was purchased from Cerilliant Corporation and diluted to 1,130 µg/mL in HPLC-grade methanol gravimetrically. Isotopically labeled internal standards (ISTD) were purchased from Cerilliant Corporation as a mixture (500 ng/mL in water for each compound; limited distribution) and include the following: EMPA, ethyl-D₅; IMPA, isopropyl-¹³C₃; MMPA, methylphosphonyl-¹³C, methylphosphonyl-D₃; PMPA, trimethylpropyl-¹³C₆ and CMPA, cyclohexyl-¹³C₆.

Organic-free 18.2 MΩ Type I water was generated in-house using a system purchased from Aqua Solutions, Inc. (Jasper, GA). Pelletized ≥99.0% ammonium acetate was purchased from Fluka (St. Louis, MO). HPLC-grade acetonitrile and methanol were purchased from Tedia (Fairfield, Ohio). A matched set of human serum, plasma and whole blood samples were purchased from Tennessee Blood Services Corporation (Memphis, TN). For each matrix, 20 individual vials, obtained from Tennessee Blood Services, were pooled in-house for a total volume of 50 mL. Lysed blood was prepared by freezing whole blood at −70°C for 5 days (24). Postmortem blood was harvested from a single individual presumably unexposed of nerve agents by Golden West Biological (Temecula, CA) 25 h and 40 min following death. Because of the limited shelf life of whole blood, postmortem blood excluded, all experiments were completed within 3 weeks of sample receipt (27, 28). Postmortem blood, plasma and serum were all stored at −70°C until use (29).

Calibrator and quality control preparation

Initial individual solutions of each analyte (20 µg/mL in 90:10 methanol–water) were prepared gravimetrically from concentrates or solids purchased from Cerilliant Corporation (Round Rock, TX) and used to make two stock solutions. Stock solution I (4.0 µg/mL per analyte) was prepared by transferring 400 µL initial solution to a vial containing 1,600 µL of 95:5 acetonitrile–water. Stock solution II (0.2 µg/mL per analyte) was prepared by pipetting 100 µL of stock solution I to a vial containing 1,900 µL of 95:5 acetonitrile–water. Calibrator spiking solutions were prepared in 95:5 acetonitrile–water from stock solutions I and II at the following concentrations: 1.0, 2.0, 5.0, 10, 25, 50 and 100 ng/mL; the final calibrator concentrations were at the same concentrations as the calibrator spiking solutions. Similarly, quality control low (QCL, 4.0 ng/mL) and quality control high (QCH, 40 ng/mL) spiking solutions were prepared by aliquoting 200 µL of stock solution II and 100 µL of stock solution I, respectively, to 10.0 mL volumetric flasks and diluting to volume with 95:5 acetonitrile–water. Concentrated samples for testing a proposed dilution scheme were prepared volumetrically at 1,000 ng/mL in serum, plasma, whole blood, lysed blood and postmortem blood. Twenty-five microliters of each 1,000 ng/mL sample was diluted with 600 µL of pooled serum, resulting in a 40.0 ng/mL sample for analysis. A working ISTD solution was prepared by diluting the purchased 500 ng/mL solution to a final concentration of 23.8 ng/mL in deionized water. Samples prepared in whole blood were stored at 4°C; all other solutions were stored at −20°C.

Sample preparation and analysis

To prepare calibrators and controls for extraction, 25 µL of the working ISTD solution, 50 µL of spiking calibrator and 50 µL of biological matrix were added to a 2-mL 96-well plate. For unknown samples, 25 µL of the working ISTD solution, 50 µL of 95/5 acetonitrile–water and 50 µL of biological matrix were added to a 2-mL 96-well plate. All calibrators, controls and samples were diluted with 1 mL acetonitrile, sealed with Adhesive polymerase chain reaction (PCR) Sealing Foil (Thermo Scientific, Hudson, NH), and vortexed for 5 min on a ThermoLab Systems Wellmix (Hudson, NH). Samples were then centrifuged for 5 min at 3,000 rpm using an Eppendorf Centrifuge 5810 R (Hauppauge, NY) to pelletize the precipitated proteins.

SPE was automated by using a Perkin–Elmer Zephyr system (Waltham, MA) using vacuum to draw the solutions through the plate. The Phenomenex Strata Si-1 SPE 96-well plate (55 µM, 70 Å, 100 mg, Torrance, CA) was pretreated with 1,000 µL of 25% water in acetonitrile, followed by 1,000 µL of acetonitrile. The supernatant was loaded onto the SPE 96-well plate and rinsed by a two-step process: (i) 1,000 µL of acetonitrile; and (ii) 1,000 µL of 7% water in acetonitrile. The analytes were then eluted with 1,000 µL of 28% water in acetonitrile and collected in a clean 96-well Nunc^® plate.

Samples were concentrated at 70°C under 25 L/min of nitrogen in a Biotage 96-Well TurboVap (Charlotte, NC) for 30 min and then evaporated to dryness after increasing the flow rate to 70 L/min. The samples were reconstituted using 100 µL of 5% water in acetonitrile, mixed by pipet twice, and transferred to a 150 µL 96-well PCR plate. The plate was sealed using Easy Pierce 20 µm Foil Sheet (Thermo Scientific, Hudson, NH) and an AB gene plate sealer (Thermo Scientific).

Instrumental analysis

The chromatographic separation was performed using an Agilent 1290 HPLC (Santa Clara, CA) with a Waters Atlantis^® HILIC 2.1-mm × 50-mm (3-μm particles with 70% porosity) HPLC column (Milford, MA) at 35°C. The sample needle was washed for 10 s prior to a 10 µL sample injection. Isocratic chromatographic separation was accomplished with a mobile phase of 88% acetonitrile and 12% 20 mM ammonium acetate. The flow rate was increased from 500 to 1,500 μL/min at 3.01 min then returned to 500 μL/min at 5.01 min for a total cycle time of 5.01 min.

The mass spectrometric (MS) analysis was performed using an API 5500 triple quadrupole QTRAP mass spectrometer from Sciex (Redwood City, CA) controlled by Analyst^® software (version 1.6.2). The mass spectrometer was operated in negative-ion mode, using multiple reaction monitoring (MRM). Two transitions were monitored for each analyte, and one transition monitored for each internal standard. The same target fragment ions were selected for identification of nerve agent metabolites as documented by Hamelin et al. (15) with the exception of the CMPA quantitation (m/z 177.0 → 79.0) and confirmation (m/z 177.0 → 95.0) transitions. The following conditions were optimized for each individual transition to maximize signal: declustering potential, collision energy and cell exit potential. The MS settings are as follows: curtain gas (CUR), 35 psi; nebulizer gas (GS1), 40 psi; turbo gas (GS2), 40 psi; GS2 temperature (TEM), 550°C; collision gas, nitrogen; collision gas (CAD), medium, with an average pressure reading of 2.8 × 10⁻⁵ Torr; ionspray potential (IS), −4,000 V; entrance potential (EP), −10 and interface heater (IHE), on.

Data analysis

Quantitation was based on a standard curve comprising of eight calibrators ranging from 1 to 100 ng/mL. The calibrator response was divided by the internal standard response to normalize any sample loss that may have occurred during preparation, separation and ionization, then plotted against the concentration of the calibrators. Least-squares regression with x⁻¹ weighting of each observation was used to estimate all calibration curves: a linear regression fit was used for all analytes with the exception of MMPA, which used a quadratic regression fit. Each calibration curve must have attained a coefficient of determination (R²) of 0.990 or greater to be accepted. Least-squares regression was also used to evaluate correlation between calculated plasma concentrations to directly measured plasma concentrations.

The limits of detection (LOD) of the assay were calculated using the approach described by Taylor (30). The standard deviation was determined from the quantitated values (n = 15) for the lowest calibrator obtained over the course of the characterization for each analyte. These standard deviations were then multiplied by three to estimate the LOD.

Stability evaluation

Two separate pools of human serum were purchased from Tennessee Blood Services (Memphis, TN) and labeled matrix 1 (M1) and matrix 2 (M2). Each of these two pools was then divided into three portions and fortified to the final concentrations of 2.0, 15 and 75 ng/mL, and labeled low, medium and high, respectively. These six fortified sera samples were then aliquoted into Nalgene^® 2 mL cryogenic vials (Sigma-Aldrich, St. Louis, MO) for storage at −20°C, 4–8°C and at room temperature (∼25°C) to test analyte stability. Twelve replicates of each serum sample concentration were prepared and analyzed to establish the concentration at time zero. At time points of 1, 2, 3, 4, 6 and 8 weeks, serum samples from each of the storage conditions were prepared and analyzed in triplicate. Additionally, serum samples were stored at −70°C and defrosted up to four times for use in a freeze–thaw experiment. Freeze–thaw samples were prepared and analyzed in triplicate. Serum standards were stored at −70°C prior to use.

Hematocrit measurement

Hematocrit, a measurement of the amount of whole blood cells to plasma, was measured for all whole blood samples prior to the collection of plasma following centrifugation. This measurement was obtained using micro-hematocrit capillary tubes and a micro-hematocrit centrifuge (LW Scientific, Lawrenceville, GA).

Safety considerations

The techniques and materials in this method do not pose any special hazards. Only OPNA metabolites were used for sample fortification, and no live nerve agents were used. Blood products purchased from Tennessee Blood Services were prescreened by the vendor in accordance with FDA regulations to be free of hepatitis B, syphilis and HIV. This study used de-identified blood products acquired from commercial sources, and thus the work did not meet the definition of human subjects as specified in 45 CFR 46.102 (f) (Department of Health and Human Services, 2009). General considerations include exercising universal precautions, such as wearing appropriate personal protective equipment, when handling chemicals and biological specimens.

Results and discussion

Extracted serum samples were analyzed using the liquid chromatography tandem mass spectrometry (LC–MS-MS) conditions established by Hamelin et al. (15). To further optimize MS response, a mobile phase composition of 88:12 acetonitrile–20 mM ammonium acetate was selected. The flow rate was increased to 1,500 µL/min after elution of all analytes to remove additional matrix components. Furthermore, new quantitation and confirmation transitions were chosen for CMPA to maximize sensitivity and reproducibility.

Using these updated parameters, 21 calibration curves with QC samples in human serum were extracted and analyzed over a period of 22 days by three analysts. The assessment of two QC samples at concentrations of 4.0 and 40 ng/mL demonstrated high accuracy (98.4–103%) with a low coefficient of variation (CV = 3.5–7.0%) for the detection of these five nerve agent hydrolysis products in serum. The estimated LODs for these compounds in serum using an acetate buffer ranged from 0.13 to 0.17 ng/mL, which were comparable to previously reported limits using a fluoride buffer (15).

Fortified samples were prepared in serum, plasma, whole blood, lysed blood and postmortem blood to determine the precision, accuracy, sensitivity and specificity for each matrix using the reported method. Equivalence between each matrix and serum was evaluated to determine if serum calibrators could be used for quantitation of these matrices. Using a singular calibrant matrix would increase sample throughput and provide an accurate, cost-effective alternative to preparing and storing multiple matrix calibrators. Individual whole blood samples, fortified with metabolites, were analyzed along with their corresponding plasma samples to determine an overall B:P ratio for each compound.

Accuracy and precision

Extension of this assay to additional blood matrices required the determination of reproducibility and accuracy for these five compounds within each matrix. To accomplish this, plasma, whole blood, lysed blood and postmortem blood were fortified with nerve agent metabolites at 2.0, 25 and 100 ng/mL and subsequently extracted and analyzed. A representative chromatogram in postmortem blood is shown in Figure 1. A total of 15 replicates of each sample concentration in each matrix was analyzed over a period of 5 days. Eight calibrators, at concentrations 1.0, 2.0, 5.0, 10, 25, 50 and 100 ng/mL, were also prepared in the four matrices and analyzed along with the samples each day. Serum quality control (QC) samples and calibrators were prepared, extracted and analyzed in the same manner to serve as the reference matrix. The accuracies for the samples at 2.0, 25, and 100 ng/mL in the four additional matrices were calculated using the matrix-matched curves to be 100.1–110.1%, 100.1–105.9% and 100.1–113.2%, respectively. The corresponding CV for these values were 2.98–13.5%, 2.31–5.51% and 2.29 9.12%. All values were within the recommended limits for precision and accuracy for bioanalytical methods (31). Quantitated results for serum, lysed blood and postmortem blood are shown in Table 1.

Figure 1.

Chromatograms of (a) a 10.0-ng/mL calibrator prepared in postmortem blood and (b) unfortified postmortem blood. This figure is available in black and white in print and in color at JAT online.

Table 1.

Mean, CV and Accuracy Values Obtained for 2.0, 25, and 100 ng/mL Samples (n = 15) Using Matrix-Matched Calibrators

	Prepared (ng/mL)	Serum			Lysed blood			Postmortem blood
		Mean	CV (%)	Accuracy	Mean	CV (%)	Accuracy (%)	Mean	CV (%)	Accuracy (%)
EMPA	2	1.97	5.56	101	1.98	4.27	101	2.05	5.95	103
	25	24.8	4.16	101	24.7	4.33	101	24.7	4.64	101
	100	101	3.49	101	99.4	3.56	101	99.6	2.29	100
IMPA	2	1.91	6.97	104	1.99	4.99	105	1.80	13.50	110
	25	25.6	3.81	102	25.6	5.51	103	24.7	2.31	101
	100	98.3	5.12	102	99.7	3.46	100	98.8	3.12	101
MMPA	2	1.99	2.98	101	1.95	3.29	103	1.96	5.34	102
	25	25.5	3.95	102	24.8	4.14	101	24.2	3.21	103
	100	99.8	9.02	100	101.5	6.05	102	99.8	3.92	100
PMPA	2	2.00	7.31	100	1.82	10.29	109	2.06	8.52	103
	25	24.8	2.45	101	25.3	4.94	101	25.0	5.03	100
	100	99.6	3.65	100	101.5	3.80	102	100.9	4.06	101
CMPA	2	1.94	8.85	103	2.00	4.59	100	2.01	5.36	100
	25	25.2	4.05	100	25.2	3.78	101	24.8	4.78	101
	100	102	4.54	102	102.2	4.15	102	99.9	4.69	100

Evaluation of control samples using serum-matched calibrators

The use of one matrix for calibration of multiple blood matrices is desirable to minimize materials and simplify sample analysis. To test a single calibrant matrix, calibration curves prepared in serum were used to evaluate plasma, whole blood, lysed blood and postmortem blood samples. Fortified samples at 2.0, 25 and 100 ng/mL quantitated with serum calibrators resulted in a CV of 15.4% or less, and the mean measured values were within 9.3% of the expected value for all metabolites in the four matrices. The mean values obtained for the 25 ng/mL control sample (n = 15) in each blood matrix are shown in Figure 2. The results were also evaluated using the nonparametric Kruskall–Wallis test and parametric analysis of variance, and the differences in concentration among the matrices was not statistically significant. These data indicate that the use of serum calibrators for plasma, lysed blood, whole blood and postmortem blood was comparable to the use of the matrix-matched calibrators.

Figure 2.

Calculated concentration of a 25.0-ng/mL EMPA control sample in serum, plasma, whole blood, lysed blood and postmortem blood quantitated with serum calibrators. Dashed lines represent acceptable percent error (±15%); error bars equal to ±1 SD. This figure is available in black and white in print and in color at JAT online.

Reportable range

A linear response was confirmed for these metabolites from 1 to 100 ng/mL from the R² value, which exceeded 0.990 for all calibration curves analyzed for this study. To measure the reproducibility and confirm the lower limit of quantitation, a low-level QC sample at 1.0 ng/mL in each matrix was extracted and analyzed five times per day on three separate days for a total of 15 individual analyses. The inaccuracy for the low-level QC sample, determined using matrix-matched calibrators, was <13.5% with a CV of <16%. The standard deviations of the quantitated results were used to estimate LODs. The resulting LODs ranged from 0.09 to 0.43 ng/mL in the four matrices evaluated, similar to the LODs for these metabolites in serum (15).

Serum metabolite concentrations resulting from exposures to GB and VX have been documented above the quantitation range (10, 12) of this method. Given previous concerns with MS detector saturation with these compounds (18), dilution of samples above the highest calibrator was selected to extend the dynamic range. To evaluate this approach for samples outside the method range of the reported method, a 1,000 ng/mL sample was prepared in each of the five matrices and diluted with pooled serum to a final concentration of 40.0 ng/mL. Each of the diluted samples was prepared in triplicate and quantitated using a serum calibration curve. The un-diluted sample concentration was calculated from the analytical result multiplied by the dilution factor. The accuracy of each mean ranged from 96.2 to 113% with the corresponding variance of 0.55–6.83%. These results confirmed that dilution is an acceptable means to quantify samples above the highest calibrator in each of the tested matrices.

Sample analysis

Twenty individual samples of serum, plasma and whole blood with no known exposure to OPNAs were extracted and analyzed to evaluate for potential chromatographic interferences. One postmortem blood sample, collected from a single donor who showed no signs of exposure to OPNAs, was also evaluated for interferences (Figure 1). Adjacent peaks were initially seen in the confirmation transition for EMPA; however, this matrix contribution was chromatographically separated when the mobile phase composition was adjusted as described. No additional matrix interferences were detected at the analyte retention times for all metabolites, supporting the specificity of this analysis for these matrices.

Stability results

The stability of the nerve agent metabolites was evaluated in serum, using three spike levels of each metabolite prepared in two separate sera pools (M1 and M2). After preparation and aliquoting, 12 samples for each level were analyzed for an initial value. The remaining samples were stored in a −20°C freezer, 2–4°C refrigerator or on the bench at room temperature until analyzed in triplicate at Weeks 1, 2, 3, 4, 6 and 8. A lack of stability was defined as a mean result >10% below the initial value for two consecutive time points. For the duration of the stability study, all analytes were stable at −20°C, 2–4°C and at room temperature for all spike levels in both sera. Based on the Arrhenius approximation (32), the nerve agent metabolites in samples held at room temperature (∼20°C) would be stable for at least 2 years.

B:P ratio

Due to sample volume limitations, it may not be possible to separate whole blood into serum or plasma for analysis. To determine the nerve agent metabolite B:P ratios for each of the five nerve agent metabolites, 40 individual whole blood samples were fortified at eight concentrations across the reportable range using a 20-µg/mL OPNA metabolite stock prepared in water for a total of five individual samples per concentration. Aliquots of these whole blood samples were extracted and analyzed. The remaining fortified whole blood samples were centrifuged to obtain the corresponding plasma samples, which were also extracted and analyzed. The quantitated results from whole blood and plasma were used to calculate the B:P ratio for each analyte. The mean values for the B:P ratio were similar for all the metabolites (0.53–0.57) with CVs ranging from 19 to 22%. Estimated plasma concentrations were determined from the whole blood measurements using these B:P ratios. The estimated plasma concentrations were compared with the measured plasma concentrations graphically in Figure 3 with the resulting correlation coefficients from a linear least squares regression reported in the inset table.

Figure 3.

Comparison of estimated EMPA plasma concentration using blood to plasma ratios (B:P) to directly measured plasma concentration. Average blood to plasma ratios and coefficients of determination for five nerve agent metabolites determined from the regression of estimated (using B:P ratios and hematocrit corrected equations) vs measured concentrations are also reported (inset). This figure is available in black and white in print and in color at JAT online.

The B:P ratio values for all OPNA metabolites were <1, indicating that the metabolites had a lower solubility in the red blood cells than the plasma (25) which is supported by the polarity of these compounds (log P −0.3 to 1.4). Even so, the preferential distribution of the metabolites in plasma may not entirely remove the effect of hematocrit. For this reason, hematocrit was measured for all whole blood samples prior to fortification. Hematocrit values ranged from 15 to 48%, with the majority of the values between 35 and 45%. When estimated plasma concentrations were calculated from whole blood concentrations, a significant association with hematocrit in linear regression models was identified. To determine if an improved plasma concentration estimation which incorporated the hematocrit value in the plasma concentration estimation, a novel conversion equation was applied (Equation (1)).

$$Y={\hat{\beta }}_0+{\hat{\beta }}_1{X}_1+{\hat{\beta }}_2\cdot X_2+{\hat{\beta }}_{12}\cdot X_{12},$$ (1)

where ${\hat{\beta }}_0$ is the estimate of the intercept for plasma concentration, ${\hat{\beta }}_1$ is the estimate of the linear slope for the unit change in plasma concentration for each unit change in whole blood concentration, ${\hat{\beta }}_2$ is the estimated unit change in plasma concentration for each unit change in hematocrit concentration and ${\hat{\beta }}_{12}$ is the estimated unit change in plasma concentration for each unit change in the product of whole blood and hematocrit concentrations. The inputs for the linear regression equations for each nerve agent metabolites are defined in Supplementary data, Table SI. When the estimated plasma concentration, determined using Equation (1), was correlated to the measured plasma concentration, the R² values ranged from 0.956 to 0.968 (Figure 3, table inset). The incorporation of hematocrit resulted in a slight improvement of the estimate of plasma concentration from whole blood measurements using only the B:P ratio. Calculation of plasma concentrations from measured whole blood concentrations, with and without hematocrit adjustments, provided similar values when compared with the measured plasma concentrations. For this reason, estimates without hematocrit adjustment were deemed sufficient for potential emergency response scenarios where insufficient sample volume would preclude hematocrit measurements.

Conclusion

To evaluate potential human exposure samples, five nerve agent metabolites were quantitated in plasma, whole blood, lysed blood and postmortem blood. Control samples in each matrix demonstrated high accuracy and precision for each matrix tested across a range of concentrations from 1 to 100 ng/mL. Additionally, quantitation using serum calibrators resulted in accuracy and precision that was within FDA's recommended limits for all four matrices. This matrix equivalence allows the use of only one set of calibrators and controls, providing a cost-effective alternative to storing or procuring multiple blood matrices for use as a calibrant matrix. Nerve agent metabolite blood-to-plasma ratios were also determined for all analytes. The resulting values, ranging from 0.53 to 0.57, can be used to compare future data obtained in whole blood to previous studies in plasma. True exposure samples could be used to further validate the predictability of the results of this study; in the meantime, this ratio can provide guidance for the interpretation of results obtained solely in whole blood.

Supplementary data

Supplementary data are available at Journal of Analytical Toxicology online.

Conflict of interest statement

The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention. Use of trade names is for identification only and does not imply endorsement by the Centers for Disease Control and Prevention, the Public Health Service or the US Department of Health and Human Services.