Journal of Analytical Toxicology, Vol. 16, M a y / J u n e 1992
Determination of 3-Quinuclidinyl Benzilate (QNB) and Its Major Metabolites in Urine by Isotope Dilution Gas Chromatography/Mass Spectrometry* G a r y D. B y r d t , R o b e r t C. P a u l e , L a n e C. S a n d e r , L o r n a T. S n i e g o s k i ,
and Edward
W h i t e V**
Center for Analytical Chemistry, National/nstitute of Standards and Technology, Gaithersburg, MD 20899 Howard
T. B a u s u m
U.S. Army Biomedical Research and Development Laborato~ Fort Detrick, Frederick, Maryland 21702
Abstract
I
In response to the scheduled destruction of U.S. military stockpiles of the hallucinogenic agent 3-quinuclldinyl benzilate (QNB), a specific confirmatory test for human exposure to QNB was developed. The amount of the parent compound in the urine as well as the two major metabolites, 3-quinuclidinol (Q) and benzilic acid (BA), was determined because the relationship between QNB dose and levels of QNB and its metabolites in human urine is not known. QNB was determined in urine samples spiked at a target level of 0.5 ng/mL, and the metabolites BA and Q were determined at a target level of 5 ng/mL. The method uses solid-phase extraction to isolate each analyte from the urine and isotope dilution gas chromatography/mass spectrometry for quantitation. Each analyte is converted to its trimethylsilyl derivative for analysis. The analytical method was tested on eight different urine samples spiked with known amounts of the analytes near the target levels, at 10 times the target levels, and blank (unspiked) urine samples. The variabilities in the method are for the most part evenly distributed between three imprecision categories: GC/MS measurement, sample preparation, and the urine samples. The total imprecision (1 standard deviation) of a single measurement is about 15% of the value for each analyte.
Introduction
The United States has initiated efforts to dispose of its aging and obsolete stockpile of chemical agents and munitions (1). One of the materials scheduled for disposal is the hallucinogenic agent 3-quinuclidinylbenzilate (QNB), code-named BZ. The destruction of current stockpiles of QNB has the potential for worker exposure to this material. Consequently, a chemical test to confirm human exposure to QNB has been developed. Information "Presented, in part. at the 35th Annual Conference on Mass Spectrometry and Aflied Topics, Denver, Colorado, May 24-29, 1987, and at the Accuracy in Trace Analysis Symposium, National Bureau of Standards, Gaithersburg, Maryland, September 28-October 1,1987 Certain commercial equipment, instruments, or materials are idenlified in this report to specify adequately the experimental procedures. Such identification does nol imply recommendation or endorsement by the National InstRute of Standards and Technology, nor does it imply that the materials or equipment identified are necessarily the best available for the purpose Present address: R.J. Reynolds Tobacco Co., Bowman-Gray Technical Center, Winston-SaLem, NC "'Author to whom correspondence should be addressed
182
from such a chemical test is not intended to be indicative, but rather confirmatory, because exposure to QNB is most rapidly manifested by symptoms such as tachycardia and mydriasis (2). This paper describes the development of a specific confirmatory chemical test for human exposure to QNB. QNB is one of the most potent anticholinergic psychotomimetics known (3). It is also widely used in research on the brain and nervous system as a specific binding antagonist for the muscarinic class of cholinergic receptors. The effects of very small doses of this agent on the brain and nervous system can produce incapacitation (2,4). QNB undergoes hydrolysis in aqueous buffers to produce benzilic acid (BA) and 3-quhmclidinol (Q) as shown in Figure 1 (5). Little is known of the metabolism and dose/response effects of QNB in the body. Work by the Army (6) indicates that most of the QNB that enters the body is excreted via the kidneys, making urine the meditnn of choice for detection of the unaltered drug and its metabolhes. About 3% of an administered close is excreted unchanged in rats. and Q has been detected in rat urine (6,7), but such informaton has not been reported for humans. Because of the t, nccrt:finties in the relative amounts of QNB, BA, and Q that may appear in human urine, a method that could determine all three in urine is desirable. The maximum no-effect dose of QNB in humans has been estimated to be 0.5 to 1.0 p.g/kg (6,7). The target levels for the detection of QNB and its hydrolysis products, BA and Q, were chosen so as to have a good probability of detecting exposures of 0.5 p.g/kg or higher. It was assumed that 2C/c of a dose of QNB entering the human body will appear in the urine as QNB or its glucuronide conjugate and 40c/~, as BA and Q within the first eight hours. From this, urinary concentrations of 0.5 to 0.7 ng/mL for QNB (including conjugated QNB) and 5 to 7 ng/mL for BA and Q can be calculated (8). These target levels should provide a safety margin adequate for the determination of the analytes in the urine of a person who was exposed to QNB at a level that would produce a detectable physiological effect (8). The method described here uses solid-phase extraction of the urine and isotope dilution in conjunction with gas chromatography/mass spectrometry (GC/MS). Because QNB is hazardous, neither the correctness of the target levels chosen nor the method developed could be tested on urine samples from exposed persons. The method was developed and tested on urine samples spiked with QNB and its metabolites at known concentrations. The results indicate that any exposure to QNB resulting in concentrations of QNB, BA, and Q in the urine at the target levels or higher could be detected.
Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission.
Journal of AnalyticalToxicology,Vol. 16, May/June 1992
Experimental
Warning: Because of the harmful effects of QNB when ingested in small amounts, extreme care should be taken when working with this material. Reagents 3-Quinuclidinyl benzilate (hydrochloride salt) was obtained from Hoffmann-La Roche, Inc. Benzilic acid, 3-quinuclidinol, and N-methyl-N-trimethylsilyltrifluoroacetamide(MSTFA) were obtained from commercial sources and used as supplied. The synthesis of the labeled analogs of the three analytes, 3-180_quin_ uclidinyl ds-benzilate (94 atom % 180 and >99 atom % ds), d 5benzilic acid (>99 atom % ds), and 3-180-quinuclidinol(94 atom % 180), is described elsewhere (9). B-Glucuronidase enzyme (500,000 units per gram solid, pH 3.8) was obtained from Sigma Chemical Co. All other chemicals were reagent grade from commercial sources and used as supplied.
Apparatus Solid-phase extraction Sep-Pak TM cartridges were obtained from Waters Associates. One-mL conical vials fitted with a Teflon-valved cap and a septum seal were used for all derivatization reactions. GC/MS analyses were carried out on a quadrupole mass spectrometer equipped with a 30-m x 0.25-mm i.d., DB-17 fused-silica capillary column connected directly to the ion source of the mass spectrometer. The instrument was operated in the electron ionization mode with an ionizing energy of 70 eV and at unit resolution. Collection and treatment of s a m p l e s
The analytical method was tested on urine specimens taken from eight individuals. Urine samples of 180 mL or greater were collected over a 4-h period from five males and three females. The urine samples were subdivided into 60-mL samples according to the scheme shown in Figure 2. Samples were spiked with QNB, BA, and Q from one of two independently prepared acetonitrile solutions, each of which contained the three analytes. Some samples were spiked with the analytes at concentrations near the target level and others at 10 times this amount. Several samples were left blank. Each 60-mL urine sample was treated as follows: Immediately after the samples were spiked with the unlabeled analytes, the isotopically labeled internal standards were added at concentrations approximately twice the target levels using one of two independently prepared acetonitrile solutions, each of which contained the three analytes. To stabilize the QNB, glacial acetic acid was added (1.5 mL per 60 mL of urine) to adjust the urine to pH 3-4, which is the pH for minimum hydrolysis of QNB in aqueous media at 25~ (5). Sodium azide was added as a preservative at a concentration of 0.05% (w/w) to inhibit bacterial growth. All samples were placed in refrigerated storage (4~ for three days to mimic actual shipping conditions. Before extraction, the urine samples were treated with an enzyme solution to free any glucuronide conjugate of the analytes that might occur in natural samples. The B-glucuronidase enzyme used was from abalone entrails (500,000 units per gram solid) with an optimum activity at pH 3.8, which is in the pH range of the stored urine. To each 60-mL urine sample, 10 mg of the enzyme (approximately 5,000 units) in one mL of water was added before incubation at 37~ for 18 h.
Recovery of analytes from urine and preparation for GC/MS A different solid-phase extraction method was used for each an-
alyte and the derivatized extracts were analyzed by GC/MS. From each 60-mL urine sample, one 20-mL aliquot was used for the QNB analysis and one 20-mL aliquot was used for both the BA and Q analyses. The remaining 20 mL was held in reserve in the event that an analysis failed and had to be repeated. Recoveries of QNB, BA, and Q were determined by comparison of the GC/MS response from samples with the response from standards. Extraction ofQNB. A Cl8 Sep-Pak cartridge was prepared by wetting it with methanol followed by water. A 20-mL urine sample was adjusted to pH 9-10 with 10M NaOH and loaded onto the column at a rate of 5-10 mL/min. The cartridge was washed with 5 mL of water followed by 5 mL of 40% (v/v) CH3CN in water. The sample was eluted from the cartridge with 3 mL of methanol and the eluate was reduced in volume to 1 mL under a gentle stream of nitrogen at 70~ The eluate was transferred to a 1-mL conical vial and blown to dryness. Derivatizing solution (50 pL, 2:1 CH3CN-MSTFA) was added and the sample was capped and heated at 70~ for 3 h. The recovery of QNB at the 0.5-ng/mL level was about 40%. Extraction ofBA. A Cl8 Sep-Pak cartridge was prepared by wetting it with methanol followed by water. A 20-mL urine sample at the storage pH was loaded onto the cartridge at a rate of 5-10 mL/minute. The cartridge was washed with 5 mL of water. The eluate from the 20-mL load of the sample and the 5mL water wash were saved for the Q work-up described below. The cartridge was washed with 3 mL of 15% (v/v) CH3CN in water. The BA was eluted from the cartridge with 3 mL of CHCI3 and the eluate was reduced in volume under a gentle stream of nitrogen at 70~ to approximately 1 mL. The eluate was transferred to a 1-mL conical vial and blown to dryness. Derivatizing solution (50 IaL, 2:1 CH3CN-MSTFA) was added and the sample was capped and heated at 70~ for 15 min. The recovery of BA at the 5-ng/mL level was about 80%. Extraction of Q. The Q work-up used the eluate from the BA load and water wash which was lighter in color than unfiltered urine. This eluate was adjusted to pH 6-7 with concentrated NHaOH. A FlorisilTM Sep-Pak cartridge was prepared by wetting with 10 mL of water and the urine was loaded at a rate of 5-10 mL/minute. The cartridge was washed with 5 mL of water. Removal of the Q was accomplished by forcing 3 x 20 mL of concentrated NH4OH through the cartridge and collecting the effluent. The solvent was removed under reduced pressure on a OH 0
Figure 1. Hydrolysisof QNB.
.............
1~-7
Spiked Concentr~tlon
0
T
~ Ti'
. . . . . . . . . . . . .
SpikedC~can~ratloa ...............
8
r-Ll
I~-I ]'
s
T
~ +~ ........ t
8 1ST 1ST' FLl A A
I--~ T'
0
T
I~]---I T'
e
T
T'
t [Si"iT ]ST' ~ "~8 +i' ]ST
8
r~ A
r~l
Figure 2. Sampling scheme for validation of method. T, T' = target levels. lOT, lOT' = 10 times target levels. Samples 2, 4, and 8 are from female subjects, the others were from males.
183
Journal of AnalyticalToxicology, Vol. 16, May/June 1992
rotary evaporation apparatus while the solution was heated at 70~ The residue was transferred to a concentration tube with 3 • 5 mL of methanol and blown down to 1 mL under nitrogen at 70~ The solution was transferred to a 1-mL conical vial and blown completely dry under nitrogen at 70~ Derivatizing solution (100 HL 2:1 CH3CN-MSTFA) was added and the sample was capped and heated at 70~ for 15 min. The recovery of Q from urine at the 5-ng/mL level was approximately 50%. GC/MS of urine extracts
The GC had a split flow injector port which was maintained at 250~ The chromatographic peaks corresponding to the trimethylsilyl derivatives of QNB, BA, and Q in the spiked samples were identified by comparison of retention times with those produced by analyzing pure standards of each compound and by comparing extracts of blanks with the spiked samples. Solvent blanks were run prior to extract samples to check for analyte carry-over from previous runs. Owing to the complexity of the urine extracts, each extract was analyzed by GC/MS separately and conditions for the individual analytes are described below. Because of the low levels of analyte, the QNB extract was analyzed by splitless mode GC. The oven temperature was initially held at 60~ for 0.5 min, then taken up to 225~ at 40~ and held there for 1 min. The temperature was then increased to 273~
at 4~ The mass spectrometer was set to monitor m/z 255 for the QNB derivative and m/z 260 for the labeled QNB derivative. The BA extract was analyzed by the use of split injection with a split flow ratio of 15:1. The GC column was initially held at 175~ for 2 min and then programmed up to 201~ at a rate of 4~ The mass spectrometer conditions for the BA derivative were identical to those for the QNB derivative. The Q extract was analyzed by the use of split injection with a split flow ratio of 15:1. The GC column was initially held at 75~ for 2 min and then programmed up to 115~ at a rate of 4~ The mass spectrometer was set to monitor m/z 199 for the Q derivative and ,u'z 201 for the labeled Q derivative. Response factor solutions consisting of known amounts of all three labeled and unlabeled analytes dissolved in the derivatizing agent were prepared. Concentrations of the labeled materials were approximately twice those of the unlabeled analytes. For any series of samples run in a day, two independently prepared response factor solutions were each analyzed, once prior to and once just after, each series for a given analyte. The response factor used for a series was an average of the four measurements. Each sample extract was analyzed twice by GC/MS.
Results and Discussion TMS
900
0 0
(C6Hs)2C-O-TMS ,
10001
A GC/MS method for the quantitative determination of QNB, potentially present as an impurity in a prescription drug, has been
B00 700
600 50{} 400
TK4S *
300
7/3
o)
Ion
255.80
amu.
CTH,~N *
t/z o
200
,00] I 'c
409
366
QNB-TMS
200
100
Figure 3. Mass spectrum of TMS derivative of ONB.
Ion
260.00
amu.
LQNB-TMS j-
10001 900 800
TUS
7~
~ss ! (CiHs)iC-0-TMS ,
9
TUS 0
0
!3
7~0~
~4
Time
(min.)
15
!S
I
I(M
-- CHj
147 /
'~176 1
CO)"
L?
..LI_ t
00
b)
200 Mass/Charge
Ion
255.00
amu.
Figure 4. Mass spectrum of TMS derivative of BA.
1 0 00" 900" 800" 700~ 600"
Ion J
J
(M -
amu.
LQNB-TMS
TMSOCH2) ~ 9/6
C6HI~OS i 9
1/29
50D
400"
260.20
TMS +
(M- CH3)" M"
CTHI2N. 0
300"
45
184
199
~
P
:3
14
Time
(mln.)
15
16
200 100
~a=s,Ch."ge
Figure 5. Mass spectrum of TMS derivative of Q.
184
Figure 6. SIM chromatogram of m/z 255 and m/z 260 from extracts of urine a) spiked with 0.5 ng/mL ONB and b) a urine blank. QNB-TMS is the derivatized flNB and LQNB-TMS is the derivatized labeled internal standard. In both urine samples the labeled internal standard is 0.90 ng/mL.
Journal of Analytical Toxicology, Vol. 16, May/June 1992
reported (10). That method used extraction and thin-layer chromatography to separate QNB from the matrix, followed by GC/MS of the trimethylsilyl (TMS) derivative. The article also indicated that underivatized QNB decomposes to benzophenone when exposed to the heated surfaces in the GC injector port. BA can also decompose, through decarboxylation, and does not chromatograph as the acid. Therefore, the analytes, QNB, BA, and Q, were converted to their TMS derivatives (QNB-TMS, BA-TMS, and Q-TMS) to improve their chromatographic behavior and stability. N-Methyl-N-trimethylsilyltrifluoroacetamide (MSTFA) was selected as the derivatizing agent for the analytes to minimize chromatographic peak tailing of the solvent front because QTMS elutes at relatively low GC column temperatures. While BA and Q readily dissolved in MSTFA, QNB proved to be only partially soluble in this material. A mixture of 2:1 acetonitrileMSTFA adequately dissolved the QNB, as well as the other two analytes. BA and Q were completely derivatized in less than 15 rain following addition of the derivatizing solution, whereas QNB required 3 h at 70~ to achieve more than 98% derivatization. The mass spectra of the derivatized compounds are shown in Figures 3-5. The mass spectra of the QNB-TMS and the BATMS derivatives both have an ion at m/z 255 as the base peak which has the stable structure shown. This ion was chosen for selected ion monitoring (SIM) of both species. For the labeled materials, this ion is shifted to m/z 260. The spectrum of QNB-TMS also contains characteristic ions at m/z 1I0 and 126, which come from the quinuclidinyl portion of the molecule. A weak molecular
o)
Ion
255.Q0
ion, M +., for QNB-TMS at m/z 409 is observed. BA-TMS has two TMS groups following derivatization and, while no molecular ion is observed, two high mass fragment ions at m/z 329 and 357 occur that are characteristic of this molecule. Q-TMS undergoes extensive fragmentation, as shown in Figure 5. The molecular ion at m/z 199 has a relative intensity greater than 50% of the base peak and was selected as the representative ion for SIM. This ion is shifted to m/z 201 for the labeled material. Other ions of interest include (M - CH3) § at m/z 184, the quinuclidinyl ion at m/z 110, and (M - TMSOCH2) + at m/z 96. The composition of the ion at m/z 129 was determined to be C 6 H I 3 O S i § by exact mass measurement on a double-focusing high resolution mass spectrometer using a direct probe to introduce the sample. The structure of this ion is not known. Figures 6-8 show ion chromatograms for each derivatized analyte in (a) a urine extract with the analyte at the target level and (b) a urine blank. No significant interferences were found in any of the blank urines analyzed. Gradual decreases in retention times for the analytes, most evident for BA in Figure 7, were observed over the course of the validation test, presumably the consequence of loss of liquid phase from the colunm resulting from the large number of samples and the baking of the column needed to minimize background. The analytical method developed was tested on eight urine specimens from different individuals according to the scheme in Figure 2. This scheme was designed to provide needed information using a minimum number of samples and measurements. The following information was obtained: the variability of the sample preparation, the urine-to-urine variability, and the variability of the
tm,.
o) Ion
268.08
Ion
IB9.BB
amu.
JV Ion
6.5
mmu.
7.0 Time
7.5 (mln.)
8.0
8.5
201.B8
9.8 !8
b)
Ion
255.00
- -
~ 7.0
268.88
Time
(mtn.)
| I
12
lmu.
b)
Ion
mmu.
Ton
199.00
Ion
281.88
Imu.
amu.
~_~LBA-TMS
7.5 Tlme
8.0 (mln.)
8.5
Fioure 7. SIM chromatogram of m/z255 and m/z260 from extracts of urine a) spiked with 4.80 ng/mL BA and b) a urine blank. BA-TMS is the defivatizedBA and LBA-TMS is the derivatizedlabeled internal standard. In both urine samples the labeledinternal standard is 8.55 ng/mL
9
1~
amu.
Ttme
(mln.)
1!
12
Figure 8. SIM chromatogram of m/z 199 and m/z 201 from extracts of urine a) spiked with 500 ng/mL ONB and b) a urine blank O-TMS is the derivatized Q and LQ-TMS is the derivatized labeled internal standard. In both urine samples the labeled internal standard is 10.22 ng/mL
185
Journal of Analytical Toxicology, Vol. 16, M a y / J u n e 1992
response factor is not correct because the analyte and labeled internal standard both contribute to the signals at the masses monitored (11). The error at the target level is about 5% and thus negligible compared to other sources of error, but for the measurements at 10 times the target level the values are about 30% low. This is satisfactory for the present purpose of determining whether or not the analyte is or is not present above the target level, but the method for Q would require modification if the target level were different or if the method were to be adapted for use as an assay. For all three analytes the measured value was considered good at these concentration levels. Based on the established target levels, no false positives occurred for any of the four blank urines for any of the three analytes. The blank with the highest relative response among all the analytes, 3B for QNB, gave a measured value that was less than half the target value. For samples spiked near the target level, no measured value was less than 80% of the spiked value. On this basis, no false negatives were apparent. Of the 54 extractions performed for the validation test, 5 failed the initial work-up (no internal standard was recovered). However, none of these failed when repeated on the back-up urine sample incorporated into the method. As a "worst case" example for all analytes spiked at the target level, the maximum percent difference between measured and spiked values was +31%. Statistical calculations yielded, for each analyte, the sample averages, the three components of standard deviation (GC/MS measurement, sample preparation, and urine-to-urine variability), and the standard deviation of a single measurement. Samples 1-4, 6, and 7 had both duplicate sample preparations and GC/MS measurements on each sample so that analyses of variance (ANOVA) could be made for the corresponding components for these samples. These results, along with those from samples 5 and 8, were then used in an ANOVA for unbalanced data to calculate the components of standard deviation between urine samples (12). This was done for each analyte and each spike level (T and 10T). The 0 components of standard deviation (expressed as percent of the analyte value) arc listed in Spiked Meas. % Dil Table II. This table shows that the impreci4.22 -16 sions are more or less evenly distributed be5.00 5.05 +2 tween the three imprecision categories 4.95 5.27 +5 (GC/MS measurement, sample preparation, 5.00 6.22 +26 and urine sample). These effects are relatively 4.95 5.30 + 6 constant, considering the limited number of 5.00 4.43 -11 samples and measurements. This table also 4.95 shows that the total imprecision of a single 5.10 +2 5.00 measurement is about 15% for each analyte, 4.95 4.76 -4 and that it is not greatly different between the 5.00 6.55 +31 T and 10T spike levels.
GC/IVIS measurements at the target level and 10 times the target level. Information was desired at the higher concentrations because these levels would be expected in the event of an exposure producing a detectable physiological effect. Table I summarizes the data for all samples for the analytes QNB, BA, and Q. The sub-sample numbers correspond to those in the diagram in Figure 2. The values given for the spiked samples have not been corrected for the blank because no blank would be available for samples resulting from exposure to QNB. The upper part of the table contains the results for samples that were spiked near the target level for each analyte. The middle part of the table contains the results for the samples that were spiked at approximately 10 times the target level for each analyte. The lower part of the table contains the results for the blanks. For urine samples spiked with QNB at the target level, Table I shows that the measured QNB concentration differed from the spiked value by -18% to +14%. At 10 times the target level the measured value differed by -24% to +2% from the spiked value. The highest reading on a blank gave a value 40% of the target value for QNB (sub-sample 3B). In urine samples with BA spiked near the target level, the measured BA concentration differed from the spiked value by -15 to +19%. At 10 times the target value there was less scatter of the measurements with measured values differing by -7 to +4% from the spiked value. The four blanks measured gave values of 1 to 7% of the target value, again with sample 3B giving the highest reading. Q, when spiked near the target level, was measured at concentrations differing by -16 to +31% from the spiked value. At 10 times the target level, the measured value differed from the spiked value by -26 to -7%. The four blanks were measured from 2 to 25% of the target value with sample 3B again giving the highest reading. For Q the calculation of the concentration from a single
Table I. Summary of Validation Test Results* ONB Subsample Spiked Meas. % Dif
BA Spiked Meas. % Dif
1-1
0.450
0.487
+8
4.80
4.14
1-2
0.450
0.514
+14
4.67
4.52
-14 -3
2-1
0.450
0.403
-10
4.80
4.23
-12
2-2
0.450
0,370
-18
4.67
4.25
-9
3-1
0.450
0.471
+5
4.80
5.70
+19
3-2
0.450
0.451
+0
4.67
5.36
+15
4-1
0.450
0.386
-14
4.80
4.08
-15
4-2
0.450
0.373
-17
4.67
4.53
-3
5-1
0.450
0.465
+3
4.80
4.80
0
6-1
4.50
4.25
-6
48.0
44.7
-7
50.0
36.8
-26
6-2
4.50
4.57
+2
46.7
46.5
0
49.5
39.5
-20
7-1
4.50
3.40
-24
48.0
45.2
-6
50.0
44.4
-11
7-2
4.50
4.54
+1
46.7
46.8
0
49.5
43.3
-13
8-1
4.50
3.78
-16
48.0
49.8
+4
50.0
46.6
-7
1-B
0.000
0.000
--
0.00
0.04
--
0.00
0.40
2-B
0.000
0.013
--
0.00
0.09
--
0.00
0.67
3-B
0.000
0.198
--
0.00
0.35
--
0.00
1.23
6-B
0.000
0.000
--
0.00
0.35
--
0.00
0.10
9 The numbers in the "Spiked" column represent the concentration of analyte added to the urine sample and the numbers in the "Meas." column represent the average of two separate GO/MS measurements. Concentrations are in ng/mL. The "% Dif" is the relative difference of the measured concentration from the spiked concentration.
186
Acknowledgments This work was supported by the U. S. Army Medical Research and Development Command, Fort Detrick, MD and the U.S. Army Corp of Engineers, Office of Chemical Agent Demilitarization, Aberdeen Proving Ground, MD. We wish to thank Hoffman-La Roche, Inc. for providing a sample of QNB. We also wish to thank Dr. Walter Benson of
Journal of Analytical Toxicology, Vol. 16, May/June 1992
Table II. Distribution of Relative Uncertainties from Validation Test Results* Analyte QNB
Sample Levelt
GC/MS Prep
Sample Single
1-5
T
12
0
10
16
6-8
10T
7
13
0
15
BA
1-5 6-8
T 10T
8 6
0 1
11 3
13 7
Q
1-5 6-8
T 10T
10 3
10 2
12 10
18 10
* All numbersare expressedin percent of the analytevalue. "GC/MS", "Prep", "Sample", and "Single" refer to the componentsof standarddeviation for the GC/MS measurement,the samplepreparation,the sample-to-samplevariability, and tothe standarddeviationof a singlemeasurement,respectively. t T, 10T= targetleveland 10 timestargetlevel.
the Food and Drug Administration and Dr. Marvin Cohen of Hoffman-La Roche, Inc. for assistance with literature related to this work.
References 1. National Research Council, Disposal of Chemical Munitions and Agents, National Academy Press, Washington, DC, 1984. 2. L. AIbanus. Central and peripheral effects of anticholinergic compounds. Acta PharmacoL ToxicoL 2 8 : 3 0 5 - 2 6 (1970).
3. L.G. Abood. In Psychotropic Drugs, Part 3, F. Hoffmeister and G. Stille, Eds., Springer, Berlin, 1982, pp. 331-47. 4. N.J.M. Birdsall and E.C. Hulme. Biochemical studies on muscarinic acetylcholine receptors. J. Neurochem. 2 7 : 7 - 1 6 (1976). 5. L.A. Hull, D.H. Rosenblatt, and J. Epstein. 3-Quinuclidinyl benzilate hydrolysis in dilute aqueous solution. J. Pharm. Sci. 69: 856-59 (1979). 6. D.H. Rosenblatt, J.C. Dacre, R.N. Shiotsuka, and C.D. Rowlett. Problem definition studies on potential environment pollutants. VIII. Chemistry and toxicology of BZ (3-quinuclidinyl benzilate), Technical Report 7710, U.S. Army Medical Bioengineering Research and Development Laboratory, Fort Detrick, Frederick, MD, Aug. 1977. 7. National Academy of Sciences, Committee on Toxicology, Possible long-term health effects of short-term exposure to chemical agents, National Academy Press, Washington, D.C. June 1982. 8. G.D. Byrd, L.T. Sniegoski, and E. White V. Development of a confirmatory chemical test for exposure to 3-quinuclidinyl benzilate (BZ), National Bureau of Standards, Gaithersburg, MD, Aug. 1987. Avail. NTIS, Order No. AD-A191746. 9. L.T. Sniegoski, G.D. Byrd, and E. White V. Synthesis of 3-quinuclidinol-tsO, benzilic-d5 acid, and 3-quinuclidinyl-tsO benzilate-d5. J. Labelled Compd. Radiopharm. 2 7 : 9 8 3 - 9 3 (1989). 10. B.J. Millard and W.R. Benson. Determination of 3-quinuclidinyl benzilate impurity in Clidinium Bromide by thin-layer chromatography and single ion monitoring after gas chromatography mass spectrometry. Biomed. Mass Spectrom. 6 : 2 7 1 - 7 5 (1979). 11. J.F. Pickup and K. McPherson. Theoretical considerations in stable isotope dilution mass spectrometry for organic analysis. Anal Chem. 48:1885-90 (1976). 12. R.C. Paule and J. Mandel. Consensus values and weighting factors. J. Res. Natl. Bur. Stand (U.S.) 8 7 : 3 7 7 - 8 5 (1982). Manuscript received June 26, 1991.
187
Determination of 3-Quinuclidinyl Benzilate (QNB) and Its Major Metabolites in Urine by Isotope Dilution Gas Chromatography/Mass Spectrometry* G a r y D. B y r d t , R o b e r t C. P a u l e , L a n e C. S a n d e r , L o r n a T. S n i e g o s k i ,
and Edward
W h i t e V**
Center for Analytical Chemistry, National/nstitute of Standards and Technology, Gaithersburg, MD 20899 Howard
T. B a u s u m
U.S. Army Biomedical Research and Development Laborato~ Fort Detrick, Frederick, Maryland 21702
Abstract
I
In response to the scheduled destruction of U.S. military stockpiles of the hallucinogenic agent 3-quinuclldinyl benzilate (QNB), a specific confirmatory test for human exposure to QNB was developed. The amount of the parent compound in the urine as well as the two major metabolites, 3-quinuclidinol (Q) and benzilic acid (BA), was determined because the relationship between QNB dose and levels of QNB and its metabolites in human urine is not known. QNB was determined in urine samples spiked at a target level of 0.5 ng/mL, and the metabolites BA and Q were determined at a target level of 5 ng/mL. The method uses solid-phase extraction to isolate each analyte from the urine and isotope dilution gas chromatography/mass spectrometry for quantitation. Each analyte is converted to its trimethylsilyl derivative for analysis. The analytical method was tested on eight different urine samples spiked with known amounts of the analytes near the target levels, at 10 times the target levels, and blank (unspiked) urine samples. The variabilities in the method are for the most part evenly distributed between three imprecision categories: GC/MS measurement, sample preparation, and the urine samples. The total imprecision (1 standard deviation) of a single measurement is about 15% of the value for each analyte.
Introduction
The United States has initiated efforts to dispose of its aging and obsolete stockpile of chemical agents and munitions (1). One of the materials scheduled for disposal is the hallucinogenic agent 3-quinuclidinylbenzilate (QNB), code-named BZ. The destruction of current stockpiles of QNB has the potential for worker exposure to this material. Consequently, a chemical test to confirm human exposure to QNB has been developed. Information "Presented, in part. at the 35th Annual Conference on Mass Spectrometry and Aflied Topics, Denver, Colorado, May 24-29, 1987, and at the Accuracy in Trace Analysis Symposium, National Bureau of Standards, Gaithersburg, Maryland, September 28-October 1,1987 Certain commercial equipment, instruments, or materials are idenlified in this report to specify adequately the experimental procedures. Such identification does nol imply recommendation or endorsement by the National InstRute of Standards and Technology, nor does it imply that the materials or equipment identified are necessarily the best available for the purpose Present address: R.J. Reynolds Tobacco Co., Bowman-Gray Technical Center, Winston-SaLem, NC "'Author to whom correspondence should be addressed
182
from such a chemical test is not intended to be indicative, but rather confirmatory, because exposure to QNB is most rapidly manifested by symptoms such as tachycardia and mydriasis (2). This paper describes the development of a specific confirmatory chemical test for human exposure to QNB. QNB is one of the most potent anticholinergic psychotomimetics known (3). It is also widely used in research on the brain and nervous system as a specific binding antagonist for the muscarinic class of cholinergic receptors. The effects of very small doses of this agent on the brain and nervous system can produce incapacitation (2,4). QNB undergoes hydrolysis in aqueous buffers to produce benzilic acid (BA) and 3-quhmclidinol (Q) as shown in Figure 1 (5). Little is known of the metabolism and dose/response effects of QNB in the body. Work by the Army (6) indicates that most of the QNB that enters the body is excreted via the kidneys, making urine the meditnn of choice for detection of the unaltered drug and its metabolhes. About 3% of an administered close is excreted unchanged in rats. and Q has been detected in rat urine (6,7), but such informaton has not been reported for humans. Because of the t, nccrt:finties in the relative amounts of QNB, BA, and Q that may appear in human urine, a method that could determine all three in urine is desirable. The maximum no-effect dose of QNB in humans has been estimated to be 0.5 to 1.0 p.g/kg (6,7). The target levels for the detection of QNB and its hydrolysis products, BA and Q, were chosen so as to have a good probability of detecting exposures of 0.5 p.g/kg or higher. It was assumed that 2C/c of a dose of QNB entering the human body will appear in the urine as QNB or its glucuronide conjugate and 40c/~, as BA and Q within the first eight hours. From this, urinary concentrations of 0.5 to 0.7 ng/mL for QNB (including conjugated QNB) and 5 to 7 ng/mL for BA and Q can be calculated (8). These target levels should provide a safety margin adequate for the determination of the analytes in the urine of a person who was exposed to QNB at a level that would produce a detectable physiological effect (8). The method described here uses solid-phase extraction of the urine and isotope dilution in conjunction with gas chromatography/mass spectrometry (GC/MS). Because QNB is hazardous, neither the correctness of the target levels chosen nor the method developed could be tested on urine samples from exposed persons. The method was developed and tested on urine samples spiked with QNB and its metabolites at known concentrations. The results indicate that any exposure to QNB resulting in concentrations of QNB, BA, and Q in the urine at the target levels or higher could be detected.
Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission.
Journal of AnalyticalToxicology,Vol. 16, May/June 1992
Experimental
Warning: Because of the harmful effects of QNB when ingested in small amounts, extreme care should be taken when working with this material. Reagents 3-Quinuclidinyl benzilate (hydrochloride salt) was obtained from Hoffmann-La Roche, Inc. Benzilic acid, 3-quinuclidinol, and N-methyl-N-trimethylsilyltrifluoroacetamide(MSTFA) were obtained from commercial sources and used as supplied. The synthesis of the labeled analogs of the three analytes, 3-180_quin_ uclidinyl ds-benzilate (94 atom % 180 and >99 atom % ds), d 5benzilic acid (>99 atom % ds), and 3-180-quinuclidinol(94 atom % 180), is described elsewhere (9). B-Glucuronidase enzyme (500,000 units per gram solid, pH 3.8) was obtained from Sigma Chemical Co. All other chemicals were reagent grade from commercial sources and used as supplied.
Apparatus Solid-phase extraction Sep-Pak TM cartridges were obtained from Waters Associates. One-mL conical vials fitted with a Teflon-valved cap and a septum seal were used for all derivatization reactions. GC/MS analyses were carried out on a quadrupole mass spectrometer equipped with a 30-m x 0.25-mm i.d., DB-17 fused-silica capillary column connected directly to the ion source of the mass spectrometer. The instrument was operated in the electron ionization mode with an ionizing energy of 70 eV and at unit resolution. Collection and treatment of s a m p l e s
The analytical method was tested on urine specimens taken from eight individuals. Urine samples of 180 mL or greater were collected over a 4-h period from five males and three females. The urine samples were subdivided into 60-mL samples according to the scheme shown in Figure 2. Samples were spiked with QNB, BA, and Q from one of two independently prepared acetonitrile solutions, each of which contained the three analytes. Some samples were spiked with the analytes at concentrations near the target level and others at 10 times this amount. Several samples were left blank. Each 60-mL urine sample was treated as follows: Immediately after the samples were spiked with the unlabeled analytes, the isotopically labeled internal standards were added at concentrations approximately twice the target levels using one of two independently prepared acetonitrile solutions, each of which contained the three analytes. To stabilize the QNB, glacial acetic acid was added (1.5 mL per 60 mL of urine) to adjust the urine to pH 3-4, which is the pH for minimum hydrolysis of QNB in aqueous media at 25~ (5). Sodium azide was added as a preservative at a concentration of 0.05% (w/w) to inhibit bacterial growth. All samples were placed in refrigerated storage (4~ for three days to mimic actual shipping conditions. Before extraction, the urine samples were treated with an enzyme solution to free any glucuronide conjugate of the analytes that might occur in natural samples. The B-glucuronidase enzyme used was from abalone entrails (500,000 units per gram solid) with an optimum activity at pH 3.8, which is in the pH range of the stored urine. To each 60-mL urine sample, 10 mg of the enzyme (approximately 5,000 units) in one mL of water was added before incubation at 37~ for 18 h.
Recovery of analytes from urine and preparation for GC/MS A different solid-phase extraction method was used for each an-
alyte and the derivatized extracts were analyzed by GC/MS. From each 60-mL urine sample, one 20-mL aliquot was used for the QNB analysis and one 20-mL aliquot was used for both the BA and Q analyses. The remaining 20 mL was held in reserve in the event that an analysis failed and had to be repeated. Recoveries of QNB, BA, and Q were determined by comparison of the GC/MS response from samples with the response from standards. Extraction ofQNB. A Cl8 Sep-Pak cartridge was prepared by wetting it with methanol followed by water. A 20-mL urine sample was adjusted to pH 9-10 with 10M NaOH and loaded onto the column at a rate of 5-10 mL/min. The cartridge was washed with 5 mL of water followed by 5 mL of 40% (v/v) CH3CN in water. The sample was eluted from the cartridge with 3 mL of methanol and the eluate was reduced in volume to 1 mL under a gentle stream of nitrogen at 70~ The eluate was transferred to a 1-mL conical vial and blown to dryness. Derivatizing solution (50 pL, 2:1 CH3CN-MSTFA) was added and the sample was capped and heated at 70~ for 3 h. The recovery of QNB at the 0.5-ng/mL level was about 40%. Extraction ofBA. A Cl8 Sep-Pak cartridge was prepared by wetting it with methanol followed by water. A 20-mL urine sample at the storage pH was loaded onto the cartridge at a rate of 5-10 mL/minute. The cartridge was washed with 5 mL of water. The eluate from the 20-mL load of the sample and the 5mL water wash were saved for the Q work-up described below. The cartridge was washed with 3 mL of 15% (v/v) CH3CN in water. The BA was eluted from the cartridge with 3 mL of CHCI3 and the eluate was reduced in volume under a gentle stream of nitrogen at 70~ to approximately 1 mL. The eluate was transferred to a 1-mL conical vial and blown to dryness. Derivatizing solution (50 IaL, 2:1 CH3CN-MSTFA) was added and the sample was capped and heated at 70~ for 15 min. The recovery of BA at the 5-ng/mL level was about 80%. Extraction of Q. The Q work-up used the eluate from the BA load and water wash which was lighter in color than unfiltered urine. This eluate was adjusted to pH 6-7 with concentrated NHaOH. A FlorisilTM Sep-Pak cartridge was prepared by wetting with 10 mL of water and the urine was loaded at a rate of 5-10 mL/minute. The cartridge was washed with 5 mL of water. Removal of the Q was accomplished by forcing 3 x 20 mL of concentrated NH4OH through the cartridge and collecting the effluent. The solvent was removed under reduced pressure on a OH 0
Figure 1. Hydrolysisof QNB.
.............
1~-7
Spiked Concentr~tlon
0
T
~ Ti'
. . . . . . . . . . . . .
SpikedC~can~ratloa ...............
8
r-Ll
I~-I ]'
s
T
~ +~ ........ t
8 1ST 1ST' FLl A A
I--~ T'
0
T
I~]---I T'
e
T
T'
t [Si"iT ]ST' ~ "~8 +i' ]ST
8
r~ A
r~l
Figure 2. Sampling scheme for validation of method. T, T' = target levels. lOT, lOT' = 10 times target levels. Samples 2, 4, and 8 are from female subjects, the others were from males.
183
Journal of AnalyticalToxicology, Vol. 16, May/June 1992
rotary evaporation apparatus while the solution was heated at 70~ The residue was transferred to a concentration tube with 3 • 5 mL of methanol and blown down to 1 mL under nitrogen at 70~ The solution was transferred to a 1-mL conical vial and blown completely dry under nitrogen at 70~ Derivatizing solution (100 HL 2:1 CH3CN-MSTFA) was added and the sample was capped and heated at 70~ for 15 min. The recovery of Q from urine at the 5-ng/mL level was approximately 50%. GC/MS of urine extracts
The GC had a split flow injector port which was maintained at 250~ The chromatographic peaks corresponding to the trimethylsilyl derivatives of QNB, BA, and Q in the spiked samples were identified by comparison of retention times with those produced by analyzing pure standards of each compound and by comparing extracts of blanks with the spiked samples. Solvent blanks were run prior to extract samples to check for analyte carry-over from previous runs. Owing to the complexity of the urine extracts, each extract was analyzed by GC/MS separately and conditions for the individual analytes are described below. Because of the low levels of analyte, the QNB extract was analyzed by splitless mode GC. The oven temperature was initially held at 60~ for 0.5 min, then taken up to 225~ at 40~ and held there for 1 min. The temperature was then increased to 273~
at 4~ The mass spectrometer was set to monitor m/z 255 for the QNB derivative and m/z 260 for the labeled QNB derivative. The BA extract was analyzed by the use of split injection with a split flow ratio of 15:1. The GC column was initially held at 175~ for 2 min and then programmed up to 201~ at a rate of 4~ The mass spectrometer conditions for the BA derivative were identical to those for the QNB derivative. The Q extract was analyzed by the use of split injection with a split flow ratio of 15:1. The GC column was initially held at 75~ for 2 min and then programmed up to 115~ at a rate of 4~ The mass spectrometer was set to monitor m/z 199 for the Q derivative and ,u'z 201 for the labeled Q derivative. Response factor solutions consisting of known amounts of all three labeled and unlabeled analytes dissolved in the derivatizing agent were prepared. Concentrations of the labeled materials were approximately twice those of the unlabeled analytes. For any series of samples run in a day, two independently prepared response factor solutions were each analyzed, once prior to and once just after, each series for a given analyte. The response factor used for a series was an average of the four measurements. Each sample extract was analyzed twice by GC/MS.
Results and Discussion TMS
900
0 0
(C6Hs)2C-O-TMS ,
10001
A GC/MS method for the quantitative determination of QNB, potentially present as an impurity in a prescription drug, has been
B00 700
600 50{} 400
TK4S *
300
7/3
o)
Ion
255.80
amu.
CTH,~N *
t/z o
200
,00] I 'c
409
366
QNB-TMS
200
100
Figure 3. Mass spectrum of TMS derivative of ONB.
Ion
260.00
amu.
LQNB-TMS j-
10001 900 800
TUS
7~
~ss ! (CiHs)iC-0-TMS ,
9
TUS 0
0
!3
7~0~
~4
Time
(min.)
15
!S
I
I(M
-- CHj
147 /
'~176 1
CO)"
L?
..LI_ t
00
b)
200 Mass/Charge
Ion
255.00
amu.
Figure 4. Mass spectrum of TMS derivative of BA.
1 0 00" 900" 800" 700~ 600"
Ion J
J
(M -
amu.
LQNB-TMS
TMSOCH2) ~ 9/6
C6HI~OS i 9
1/29
50D
400"
260.20
TMS +
(M- CH3)" M"
CTHI2N. 0
300"
45
184
199
~
P
:3
14
Time
(mln.)
15
16
200 100
~a=s,Ch."ge
Figure 5. Mass spectrum of TMS derivative of Q.
184
Figure 6. SIM chromatogram of m/z 255 and m/z 260 from extracts of urine a) spiked with 0.5 ng/mL ONB and b) a urine blank. QNB-TMS is the derivatized flNB and LQNB-TMS is the derivatized labeled internal standard. In both urine samples the labeled internal standard is 0.90 ng/mL.
Journal of Analytical Toxicology, Vol. 16, May/June 1992
reported (10). That method used extraction and thin-layer chromatography to separate QNB from the matrix, followed by GC/MS of the trimethylsilyl (TMS) derivative. The article also indicated that underivatized QNB decomposes to benzophenone when exposed to the heated surfaces in the GC injector port. BA can also decompose, through decarboxylation, and does not chromatograph as the acid. Therefore, the analytes, QNB, BA, and Q, were converted to their TMS derivatives (QNB-TMS, BA-TMS, and Q-TMS) to improve their chromatographic behavior and stability. N-Methyl-N-trimethylsilyltrifluoroacetamide (MSTFA) was selected as the derivatizing agent for the analytes to minimize chromatographic peak tailing of the solvent front because QTMS elutes at relatively low GC column temperatures. While BA and Q readily dissolved in MSTFA, QNB proved to be only partially soluble in this material. A mixture of 2:1 acetonitrileMSTFA adequately dissolved the QNB, as well as the other two analytes. BA and Q were completely derivatized in less than 15 rain following addition of the derivatizing solution, whereas QNB required 3 h at 70~ to achieve more than 98% derivatization. The mass spectra of the derivatized compounds are shown in Figures 3-5. The mass spectra of the QNB-TMS and the BATMS derivatives both have an ion at m/z 255 as the base peak which has the stable structure shown. This ion was chosen for selected ion monitoring (SIM) of both species. For the labeled materials, this ion is shifted to m/z 260. The spectrum of QNB-TMS also contains characteristic ions at m/z 1I0 and 126, which come from the quinuclidinyl portion of the molecule. A weak molecular
o)
Ion
255.Q0
ion, M +., for QNB-TMS at m/z 409 is observed. BA-TMS has two TMS groups following derivatization and, while no molecular ion is observed, two high mass fragment ions at m/z 329 and 357 occur that are characteristic of this molecule. Q-TMS undergoes extensive fragmentation, as shown in Figure 5. The molecular ion at m/z 199 has a relative intensity greater than 50% of the base peak and was selected as the representative ion for SIM. This ion is shifted to m/z 201 for the labeled material. Other ions of interest include (M - CH3) § at m/z 184, the quinuclidinyl ion at m/z 110, and (M - TMSOCH2) + at m/z 96. The composition of the ion at m/z 129 was determined to be C 6 H I 3 O S i § by exact mass measurement on a double-focusing high resolution mass spectrometer using a direct probe to introduce the sample. The structure of this ion is not known. Figures 6-8 show ion chromatograms for each derivatized analyte in (a) a urine extract with the analyte at the target level and (b) a urine blank. No significant interferences were found in any of the blank urines analyzed. Gradual decreases in retention times for the analytes, most evident for BA in Figure 7, were observed over the course of the validation test, presumably the consequence of loss of liquid phase from the colunm resulting from the large number of samples and the baking of the column needed to minimize background. The analytical method developed was tested on eight urine specimens from different individuals according to the scheme in Figure 2. This scheme was designed to provide needed information using a minimum number of samples and measurements. The following information was obtained: the variability of the sample preparation, the urine-to-urine variability, and the variability of the
tm,.
o) Ion
268.08
Ion
IB9.BB
amu.
JV Ion
6.5
mmu.
7.0 Time
7.5 (mln.)
8.0
8.5
201.B8
9.8 !8
b)
Ion
255.00
- -
~ 7.0
268.88
Time
(mtn.)
| I
12
lmu.
b)
Ion
mmu.
Ton
199.00
Ion
281.88
Imu.
amu.
~_~LBA-TMS
7.5 Tlme
8.0 (mln.)
8.5
Fioure 7. SIM chromatogram of m/z255 and m/z260 from extracts of urine a) spiked with 4.80 ng/mL BA and b) a urine blank. BA-TMS is the defivatizedBA and LBA-TMS is the derivatizedlabeled internal standard. In both urine samples the labeledinternal standard is 8.55 ng/mL
9
1~
amu.
Ttme
(mln.)
1!
12
Figure 8. SIM chromatogram of m/z 199 and m/z 201 from extracts of urine a) spiked with 500 ng/mL ONB and b) a urine blank O-TMS is the derivatized Q and LQ-TMS is the derivatized labeled internal standard. In both urine samples the labeled internal standard is 10.22 ng/mL
185
Journal of Analytical Toxicology, Vol. 16, M a y / J u n e 1992
response factor is not correct because the analyte and labeled internal standard both contribute to the signals at the masses monitored (11). The error at the target level is about 5% and thus negligible compared to other sources of error, but for the measurements at 10 times the target level the values are about 30% low. This is satisfactory for the present purpose of determining whether or not the analyte is or is not present above the target level, but the method for Q would require modification if the target level were different or if the method were to be adapted for use as an assay. For all three analytes the measured value was considered good at these concentration levels. Based on the established target levels, no false positives occurred for any of the four blank urines for any of the three analytes. The blank with the highest relative response among all the analytes, 3B for QNB, gave a measured value that was less than half the target value. For samples spiked near the target level, no measured value was less than 80% of the spiked value. On this basis, no false negatives were apparent. Of the 54 extractions performed for the validation test, 5 failed the initial work-up (no internal standard was recovered). However, none of these failed when repeated on the back-up urine sample incorporated into the method. As a "worst case" example for all analytes spiked at the target level, the maximum percent difference between measured and spiked values was +31%. Statistical calculations yielded, for each analyte, the sample averages, the three components of standard deviation (GC/MS measurement, sample preparation, and urine-to-urine variability), and the standard deviation of a single measurement. Samples 1-4, 6, and 7 had both duplicate sample preparations and GC/MS measurements on each sample so that analyses of variance (ANOVA) could be made for the corresponding components for these samples. These results, along with those from samples 5 and 8, were then used in an ANOVA for unbalanced data to calculate the components of standard deviation between urine samples (12). This was done for each analyte and each spike level (T and 10T). The 0 components of standard deviation (expressed as percent of the analyte value) arc listed in Spiked Meas. % Dil Table II. This table shows that the impreci4.22 -16 sions are more or less evenly distributed be5.00 5.05 +2 tween the three imprecision categories 4.95 5.27 +5 (GC/MS measurement, sample preparation, 5.00 6.22 +26 and urine sample). These effects are relatively 4.95 5.30 + 6 constant, considering the limited number of 5.00 4.43 -11 samples and measurements. This table also 4.95 shows that the total imprecision of a single 5.10 +2 5.00 measurement is about 15% for each analyte, 4.95 4.76 -4 and that it is not greatly different between the 5.00 6.55 +31 T and 10T spike levels.
GC/IVIS measurements at the target level and 10 times the target level. Information was desired at the higher concentrations because these levels would be expected in the event of an exposure producing a detectable physiological effect. Table I summarizes the data for all samples for the analytes QNB, BA, and Q. The sub-sample numbers correspond to those in the diagram in Figure 2. The values given for the spiked samples have not been corrected for the blank because no blank would be available for samples resulting from exposure to QNB. The upper part of the table contains the results for samples that were spiked near the target level for each analyte. The middle part of the table contains the results for the samples that were spiked at approximately 10 times the target level for each analyte. The lower part of the table contains the results for the blanks. For urine samples spiked with QNB at the target level, Table I shows that the measured QNB concentration differed from the spiked value by -18% to +14%. At 10 times the target level the measured value differed by -24% to +2% from the spiked value. The highest reading on a blank gave a value 40% of the target value for QNB (sub-sample 3B). In urine samples with BA spiked near the target level, the measured BA concentration differed from the spiked value by -15 to +19%. At 10 times the target value there was less scatter of the measurements with measured values differing by -7 to +4% from the spiked value. The four blanks measured gave values of 1 to 7% of the target value, again with sample 3B giving the highest reading. Q, when spiked near the target level, was measured at concentrations differing by -16 to +31% from the spiked value. At 10 times the target level, the measured value differed from the spiked value by -26 to -7%. The four blanks were measured from 2 to 25% of the target value with sample 3B again giving the highest reading. For Q the calculation of the concentration from a single
Table I. Summary of Validation Test Results* ONB Subsample Spiked Meas. % Dif
BA Spiked Meas. % Dif
1-1
0.450
0.487
+8
4.80
4.14
1-2
0.450
0.514
+14
4.67
4.52
-14 -3
2-1
0.450
0.403
-10
4.80
4.23
-12
2-2
0.450
0,370
-18
4.67
4.25
-9
3-1
0.450
0.471
+5
4.80
5.70
+19
3-2
0.450
0.451
+0
4.67
5.36
+15
4-1
0.450
0.386
-14
4.80
4.08
-15
4-2
0.450
0.373
-17
4.67
4.53
-3
5-1
0.450
0.465
+3
4.80
4.80
0
6-1
4.50
4.25
-6
48.0
44.7
-7
50.0
36.8
-26
6-2
4.50
4.57
+2
46.7
46.5
0
49.5
39.5
-20
7-1
4.50
3.40
-24
48.0
45.2
-6
50.0
44.4
-11
7-2
4.50
4.54
+1
46.7
46.8
0
49.5
43.3
-13
8-1
4.50
3.78
-16
48.0
49.8
+4
50.0
46.6
-7
1-B
0.000
0.000
--
0.00
0.04
--
0.00
0.40
2-B
0.000
0.013
--
0.00
0.09
--
0.00
0.67
3-B
0.000
0.198
--
0.00
0.35
--
0.00
1.23
6-B
0.000
0.000
--
0.00
0.35
--
0.00
0.10
9 The numbers in the "Spiked" column represent the concentration of analyte added to the urine sample and the numbers in the "Meas." column represent the average of two separate GO/MS measurements. Concentrations are in ng/mL. The "% Dif" is the relative difference of the measured concentration from the spiked concentration.
186
Acknowledgments This work was supported by the U. S. Army Medical Research and Development Command, Fort Detrick, MD and the U.S. Army Corp of Engineers, Office of Chemical Agent Demilitarization, Aberdeen Proving Ground, MD. We wish to thank Hoffman-La Roche, Inc. for providing a sample of QNB. We also wish to thank Dr. Walter Benson of
Journal of Analytical Toxicology, Vol. 16, May/June 1992
Table II. Distribution of Relative Uncertainties from Validation Test Results* Analyte QNB
Sample Levelt
GC/MS Prep
Sample Single
1-5
T
12
0
10
16
6-8
10T
7
13
0
15
BA
1-5 6-8
T 10T
8 6
0 1
11 3
13 7
Q
1-5 6-8
T 10T
10 3
10 2
12 10
18 10
* All numbersare expressedin percent of the analytevalue. "GC/MS", "Prep", "Sample", and "Single" refer to the componentsof standarddeviation for the GC/MS measurement,the samplepreparation,the sample-to-samplevariability, and tothe standarddeviationof a singlemeasurement,respectively. t T, 10T= targetleveland 10 timestargetlevel.
the Food and Drug Administration and Dr. Marvin Cohen of Hoffman-La Roche, Inc. for assistance with literature related to this work.
References 1. National Research Council, Disposal of Chemical Munitions and Agents, National Academy Press, Washington, DC, 1984. 2. L. AIbanus. Central and peripheral effects of anticholinergic compounds. Acta PharmacoL ToxicoL 2 8 : 3 0 5 - 2 6 (1970).
3. L.G. Abood. In Psychotropic Drugs, Part 3, F. Hoffmeister and G. Stille, Eds., Springer, Berlin, 1982, pp. 331-47. 4. N.J.M. Birdsall and E.C. Hulme. Biochemical studies on muscarinic acetylcholine receptors. J. Neurochem. 2 7 : 7 - 1 6 (1976). 5. L.A. Hull, D.H. Rosenblatt, and J. Epstein. 3-Quinuclidinyl benzilate hydrolysis in dilute aqueous solution. J. Pharm. Sci. 69: 856-59 (1979). 6. D.H. Rosenblatt, J.C. Dacre, R.N. Shiotsuka, and C.D. Rowlett. Problem definition studies on potential environment pollutants. VIII. Chemistry and toxicology of BZ (3-quinuclidinyl benzilate), Technical Report 7710, U.S. Army Medical Bioengineering Research and Development Laboratory, Fort Detrick, Frederick, MD, Aug. 1977. 7. National Academy of Sciences, Committee on Toxicology, Possible long-term health effects of short-term exposure to chemical agents, National Academy Press, Washington, D.C. June 1982. 8. G.D. Byrd, L.T. Sniegoski, and E. White V. Development of a confirmatory chemical test for exposure to 3-quinuclidinyl benzilate (BZ), National Bureau of Standards, Gaithersburg, MD, Aug. 1987. Avail. NTIS, Order No. AD-A191746. 9. L.T. Sniegoski, G.D. Byrd, and E. White V. Synthesis of 3-quinuclidinol-tsO, benzilic-d5 acid, and 3-quinuclidinyl-tsO benzilate-d5. J. Labelled Compd. Radiopharm. 2 7 : 9 8 3 - 9 3 (1989). 10. B.J. Millard and W.R. Benson. Determination of 3-quinuclidinyl benzilate impurity in Clidinium Bromide by thin-layer chromatography and single ion monitoring after gas chromatography mass spectrometry. Biomed. Mass Spectrom. 6 : 2 7 1 - 7 5 (1979). 11. J.F. Pickup and K. McPherson. Theoretical considerations in stable isotope dilution mass spectrometry for organic analysis. Anal Chem. 48:1885-90 (1976). 12. R.C. Paule and J. Mandel. Consensus values and weighting factors. J. Res. Natl. Bur. Stand (U.S.) 8 7 : 3 7 7 - 8 5 (1982). Manuscript received June 26, 1991.
187
