Validated HPLC-MS/MS Method To Quantify Low Levels of Domoic Acid in Plasma and Urine after Subacute Exposure.

Domoic acid (DA) is a marine neurotoxin produced by several species of Pseudo-nitzschia. DA causes severe neurological toxicity in humans and animals. To address the current analytical need to quantify low levels of DA in human and animal body fluids, a sensitive and selective high performance liquid chromatography-tandem mass spectrometry method was developed to measure DA in plasma and urine. This method was fully validated to accurately and precisely quantify DA between 0.31 and 16 ng/mL in plasma and between 7.8 and 1000 ng/mL in urine. Our group introduced the use of a novel internal standard, tetrahydrodomoic acid to control for matrix effects and other sources of variability. This validated method will be useful to assess DA concentrations in biological samples of human or animal origin after suspected DA exposure from contaminated food. It will also be applicable to sentinel programs and research studies to analyze body fluids with low levels of DA.

Introduction

Domoic acid (DA) is a marine neurotoxin produced by several species of diatoms of Pseudo-nitzschia.[1] DA is readily filtered and taken up by shellfish that is subsequently consumed by predators and humans. Exposure to DA via consumption of contaminated shellfish is a risk to human health. During a marked algal bloom in 1987, 107 people experienced acute neurological symptoms, and 3 died shortly after eating mussels contaminated with DA.[2,3] To protect human health, DA monitoring programs have been established, and extensive efforts have been made to develop sensitive and selective methods to measure DA concentrations in seafood and seawater.[4] The recently developed methods include indirect measurement of DA via competitive enzyme-linked immunosorbent assay (ELISA)[5] and direct measurement of DA via liquid chromatography-tandem mass spectrometry (LC-MS/MS)[6−14] or laser ablation-MS.[15] These methods have been successful in monitoring DA in matrices that are available in large quantities, such as seawater and mussels. However, no methods have been published to determine DA concentrations in human blood or urine after potential exposures to subacute doses of DA, and plasma DA concentrations have not been reported in humans following contaminated shellfish consumption.[16,17] DA exposures have, however, been measured using indirect competitive ELISA followed by confirmatory LC-MS/MS in marine mammals, such as California sea lions (CSLs),[18] exposed to DA via food and showing neurological symptoms of acute or chronic toxicity. The DA concentrations ranged from undetectable to 200 ng/mL in serum and from undetectable to 3700 ng/mL in urine. On the basis of estimates of DA exposure and toxic effects observed therein, and from results of laboratory studies in nonhuman primates, researchers and environmental health agencies have proposed estimates for tolerable daily intake (TDI) of DA from 0.075 to 0.1 mg/kg/day.[19−22] However, recent studies have shown that some recreational harvesters consume DA in excess of the proposed TDI[16] and that chronic low-level exposure to DA is associated with memory loss.[17] On the basis of the dose–exposure relationship in nonhuman primates and physiologically based pharmacokinetic (PBPK) modeling and simulations,[23] we predicted that a bioanalytical method with a lower limit of quantification (LLOQ) around 0.2−0.4 ng/mL is needed to detect DA in human or nonhuman primate plasma after exposure to DA at or above the proposed TDI. None of the currently available methods to detect DA in serum by LC-MS/MS reaches this sensitivity, and a single method has been reported to detect DA in urine at this level (Table ). Hence, more sensitive, validated, LC-MS/MS methods are needed to assess human DA exposures.

Table 1

Summary of Published Bioanalytical Methods and the Available Validation Data for Measurement of DA in Serum and Urinea

bioanalytical method	matrix	species	extraction	sample required	LOD (ng/mL)	LLOQ (ng/mL)	recovery (%)	intraday variability (%)	interday variability (%)	references
LC-MS	serum	bovine	SPE-HLB	1 g	5 ng/g		92–95			(25)
LC-MS	serum	rat	SPE-C18	200 μL	3		95			(26)
LC-MS	serum	marine mammals	SPE-C18	500 μL			89–93			(6)
LC-MS	serum	CSL	MeOH + filtration	1–4 g		>20				(5)
LC-MS	serum	CSLb	SPE-C18		0.48					(27)
ELISAb	serum	rat	MeOH				>100	11.50	7.80	(26)
ELISAb	serum	CSLb	MeOH + filtration	1–4 g		0.4				(5)
ELISAb	serum	CSLb	n/a		0.25			18		(27)
ELISAc	serum	CSLb	SPE-C18		2.5			4		(27)
LC-MS	urine	bovine	SPE-HLB	1 g	5 ng/g		90–98			(25)
LC-MS	urine	marine mammals	SPE-C18	1000 μL			79–104			(6)
LC-MS	urine	CSLb	MeOH + filtration	1–4 g		>20				(5)
LC-MS	urine	CSLb	SPE-C18		0.48					(27)
LC-MS	urine	human	SPE-PAX	100 μL	0.12	0.37	88–103	2.1–7.6	2.6–12.7	(28)
ELISAb	urine	CSLb	MeOH + filtration	1–4 g		0.4				(5)
ELISAb	urine	CSLb	n/a		0.25			14		(27)
ELISAc	urine	CSL	SPE-C18		2.5			4		(27)

Empty spaces indicate lack of available data. CSL, California sea lion.

Using polyclonal antibody (Biosense).

Using monoclonal antibody (MeS).

DA is a hydrophilic molecule (clog P = −0.23, value obtained from Drug Bank on June 12, 2018, https://www.drugbank.ca/drugs/DB02852) with three carboxyl groups (pKa: 1.85, 4.47, 4.75) and an amine group (pKa: 10.60),[24] as shown in Figure A. The ionized and polar nature of DA makes extraction from biological matrices challenging. Solid-phase extraction (SPE) has been used to obtain DA from complex samples,[6,25−28] but this step can introduce variability into the assay via sample recovery.

Figure 1

UV spectra of (A) 30 μM DA in water and (B) 20 μM tetrahydrodomoic acid (THDA) in 10% aqueous methanol.

Several bioanalytical methods have been developed or optimized to measure DA concentrations directly or indirectly in serum and urine samples from various mammalian species, as listed in Table . However, validation data have not been provided for any of the LC-MS/MS methods for serum analysis and for only one of the urine analysis methods.[28] Despite the extraction methods used in the published assays and the known variability in analyte ionization in LC-MS/MS methods, none of these published methods for serum and urine analyses includes an internal standard. In mussel and seawater analyses, the deuterated dansyl chloride derivative of DA,[29] kainic acid (KA),[30] and leucine-enkephalin (ENK)[7,9] have been used as internal standards for LC-MS/MS methods. These internal standards have distinctly different chromatographic and spectrometric properties than DA, which decrease their usefulness. They have also not been applied to plasma, serum, or urine analysis. To address the lack of internal standards in published methods, we explored the feasibility to use fully reduced tetrahydrodomoic acid (THDA) (Figure B) as a suitable internal standard for DA measurement by LC-MS/MS.

The goal of this study was to develop and validate a high performance liquid chromatography-tandem mass spectrometry (HPLC-MS/MS) method for quantification of DA in urine and plasma with sufficient sensitivity to measure DA following low-level exposure to DA. The applicability of this method was confirmed in a cohort of Macaca fascicularis (cynomolgus) monkeys following a single oral dose of 0.075 mg/kg DA (proposed human TDI for DA). To further enhance the applicability, the method was cross-validated in human plasma.

Results

Fragmentation Patterns and HPLC-MS/MS Optimization

To characterize the ionization and fragmentation of DA and THDA, MS2 scans of the precursor ions and MS3 scans of the two product ions were collected. The enhanced product ion (EPI) mass spectra are shown in Figure . The product ions observed in the MS2 spectra for DA ([M + H]+m/z 312) are consistent with published spectral data.[31,32] The MS3 spectra of the two product ions of DA (m/z 312 > 294 and m/z 312 > 266) and THDA (m/z 316 > 298 and m/z 316 > 270) support the analogous fragmentation of the two compounds. HPLC-MS/MS multiple reaction monitoring (MRM) was optimized for quantification of DA. The three most sensitive MS2 product ions for DA (m/z 312 > 266, m/z 312 > 248, and m/z 312 > 220) and THDA (m/z 316 > 270, m/z 316 > 252, and m/z 316 > 224) were chosen for each compound. The final MS/MS conditions are described in Materials and Methods section. Representative chromatograms of DA and THDA are shown in Figure .

Figure 2

Mass spectral analyses of DA and THDA. MS2 EPI scans of (A) 3.1 ng/mL (10 nM) DA and (B) 0.63 ng/mL (2 nM) THDA. MS3 EPI scans of DA product ions (C) m/z 312 > 294 and (E) m/z 312 > 266. MS3 EPI scans of THDA product ions (D) m/z 316 > 298 and (F) m/z 316 > 270.

Figure 3

MRM chromatogram of 3.1 ng/mL (10 nM) DA and 0.63 ng/mL (2 nM) of THDA. DA elutes at 2.4 min, whereas THDA elutes at 2.6 min.

The on-column limit of detection (LOD) of DA was determined to be 0.52 pg (1.7 fmol) (Figure A). Signal-to-noise ratio (S/N) was greater than 3 for transition m/z 312 > 266, whereas the S/N was less than 3 for transitions m/z 312 > 248 and m/z 312 > 220 at LOD. The on-column lower limit of quantification (LLOQ) was 1.0 pg (3.3 fmol) (Figure B). The S/N was greater than 3 for the all three transitions at LLOQ.

Figure 4

MRM chromatograms of DA at (A) on-column LOD at 0.52 pg (1.7 fmol) and (B) on-column LLOQ at 1.0 pg (3.3 fmol).

Method Validation

Selectivity of the method was assessed by analyzing blank plasma and urine samples from six treatment-naïve monkeys to determine potential matrix interference. An interference peak was observed at 2.4 min in both plasma and urine chromatograms with the MRM transition m/z 316 > 270 and hence this MRM transition was not used. No interference peak was observed in the chromatograms with the other MRM transitions (Figure A,B). Specificity of the method was assessed by analyzing plasma and urine samples with THDA as an internal standard and without DA to determine potential interference from the internal standard. No interference peak was observed at any of the DA MRM transitions. (Figure C,D). DA recovery was complete in all samples with measured recoveries of 105 ± 10, 108 ± 2, 107 ± 10, and 101 ± 2% at 0.31, 0.93, 7.8, and 12 ng/mL (1, 3, 25, and 40 nM), respectively, in plasma and 113 ± 9, 110 ± 8, 107 ± 3, and 106 ± 5% at 7.8, 23, 500, and 780 ng/mL (25, 75, 1600, and 2500 nM), respectively, in urine. Extracted samples were stable for up to 5 days in the autosampler and up to a day on bench top. Plasma and urine samples were stable at −20 °C storage for at least five freeze–thaw cycles and for up to 28 weeks in plasma and 34 weeks in urine.

Figure 5

MRM chromatograms monitored for DA transitions in plasma and urine. The transitions monitored were m/z 312 > 266, 248, 220 for DA and m/z 316 > 252 for THDA. (A) Blank plasma, (B) blank urine, (C) plasma with IS, (D) urine with IS, (E) plasma spiked with 0.31 ng/mL (1 nM) DA (LLOQ), and (F) urine spiked with 7.8 ng/mL (25 nM) DA (LLOQ).

The DA/THDA peak area ratio was linear between 0.31 and 16 ng/mL (1 and 50 nM) of DA in plasma and between 7.8 and 1000 ng/mL (25 and 3200 nM) of DA in urine. All calibration curves satisfied the predefined acceptance criteria with greater than 80% nonzero calibration standards within 15% of nominal concentration. The LLOQ in plasma was 0.31 ng/mL (1 nM), at which concentration all three mass transitions (m/z 312 > 266, 248, 220) were detected with S/N > 3. The LOD in plasma was 0.16 ng/mL (0.5 nM), at which concentration only mass transition m/z 312 > 266 was detected with S/N > 3. The LLOQ in urine was 7.8 ng/mL (25 nM), at which concentration all three mass transitions were detected with S/N > 3. The representative LC-MS chromatograms at LLOQ in plasma and urine are shown in Figure E,F. The intraday and interday accuracy (% error) and precision (% CV) in plasma and urine are shown in Table .

Table 2

Accuracy and Precision Data for Monkey and Human Plasma and Urinea

	LLOQ		LQC		MQC		HQC
	Plasma
	0.31 ng/mL		0.93 ng/mL		7.8 ng/mL		12 ng/mL
	% error	% CV	% error	% CV	% error	% CV	% error	% CV
monkey intraday	–2.3	5.6	4.4	2.6	2.4	3.7	–2.9	3.1
monkey interday	–5.6	6.9	–1.2	5.9	0.6	4.7	–2.3	4.7
human intraday	7.3	11	–6.2	6.9	–7.0	3.5	0.0	3.8

Intraday and interday accuracy (% error) and precision (% CV) of plasma and urine quality control (QC) samples. LQC, low QC; MQC, middle QC; HQC, high QC.

Overall, the accuracy and precision parameters would have passed assay validation criteria if samples were quantified without an internal standard. However, the variability in the quantification of extracted samples nearly doubled in the absence of an internal standard, and in long LC-MS/MS run batches spanning more than a day (>150 samples), the variability exceeded 15% across the run without internal standard normalization. In addition, occasional samples prepared from plasma of individual animals had approximately 50% error in accuracy when quantified without internal standard normalization. This poor accuracy was corrected to 6% via the normalization to the internal standard. This error is likely due to matrix effects on DA and THDA. Based on an analysis of calibration curves of DA and THDA in solvent and matrix, matrix effects were present for both compounds, and the matrix effect was not different between the compounds.

This LC-MS method was also cross-validated in pooled human plasma and urine. Although different background peaks were detected in human plasma and urine when compared to monkey plasma and urine, none of these interfered with the quantification of DA or THDA. The human plasma LLOQ was 0.31 ng/mL (1 nM) and the urine LLOQ was 7.8 ng/mL (25 nM). Plasma and urine calibration standards satisfied the predefined acceptance criteria with 89 and 100% nonzero calibration standards of plasma and urine, respectively, falling within 15% of nominal concentration. The accuracy and precision data are shown in Table .

Measuring DA in Monkey Plasma and Urine after Oral Exposure to DA

Plasma and urine DA concentrations were determined in samples collected over a 24 h period from healthy female cynomolgus monkeys (n = 10) following single oral doses of DA at the human TDI of 0.075 mg/kg. Plasma and urine concentrations from these monkeys are shown in Figure . A total of 77 plasma samples were collected and analyzed to measure DA concentrations. Sixty three samples (82%) contained DA above the LLOQ at 0.31 ng/mL (1 nM). The range of determined plasma concentrations was between 0.31 and 11 ng/mL (1 and 35 nM). In the remaining 18% of samples (n = 14), DA was detected, indicating that the concentration of DA in these samples was between 0.16 ng/mL (0.5 nM, LOD) and 0.31 ng/mL (1 nM, LLOQ). A total of 60 urine samples were collected and analyzed to measure DA concentrations and DA excretion into urine. All 60 samples contained DA above the LLOQ of 7.8 ng/mL (25 nM). The range of determined urine concentrations was between 9.4 and 745 ng/mL (30 and 2400 nM). Pharmacokinetic parameters were estimated from nine monkeys (n = 9) because the AUC could not be reliably estimated from one monkey, as more than 50% of the plasma concentrations were below LLOQ. A geometric mean peak plasma concentration (Cmax) of 2.3 ng/mL (95% confidence interval (CI): 1.2–4.4) was observed between 1 and 12 h after the dose (tmax). The geometric mean area under the plasma concentration versus time curve (AUC0–24) was 19 ng·h/mL (95% CI: 13–28), the geometric mean amount of DA excreted unchanged in urine (Ae,0–24) was 5.2 μg (95% CI: 3.6–7.4), and the geometric mean renal clearance (CLr) was 4.5 mL/min (95% CI: 2.5–8.2). The potential presence of DA or any potential interference in the assay was also measured in plasma samples collected over a 24 h period from control healthy female cynomolgus monkeys (n = 10) following a single dose of 5% sucrose vehicle solution. A total of 20 plasma samples were analyzed, and none of the samples contained detectable DA.

Figure 6

DA pharmacokinetics following a single 0.075 mg/kg dose of DA to a group of 10 healthy adult female cynomolgus monkeys. (A) Geometric mean plasma concentration–time curve (black line) with individual data above the LLOQ (82%, n = 63) shown as symbols. Mean maximum plasma concentration (Cmax) and area under the plasma concentration versus time curve (AUC) from 0 to 24 h are reported on the graph; (B) Individual urine concentration data measured above the LLOQ (100%, n = 60); and (C) individual cumulative % dose excreted unchanged in urine. Geometric mean amount of unchanged DA excreted (Ae) from 0 to 24 h and renal clearance (CLr) are reported as inset in the graph.

Discussion

Recent reports of DA consumption exceeding the proposed TDIs in recreational shellfish harvesters,[16] and the health risks associated with chronic exposure to subacute levels of DA[17] highlight the need to investigate the exposure–effect relationship of DA. A sensitive method to detect low DA concentrations in plasma and urine, such as the one described here, would provide quantitative biomarker data to define this exposure–effect relationship. On the basis of DA pharmacokinetics observed in monkeys and PBPK modeling, a Cmax of 2 ng/mL (6.5 nM) was predicted in humans following a single oral DA consumption at the TDI (0.075 mg/kg).[23] This prediction set the required sensitivity threshold of analytical methods for DA quantification at about 5–10-fold below the predicted Cmax (i.e., 0.2–0.4 ng/mL) to detect and quantify DA in samples from people with DA exposure near the proposed TDIs.

Despite many analytical methods having been developed for DA, only a few of them have been developed to determine DA in biological fluids, such as plasma and urine.[5,6,25−28] Current published LC-MS/MS methods have largely been used to confirm the presence of DA in samples that were positive by ELISA detection,[6,18] although the analytical sensitivity of most reported LC-MS/MS methods is lower than ELISA (Table ). ELISA methods have been validated for DA quantification in serum and urine,[5,27] whereas no validation data exist for LC-MS/MS methods used for DA quantification in plasma and only a single LC-MS/MS assay has been validated for measuring DA in urine samples.[28] The method reported here is the first LC-MS/MS method for DA analysis that has been validated for measurement of DA in plasma samples. Importantly, the assay was validated to detect concentrations at 0.16 ng/mL (0.5 nM, LOD) and to reproducibly quantify concentrations at or above 0.3 ng/mL (1 nM, LLOQ) of DA in plasma, demonstrating sufficient sensitivity to assess subchronic exposures to DA. As shown by our study in cynomolgus monkeys, the sensitivity of this validated method was sufficient to quantify DA in plasma following a single oral dose at the proposed TDI. Similarly, our method was validated to detect and quantify DA at or above 7.8 ng/mL (25 nM, LLOQ) in urine. All urine samples collected from the monkeys dosed with DA contained quantifiable DA concentrations. Yet, the LLOQ for urine samples can be further reduced with less dilution to increase the sensitivity of the assay if needed.

One of the major challenges when developing analytical methods to detect DA in biological samples is variable recovery following complex extraction or variable MS response due to matrix effects. The recovery of DA from different marine animal samples following SPE ranged between 20 and 100% resulting in challenging quantitation of DA.[6] A few published methods used KA and ENK as internal standards in seawater and mussel analyses to control for assay variability,[7,9,30] but the distinct chromatographic and mass spectrometric properties of KA and ENK may decrease their usefulness as internal standards for DA. Another recently published method introduced the use of the labeled dansyl chloride derivative of DA as an internal standard,[29] but the sample and internal standard are derivatized separately. To address these issues with selection of an internal standard, we synthesized and validated THDA as an internal standard. THDA is structurally similar to DA and shares similar chromatographic and mass spectrometric properties. We demonstrated the appropriate chromatographic and mass spectrometric performance of THDA to control for assay variability. The use of THDA improved the reproducibility of the method as demonstrated by acceptable accuracy and precision when the internal standard was used, but inadequate accuracy and precision in the absence of an internal standard during long LC-MS/MS runs. Importantly, although this method uses a simple single methanol extraction step, THDA can easily be applied to other LC-MS/MS and sample preparation methods that quantify DA in more complex matrices.

ELISA-based methods to determine DA concentrations in samples from marine mammals are convenient to use with minimal lab equipment requirements. However, discrepancies between results from ELISA and LC-MS/MS methods have been reported.[27] One of the possible reasons for these discrepancies is that marine mammals chronically exposed to DA may produce antibodies against DA, which may interfere with the ELISA assays, and hence ELISA assays would underestimate the DA content. Another possible reason is that complex biological matrices, like plasma or serum, may contain interfering components that bind to the anti-DA antibody used in the ELISA assay, which would overestimate the DA content. The validated HPLC-MS/MS method described here provides a key advantage over the ELISA by obviating the above issues and allowing direct detection of DA concentrations and exposure in chronically exposed marine animals. However, the assay described will require the investment or access to a state-of-the-art LC-MS/MS instrumentation.

In summary, the developed HPLC-MS/MS method described here has sufficient sensitivity to determine DA concentrations in monkeys following low-level (∼TDI) exposures. It also meets the reproducibility criteria for method validation providing improved quality of DA quantification in biological matrices. This validated HPLC-MS/MS method and the new internal standard THDA can be widely applicable to research studies and sentinel programs related to the health risks and exposure levels of DA.

Materials and Methods

Chemicals and Reagents

A certified calibration standard of DA (332 μM) in acetonitrile/water (1:19, v/v) was purchased from National Research Council Canada (Halifax, NS, Canada). HPLC solvents, including Optima LC/MS grade water, acetonitrile, methanol, and formic acid, were purchased from Fisher Scientific (Pittsburgh, PA). DA, dihydrokainic acid (DHKA), acetaldehyde, sodium nitroprusside, sodium carbonate, and sodium bicarbonate were purchased from Millipore Sigma (St. Louis, MO). Frozen treatment-naïve monkey plasma and urine were obtained from the Washington National Primate Research Center (WaNPRC) at the University of Washington (Seattle, WA). Frozen human plasma and urine were obtained from banked tissue at the University of Washington (Seattle, WA).

The internal standard THDA was synthesized, as previously described.[33] Briefly, 1.5 mg of DA powder was dissolved in 5 mL of aqueous methanol (90%, v/v) prior to the addition of 10 mg of platinum dioxide. The mixture was placed in a round-bottom flask evacuated on a Schlenk line flushing repeatedly with argon. Hydrogen, generated by drop-wise addition of aqueous 10% sulfuric acid to an aqueous solution of sodium borohydride (37 mg, 1.0 mmol, in 1.5 mL of water), was introduced to the DA-containing flask via a hypodermic needle. After 5 h, the reaction mixture was flushed with argon a few times, filtered through Celite, followed by two methanol washes. The product is a mixture of two diastereomers because the reduction process introduces a new chiral center. This stock solution of THDA was stored at 4 °C to be used as an internal standard.

The concentration of THDA was determined by a colorimetric assay using a mixture of acetaldehyde and sodium nitroprusside, commonly referred to as Simon’s reagent.[34,35] This assay is used to detect certain secondary amines based on their reaction with acetaldehyde and sodium nitroprusside.[35] On the basis of their structural similarity, DHKA was chosen as the reference standard to quantify THDA. The colorimetric response at 590 nm was linear between 100 and 500 μM of DHKA under the assay conditions. For the assay, duplicate 100 μL samples of DHKA in water or THDA in 10% aqueous methanol were added to 100 μL of 50 mM sodium carbonate–bicarbonate buffer at pH 9.8 in a clear-bottom 96-well plate. To initiate the reactions, 50 μL of Simon’s reagent containing 10% w/w acetaldehyde and 1% w/w sodium nitroprusside in water were added to each well. Absorbance at 590 nm was determined on a Tecan Infinite 200 PRO spectrophotometer (San Jose, CA) between 8 and 20 min following spiking of the reagent. The experiment was repeated on three different days. The purity of the product was assessed by HPLC-MS/MS to be greater than 98%. Ultraviolet absorbance was determined on Olis Modernized Aminco DW-2 spectrophotometer (Bogart, GA), and the UV spectra are shown in Figure .

HPLC-MS/MS

Aqueous stock solutions of the certified DA standard and THDA stock at 3.1 ng/mL (10 nM) and 0.63 ng/mL (2 nM), respectively, were used to characterize their fragmentation patterns and to optimize their MS/MS detection parameters on a Sciex 6500 QTRAP system (Foster City, CA) by positive ion electrospray ionization. The optimized MS parameters for both DA and THDA were ion source temperature at 550, curtain gas at 50, nebulizing gas (GS1) at 70, drying gas (GS2) at 70, collision activated dissociation gas at −2, ion-spray voltage at 5500, declustering potential at 15, and entrance potential at 10. Collision energy (CE) for MS2 analysis was set as 24, and excitation energy (AF2) for MS3 EPI scans was set at 0.05. For MS2 EPI scans, Q1 was set to filter for the precursor ions [M + H]+m/z 312 for DA and m/z 316 for THDA. For MS3 EPI scans, the QTRAP function was used, and Q3 filter was set up to filter for m/z 312 > 294 (CE 15) and m/z 312 > 266 (CE 20) for DA and m/z 316 > 298 (CE 15) and m/z 316 > 270 (CE 20) for THDA.

The LC-MS method was developed and validated using a Shimadzu UFLC XR DGU-20A5 (Kyoto, Japan) equipped with a Phenomenex Synergi Hydro-RP 100 Å (2.5 μm, 50 × 2 mm2) LC column and a guard cartridge (2 × 2.1 mm2, sub 2 μm) (Torrance, CA) coupled to a Sciex 6500 QTRAP system (Foster City, CA). A 9 min gradient elution was employed using (A) water with 0.1% formic acid and (B) 95% acetonitrile with 0.1% formic acid at a flow rate of 0.5 mL/min. Sample injection volume was 10 μL. The gradient was initiated at 5% B for 1 min, increased to 100% B over the next 3 min, kept at 100% B for 30 s before decreasing to 5% B over 30 s, and re-equilibrating at 5% B for another 4 min. Three product ions for DA (m/z 312 > 266, m/z 312 > 248, and m/z 312 > 220) and for THDA (m/z 316 > 270, m/z 316 > 252, and m/z 316 > 224) were monitored to confirm the identity of the analytes. MRM conditions were optimized to collision energy (CE) at 24 eV, and collision cell exit potential at 10 eV.

The on-column LOD for DA was determined by injecting 10 μL of DA standard in water at concentrations ranging from 0.031 to 0.31 ng/mL (0.1–1 nM). The LOD was defined as the lowest amount of DA injected producing a S/N > 3 for the DA transition m/z 312 > 266. The on-column LLOQ was defined as the lowest amount of injected DA producing a S/N > 3 for all three DA mass transitions m/z 312 > 266, 248, 220. Calibration standards covering 4 orders of magnitude at concentrations ranging from 0.031 to 310 ng/mL (0.1–1000 nM) of DA were prepared in water to determine the linearity of response (peak area of DA mass transition m/z 312 > 266).

DA Extraction from Plasma and Urine Samples

DA was extracted from plasma by mixing 60 μL of plasma standards or samples with 120 μL LC/MS grade methanol containing 0.63 ng/mL (2 nM) THDA and vortexed for 30 s to precipitate proteins. The precipitated plasma standards and samples were centrifuged at 16 100g for 60 min, and the supernatant was removed for HPLC-MS/MS analysis. DA was extracted from urine by mixing 10 μL of urine standards or samples with 490 μL of LC/MS grade water containing 0.63 ng/mL (2 nM) THDA and subsequently 500 μL of methanol. The urine was vortexed for 30 s to precipitate proteins and centrifuged at 16 100g for 15 min. The supernatant was removed for HPLC-MS/MS analysis.

The method was validated according to the FDA Guidance for Industry Bioanalytical Method Validation[36] using pooled cynomolgus monkey plasma and urine. The plasma calibration standards were prepared by spiking plasma with the DA certified standard and serially diluting with plasma to nine concentrations ranging between 0.16 and 16 ng/mL (0.5–50 nM). The urine calibration standards were prepared by spiking urine with DA certified standard and serially diluting with urine to nine concentrations ranging between 7.8 and 1000 ng/mL (25–3200 nM). The calibration standards were processed and analyzed by HPLC-MS/MS, as described above. The MRM chromatograms were integrated using AB Sciex MultiQuant software version 2.1.1 (Foster City, CA). DA was quantified using the peak area ratio of m/z 312 > 266 (DA) and m/z 316 > 252 (THDA), and a weighted (1/y2) calibration curve was fitted linearly to the data. The 1/y2 weighing was used as the simplest weighing scheme providing sufficient accuracy over the concentration range used. The 1/y2 weighing provided better accuracy over the concentration range studied in comparison to uniform weighing. The acceptance criteria for each calibration curve were defined as greater than 75% of all nonzero concentrations determined with less than 15% error from the nominal concentrations, except at the LLOQ, which was accepted with less than 20% error.

Plasma QCs were prepared by spiking blank plasma with DA certified standard to 0.31 ng/mL (LLOQ), 0.93 ng/mL (LQC), 7.8 ng/mL (MQC), 12 ng/mL (HQC) (1, 3, 25, 40 nM). Urine QCs were prepared by spiking blank urine with DA certified standard to 7.8 ng/mL (LLOQ), 23 ng/mL (LQC), 500 ng/mL (MQC), 780 ng/mL (HQC) (25, 75, 1600, 2500 nM). Blank plasma and urine from six treatment-naïve cynomolgus monkeys with and without internal standard were analyzed for potential interference. The accuracy and precision of the method were determined on three different days. Interday variability was calculated with at least 12 replicates of each QC analyzed on 3 different days. Intraday variability was determined with five replicates analyzed on the same day. Calibration standards were analyzed in duplicates along with blanks and replicates of QCs (LLOQ, LQC, MQC, HQC) in each run. The LLOQ was defined as the lowest concentration in plasma or urine with S/N > 3 for DA transition m/z 312 > 266 that could be repeatedly determined with less than 20% error and within 20% CV. The % error and % CV were calculated according to eqs and 2The LOD was defined as the lowest concentration in plasma and urine with S/N > 3 for the DA transition m/z 312 > 266, but did not meet the LLOQ acceptance criteria for reproducibility. QCs with less than 15% error from the nominal concentration or less than 20% error at LLOQ were accepted.

Stability of extracted samples was determined by repeated injections of QC samples (LQC, MQC, HQC) stored either in the autosampler at 4 °C for up to 5 days or on bench top at room temperature for a day. Acceptance criterion of extracted sample stability was defined as less than 15% CV from repeated injections. Long-term stability was determined by repeated analysis of spiked plasma controls at 0.93, 1.9, 3.7 ng/mL (3, 6, 12 nM) and spiked urine controls at 16, 160, 620 ng/mL (50, 500, 2000 nM) stored at −20 °C and that were subjected to at least five freeze–thaw cycles. Acceptance criterion of long-term stability was defined as less than 15% CV of measured concentration from repeated analysis.

Recovery of DA was determined at 0.31, 0.93, 7.8, 12 ng/mL (1, 3, 25, 40 nM) of DA in plasma and 7.8, 23, 500, 780 ng/mL (25, 75, 1600, 2500 nM) of DA in urine in triplicates. The recovery was calculated by comparing the DA/THDA peak area ratio of samples spiked with DA prior to methanol extraction to DA/THDA peak area ratio in samples extracted with methanol and subsequently spiked with the DA at matching amount. The THDA (0.32 ng/mL) was added to all samples following extraction. The percent recovery was reported as the percentage of peak area ratio of recovery samples divided by the mean peak area ratio of control samples.

Cross-validation using human plasma and urine was performed in a single accuracy and precision validation run. Plasma and urine calibration standards and QCs were prepared as described for monkey standards and QCs. Plasma and urine were pooled from banked tissue from six individuals. Potential matrix interferences were assessed by analyzing blank plasma and urine. Duplicates of calibration standards and five replicates of LQC, MQC, HQC, and four replicates of LLOQ were included in the cross-validation to assess variability.

Animal Study Samples and Application of the Method

Monkey plasma and urine samples from a DA toxicokinetic study were analyzed to evaluate the applicability of the assay to determine DA exposure after a single oral dose at the proposed TDI of 0.075 mg/kg. All animal procedures followed the guidelines set by the Animal Welfare Act and the Guide for Care and Use of Laboratory Animals of National Research Council. The research protocol was approved by the University of Washington Institutional Animal Care and Use Committee. Fresh blood and urine samples were obtained at 1, 2, 4, 6, 8, 10, 12, and 24 h from 20 healthy adult female cynomolgus monkeys following a single oral dose of 0.075 mg/kg DA in 5% sucrose in water (n = 10) or a single oral dose of 5% sucrose in water (n = 10). Training protocols using positive reinforcement were implemented to collect blood without sedation. Fresh blood was collected from the peripheral vein using sodium heparin-coated collection tubes and centrifuged at 3000g for 15 min within 1 h of collection to isolate plasma from blood cells. The plasma samples were stored at −20 °C until analysis. Urine was collected from the cage pans equipped with metabolic urine collection trays at the same time fresh blood was collected. The total volume of urine in the pan was measured, and a 2 mL sample was collected before discarding the rest of the urine. The urine samples were stored at −20 °C until analysis.

Pharmacokinetic Analysis

Pharmacokinetic parameters, including area under the plasma concentration–time curve (AUC) and maximum plasma concentration (Cmax), were estimated by noncompartmental analysis using Phoenix WinNonlin (St. Louis, MO). The plasma concentration was assigned to be 0.2 ng/mL when DA was detected but was below the LLOQ for AUC calculation. The cumulative amount of unchanged DA excreted in urine (Ae) was calculated from the determined concentration and the volume of urine at all collection time intervals using eq . Renal clearance (CLr) was estimated from the cumulative amount excreted unchanged in urine and the plasma AUC using eq