Quantification of Nerve Agent Biomarkers in Human Serum and Urine

Article pubs.acs.org/ac

Quantification of Nerve Agent Biomarkers in Human Serum and Urine Bent Tore Røen,*,†,‡ Stig Rune Sellevåg,† and Elsa Lundanes‡ †

Norwegian Defence Research Establishment, P.O. Box 25, NO-2027 Kjeller, Norway Department of Chemistry, University of Oslo, P.O. Box 1033, Blindern, NO-0315 Oslo, Norway

‡

ABSTRACT: A novel method for rapid and sensitive quantification of the nerve agent metabolites ethyl, isopropyl, isobutyl, cyclohexyl, and pinacolyl methylphosphonic acid has been established by combining salting-out assisted liquid−liquid extraction (SALLE) and online solid phase extraction−liquid chromatography−tandem mass spectrometry (SPE-LC-MS/MS). The procedure allows confirmation of nerve agent exposure within 30 min from receiving a sample, with very low detection limits for the biomarkers of 0.04− 0.12 ng/mL. Sample preparation by SALLE was performed in less than 10 min, with a common procedure for both serum and urine. Analyte recoveries of 70−100% were obtained using tetrahydrofuran as extraction solvent and Na2SO4 to achieve phase separation. After SALLE, selective analyte retention was obtained on a ZrO2 column by Lewis acid−base and hydrophilic interactions with acetonitrile/1% CH3COOH (82/18) as the loading mobile phase. The phosphonic acids were backflush-desorbed onto a polymeric zwitterionic column at pH 9.8 and separated by hydrophilic interaction liquid chromatography. The method was linear (R2 ≥ 0.995) from the limits of quantification to 50 ng/mL, and the within- and between-assay repeatability at 20 ng/mL were below 5% and 10% relative standard deviation, respectively.

N

AMPAs in biofluids employ solid phase extraction (SPE) or liquid−liquid extraction (LLE) for analyte enrichment and sample cleanup. The ionic character of the AMPAs (pKa = 2.2− 2.3) makes them well suited for anion exchange SPE.8,9 Reversed-phase interaction columns6,7 and a hydrophilicliphophilic balanced polymer10 have been employed after acidifying the samples to bring the analytes on nonionic form. Moreover, the AMPAs have been isolated by hydrophilic interactions on silica sorbents after diluting the clinical samples in an excess of acetonitrile (ACN).11−13 Noort et al.5 and Barr et al.14 have performed LLE for extraction of the AMPAs into hydrophobic organic solvents from acidified serum and urine, respectively. Miki et al. performed derivatization under liquid− liquid−solid phase-transfer conditions followed by GC-MS/MS for determination of the AMPAs in biofluids.15 Recently, Lin et al. performed solid phase supported derivatization followed by LLE and GC-MS/MS determination of the derivatives.16 Also, salting-out assisted LLE (SALLE) has been utilized for extraction of the AMPAs into polar solvents such as isopropanol17 and ACN,18 followed by change of solvent and derivatization prior to GC-MS determination. Most of the above-mentioned techniques include multistage sample preparation to achieve sensitivity at low to sub-ppb levels, like deproteinization, evaporation and change of solvent, and derivatization of the AMPAs if GC is performed. This makes

erve agents are organophosphorus compounds extremely toxic to humans by disrupting the mechanism for transfer of nerve messages to organs. Following exposure, the major part of the nerve agents is enzymatically hydrolyzed to their corresponding alkyl methylphosphonic acids (AMPAs); see Figure 1.1 These metabolites will slowly undergo further hydrolysis by loss of their alkyl group to methylphosphonic acid (MPA).2 Although MPA may serve as a general, less-specific biomarker for possible nerve agent exposure, the primary hydrolysis products are agent specific. After the terrorist attacks with sarin in Matsumoto (1994) and Tokyo (1995), isopropyl methylphosphonic acid was found in urine from surviving victims at sub-ppm levels during the first 24 h.3,4 Lower levels (2−135 ng/mL) were found in serum collected within 1.5−2.5 h after exposure.5 Elimination of the metabolites from blood is known to occur within 1−3 days,6,7 whereas urinary metabolites may be found at low to sub-ppb levels up to 2 weeks after an exposure.1 Consequently, rapid determination is important to guide medical countermeasures in emergency cases, and method sensitivity at sub-ppb levels may be needed if retrospective determination is performed for forensic purposes. Liquid and gas chromatography in combination with mass spectrometry (LC-MS and GC-MS) are the main analytical techniques for determination of the AMPAs in serum and urine.1,2 For GC-MS determination, a derivatization step is required to transform the AMPAs to their respective phosphonic esters. Due to the complexity of the sample matrices and expected low level of target compounds, sample cleanup and analyte enrichment may be needed prior to instrumental analysis. Most methods for determination of the © XXXX American Chemical Society

Received: September 11, 2014 Accepted: November 5, 2014

A

dx.doi.org/10.1021/ac503408x | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

Figure 1. Structures of selected nerve agents and their primary hydrolysis products. The pKa and log Kow values were calculated using Advanced Chemistry Development software V11.02 (ACD/Laboratories, 1994−2012).

98%) were purchased from Sigma-Aldrich Chemie GmbH, Steinheim, Germany. Isopropyl methylphosphonic acid (iPMPA), isobutyl methylphosphonic acid (iBMPA, 1000 μg/ mL in methanol), and cyclohexyl methylphosphonic acid (CMPA, 1000 μg/mL in methanol) were delivered by Cerilliant Corp., Round Rock, TX, USA. The isotopic labeled standards ethyl-D5 methylphosphonic acid (100 μg/mL in methanol), isopropyl-D7 methylphosphonic acid (1000 μg/mL in methanol), and pinacolyl-13C6 methylphosphonic acid (100 μg/mL in methanol) were also delivered by Cerilliant Corp. Stock solutions of EMPA, iPMPA, and PMPA were prepared separately at 0.5 mg/mL by diluting 25 mg of the neat agents in 50 mL ACN. All AMPAs were diluted separately to 10 μg/mL in ACN, and joint spiking solutions were prepared in type I water (classified according to the American Society of Testing and Materials, D1193-91). The spiking solutions were prepared so that the dilution of serum and urine by addition was less than 5% (v/v). The isotopic labeled compounds were prepared in ACN in a joint solution at 5 μg/mL and were further diluted in type I water to an internal standard (IS) working solution of 175 ng/mL of each. All solutions were stored at 4 °C until use. Human serum and urine were kindly provided by three women and three men and stored at 4 °C for maximum 14 days prior to use. Aliquots were allowed to reach room temperature before being spiked with the AMPAs. After spiking, the samples were thoroughly mixed on a whirl mixer and stored at room temperature for at least 1 h prior to extraction. Salting-Out Assisted Liquid−Liquid Extraction. Aliquots of 150 mg Na2SO4 were weighed into 1.5 mL polypropylene microcentrifuge tubes. Spiked serum and urine samples of 300 μL were added to the tubes, followed by 40 μL of the 175 ng/mL IS solution and 16 μL 30% HCl (final concentrations of 20 ng/mL IS and 0.4 M HCl). The samples were mixed for a few seconds on a whirl mixer, and aliquots of 300 μL tetrahydrofuran (THF) were added. After whirl mixing for 30 s, the samples were centrifuged at 3000g for 3 min. Finally, 160 μL of the THF phase was transferred to autosampler vials with 300 μL fused insert, 40 μL of 1% CH3COOH was added, and the vials were whirl mixed. Instrumental Configuration. The online SPE-LC-MS/MS analyses were performed on an Ultimate 3000 RS LC (Dionex Corp., Idstein, Germany) and a MicroTof-Q III mass spectrometer with IonBooster ESI source (Bruker Daltonics, Bremen, Germany). A schematic diagram for sample loading and chromatographic separation is shown in Figure 2, together with a description of the solvents delivered by the LC-pumps. The column switching system was located inside an FLM-3100

the overall procedures quite laborious and time-consuming. Subramaniam et al. reported a more rapid procedure using direct derivatization of the AMPAs in urine with a fluorinated phenyldiazomethane reagent followed by GC-MS/MS determination.19 Unfortunately, the diazo-reagent is not commercially available and should be handled with care due to its carcinogenic and energetic properties. The aim of the present study was to establish a rapid, sensitive, and reliable method for determination of the AMPAs in serum and urine by the use of automated SPE in combination with LC-MS. Combining high sensitivity with less laborious methods may be accomplished by online SPE techniques.20−23 The complexity and high salt content of serum and urine, however, make it challenging to isolate ionic compounds like the AMPAs in online SPE procedures. In contrast to single use offline SPE cartridges, the bed volume of an online SPE column must be kept as small as possible to minimize void volumes in the system, and the column must be reconditioned for repeated sample injections. We have therefore developed a novel method using SALLE for sample cleanup followed by online SPE-LC-MS/MS of the organic extract. In SALLE, water miscible organic solvents are used for extraction, and phase separation is achieved by saturating the mixture with an appropriate salt. The technique is simple, rapid, and well suited for extraction of polar compounds from aqueous matrices like biofluids.24,25 After SALLE, selective enrichment of the AMPAs was achieved by online SPE on a zirconium dioxide (ZrO2) cartridge. ZrO2 exhibits Lewis acid properties and is able to undergo both ion- and ligand-exchange interactions with Lewis bases,26,27 as well as providing hydrophilic interactions.28,29 Enrichment of the AMPAs on ZrO2 is usually achieved by Lewis acid−base interactions;30−32 the retention mechanism is described in detail in a previous study.21 With the AMPAs extracted into a water miscible organic solvent, hydrophilic interactions were utilized for preconcentration of the analytes as well. Analyte separation was performed by hydrophilic interaction liquid chromatography (HILIC) and electrospray ionization (ESI)-high resolution tandem MS. The developed method was employed for determination of five AMPAs (Figure 1) at sub-ppb levels in human serum and urine. To our knowledge, the successful combination of SALLE and online SPE-LC-MS/MS is presented for the first time.

■

EXPERIMENTAL SECTION Chemicals and Solutions. Pinacolyl methylphosphonic acid (PMPA, 97%) and ethyl methylphosphonic acid (EMPA, B



Article

40 μL/min to save solvents. At the same time, the prefilter was backflushed to waste with P1 for removal of particles at the filter inlet after the model of Svendsen et al.33 The mobile phase composition of P1 was changed to 80% H2O during prefilter backflush to save organic solvent. Finally, the system was equilibrated at initial conditions (18.5−25 min). The high-resolution MicroTof-Q III (R > 18 000 at m/z 922) was operated in MS/MS mode with the settings described in Table 2. Compressed N2 (99.9999%) was used as collision gas, Table 2. Optimized MS/MS Acquisition Parametersa retention time window (min) 9.55−10.30 10.30−11.10

Figure 2. Diagram of the online SPE-LC-MS/MS setup (see text and Table 1 for details), together with solvents and flow delivered by the three LC pumps (% in v/v).

11.10−11.85 11.85−12.55

flow manager supported with two 10-ports, two-position micro switching valves operated at 35 °C. The loading flow (P1) was delivered from a DGP-3600M dual gradient pump via a WPS3000 autosampler with variable volume split-loop injection and a 100 μL sample loop. The LC flow was delivered from channel 2 of the DGP-3600M pump (P2), and the SPE washing solution was delivered by an AXP-MS pump from Dionex Corp. (P3). Preconcentration was performed on a ZrO2 column (2.0 mm × 10 mm, 3 μm) from ZirChrom Separations, Inc., Anoka, MN, USA. Separation was achieved with a ZIC-pHILIC column (2.1 mm × 150 mm, 5 μm) from Merck KGaA, Darmstadt, Germany. The stainless steel prefilter (0.2 μm) was from Thermo Fisher Scientific Inc., Bellefonte, PA, USA. The mobile phase compositions and valve settings during the SPE-LC-MS/MS procedure are described in Table 1. Organic

P1

VR

%B

%C

%B

%C

0 7 7.5 8 13 13.5 14 18 18.5 25

1

1 2

9 9 80

9 9 0

10.5

7.5

2

1

10.5 32.5

7.5 7.5

32.5 10.5 10.5

7.5 7.5 7.5

80 9

0 9

1 1

1

9

9

PMPA PMPA-13C6 CMPA iBMPA iPMPA iPMPA-D7 EMPA EMPA-D5

179.1 185.1 177.1 155.1 137.0 144.0 123.0 128.0

14 14 16 14 12 12 12 12

and the collision RF peak-to-peak voltage was 120 V. Precursor ions of [M − H]− were selected, and complete fragmentation ion spectra were acquired in the TOF mass analyzer. The [M − H − alkyl]− product ion of m/z 94.989 was chosen for confirmation of the presence of the analytes when extracted at a mass accuracy of ±5 mDa. Because of the high mass selectivity in extracting the major product ion, a second confirmation ion was not considered necessary. The quantitative calculations performed in the method validation (described below), were based on the obtained peak areas of the m/z 94.989 ± 5 mDa ions. Method Optimization. Different combinations of organic solvent and salts for extraction of the AMPAs were investigated with EMPA, iPMPA, and PMPA added to type I water at 1 μg/ mL (n = 6). The organic extracts were diluted 1:10 with the mobile phase, added iPMPA-D7 for IS correction, and injected (1 μL) onto the ZIC-pHILIC column. The efficiency of the different extractions was calculated by comparing the obtained peak areas of the m/z [M − H]− ions with those where the AMPAs were injected at concentrations corresponding to 100% extraction yield (n = 6). For investigation of ZrO2 and TiO2 (2.0 mm × 10 mm, 5 μm from ZirChrom) as SPE sorbents in HILIC mode, the columns were mounted in the setup as described for the switching valve to the right in Figure 2. The “Waste” and “LC-MS” outlets were connected to the MS during sample loading and backflush desorption, respectively, via the left switching valve. In this way, both potential breakthrough of the analytes during sample loading and the desorption rate could be measured. Investigation of the SPE columns and optimization of the online SPE-LC-MS/MS procedure were performed with the AMPAs added to type I water and to pooled serum and urine at 10−20 ng/mL. Method Validation. Method validation was performed with the AMPAs added to pooled serum and urine that were collected from the six volunteers. Determination of the limits of detection (LODs) was performed by adding the analytes to serum and urine at 0.05, 0.10, and 0.20 ng/mL (n = 3). The

P2

VL

collision energy (eV)

The [M − H − alkyl]− product ion of m/z 94.989 ± 5 mDa was used for conformation for all analytes.

mobile phase composition valve position

precursor ion m/z [M − H]− ± 3 Da

a

Table 1. Time Settings for Valve Positions and Mobile Phase Compositions during the Online SPE-LC-MS/MS Procedure

time (min)

analyte

extracts (50 μL) of serum and urine were injected onto the ZrO2 column followed by 7 min washing with P1. The analytes were backflush-desorbed onto the separation column for 7−8 min with P2, and gradient separation was subsequently performed (82−60% ACN for 8−13 min). During LC-MS/ MS, the ZrO2 column was washed at pH 13 with P3 by switching the left valve to position 2. When not performing column washing, the flow delivered by P3 was programmed to C



Article

LODs were calculated from the regression lines of the measured peak heights of the analyzed samples. The linearity was investigated at six concentration levels: 0.3 (0.15 for iPMPA in serum), 3, 10, 20, 35, and 50 ng/mL. The method repeatability was investigated at the limits of quantification (LOQs) and at 20 ng/mL (10 ng/mL for iPMPA in serum). Spiked serum and urine were each divided into six subsamples which were analyzed subsequently the same day (within-assay), and daily spiked samples were analyzed for six consecutive days (between-assay). Recoveries from the SALLE procedure were investigated by comparing the obtained peak areas with those where the analytes were added to the vial after the extraction procedure (n = 6).

Figure 3. Extraction yields of EMPA, iPMPA, and PMPA from SALLE using two different extraction solvents (ACN and THF) and two additives for phase separation (NaCl and Na2SO4), presented as mean values ± SD (n = 6).

■

RESULTS AND DISCUSSION Online SPE-LC-MS techniques offer the advantages of high method sensitivity and reduced sample preparation time.22,23 We have previously successfully employed the technique for determination of the AMPAs in environmental samples.20,21 Because of the high amounts of inorganic ions and organic acids in serum and urine, pretreatment of the samples was necessary prior to online SPE-LC-MS. In order to obtain a short analysis time, a simple and rapid sample preparation procedure was sought using SALLE. With the AMPAs extracted into a water-soluble organic solvent, hydrophilic interactions could be utilized for preconcentration and separation of the analytes. Hence, metal oxides like ZrO2 and TiO2 were obvious candidates as SPE sorbents due to high selectivity toward strong Lewis bases like the AMPAs21 as well as the ability to undergo hydrophilic interactions.28,34 The secondary hydrolysis product of the nerve agents, MPA, was not included in the present investigation because of poor compatibility with the method. In an earlier investigation we reported strong retention of MPA on the metal oxides resulting in very slow desorption from the SPE columns at LC-MS compatible conditions.21 Moreover, we have observed poor chromatographic performance for MPA on HILIC columns, probably because of its very high polarity. Salting-Out Assisted Extraction Conditions. To find the optimal conditions for extracting the AMPAs from aqueous samples by SALLE, the extraction yields of EMPA, iPMPA, and PMPA were investigated from type I water. Different combinations of two organic solvents (ACN and THF) and two salts (NaCl and Na2SO4) to achieve phase separation were investigated. ACN has shown the ability of phase separation from water with use of various salts.25 THF was included for investigation due to its proton acceptor properties, suitable for extracting proton donors such as the AMPAs. Extraction by SALLE is known to occur fast because the organic solvent is completely miscible with the aqueous phase.24 A short extraction time of 30 s was therefore used in the experiments, and the phase ratio was 1:1. Prior to extraction, the samples were acidified with 0.4 M HCl to protonate the AMPAs, making them less hydrophilic. Centrifugation was performed after extraction to enhance phase separation. The time needed for preparing one sample was approximately 10 min, but time per sample could be considerably reduced when treating several samples simultaneously (e.g., in 96-well plates). Figure 3 shows the extraction yields obtained with the different combinations of additives and extraction solvents. Higher extraction yields of EMPA and iPMPA were obtained when using Na2 SO4 compared to the use of NaCl. For the less polar PMPA, the effect of the salts was opposite with ACN as solvent. No

significant difference in extraction efficiency was observed between the solvents with Na2SO4 as additive, but higher amount of coextracted sulfate was seen with use of ACN. The sulfate ion is a strong Lewis base possibly interfering with analyte retention on metal oxides.35 Na2SO4 and THF were therefore chosen for the SALLE procedure, because Na2SO4 gave the overall highest extraction yield and THF the lowest amount of coextracted sulfate. Stationary Phase for Solid Phase Extraction. The retention of the AMPAs on ZrO2 and TiO2 was investigated in HILIC mode to utilize secondary hydrophilic interactions between the analytes and the sorbents in addition to Lewis acid−base interactions. Even though the AMPAs were solved in THF from SALLE, ACN was preferred as organic eluent because of its widespread use in LC-MS applications. Prior to be injected on the SPE columns, the THF samples were diluted with water to contain the same concentration of H2O that was used in the loading mobile phase. Complete retention was observed for all the AMPAs on both sorbents after eluting with 130 column volumes with 80% ACN in the mobile phase. When pure H2O was used for sample loading in previous experiments, analyte breakthrough occurred on ZrO2 after eluting with 70−110 column volumes.21 Hence, secondary hydrophilic interactions evidently occurred between the AMPAs and the ZrO2 stationary phase when loading with 80% ACN.28,34 No significant difference in the retention strength was observed between the ZrO2 and TiO2 sorbents in HILIC mode. Analyte desorption from TiO2 was slower compared to ZrO2, however, leading to broader chromatographic peaks after eluting through the separation column. The ZrO2 column was therefore chosen for further method development. Sample Loading. In addition to the AMPAs, the THF phases from SALLE contained many coextracted organic compounds naturally present in serum and urine, and minor amounts of inorganic ions. Hydrophilic interactions alone were not strong enough to retain polar compounds for prolonged time with 80% ACN in the loading mobile phase. Compounds not containing Lewis base functionality could therefore be washed out of the ZrO2 column. Carboxylic acids and inorganic anions like sulfate and phosphate exhibit Lewis base properties and were retained on the column together with the AMPAs. When adjusting the aqueous part of the loading mobile phase below pH 5 with CH3COOH, carboxylic acids like hippuric acid started to elute. The poorer retention of the carboxylic acids on ZrO2 at lower pH was probably caused by displacement with the CH3COO− ion, which is a moderately D



Article

strong Lewis base,26 and loss of their Lewis base properties due to protonation of the carboxylate groups. This effect of pH on retention of carboxylic acids was consistent with what Kučera et al. found for benzoic acids on ZrO2 at HILIC conditions.28 The AMPAs are stronger Lewis bases compared to the carboxylic acids, and no loss of analyte was observed until the pH was adjusted below 2.5 where protonation of the analytes take place. The pH was therefore set at 2.7 to effectively elute the carboxylic acids together with contaminants not exhibiting Lewis base properties, without eluting the AMPAs. The ACN/ H2O ratio used in the loading mobile phase did not influence the selectivity on ZrO2, as both the AMPAs and the carboxylic acids were able to undergo hydrophilic interactions. With 80% ACN and pH 2.7 (measured in the aqueous phase), the major part of the contaminants was eluted after 6 min (400 μL/min). A loading time of 7 min was therefore chosen, corresponding to approximately 120 column volumes. Analyte Desorption and SPE Column Washing. After sample loading, analyte desorption onto the separation column was performed in backflush mode. Rapid desorption was needed because refocusing on the separation column was poor for the least retentive of the AMPAs (PMPA). Retention of solutes on ZrO2 is known to decrease rapidly at basic conditions due to the competing hydroxide ion which is a very strong Lewis base.26 The aqueous part of the mobile phase was therefore adjusted to near pH 10 which was the operation limit for the separation column. Di- and trivalent organic acids and inorganic anions are known to be strongly retained on ZrO2 due to their ability to form chelates on the surface.35 This was observed for sulfate and phosphate as well as the di- and trivalent organic acids present in serum and urine, like succinic acid and citric acid. These compounds were not eluted during sample loading, and only minor amounts were desorbed onto the separation column at pH 10. High selectivity toward the AMPAs onto the separation column was then maintained, and the more strongly retained matrix components could be eluted from the SPE column in a second washing step while analyte separation was performed. Effective elution of these matrix components was obtained by washing at pH 13 using 0.25% (v/v) tetramethylammonium hydroxide and 20% ACN between injections. For this purpose a separate LC-pump was employed, compatible for use with high pH. Optimization of the Online SPE-LC-MS/MS Procedure. The final optimization of the automated SPE-LC-MS/MS procedure was performed with the setup as described in Figure 2. A polymeric HILIC column was used in the configuration, compatible for use with pH ≤ 10. To ensure rapid analyte desorption from the SPE column, the aqueous part of the mobile phase was adjusted to pH 9.8 with NH4OH. Higher amount of ammonium acetate in the mobile phase gave more rapid desorption as well, but was kept at 15 mM to minimize ion suppression. Decreasing the amount of ACN was another way to increase the desorption rate from ZrO2, but a high ACN concentration was needed to ensure refocusing of the analytes on the separation column. A start gradient of 82% ACN was therefore the best compromise between the two latter mechanisms counteracting each other. The injection volume should be set to obtain high analyte enrichment on the SPE column while buildup of contaminants is kept at a minimum. Possible breakthrough of analytes and matrix effects in ESI should also be considered. Even though high selectivity was achieved for the AMPAs, coeluting

components were not completely absent. Prior to injection, to 160 μL aliquots of the serum and urine extracts were added 40 μL of 1% CH3COOH to give them nearly identical concentration of water as the loading mobile phase. For urinary extracts, a linear increase in signal height and peak area was seen for the analytes at increasing injection volumes up to 50 μL, and with a flattening of the plotted signal heights at 100 μL injected (results not shown). For serum extracts, the signal height and peak area were linear for EMPA and iPMPA up to 100 μL injected, but nonlinear for PMPA above 20 μL injected. The nonlinear signal of PMPA was probably caused by enhanced ion suppression rather than analyte breakthrough as one dominating signal from one or more coeluting compounds was observed when operated in single TOF mode. The m/z 129.054 of the dominating signal was consistent with [M − H]− for ketoisocaproate and ketomethylvalerate, which are known to be present in blood plasma as intermediates from the metabolism of leucine and isoleucine, respectively.36 The reason for these compounds not being eluted during sample loading may be that the alfa-position of the keto group make them stronger Lewis bases compared to other monovalent carboxylic acids. To ensure high method robustness and accuracy, 50 μL was chosen as injection volume for both the serum and urine extracts. More than 350 extracts could be injected on the system with this sample volume before breakthrough of the AMPAs occurred during sample loading. Repeatable analyte recoveries were re-established immediately after replacement of the ZrO2 column. Extensive flushing of the used ZrO2 column with NaOH may re-establish the performance,37 but this was not investigated. No pressure build-up occurred on the SPE or separation column during method optimization and validation with the prefilter backflush system described in Figure 2. Memory effects were investigated by analyzing blank samples after injecting extracts of serum and urine containing the AMPAs at 20 ng/mL. Carry-over was observed at levels of 0.2−0.5% for five injections. Different washing procedures for the autosampler and SPE column were investigated without eliminating this memory effect. Hence, blank samples should always be analyzed in advance for determination of the AMPAs at trace levels in unknown samples. Figure 4 shows the extracted ion chromatograms (EICs) from SPE-LC-MS/MS determination of an extract from serum where the AMPAs were added at 20 ng/mL (10 ng/mL for iPMPA). The analytes were backflush-desorbed onto the separation column after 7 min, and the compounds of interest were eluted within 12.5 min after sample injection. With the rapid SALLE procedure prior to instrumental analysis, possible

Figure 4. EICs of [M − H-alkyl]− ± 5 mDa from online SPE-LC-MS/ MS determination of a SALLE extract of a serum sample with the AMPAs added at 20 ng/mL (10 ng/mL for iPMPA). E



Article

known to be more sensitive compared to the Q-TOF instrument when operated in tandem MS mode.38 A generally accepted approximation of the LOQs is 3 times the LODs. For simplicity, the lower quantification level in the present work was set to 0.3 ng/mL for all analytes except iPMPA in serum (0.15 ng/mL). Good linear relationship was observed for all analytes when investigated in the concentration range of LOQ-50 ng/mL. High method repeatability was proven at 20 ng/mL, and the variations at the estimated LOQs were within the generally accepted 20%. The recoveries from SALLE reflected the lipophilic properties of the analytes, ranging from 98−101% for PMPA and CMPA to 70−74% for the more polar EMPA. Interestingly, the recoveries using SALLE were higher from serum and urine compared to the extraction yields obtained from type I water (Figure 3). This effect may be due to the combination of different salts in the sample matrices, enforcing analyte partitioning toward the organic phase. The method accuracy and precision were investigated by adding the analytes at LOQ and at 20 ng/mL (10 ng/mL for iPMPA in serum) to the six serum and urine samples individually. Quantification of the analytes with isotopic labeled IS was performed on the basis of the calibration curves obtained in the linearity test. For iBMPA and CMPA, the obtained peak areas were compared with those where the same amount of analytes were added to pooled serum and urine (n = 4). Figure 6 shows the quantified values for each of the AMPAs in the six individual samples. When the analytes were added at 20 (and 10) ng/mL, only one out of 60 measurements was outside ±10% of the true value. In general, higher accuracy was obtained in determination of the analytes in serum compared to urine at the highest concentration level. Determination of CMPA and iBMPA in urine suffered from the lack of isotopic labeled standard, resulting in poorer precision compared to the other analytes. At the estimated LOQs, five out of the 60 measurements were outside ±20% of the true value. A systematic bias was observed for determination of iPMPA in serum at the LOQ, but the reason for this was not fully understood. Still, the results should be considered adequate for determination of the AMPAs at sub-ppb levels. Possible matrix effects were investigated by comparing the noncorrected peak areas of each analyte in the individual samples. The peak areas showed little variations from sample to sample when added to serum (3−4% RSD), proving that the method was not prone to measurable matrix effects for this type of sample. Higher variations were observed for the urine samples (5−30% RSD), probably due to the larger variations in the amount of matrix components. Lowest peak areas were measured in the extract of the most colored urine sample, and this sample also showed the highest background signal of matrix components when measured in single TOF MS mode. The two compounds with no labeled internal standard had the lowest variation in peak areas from sample to sample (CMPA, 5% RSD and iBMPA, 7% RSD) and eluted in a retention time window with low background. The variations in peak areas of the other compounds indicate that matrix effects may occur for these analytes. The use of labeled internal standard will correct for this in quantitative measurements, but the LOD can vary somewhat from sample to sample.

nerve agent exposure could be confirmed in less than 30 min from receiving a sample. The present technique also offers the advantage of one common sample preparation procedure and instrumental method for serum and urine. The tandem MS procedure was chosen because the signal-to-noise was somewhat higher compared to extracting the quasi-molecular ions ([M − H]− ± 5 mDa) in single TOF application. Still, the signal intensities using single TOF mode indicated sensitivity at sub-ppb levels. This suggests that generic screening of AMPAs originating from nerve agents other than those reported in the present study is possible down to at least 1 ng/mL. Method Validation. The developed method was validated with the AMPAs added to pooled serum and urine, and to the six individual samples for determination of method accuracy and precision. No significant background signal was measured at the retention time of the analytes when blank samples of individual serum and urine were subjected to SALLE and analyzed according to the described procedure. Linearity and repeatability were calculated with IS correction for PMPA, iPMPA, and EMPA, whereas no peak area correction was performed for CMPA and iBMPA due to the lack of isotopiclabeled standards for the two latter compounds. Data from the method validation are shown in Table 3. The LODs were determined as the concentrations giving signal intensity for the product ions of approximately 100 counts. This is 3−5 times the signal of arbitrary noise in MS/MS mode when extracted at a mass accuracy of ±5 mDa. Figure 5 shows the signals of the

Figure 5. EICs ([M − H − alkyl]− ± 5 mDa) from online SPE-LCMS/MS of a THF extract from urine where the AMPAs were added at 0.1 ng/mL. From top: EMPA, iPMPA, iBMPA, CMPA, and PMPA.

AMPAs when added to urine at 0.1 ng/mL, close to the LODs obtained in the two sample matrices. The only reported method obtaining LODs of the AMPAs in serum at sub-ppb levels using LC-MS/MS (0.3−0.5 ng/mL) is the recent work of Hamelin et al.,13 who utilized hydrophilic interactions in both sample preparation (offline SPE) and chromatographic separation. The obtained LODs in urine are comparable to those reported by Mawhinney et al. (0.03−0.24 ng/mL) using offline SPE with a silica column followed by LC-MS/MS.12 Lower LODs have recently been reported for determination of the analytes in urine (0.02 ng/mL) using pentafluorobenzyl derivatization and GC-MS/MS determination with the highly sensitive negative ion chemical ionization technique.16 Our method, however, offers a simpler and more rapid sample preparation procedure compared to the others, applicable for both serum and urine. Moreover, in the compared methods triple quadrupole mass analyzers have been applied, which are

■

CONCLUSION The use of SALLE offered rapid and efficient isolation of the AMPAs in one procedure that was applicable for both serum F



Article

Table 3. Data from Method Validation for Determination of the AMPAs in Serum (S) and Urine (U) PMPA

CMPA

iBMPA

iPMPA

EMPA

LOD, ng/mL

S U

0.09 0.11

0.08 0.08

0.08 0.09

0.04 0.10

0.11 0.12

linearity (R2), 0.3a−50 ng/mL

S U

0.998 0.995

0.998 0.996

0.997 0.996

0.998 0.998

0.997 0.997

within-assay repeatability, RSD (n = 6) LOQ S U 20 ng/mLb S U

11 16 4 2

14 14 4 3

11 7 3 3

14 17 3 3

20 17 3 3

between-assay repeatability, RSD (n = 6) LOQ S U 20 ng/mLb S U

12 11 3 3

11 12 6 7

14 9 5 8

14 10 4 5

16 9 4 4

98 ± 4 101 ± 2

98 ± 4 98 ± 3

92 ± 3 97 ± 3

89 ± 3 89 ± 2

70 ± 2 74 ± 2

recovery (SALLE) ± SD (n = 6) 20 ng/mLb a

S U

0.15 ng/mL for iPMPA in serum. b10 ng/mL for iPMPA in serum.

Figure 6. Quantification of the AMPAs when added to six individual samples of serum (S) and urine (U) at 20 ng/mL (10 ng/mL for iPMPA in serum) and at LOQ, plotted as mean percent of true values ± SD (n = 4).

rapid determination is crucial to guide medical countermeasures in emergency cases.

and urine. No more than 10 min was needed for pretreatment of one sample, and even the most polar of the analytes (EMPA) were extracted with 70% recovery. By transferring the analytes into the water-soluble organic solvent, secondary hydrophilic interactions could be utilized for enrichment of the AMPAs together with Lewis acid−base interactions on ZrO2. High selectivity toward the AMPAs was obtained in the SPE procedure by introducing the organic extracts at weak acidic environments, followed by desorption at pH 9.8. High method robustness was achieved by washing the SPE column at pH 13 between injections, since more than 350 sample injections could be performed before analyte breakthrough occurred. This new combination of SALLE and online HILIC SPE-LC-MS/ MS resulted in very low LODs of 0.04−0.12 ng/mL for the biomarkers, and the time needed to confirm nerve agent exposure was less than 30 min from receiving a sample. Detection limits at sub-ppb levels may be essential to verify nerve agent poisoning for prolonged time after exposure, and

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: Bent-Tore.Roen@ffi.no. Tel.: +47 6380 7881. Fax: +47 6380 7509. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was funded by the Norwegian Defence Research Establishment.

■

REFERENCES

(1) Black, R. M.; Read, R. W. Arch. Toxicol. 2013, 87, 421−437. (2) Black, R. M. J. Chromatogr. B 2010, 878, 1207−1215.

G



Article

(38) Choi, B. K.; Hercules, D. M.; Zhang, T.; Gusev, A. I. Curr. Trends Mass Spectrom. 2003, 18, 524−531.

(3) Minami, M.; Hui, D. M.; Katsumata, M.; Inagaki, H.; Boulet, C. A. J. Chromatogr. B 1997, 695, 237−244. (4) Nakajima, T.; Sasaki, K.; Ozawa, H.; Sekijima, Y.; Morita, H.; Fukushima, Y.; Yanagisawa, N. Arch. Toxicol. 1998, 72, 601−603. (5) Noort, D.; Hulst, A. G.; Platenburg, D. H. J. M.; Polhuijs, M.; Benschop, H. P. Arch. Toxicol. 1998, 72, 671−675. (6) Shih, M. L.; McMonagle, J. D.; Dolzine, T. W.; Gresham, V. C. J. Appl. Toxicol. 1994, 14, 195−199. (7) Evans, R. A.; Jakubowski, E. M.; Muse, W. T.; Matson, K.; Hulet, S. W.; Mioduszewski, R. J.; Thomson, S. A.; Totura, A. L.; Renner, J. A.; Crouse, C. L. J. Anal. Toxicol. 2008, 32, 78−85. (8) Ciner, F. L.; McCord, C. E.; Plunkett, R. W., Jr.; Martin, M. F.; Croley, T. R. J. Chromatogr. B 2007, 846, 42−50. (9) Kataoka, M.; Seto, Y. J. Chromatogr. B 2003, 795, 123−132. (10) Riches, J.; Morton, I.; Read, R. W.; Black, R. M. J. Chromatogr. B 2005, 816, 251−258. (11) Swaim, L. L.; Johnson, R. C.; Zhou, Y. T.; Sandlin, C.; Barr, J. R. J. Anal. Toxicol. 2008, 32, 774−777. (12) Mawhinney, D. B.; Hamelin, E. I.; Fraser, R.; Silva, S. S.; Pavlopoulos, A. J.; Kobelski, R. J. J. Chromatogr. B 2007, 852, 235− 243. (13) Hamelin, E. I.; Schulze, N. D.; Shaner, R. L.; Coleman, R. M.; Lawrence, R. J.; Crow, B. S.; Jakubowski, E. M.; Johnson, R. C. Anal. Bioanal.Chem. 2014, 406, 5195−5202. (14) Barr, J. R.; Driskell, W. J.; Aston, L. S.; Martinez, R. A. J. Anal. Toxicol. 2004, 28, 372−378. (15) Miki, A.; Katagi, M.; Tsuchihashi, H.; Yamashita, M. J. Anal. Toxicol. 1999, 23, 86−93. (16) Lin, Y.; Chen, J.; Yan, L.; Guo, L.; Wu, B.; Li, C.; Feng, J.; Liu, Q.; Xie, J. Anal. Bioanal.Chem. 2014, 406, 5213−5220. (17) Stan’kov, I. N.; Kondrat’ev, V. B.; Derevyagina, I. D.; Morozova, O. T.; Sadovnikov, S. V.; Selivanova, V. I.; Mylova, S. N.; Karpel’tseva, Y. A.; Karaseva, I. E.; Kuz’mina, N. E.; Krylov, V. V. J. Anal. Chem. 2011, 66, 626−632. (18) Tsuchihashi, H.; Katagi, M.; Nishikawa, M.; Tatsuno, M. J. Anal. Toxicol. 1998, 22, 383−388. (19) Subramaniam, R.; Ostin, A.; Nilsson, C.; Astot, C. J. Chromatogr. B 2013, 928, 98−105. (20) Røen, B. T.; Sellevåg, S. R.; Lundanes, E. Anal. Chim. Acta 2013, 761, 109−116. (21) Røen, B. T.; Sellevåg, S. R.; Dybendal, K. E.; Lundanes, E. J. Chromatogr. A 2014, 1329, 90−97. (22) Pan, J. L.; Zhang, C. J.; Zhang, Z. M.; Li, G. K. Anal. Chim. Acta 2014, 815, 1−15. (23) Rogeberg, M.; Malerod, H.; Roberg-Larsen, H.; Aass, C.; Wilson, S. R. J. Pharm. Biomed. Anal. 2014, 87, 120−129. (24) Tang, Y. Q.; Weng, N. D. Bioanalysis 2013, 5, 1583−1598. (25) Valente, I. M.; Goncalves, L. M.; Rodrigues, J. A. J. Chromatogr. A 2013, 1308, 58−62. (26) Blackwell, J. A.; Carr, P. W. Anal. Chem. 1992, 64, 853−862. (27) Nawrocki, J.; Dunlap, C.; McCormick, A.; Carr, P. W. J. Chromatogr. A 2004, 1028, 1−30. (28) Kucera, R.; Kovarikova, P.; Klivicky, M.; Klimes, J. J. Chromatogr. A 2011, 1218, 6981−6986. (29) Randon, J.; Huguet, S.; Demesmay, C.; Berthod, A. J. Chromatogr. A 2010, 1217, 1496−1500. (30) Kanaujia, P. K.; Pardasani, D.; Tak, V.; Purohit, A. K.; Dubey, D. K. J. Chromatogr. A 2011, 1218, 6612−6620. (31) Li, P. J.; Hu, B.; Li, X. Y. J. Chromatogr. A 2012, 1247, 49−56. (32) Xu, L.; Lee, H. K. Anal. Chem. 2007, 79, 5241−5248. (33) Svendsen, K. O.; Larsen, H. R.; Pedersen, S. A.; Brenna, I.; Lundanes, E.; Wilson, S. R. J. Sep. Sci. 2011, 34, 3020−3022. (34) Zhou, T.; Lucy, C. A. J. Chromatogr. A 2010, 1217, 82−88. (35) Blackwell, J. A.; Carr, P. W. Anal. Chem. 1992, 64, 863−873. (36) Hoffer, L. J.; Taveroff, A.; Robitaille, L.; Mamer, O. A.; Reimer, M. L. J. J. Nutr. 1993, 123, 1513−1521. (37) Dunlap, C. J.; McNeff, C. V.; Stoll, D.; Carr, P. W. Anal. Chem. 2001, 73, 598A−607A. H


Quantification of Nerve Agent Biomarkers in Human Serum and Urine

Recommend Documents