A Rapid, Sensitive Method for the Quantitation of ... - Semantic Scholar

34 downloads 2862 Views 964KB Size Report
urine after exposure to HD (3,5-7), the detection of 13-1yase metabolites ..... 50 ng/mL), were compared to calculate percentage recovery of the metabolite.
Journal of Analytical Toxicology, Vol. 28, July/August 2004

A Rapid, SensitiveMethod for the Quantitation of Specific Metabolites of Sulfur Mustard in Human Urine Using Isotope-Dilution Gas Chromatography-TandemMass Spectrometry Carrie U Young1, Doris Ash1, W. Jack Driskell 1, Anne E. BoyeP, Rodolfo A. Martinez 2, UA. Silks2, and John R. BarP,* 1Centersfor Disease Control and Prevention, National Center for Environmental Health, 4770 Buford Highway NE, Mailstop F-47, Atlanta, Georgia 30341 and 2National Stable Isotope Resource, Los Alamos National Laboratory, Los Alamos, New Mexico

Abstract I Sulfur mustard agent (HD) (2,2'-dichloroethyl sulfide), a Schedule I compound on the Chemical Weapons Convention Schedule of Chemicals, remains a public health concern because it is simple to synthesize and it is in the chemical weapon stockpiles of several countries. A sensitive, rapid, accurate, and precise method was developed to quantitate trace levels of 1,1 '-sulfonylbis [2-(methylthio) ethane] (SBMTE) in human urine as a means of assessing exposure to HD. The method used immobilized liquid-liquid extraction with diatomaceous earth, followed by the analysis of the urine extract using isotope-dilution gas chromatography-tandem mass spectrometry. Relative standard deviations were less than 8.6% at 1 ng/mL and 3.6% at 20 ng/mU The limit of detection for SBMTE was 0.038 ng/mL in 0.5 mL of urine.

Introduction Sulfur mustard agent (HD) [bis-(2-chloroethyl) sulfide also called 2,2'-dichloroethyl sulfide] is a powerful vesicant and a potent biological alkylating agent. HD was first employed as a chemical weapon during the First World War. Even though it was introduced as a weapon relatively late in the war, it was responsible for more chemical casualties than all other chemical agents combined (including chlorine, phosgene, and cyanogen chloride) (1). HD has been contained in the arsenals of various countries since it was first used in 1917, and its existence was * Autho," to whom correspondence and reprint requests should be addressed: John R. Barr, Centers for Disease Control and Prevention, 4770 Butord Highway NE, MS F47, Atlanta, GA 30341. E-mail: [email protected].

most recently confirmed when employed by Iraq during the Iran-Iraq War (1982-1988) (1). HD is a Schedule I compound on the Chemical Weapons Convention Schedule of Chemicals (2). Because it is relatively simple to synthesize and many nations have large stockpiles of the agent, HD remains an important threat for military use and as a terrorist weapon. The theories of the biochemical mechanism of injury of the highly reactive electrophilic HD can be divided into two major pathways that may potentially increase protease activity and cell death: metabolic changes due to DNA alkylation and chemical depletion of glutathione. A discussion of these biochemical mechanisms recently has been reviewed in detail (1). In the first part of the metabolic theory, the alkylation of DNA is believed to create strand breaks that trigger DNA repair enzymes; excessive activity of these enzymes would deplete the NAD§ stores required for glycolysis. The inhibition of glycolysis leads to a buildup of glucose-6 phosphate, which activates cellular proteases. The second part of the metabolic theory is that the alkylation of DNA leads to the inhibition of transcription and protein synthesis and this inhibition increases protease activity. In either case, the increased protease activity in basal epidermal cells could account for cleavage of the fibrils that connect the basement membrane to the basal epidermal cells, leading to severe tissue damage. The second major hypothesis for the biochemical mechanism of injury is the rapid reaction of the electrophilic HD with glutathione leading to its depletion. This depletion of glutathione would have two major effects. First, because glutathione is an important free radical scavenger, the depletion would cause loss of protection against oxygen-derived free radicals that induce lipid peroxidation. Second, glutathione depletion would result in a loss of the protein thiol status, and consequently, a loss of activity of sulfhydryl-containing enzymes, including calcium and mag-

Reproduction (photocopying)of editorial content of this journal is prohibited without publisher'spermission.

339

Journal of Analytical Toxicology, Vol. 28, July/August 2004

nesium adenosine triphosphatases that regulate calcium homeostasis. A decrease in calcium homeostasis would increase cellular calcium levels triggering an increase in protease, phospholipase, and endonuclease activity, and a subsequent breakdown of cell membranes and DNA. All of these mechanisms would lead to cell death and acute tissue damage and a combination of these biochemical mechanisms is likely responsible for injuries induced by HD exposure. The severe tissue damage caused by HD and the availability of simple synthetic methods make HD a dangerous chemical warfare agent. In the event of military or terrorist use of HD or unintentional human exposure during the destruction of chemical stockpiles, large numbers of samples may need to be analyzed during a short period of time to evaluate exposures. The principal objective for this analysis would be to determine levels of exposure for a population and distinguish the most highly exposed persons (those in need of immediate medical treatment) and persons with lower exposure (those who will require continued monitoring for long-term health effects) from the worried well (those who are convinced they have been exposed, even though they are physically well). 3~vo major metabolic pathways have been identified in rats (3). The first major metabolic pathway involves the simple hydrolysis of HD to form thiodiglycol (TDG) and its sulfoxide while the second major pathway involves reactions with glutathione. The action of 13-1yaseon glutathione-derived conjugates of HD with N-acetylcysteine (4-7) is believed to form 1,1'-sulfonylbis[2-(methylthio)ethane] (SBMTE)and its oxidizedcounterparts 1,1~-sulfonylbis[2-(methyl-sulfinyl)ethane] (SBMSE) and 1-methylsulfinyl-2-[2-(methylthio) ethylsulfonyl] ethane (MSMTESE). Although a complete understanding of HD metabolism in humans is lacking, the detection of hydrolysis and glutathione-[3-1yasederived metabolites of HD in human urine indicates some similarities in the metabolism of HD in rats and humans (6) and glutathione-derived metabolites are often formed after exposure to electrophilic compounds (6,8). Because these metabolites have been found in rat and human urine after exposure to HD (3,5-7), the detection of 13-1yase metabolites provides a forensic and diagnostic indicator of HD poisoning. Methods to determine and quantitate human and rat urinary extraction profiles of HD metabolites have been described (3-7,9-12). Most methods quantitate the major urinary metabolite TDG sulfoxide and its unoxidized analogue TDG along with their glucuronidated conjugates (13,14). More recently, methods have been introduced for the quantitative analysis of SBMTE, SBMSE, and MSMTESE (4-7). Black et al. (4,7,12) have developed several methods whereby SBMTE, SBMSE, MSMTESE, TDG, and TDG sulfoxide metabolites can be reduced to the two analytes, TDG and SBMTE,by treatment with acidic titanium trichloride (TiCl3); one of these methods includes a very powerful gas chromatographic-tandem mass spectrometric (GC-MS-MS) approach for analysis (7). Similarly, we developeda method to accurately and precisely analyze TDG and SBMTE (along with their sulfoxides) in human urine (15) that could be used in the initial stages of a public health response because additional metabolites can give further evi-

340

dence of HD exposure. The 13-1yasemetabolites are considered more selective for HD exposure than TDG and their concentrations in urine are on the same order of magnitude as the hydrolysis products. Public health decisions will most likely be made on the detection of the [3-1yasemetabolites because a natural background (1-12 ng/mL) of TDG sulfoxide and TDG has consistently been observed in urine from persons with no known HD exposure. The quantitative analyses of TDG and its sulfoxide, therefore, lack both specificity and sensitivity, limiting the value of this metabolite for identifying HD poisoning (4). Additionally, methods used to quantitate both TDG and SBMTE metabolites involve labor-intensive and time-consuming sample preparations, including a lengthy deconjugation step to liberate the glucuronidate-bound analytes. [3-Lyase metabolites are not expected to be detected in urine from unexposed persons, making SBMTEan unambiguous indicator of exposure to HD (6,7,15). Thus, for large-scale public health responses where prompt analyses are required, a method for the rapid quantitation of the specific 13-1yasemetabolites is useful. We describe a sensitive, selective, and rapid method for the quantitative analysis of SBMTE in human urine. Our method uses a simple immobilized liquid-liquid extraction step with diatomaceous earth followed by analysis using isotope-dilution GC-chemical ionization-MS-MS. We enhanced the accuracy and precision of the method using an isotopically labeled analogue as an internal standard. The sample preparation employed in this method also yielded remarkably clean extracts that dramatically decreased the need for and frequency of routine maintenance on the GC and ion source for the triplequadrupole MS.

Experimental Chemicals and materials All solvents and reagents were analytical grade. Dichloromethane was purchased from Tedia (Fairfield, OH). Acetonitrile, the reagent TiCI3 (titanium(III) chloride, about 10% solution in 20-30% hydrochloric acid, w/w), and all other chemicals unless otherwise indicated were fromAldrich Chemical (Milwaukee,WI). Sodium hydroxide (NaOH)was purchased from Caledon Laboratories, Ltd. (Georgetown,ON, Canada).Analytical standards and isotopically labeled standards were not commercially available and were synthesized at Los Alamos National Laboratory.The positions of the four 13Clabels were on the two ethyl groups. Research-grade helium and argon gases were obtained from Airgas (Radnor, PA), and instrument-grade isobutane gas was obtained from Specialty Gas Concepts (Pearland, TX). All gases used for GC-MS-M8 had a minimum purity of 99.999%.

Standard preparation and characterization Native standards. A 50-mg sample of SBMTEwas weighed and dissolved in 10 mL of acetonitrile in a 10-mL volumetric flask to a concentration of 5 m~dmL.This stock solution was then diluted to 50 IJ~dmL, 1 IJ~mL, and 10 n~dmLin acetonitrile. These working standards were used to prepare standards

Journal of Analytical Toxicology, Vol. 28, July/August 2004

in urine by spiking them into blank urine in 100-mL volumetric flasks to give eight standard solutions with concentrations of 0.1, 0.25, 0.5, 1, 5, 20, 50, and 100 ng/mL. All standard pools were mixed thoroughly for I h after enrichment. These eight enriched urine standards were aliquoted into 0.5-mL samples, stored at-70~ and analyzed to establish a calibration plot for quantitation. Labeled standards. A 50-rag sample of the labeled standard was weighed and dissolved in 10 mL of acetonitrile in a 10-mL volumetric flask to a concentration of 5 mg/mL. A 20-1JLaliquot of the 5-mg/mL stock was added to 980 IJL of acetonitrile to yield a 100-1Jg/mLsolution. A 500-1JLaliquot of the 100-pg/mL 13C-labeledstandard was combined in a 50-mL volumetric flask and diluted with acetonitrile to give a final concentration of 1 IJg/mL. Stock and working labeled reference solutions were stored at-70~

Quality control (QC) materials Urine for QC pools was collected from multiple volunteers, pooled, and prescreened to establish that neither the HD metabolite SBMTEnor interferences were detectable. The main urine pool was divided into three aliquots. Two of these aliquots were enriched with the SBMTEstock solution to create low (1 ng/mL) and high (20 ng/mL) QC pools. The third aliquot was reserved as a method blank and was used as the matrix to prepare the calibration standards. All QC pools were mixed thoroughly for 1 h following enrichment, dispensed in 0.5-mL aliquots into 15-mL vials, and stored at -70~ until needed. Each QC material was characterized by at least 20 analyses to define the mean concentrations and the 95th and 99th confidence intervals of SBMTE. QC materials reanalyzed after the initial characterization showed that SBMTEremained stable in the QC materials at-70~ for at least 6 months. It is important to note that this method begins with a TiCl3reaction which results in the reduction of sulfoxides. Thus, oxidation upon storage of the two sulfides in SBMTE would not be seen and would not affect the final results.

Sample preparation A typical sample set included 8 calibration standards and 12 unknowns, followed by a urine blank and 2 QC samples (1 ng/mL and 20 ng/mL). Larger batches contained additional QC samples to bracket the unknowns, ensuring the run was in control. Urine samples (0.5 mL each) were dispensed into a 15mL screw-cap glass culture tubes and a 20-1JLaliquot of the 1I@/mLinternal standard stock solution was added to each tube to give an internal standard concentration of 40 ng/mL. TiCl3(1 mL) was added to the urine samples, and they were incubated for 1 h at 75~ to induce reduction. A 2-mL volume of 6N sodium hydroxidewas added, and the samples were well mixed using a multiposition vortex mixer. The addition of the base resulted in the precipitation of most of the TiCl3. The samples were centrifuged at 4000 • for 5 min and the supernatant was poured directly onto the unbuffered 3-mL Chem Elut TM columns (Varian, Harbor City, CA), leaving the precipitated TiCl3 behind. Samples were immediately eluted from the columns by the addition of 16 mL of dichloromethane/acetonitrile (3:2). The eluates containing SBMTEwere collected in

15-mL conical tubes and then evaporated to dryness in the TurboVap| LVevaporator (Zymark Corp., Framingham, MA)at 40~ The nitrogen stream was initially adjusted to a gentle flow (1-5 psi) and was increased to 15-20 psi as the solvent volume decreased. Evaporation took about 20 min. Samples were reconstituted by adding 20 IJL of toluene, mixed with a vortex mixer, and transferred to autosampler vials for GC-MS-MS analysis.

Instrumentalanalysis The GC-MS-MS analyses were performed on a Finnigan TSQ~ triple-quadrupole MS (Finnigan MAT,San Jose, CA) that was interfaced to a Trace GC (CE Instruments, Thermoquest Corp., Austin, TX). Injection was automated using the A200S autosampler (Finnigan MAT, San Jose, CA). The GC column was a 30-m DB-5MS capillary column (0.25-ram i.d., 0.25-pro film thickness, J&W Scientific, Folsom, CA). Sample injection was in the splitless and the surge modes using a 3-ram internal diameter glass liner. Surge pressure was set at 43.5 psi for 2.0 rain. The injector temperature was 320~ Thermogreen LB-2 17-ram conditioned septa (Supelco, Bellefonte, PA) were used to ensure no degradation of the inlet septa at the high temperature profile. The helium carrier gas flow was set at 1 mL/min. Injection volume was 1 IJL. The GC temperature profile was (1) 90~ 2.0 rain; (2) 90~ to 320~ 70~ (3) 320~ 3 min. Mass analyses were by multiple reaction monitoring (MRM)for the m/z 215 ~ rn/z 75 and m/z 167 ~ m/z 75 transition for SBMTEand m/z 219 ~ rn/z 77 and m/z 171 m/z 77 transition for the ]3C4-1abeledSBMTE. The full-scan product ion spectrum, along with the structure of SBMTEand the collision-induced decomposition (CID) fragment ion, are shown in Figure 1. For quantitation, one decomposition was monitored for the native SBMTE (m/z 215 ~ m/z 75) and its labeled internal standard (rn/z 219 ~ m/z 77). Ionization was by chemical ionization (CI) with isobutane as the reagent gas at a source pressure of 1.5 T. (However,setting it to about 0.9 T for tuning the MS with FC43 was necessary because the higher isobutane pressure caused too many background mass peaks.) Argon was used as the collision gas for the CID. The pressure in 76.1 +CI~I2SCH~ D~ QO~ a~ 7~

so~ 4~

+HO

C

SCH~

a ~ "CJ~SCH, 21s,1

1o-~

5

~

,,',,,,,,,,,

......

,,,,

,

. . . .

8o go ,co ,~o ,~o ~3o ;~ ~.~

,. . . . . . . .

,80

~o ~ao ~

,,,,,,,,,

200 2~o

ndz

Figure 1. Full-scan tandem mass spectrum of 1,1'-sulfonylbis[2(methylthio) ethanel (SBMIE). The transition m/z 215 to m/z 75 was monitored for quantitation, lhe isotopically labeled SBMIE used for an internal standard included 13Cat the four carbons at each ethyl group. Thus, the transition of m/z 219 to m/z 77 was monitored for the internal standard. 341

Journal of Analytical Toxicology, Vol. 28, July/August 2004

the collision cell was approximately 2 mT. The collision energy was -10 eV for the quantitative ions and -15 eV for the confirmation ions. The electron multiplier, electron energy, and electron current were set at ]800 V,200 eV, and 300 I~, respectively. The electron lens voltage was 10 V. The source temperature was 150~ The scan rate was set at 0.02 s/scans. Acquisition was started at 3 rain, and the retention time of SBMTE was 5.24 rain (Figure 2). Data processing and analysis Data were initially processed by Xcalibur TM (Finnigan MAT, San Jose, CA) software supplied with the MS data system. Each ion of interest was automatically selected, the retention times calculated, and the area integrated. All data were checked for peak selection, interferences, and baseline determination and were corrected if found in error. After the data were transferred to a PC, statistical analyses were performed using SAS statistical software (SAS Institute, Inc., Cary, NC).

Quantitation Samples, blanks, standards, and QC materials were processed identically. Eight standard SBMTE concentrations (i.e., nomi-

RT: 5.24 ,oo

AA 7289

70 607

30

20

'OoL Time (rain) RT 5 24

100 90

SN 4225

~, 8o "to

B

1 ], AA 1013524

~ 50 40 I

30

\

,o 0

ii

47

rrl

,i

48

i i 1 ~ 1 1 ~

49

,i

ii

50

ii

51

i

I

III

,

~

i

52 5:~ T i m e 4min)

54

i i

i

i

,

55

, ~--~-'7~T~

56

57

Figure 2. A typical multiple-reaction monitoring ion chromatogramof a urine sample. Native SBMTEspiked at 100 pg/mL (m/z215 -~ m/z75) (A) and labeled SBMTE(m/z219 -> m/z 77) (B).

0.02 0 ~r"

~~ 2.5 2

~.

Results Method validation As shown in Figure 3, the SBMTE analysis was linear (R = 0.9996) over three orders of magnitude (from 0.100 to 100 ng/mL). The method LOD and limit of quantitation (LOQ) for the analyte were calculated according to the definition of the American Chemical Society (16) as 3So and 10s0, respectively, where So is the value of "the standard deviation as the concentration approaches zero" (17). The so was estimated as the y-intercept of the linear regression of the plot of the standard deviation of the three lowest standards used in the calibration plot versus the concentration (17). The standard deviations were calculated from 20 standard runs during a 3-week period. The LOD and LOQ for SBMTE were 0.038 ng/mL and 0.126 ng/mL, respectively.The recovery of SBMTE from urine was calculated as the percentage of RFR/RFc, where RFR and RFc are the response factors obtained from spiking half the samples with isotopically labeled analogues before the TiCI3 reduction procedure ("recovery") and half before injection ("control"), respectively. Three concentrations, run in triplicate (0.5, 15, and 50 ng/mL), were compared to calculate percentage recovery of the metabolite. The recovery efficiency varied from 31% to 38% for 15 and 50 ng/mL and from 18% to 22% for 0.5 ng/mL. Because isolation of low molecular weight polar compounds from polar matrices is difficult, the recoveries reported represent an acceptable compromise between recoveries and cleanup. More than likely, such a significant recovery loss results from loss of analyte in the precipitate after the addition of NaOH. The inherent variation of the extraction process and the concept of diminishing recoveries with decreasing concentrations can account for the variation seen in recovery efficiencies at the different concentrations. Isotope dilution analysis corrected for the recovery of the analyte because the labeled isotope behaves Table I. Precision and Accuracy of SBMTE Measurements

0.04-

3

nally 0.100, 0.250, 0.500, 1, 5, 20, 50, and 100 ng/mL in human urine) encompassing the linear range of the method were used to construct a calibration plot of response factor (RF; observed ratio of native peak area to labeled peak area) versus concentration. The lowest standard concentration was near the limit of detection (LOD) to ensure linearity at the low concentration end. The eight-point calibration curve, weighted by the reciprocal of the standard concentration, was used for quantitation.

"

Level1 Level2 Level3 Level4 (pg/L) (pg/L) (pg/L) (pg/L) Mean

"

1.5

Figure3. Calibration curve (n = 5) for SBMTEin concentratedurine. The

N 5 Spike concentration 0.80 Mean measured 0.81 concentration CV (%) 4.38 Accuracy (%) 101

calibration curve was linear (R2 = 0.9996)in the 0.100 to 100 ng/mL range. The insert shows the lower concentration data points.

* NA = not applicable.

1 ~0.5 0

0

20

40

60

80

100

120

Concentration (ng/mL)

342

5 3.00 3.07 3.67 102

5 10.0 10.2

5 30.0 30.8

NA* NA NA

2.70 102

4.02 103

3.69 102

Journal of Analytical Toxicology, Vol. 28, July/August 2004

chemically almost identically to the native compound and thus the recoveries do not affect the accuracy of the quantitation. The day-to-dayprecision of the calibration standards was evaluated by calculating the percent relative standard deviation (%RSD) of the calculated concentrations of spiked urine (n = 20) at each QC concentration level.Accuracywas evaluated in terms of a linear regression analysis plot of the response factor versus the expected concentration. When the method error is considered, the accuracy is indistinguishable from 100%. Accuracies and precisions were averaged over the concentration range to provide mean values covering a range of 0.80 to 30 ng/mL (Table I). The mean relativestandard deviation (RSD)was < 5% and the mean relative recovery was 102%, indicating the excellent dayto-day precision and relative recovery of the method. Standard QC charts for the concentration of SBMTE in the low-level (1 ng/mL) and high-level (20 ng/mL) QC materials are shown in Figure 4. Each point represents the single analysis of 20 QC materials analyzed over a 3-week period. RSDs for all QC materials were < 10%, reflecting excellent run-to-run precision over this period. Analysisof the QC materials 6 months after the initial characterization resulted in values within the 95% confidence intervals at each concentration. The concentrations of unknown samples were determined using the slope and intercept calculated by linear regression analysis of the calibration curves. An analytical run was considered out of control if any Westgard multirule for QC was violated (18).

Storage and heat stability To test for degradation of the SBMTEanalyte in urine, samples were stored for a seven-day period at three different tern-

t.$ 9

A

peratures and analyzed. Concentrations of I ng/mL, 5 ng/mL, and 20 ng/mL were kept for 1 week at ambient temperature, 4~ and -70~ The SBMTEanalyte was found to be stable for at least I week at all concentrations and temperature variations. Because of the potential biological infectivity of urine samples and because of the possibilityof the use of mixed chemical and biological agents, heat sterilizing unknown samples before analysis might be necessary. Sterilization by heat requires the urine to be incubated for I h at 90~ Duplicate samples of the SBMTE analyte in urine were subjected to 90~ for 1 h at concentrations of 1 ng/mL and 20 ng/mL and analyzed by this method as if they were unknown samples. The results showed no degradation of the SBMTEanalyte and no negative effect on analysis. It is important to note, however, that this method begins with a TiCI3reduction so that oxidation of the two sulfides in SBMTEto sulfoxidesupon storage or treatment would not be noticed or affect the quantitative results since they would be again reduced to sulfides.

Reference ranges To look for background levels of SBMTE and to test for interferences, we analyzed 120 samples of randomly collected urine from humans. These samples were a subset of 1000 sampies collected by Tennessee Blood Services (Memphis, TN) for the purpose of establishing reference ranges for various chemical warfare agent (CWA) methods. No demographic information was available for the participants and the collection of these samples was determined to be exempt from human subjects review.All samples were stored at-70~ until the time of analysis. Results showed no background levels (< 0.038 ng/mL) of SBMTE were present in any of the unexposed urine samples and that there were not interferences to the analysis.

99th confidence Interval 95th confidence interval

i , =t . . . . . . . . . . . . . . . . " . . . . . . . * . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1"11 ........................................................................................

\/%~

o.~t .................................... ....................................... ' ~ . . . . 9Sth confidence Interval o o., { . . . . . . . . . . . . . . . . . . . . . . . . . - . . . . . . . . . - . . . . . . . . . . . . . . . . ". . . . . . . . . . . 99th confidence Interval

o.,t

Discussion

[ " 0.8

~

I '

. 2 " 3

" 4

6

6

7

$

$

10

1t

ANd~lell

[..... m.m ...... LCUm

12

t3

14

1~

111

1;' " 15 " 10

2O "

run #

V,, ....... UCUm. . . . UCUm j. sameto~) i

23.0

B

22.0

confidence Interval 9Sth confidence interval

.....................................................

o . ~,,.o o 2

~

~

~

99tb

~

mean

I ,,.o

95th confidence interval 99rE confidence interval

,a.0 17~ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18.0

18.0

1

; .....

LCUm

2

3

.......

4

5

'CL.

8

7

8

--

g t0 11 12 18 14 18 18 17 Amdytkal nm #

.......

ucL.

....

ucc.

.

18 lg 20

.mle(~)l l

Figure 4. Quality control (QC) plots for 1,1 '-sulfonylbis[2-(methylthio) ethane] (SBMTE) in urine. The mean concentration of QC Low is 1 ng/mL with a relative standard deviation (RSD) of 9%. The mean and 95th and 99th confidence intervals are identified (A). The mean concentration of QC High is 20 ng/mL with an RSD of 4%. The mean and 95th and 99th confidence intervals are identified (B).

Because of the potential use of HD in both military and civilian settings, a rapid, reliable method is needed that can assess human exposure to HD. We have developed a rapid, sensitive method for the quantitation of a specific urinary metabolite of HD (SBMTE).The urine samples were first treated with TiCI3to reduce the mono and di-sulfoxides (SBMSE, MSMTESE)to the disulfide (SBMTE).This step takes what would be three analytes and reduces it into only one. Previously reported data (4) has shown that reduction of SBMSE and MSMTESE to SBMTE occurred in excess of 95% yield. Because the metabolites are no longer spread among three species, the analysis is more sensitive and sample preparation is easier since the disulfide is less polar than the mono- or di-sulfoxides. Isolation of any polar, low molecular weight compound such as SBMTE from a polar, corn-

343

Journal of Analytical Toxicology,Vol. 28, July/August2004

plex matrix, such as urine, can be an analyticalchallenge.In this method, immobilized liquid-liquid extraction using diatomaceous earth was used to isolate the SBMTEfrom the urine matrix and to remove urinary interferences. Elution of the analyte was achieved with methanol, dichloromethane, acetone, ethyl acetate, and acetonitrile (7). In our experimentation, a solvent mix of dichloromethane/acetonitrile (3:2) gave the highest percent recovery and cleanest eluate of SBMTE. We propose that acetonitrile acts as a bridging solvent between the dichloromethane and aqueous sample, thereby increasing recoveries by eluting SBMTEat both the surface of the diatomaceous earth and aqueous interfaces. In stability tests, the analyte was not affectedwhen subjected to heat sterilization (90~ 1 h). Furthermore, it was stable for at least 7 days at ambient temperature, 4~ and -70~ Other researchers have detected SBMTE in urine samples stored at -20~ for up to 10 years (7). Our method provides a rapid sample preparation procedure where 24 samples could be prepared in 3 h. It also yieldedclean extracts that were easilyanalyzed and did not require extensive maintenance of the GC interface or the MS. In most cases, routine GC maintenance was required only after running 500 samples, and the ion volume needed to be cleaned after 100 clinical samples. Thus, this cleanup procedure produced extracts that required less maintenance of the GC-triple quadrupole MS than other methods for the analysis of the urinary metabolites of HD. The sample preparation procedure yielded clean extracts but the extraction procedure was not only selective for SBMTE.To increase the sensitivity and the selectivity of this analysis, we employedGC-MS-MS.We observeda lower LODwith GC coupled to chemical ionization MS-MS than we did with liquid chromatography (LC) coupled to either atmospheric pressure chemical ionization (APCI) or electrospray ionization (ESI) MS-MS. A combination of high-resolution capillary GC and MS-MS provided good selectivityand excellent sensitivitywith an LOD of 38 pg/mL. In addition, we did not detect SBMTEin any of the 120 urine samples that were evaluated for background SBMTE concentrations, which suggests that urinary SBMTE is not produced after common environmental exposures but rather is a selective marker of HD exposure. The method's high sensitivity and improved LOD confirms concentrations of SBMTE analyte in urine at concentrations lower than any values previously reported in literature (7,15). Black and Read (6) have identified ~-lyase metabolite levels ranging from 42 to 56 ng/mL in human urine from two persons collected 2-3 days followingunintentional exposure to HD. They have also detected levels of 0.1-220 ng/mL in urine from seven people collected approximately 10-13 days after exposure (7), indicating this method should be sufficient to detect the analytes in the event of an exposure incident (6). The simplified sample preparation and improved instrument parameters enable detection of trace levels of the specific HD metabolite, SBMTE,for at least 14 daysfollowingexposureto HD depending on the dose absorbed. In the event of a biological and/or chemical attack, minimizing the amount of urine needed for analysis is crucial to ensure sufficient urine is available to perform several different

344

analysesto identifythe agent that was releasedand to assess the extent of exposure. The 0.5-mL volume of urine needed for SBMTEanalysisusing this method is more suitable for a chemical terrorism response than previously existing ones (7). Because this method is simple, sensitive, rapid, and robust, it can be used for the analysisof urine samples collectedin a chemical terrorism response. As part of an initial response where symptoms indicate the use of a vesicant or no information about symptoms is available, this method can help distinguish HD exposure from exposure to other vesicants.

Conclusions We report a sensitiveand selectivemethod for assessingsulfur mustard exposure. We have obtained specificityby using a selective tandem MS method and by measuring an unambiguous marker of HD exposure. Our method is simpleand robust with a limit of detection in the low picograrn-per-milliliterrange. Because our method is more robust than those measuring TDG alone, is sensitiveand selective for a specificmetabolite of mustard HD, and is more rapid than those detecting TDG and SBMTE in a single urine sample, it will be an important tool in detecting exposure to HD.

Acknowledgments The use of trade names is for identificationonly and does not constitute endorsement by the Public Health Services, the Department of Health and Human Services,the Centers for Disease Control and Prevention, or the U.S. Environmental Protection Agency.We thank Ritchard Parry for his assistance in running reference range samples, Dr. Roberto Bravofor assistance on the MS-MS method, and R. Donnie Whitehead for help with the sample preparation. We also thank Dr. Dana B. Barr, Dr. Sharon W. Lemire, and Dr. DavidL. Ashley for helpful discussions during the method developmentand method validation portions of this study.

References I. F.R.Sidell, J.S. Urbanetti, W.J. Smith, and C.G. Hurst. Vesicants. In Medical Aspectsof Chemical and Biological Warfare, E.T.Takafuji and D.R. Franz, Eds.Office of The Surgeon General, TMM Publications, Washington, D.C., 1997, pp 198-200. 2. Conference on Disarmament. The Convention on the Prohibition of the Development, Production, Stockpilingand Use of Chemical Weapons and on their Destruction, United Nations Document CD/1170, Geneva, Switzerland, August 1992. 3. R.M. Black, K. Brewster, R.J. Clarke, J.L. Hambrook, J.M. Harrison, and D.J. Howells. Biological fate of sulphur mustard, 1,1 'thiobis(2-chloroethane): isolation and identification of urinary metabolites following intraperitoneal administration to rat. Xenobiotica 22:405-418 (1992). 4. R.M. Black, R.J. Clarke, and R.W. Read. Analysis of 1,1'-sulphonylbis[2-methylsulphinyl)ethane] and 1-methylsulphinyl-2-[2-

Journal of Analytical Toxicology,Vol. 28, July/August2004

5.

6.

7.

8.

9. 10. 11.

(methylthio)ethylsulphonyl]ethane, metabolites of sulphur mustard, in urine using gas chromatography-mass spectrometry. ]. Chromatogr. 558:405-414 (1991 ). R.M. Black, J.L. Hambrook, D.J. Howells, and R.W. Read. Biological fate of sulfur mustard, 1,1' thiobis(2-chloroethane). Urinary excretion profiles of hydrolysis products and 13-1yasemetabolites of sulfur mustard after cutaneous application in rats. J. Anal Toxicol. 16:79-84 (1992). R.M. Black and R.W. Read. Biological fate of sulphur mustard, 1,1 'thiobis(2-chloroethane): identification of [3-1yase metabolites in hydrolysis products in human urine. Xenobiotica 25:167-173 (1995). R.M. Black and R.W. Read. Improved methodology for the detection and quantitation of urinary metabolites of sulphur mustard using gas chromatography-tandem mass spectrometry. ]. Chromatogr. B 665:97-105 (1995). R.T.H.Van Weile, R.G.J.M. Van Dijick, N.J. Van Sittert, and N.P.E. Vermeulen. Mercapturic acids, protein adducts, and DNA adducts as biomarkers of electrophilic chemicals. Crit. Rev. Toxicol. 22: 271-306 (1992). E.R. Wils, A.G. Hulst, A.L. de Jong, A. Verweij, and H.L. Boter. Analysis of thiodiglycol in urine of victims of an alleged attack with mustard gas. J. Anal. Toxicol. 9:254-257 (1985). E.R.Wils, A.G. Hulst and J. van Laar. Analysis of thiodiglycol in urine of victims of an alleged attack with mustard gas, part II. ]. Anal. Toxicol. 12:15-19 (1988). R.M. Black and R.W. Read. Detection of trace levels of thiodiglycol in blood, plasma and urine using gas chromatography-electron

12.

13.

14.

15.

16. 17. 18.

capture negative-ion chemical ionization mass spectrometry. J. Chromatogr. 449:261-270 (1988). R.M. Black and R.W. Read. Methods for the analysis of thiodiglycol sulphoxide, a metabolite of sulphur mustard in urine using gaschromatography-mass spectrometry. J. Chromatogr. 558:393-404 (1991). J.S.Graham, F.M. Reid, J.R. Smith, R.R. Stotts, F.S.Tucker, S.M. Shumaker, N.A. Niemuth, and S.J. Janny. A cutaneous full-thickness liquid sulfur mustard burn model in weanling swine: clinical patholgy and urinary excretion of thiodiglycol. J. AppL Toxicol. 20: $161-172 (2000). E.M. Jakubowski, F.R. Sidell, R.A, Evans, M.A. Carter, J.R. Keeler, J.D. McMonagle, A. Swift, J.R. Smith, and T.W. Dolzine. Quantification of thiodiglycol in human urine after an accidental sulfur mustard exposure. Toxicol. Methods 10:143-150 (2000). A.E. Boyer, D. Ash, D. Barr, C.L. Young, W.J. Driskell, R.D. Whitehead, Jr., M. Ospina, K.E. Preston,A.R. Woolfitt, R.A. Martinez, L.A. Silks, and J.R. Barr. Quantitation of the sulfur mustard metabolites 1,1'-sulfonylbis[2-(methylthio)ethane] and thiodiglycol in urine using isotope-dilution gas chromatography-tandem mass spectrometry. J. Anal ToxicoL 28:327-332 (2004). American Chemical Society Committee Report. Principles of environmental measurement. Anal. Chem. 55:2210-2218 (1983). J.K. Taylor. Quality Assurance of Chemical Measurements, CRC Press, Inc., Boca Raton, FL, 1987, pp 89--83. J.O. Westgard. Basic QC Practices: Training in Statistical Quality Control for Health Care Laboratories. Westgard QC, Inc., Madison, Wl, 2002.

345