Anal. Chem.

2004, 76, 2453-2461

A Liquid Chromatography-Tandem Mass

Spectrometry Multiresidue Method for
Quantification of Specific Metabolites of
Organophosphorus Pesticides, Synthetic
Pyrethroids, Selected Herbicides, and DEET in
Human Urine
Anders O. Olsson,* Samuel E. Baker, Johnny V. Nguyen, Lovisa C. Romanoff, Simeon O. Udunka,
Robert D. Walker, Kathryn L. Flemmen, and Dana B. Barr

Division of Laboratory Sciences, National Center for Environmental Health, Centers for Disease Control and Prevention,
4770 Buford Highway, N.E., MS F-17, Atlanta, Georgia 30360

The ability to estimate low-dose human exposure to estimated use of 3000-4000 tons in those applications, giving an
commonly used pesticides often is requested in epide- average domestic use per capita of ∼15 g/year. For the same year,
miologic studies. Therefore, fast and robust methods are the most abundantly applied pesticide in U.S. agriculture was
necessary that can measure many analytes in the same atrazine, with applications of 35 000 tons. The widespread use of
sample. We have developed a method for high-throughput pesticides and the scientific interest in potential adverse health
analysis of 19 markers of commonly used pesticides in effects of pesticide exposure have increased the demand for fast
human urine. The analytes were seven specific metabo- and robust analytical methods for measuring markers of pesticide
lites of organophosphorus pesticides, five metabolites of exposure.
synthetic pyrethroids, six herbicides or their metabolites, Many of the methods available today focus on specific pesticide
and one insect repellant. Human urine (2 mL) was spiked classes or individual pesticides or metabolites.2-10 Few methods
with stable isotopically labeled analogues of the analytes, are available for the analysis in the low-nanogram per milliliter
enzymatically hydrolyzed, extracted using solid-phase range of markers of several classes of pesticides in the same
extraction, concentrated, and analyzed using high-perfor- sample.5-11 However, in epidemiologic studies of nonoccupation-
mance liquid chromatography-tandem mass spectrom- ally exposed persons, measuring biomarkers of many different
etry. The sample was divided into two portions and pesticides often is important for obtaining an accurate representa-
analyzed on two different mass spectrometers, one using tion of a person’s exposure. We previously accomplished this in
atmospheric pressure chemical ionization (APCI) and the our laboratory by analyzing aliquots of the same urine sample
other using turbo ion spray atmospheric pressure ioniza- using a number of different methods.12
tion (TIS). All analytes except the pyrethroid metabolites
were analyzed using APCI. The detection limits for all (1) Donaldson, D.; Kiely, T.; Grube, A. 1998 and 1999 market estimates.
analytes ranged from 0.1 to 1.5 ng/mL of urine, with the Pesticides industry sales and usage report. U.S. Environmental Protection
Agency, Washington, DC, 2002.
majority (17) below 0.5 ng/mL. The analytical precision (2) Angerer, J.; Ritter, A. J. Chromatogr., B 1997, 695, 217-226.
for the different analytes, estimated as both the within- (3) Aprea, C.; Sciarra, G.; Bozzi, N. J. Anal. Toxicol. 1997, 21, 262-267.
day and between-day variation, was 3-14 and 4-19%, (4) Aprea, C.; Stridori, A.; Sciarra, G. J. Chromatogr., B 1997, 695, 227-236.
(5) Baker, S. E.; Barr, D. B.; Driskell, W. J.; Beeson, M. D.; Needham, L. L. J.
respectively. The extraction recoveries of the analytes Exp. Anal. Environ. Epidemiol. 2000, 10, 789-798.
ranged from 68 to 114%. The throughput, including (6) Bravo, R.; Driskell, W. J.; Whitehead, R. D.; Needham, L. L.; Barr, D. B. J.
calibration standards and quality control samples, is ∼50 Anal. Toxicol. 2002, 26, 245-252.
(7) Leng, G.; Kühn, K.-H.; Leng, A.; Gries, W.; Lewalter, J.; Idel, H. Chro-
samples a day. However, the analysis time with the TIS matographia 1997, 46, 265-274.
application is much shorter, and if only pyrethroid me- (8) Olsson, A. O.; Nguyen, J. V.; Sadowski, M. A.; Barr, D. B. Anal. Bioanal.
tabolite data are of interest, the throughput can be Chem. 2003, 376, 808-815.
(9) Sancho, J. V.; Pozo, O. J.; Hernandez, F. Rapid Commun. Mass Spectrom.
increased to 100-150 samples/day. 2000, 14, 1485-1490.
(10) Sancho, J. V.; Pozo, O. J.; Lopez, F. J.; Hernandez, F. Rapid Commun. Mass
In 1999, an estimated 415 000 tons of conventional pesticides Spectrom. 2002, 16, 639-645.
(11) Hill, R. H.; Shealy, D. B.; Head, S. L.; Williams, C. C.; Bailey, S. L.; Gregg,
was applied in the United States.1 The most commonly used M.; Baker, S. E.; Needham, L. L. J. Anal. Toxicol. 1995, 19, 323-329.
pesticide for home and gardens was the herbicide 2,4-D with an (12) Second National Report on Human Exposure to Environmental Chemicals.
http://www.cdc.gov/exposurereport/. U.S. Centers for Disease Control and
* To whom correspondence should be addressed. E-mail: Prevention, National Center for Environmental Health, Division of Laboratory
[email protected]. Telephone: +1-770-488-4463. Fax: +1-770-488-0142. Sciences, Atlanta, 2003.

10.1021/ac0355404 Not subject to U.S. Copyright. Publ. 2004 Am. Chem. Soc. Analytical Chemistry, Vol. 76, No. 9, May 1, 2004 2453
Published on Web 04/03/2004
Table 1. Analyzed Markers, Their Abbreviations, Parent Pesticide and Class, and Source for Native and Stable
Isotopically Labeled Standards

indicator of exposure source of standardb

analyte name abbrev to (pesticide classa) native labeled
atrazine mercapturate; N-acetyl-S-[4-(ethylamino)- ATZ atrazine (TAH) UCD CIL
6-[(1-methylethyl)amino]-1,3,5-triazin-2-yl]-L-cysteine
acetochlor mercapturate; N-acetyl-S-[2-(2-ethyl-6- ACE acetochlor (CAH) CIL CIL
methylphenyl)(ethoxymethyl)amino]-2-oxoethyl-L-cysteine
alachlor mercapturate; N-acetyl-S-[2-(2,6-diethyl- ALA alachlor (CAH) CIL CIL
phenyl)(methoxymethyl)amino]-2-oxoethyl-L-cysteine
metolachlor mercapturate; N-acetyl-S-[2-(2-ethyl-6-methyl- MET metolachlor (CAH) CIL CIL
phenyl)(2-methoxy-1-methylethyl)amino]-2-oxoethyl-L-cysteine
2,4-dichlorophenoxyacetic acid 2,4-D 2,4-D (PAAH) Aldrich CIL
2,4,5-trichlorophenoxyacetic acid 2,4,5-T 2,4,5-T (PAAH) Aldrich CIL
5-chloro-1-isopropyl-[3H]-1,2,4-triazol-3-one CIT isazophosc (OPI) CIL CIL
3-chloro-4-methyl-7-hydroxycoumarin CMHC coumaphosc (OPI) CIL CIL
2-diethylamino-6-methyl pyrimidin-4-ol DEAMPY pirimiphosc (OPI) CS CIL
2-[(dimethoxyphosphorothioyl)sulfanyl]succinic acid MDA malathionc (OPI) Cheminova CADHS
2-isopropyl-6-methyl-4-pyrimidiol IMPY Dialzinonc (OPI) CS CIL
4-nitrophenol PNP parathionc (OPI) Aldrich CIL
3,5,6-trichloro-2-pyridinol TCPY chlorpyriphosc (OPI) CS CIL
cis-3-(2,2-dibromovinyl)-2,2-dimethylcyclo- DBCA deltamethrin (PI) EQ lab
propane-1-carboxylic acid
cis- and trans-3-(2,2-dichlorovinyl)-2,2-dimethyl- cis-/trans- cyfluthrin, permethrin, JA CIL (only
cyclopropane-1-carboxylic acids DCCA cypermethrin (PI) trans)
4-fluoro-3-phenoxybenzoic acid 4-F-3-PBA cyfluthrin (PI) JA
3-phenoxybenzoic acid 3-PBA 10 of 18 commercially EQ lab CIL
available pyrethroids
in the U.S. (PI)
N,N-diethyl-m-toluamide DEET DEET (IR) CS CIL
aTAH, triazine herbicides; CAH, chloroacetoanilide herbicides; PAAH, phenoxyacetic acid herbicides; OPI, organophosphate insecticides; PI,
pyrethroid insecticides; IR, insect repellant. b Aldrich, Aldrich Chemical Co. (Milwaukee, WI); CADHS, California State Department of Health
Services (Berkeley, CA); Cheminova, Cheminova Agro A/S (Lemvig, Denmark); CIL, Cambridge Isotope Laboratories (Andover, MA); CS, Chem
Services (West Chester, PA); EQ lab, EQ Laboratories (Augsburg, Germany); JA, gift from Dr. Jürgen Angerer (Erlangen University, Erlangen,
Germany); UCD, gift from Dr. Bruce Hammock (University of California at Davis, CA). c Or their methyl counterparts.

To achieve a more cost-effective analysis, we sought to combine of isazophos has never been confirmed in urine. We measured
three previously reported analytical methods5,8,13 and to include five metabolites to evaluate exposure to the synthetic pyrethroids.
additional analytes. The analytes represent pesticides from several Two of these metabolites are specific for cyfluthrin and delta-
classessorganophosphorus insecticides, synthetic pyrethroid methrin;18,19 two are common metabolites for three pyrethroidss
insecticides, triazine herbicides, chloroacetanilide herbicides, cyfluthrin, permethrin, and cypermethrin;18,20,21 and one is a
phenoxyacetic acid herbicides, and the topical insect repellant N,N- possible metabolite of at least 10 different synthetic pyrethroids
diethyl-m-toluamide (DEET) (Table 1). All classes are of interest commercially available in the United States. All five pyrethroid
because of their extensive use or potential adverse health effects. metabolites have been used as markers of exposure in human
Although certain uses of organophosphorus insecticides are being studies.2,22,23 Furthermore, we measured urinary markers of
phased out or revised because of the pesticides’ relatively high exposure to six herbicides: atrazine, acetochlor, alachlor, and
toxicity, some organophosphorus insecticides are still used in large metolachlor by measuring their mercapturic acid conjugates and
quantities in agriculture. The use of pyrethroid insecticides has 2,4-D and 2,4,5-T as the parent herbicides. The primary human
increased dramatically during the last 20 years, both for domestic metabolite of atrazine is reportedly the mecapturate conjugate;24,25
use and in agriculture. Chloroacetanilide herbicides are some of both alachlor and metolachlor mercapturate have been identified
the most used conventional pesticides in U.S. agriculture; ac- in urine from exposed persons.26,27 Acetochlor has not been shown
etochlor, metolachlor, and alachlor make up ∼10% of the total to form the mercapturic acid conjugate; however, because of its
agricultural amount used.1 The phenoxyacetic acid herbicide, 2,4- structural similarities to alachlor and metolachlor, acetochlor is
D, and the triazine herbicide, atrazine, are the most common
pesticides for domestic and agricultural use, respectively. DEET (14) Bradway, D. E.; Talaat, M. S. J. Agric. Food Chem. 1977, 25, 1342-1344.
(15) Brealey, C. J.; Lawrence, D. K. J. Chromatogr. 1979, 168, 461-469.
is probably the most commonly applied topical insect repellent. (16) Nolan, R. J.; Rick, D. L.; Freshour, N. L.; Saunders, J. H. Toxicol. Appl.
To represent these pesticide classes, we selected 19 urinary Pharmacol. 1984, 73, 8-15.
markers of exposure (Figure 1). Seven of these represent (17) Oneto, M. L.; Basack, S. B.; Kesten, E. M. Sci. Justice 1995, 35, 207-211.
(18) Leng, G.; Leng, A.; Kühn, K.-H.; Lewalter, J.; Pauluhn, J. Xenobiotica 1997,
organophosphorus insecticides. The markers are specific hydroly- 27, 1273-1283.
sis products of chlorpyriphos, coumaphos, diazinon, isazophos, (19) Moretti, M.; Villarini, M.; Scassellati-Sforzolini, G.; Pasquini, R.; Monarca,
malathion, parathion, and pirimiphos or their methyl counterparts. S. Arch. Environ. Contam. Toxicol. 1997, 33, 323-328.
(20) Eadsforth, C. V.; Baldwin, M. K. Xenobiotica 1983, 13, 67-72.
Most of these markers have been detected in human or animal (21) Woollen, B. H.; Marsh, J. R.; Laird, W. J. D.; Lesser, J. E. Xenobiotica 1992,
urine,5,14-17 but to our knowledge, the specific hydrolysis product 22, 983-991.
(22) Heudorf, U.; Angerer, J. Environ. Health Perspect. 2001, 109, 213-217.
(13) Baker, S. E.; Olsson, A. O.; Barr, D. B. Arch. Environ. Contam. Toxicol. 2004, (23) Leng, G.; Ranft, U.; Sugiri, D.; Hadnagy, W.; Berger-Preib, E.; Idel, H. Int.
46, 281-288. J. Hyg. Environ. Health 2002, 206, 85-92.

2454 Analytical Chemistry, Vol. 76, No. 9, May 1, 2004

Figure 1. Structures of the measured pesticide biomarkers. Abbreviations are according to Table 1.

also probably excreted, in part, as a mercapturate conjugate. The Quality Control Materials. Urine was collected from multiple
marker for DEET exposure is the parent compound. DEET has anonymous donors, combined, diluted with water (1:1 v/v) to
been reported in urine from exposed rats.28 Altogether, these reduce endogenous concentrations of the analytes of interest, and
target compounds represent some of the most used pesticides in mixed overnight at 20 °C. After being pressure filtered through a
the United States and give a general, though not total, evaluation 0.45-µm SuporCap-100 Capsule (Pall Corp., Ann Arbor, MI), the
of a person’s pesticide exposure. urine was divided into three pools. The first quality control (QC)
pool (low concentration) was spiked with the native materials to
MATERIALS AND METHODS yield an approximate analyte concentration of 5 ng/mL (QCL).
Chemicals. All solvents used were of analytical grade. Aceto- The second QC pool (high concentration) was spiked with the
nitrile was obtained from Burdick & Jackson Inc. (Muskegon, MI)
native materials to yield an approximate analyte concentration of
and methanol from Caledon Laboratories (Georgetown, ON.,
15 ng/mL (QCH). The third pool was not spiked (blank urine).
Canada). The glacial acetic acid and sodium acetate were
This pool was used as matrix material for calibration standards
purchased from J.T. Baker (Phillipsburg, NJ). Deionized water
and blanks.
was organically and biologically purified with a NANOpure Infinity
UF from Barnstead International (Dubuque, IA). Nitrogen was Standard Preparation. Approximately 0.5 mg of each labeled
purchased from Airgas Inc. (Radnor, PA). Gases used by the Sciex standard was mixed and diluted with acetonitrile to obtain a
instrument were generated from a high-purity nitrogen and zero concentration of 20 µg/mL. This solution was further diluted to
air gas generator (NM20ZA, Peak Scientific Instruments Ltd., give an approximate concentration of the individual labeled
Chicago, IL) and had a minimum purity of 99.999%. The measured compounds of 1 µg/mL. This solution was used to spike the
analytes are listed in Table 1, as well as the specific source for samples.
the native and isotopically labeled standards from where they were Individual stock solutions (∼200 µg/mL in acetonitrile) of the
obtained. unlabeled analytes were prepared from pure substance, except
for DBCA, which was purchased as a 10 µg/mL acetonitrile
(24) Buchholz, B. A.; Fultz, E.; Haack, K. W.; Vogel, J. S.; Gilman, S. D.; Gee, S.
solution. From the stock solutions, adequate volumes were taken
J.; Hammock, B. D.; Hui, X.; Wester, R. C.; Maibach, H. I. Anal. Chem.
1999, 71, 3519-3525. by pipet, and eight calibration standard solutions were prepared
(25) Lucas, A. D.; Jones, A. D.; Goodrow, M. H.; Saiz, S. G.; Blewett, C.; Seiber, with the following concentrations of the individual analytes: 0.020,
J. N.; Hammock, B. D. Chem. Res. Toxicol. 1993, 6, 107-116.
(26) Driskell, W. J.; Hill, R. H.; Shealy, D. B.; Hull, R. D.; Hines, C. J. Bull. Environ. 0.040, 0.080, 0.20, 0.40, 0.80, 2.0, and 4.0 µg/mL. DBCA was not
Contam. Toxicol. 1996, 56, 853-859. added to the two highest standard solutions because of the diluted
(27) Driskell, W. J.; Hill, R. H. Bull. Environ. Contam. Toxicol. 1997, 58, 929-
933.
stock solution. To prepare a calibration curve, 25 µL of each
(28) Taylor, W. G.; Spooner, R. W. J. Agric. Food Chem. 1990, 38, 1422-1427. standard solution was added to each 2-mL blank urine sample.
Analytical Chemistry, Vol. 76, No. 9, May 1, 2004 2455
Table 2. Optimized Precursor/Product Ion Pairs, Stable Isotope Label of Internal Standard, Declustering Potential
(DP), Collision Cell Exit Potential (CXP), and Collision Offset Energy (CE) Settings for the Different Analytes on the
Sciex API4000 Instrument

isotope precursor ion product ion DP CE CXP

compound label native labeled native labeled (V) (V) (V)
trans-DCCA 13C
3 207 210 35 35 -50 -30 -5
cis-DCCA a 207 a 35 a -50 -30 -5
DBCA a 294 a 79 a -35 -18 -3
3-PBA 13C
6 213 219 93 99 -55 -28 -7
4-F-3-PBA b 231 b 93 b -60 -36 -17
a Labeled trans-DCCA used as internal standard. b Labeled 3-PBA used as internal standard.

Biological Samples. Urine samples were frozen within 4 h acid was used for the analysis on the TSQ 7000. The chromato-
of collection and stored at -20 °C before analysis. All protocols graphic separation was performed on a Betasil phenyl column (5-
were reviewed and approved by a human subjects review com- µm particle size, 100-Å pore size, and 4.6 mm i.d. × 100 mm
mittee and complied with all institutional guidelines for the length) from Keystone Scientific Inc. (Bellefonte, PA). The flow
protection of human subjects. rate was 1.0 mL/min, and the injection volume was 10 µL. The
Instruments. Samples were concentrated to dryness using a column temperature was kept at 25 °C during the analysis.
TurboVap LV evaporator (Zymark, Farmingham, MA), where the Also, for the analysis on the Sciex API4000, an isocratic elution
water temperature was set to 40 °C and nitrogen (10 psi pressure) was used. The solvent mixture was 51% acetonitrile in water with
was used as the evaporating gas. The high-performance liquid 0.1% acetic acid. The column was a Betasil C18 column (5-µm
chromatography (HPLC) was performed on an Agilent 1100 particle size, 100-Å pore size, and 1 mm i.d. × 100 mm length)
system (Agilent Tech., Waldbronn, Germany) consisting of a from Keystone Scientific Inc. The flow rate was 0.05 mL/min, and
binary pump, a degasser, an autosampler, and a temperature-stable the injection volume was 2 µL. The column was kept at 35 °C.
column compartment. For the atmospheric pressure chemical Mass Spectrometry Operating Conditions. The Sciex
ionization (APCI) application, a TSQ 7000 triple quadrupole mass API4000 was operated in the multiple reaction monitoring mode
spectrometer (ThermoQuest, San Jose, CA) was used. For the using negative ion TIS. Zero air was used for collision-activated
turbo ion spray atmospheric pressure ionization (TIS) application, dissociation, nebulizer, heater, and exhaust gases. The curtain
a Sciex API4000 triple quadrupole mass spectrometer (Applied gas was nitrogen. Ion source, collision, and curtain gases had
Biosystems/MDS Sciex, Foster City, CA) was used. optimum pressures of 16, 6, and 10 psi, respectively. The heater
Sample Preparation and Cleanup. Two milliliters of urine gas (450 °C) was operated with a pressure of 16 psi. The ion spray
was taken for analysis, and a mixture of the available isotopically current was -4.5 kV, and the entrance potential was -10 V.
labeled internal standards was added (25 µL) giving an ∼12.5 ng/ Optimized filter parameters for the precursor/product ion pairs
mL concentration of the standards in the urine. To hydrolyze are listed in Table 2. The samples were analyzed in negative
possible glucuronide or sulfate conjugated metabolites, â-glucu- ionization mode, and all analytes were measured in one segment.
ronidase type H-1 from Helix pomatia (Sigma Chemical Co., St. The total run time was less than 9 min.
Louis, MO), with a specific activity of ∼500 units/mg, was used. The TSQ 7000 was operated with the heated capillary at 450
To each sample, an amount of enzyme giving 800 units of activity °C, the corona discharge at 4.0 kV, and a capillary temperature
dissolved in 1.5 mL of a 0.2 M acetate buffer (3.1 mL of glacial of 250 °C. The sheath gas (N2) pressure was set to 50 psi and the
acetic acid, 9.7 g of sodium acetate, 1 L of water) was added. The collision gas (Ar) pressure to 2 mTorr. The precursor/product
samples were incubated for 17 h at 37 °C and then extracted using ion pairs and the collision offset energy used for the analysis of
solid-phase extraction (SPE). The SPE cartridge (Oasis HLB 3 the different compounds are summarized in Table 3. All samples
cm3, Waters, Milford, MA) was preconditioned with 1 mL of were injected twice. For the first injection, data were acquired in
methanol, followed by 1 mL of 1% acetic acid. The sample was positive ionization mode, and the total run time was 7.25 min. The
added and passed through the cartridge. To reduce interfering positive run was divided into two distinct timed segments, 0-3.5
components, the cartridge was washed with a 5% methanol in 1% and 3.5-7.25 min. For the second injection, data were acquired
acetic acid solution (1 mL). The cartridge was dried for ∼30 s in the negative ionization mode, and the total run time was 13
using vacuum. Methanol (1.5 mL) was eluted through the min. Again, the run was divided into distinct timed segments. In
cartridge and collected. Two milliliters of acetonitrile was added total, five time segments were used: 0-3.2, 3.2-4.3, 4.3-6.8, 6.8-
to the methanol fraction, and the combined extract was concen- 9, and 9-13 min.
trated to dryness and reconstituted in 50 µL of acetonitrile. The Quantification and Quality Control of Analytical Runs. An
sample was separated into two fractions, 10 and 40 µL, for the eight-point calibration plot for quantification was made up for every
analysis using HPLC-TIS-MS/MS and HPLC-APCI-MS/MS, re- analytical run using blank urine. The concentrations of the eight
spectively. calibration points ranged from 0.25 to 50 ng/mL for all analytes
HPLC Operating Conditions. An isocratic elution with a except DBCA, which had a maximum concentration of 10 ng/
mobile-phase mixture of 36% acetonitrile in water with 0.1% acetic mL. For each analytical run, eight calibration samples, two fortified
2456 Analytical Chemistry, Vol. 76, No. 9, May 1, 2004
Table 3. Precursor/Product Ion Pairs, Stable Isotope analyzed every day during a 5-day period, and the results were
Label of Internal Standard, Ion Mode, and Collision used to determine the within-day precision. The between-day
Offset Energy (CE) on the TSQ 7000 Instrument precision was determined by analyzing each QC pool in a
precursor ion product ion minimum of 93 analytical runs over a 3-month period. Two TSQ
7000 instruments were used to determine the between-day
isotope CE ion
compd label native labeled native labeled (V) mode precision of the analyses.
Matrix Effects. According to the method, the SPE cartridge is
IMPY 13C
4 153 157 84 88 22 +
DEAMPY d6 182 188 154 158 22 + washed with 0.1% acetic acid after sample addition and dried by
CIT d7 203 210 120 121 24 + vacuum to minimize unretained interfering substances that may
ATZ 13C
3 343 346 214 217 23 +
be present in residual water. Despite the drying process, some
ACE 13C
6 351 357 130 130 15 +
ALA 13C
6 365 371 162 168 25 + water remained on the cartridge. This water likely contains salts
DEET d6 192 198 119 119 22 + and polar biomolecules that could interfere with the ionization
MDA d7 273 280 141 147 13 -
PNP 13C 138 144 108 114 22 -
process. To evaluate the effect on the ionization process of
6
CMHC 13C
4 209 213 145 148 24 - coeluting matrix components in the methanol eluate, five blank
MET 13C
6 409 415 280 286 21 - urine samples (2 mL) were passed through SPE cartridges, and
2,4-D 13C
6 219 225 161 167 18 -
TCPY 1 13C 15N 198 204 198 204 15 -
the methanol eluate was collected as five 200-µL fractions for each
5
TCPY 2a 13C 15N
5 196 202 196 202 15 - blank urine sample. The anlytes were spiked into all fractions,
2,4,5-T 13C
6 255 261 197 203 16 - and the peak intensity of the different analytes was used as a
3-PBA 13C
6 213 219 93 99 25 -
measurement of positive or negative matrix effects on sensitivity.
a Confirmation ion. Intramethod and Intermethod Comparison. TCPY and
PNP can be analyzed using a gas chromatographic (GC)-MS/
MS method with a liquid-liquid extraction and chemical deriva-
tization.11 To measure the accuracy of our method, we reanalyzed
urine samples (one high and one low dose), one blank urine
with the present method urine samples (n ) 115) previously
sample, and one solvent blank were prepared, extracted, and
analyzed with the GC method. The two calibration curves used
analyzed in parallel with the unknown samples.
for quantification were prepared from stock solutions obtained
Method Validation. Limits of Detection. The limit of detection
independently.
(LOD) was calculated as three times the standard deviation of
3-PBA concentrations were determined in 115 urine samples
the noise at zero concentration.29 The estimate of the noise was
both on the Sciex API4000 and on the TSQ 7000. The results were
based on the variation in precision at concentrations close to the
used to determine agreement between the two ionization tech-
LOD. This was calculated using the four lowest calibration
niques and instruments used in our method.
standards from validation and analytical runs. This gave an
integrated LOD value over several (n > 7) runs. Furthermore,
RESULTS
the LODs were compared with the results of the calibration The optimized precursor daughter ion pairs for the different
standard samples to ensure that the calculated values agreed with analytes are summarized in Tables 2 and 3. All transitions were
the peaks observed in the lowest calibration standards. based on the [M + H]+ or [M - H]- precursor ions, except for
Extraction Efficiency. We determined the extraction recovery CIT, which was based on a cluster ion with acetonitrile, and ACE
of the method at two concentrations (5 and 50 ng/mL) by spiking and ALA, which were based on [M + H - 32]+ and [M + H -
“blank” urine samples (n ) 19 and n ) 10 with low and high 16]+ precursor ions, respectively. A confirmation ion was included
concentrations, respectively) and extracting according to the for TCPY to improve the selectivity of the analysis because a
method. Additional unspiked “blank” urine samples (n ) 30) were pseudo-MS/MS transition was used for this analyte.
extracted concurrently. Before the evaporation step, the unspiked The liquid chromatography was optimized to achieve the best
samples were spiked with the appropriate native standard (n ) possible separation and retention for all analytes in the same
20 and n ) 10 with low and high concentrations, respectively) to chromatographic system. The separation was performed under
serve as control samples representative of 100% recovery. Fur- isocratic conditions to eliminate equilibration times and thereby
thermore, all samples were spiked with a known amount of labeled reduce the total run time. The chromatographic separation of a
internal standard to correct for instrument variation, resulting in 2-mL urine sample spiked with 2 ng (1 part per billion (ppb)) of
a more accurate extraction recovery calculation. After concentrat- each analyte and extracted according to the method is shown in
ing and reconstituting, the samples were analyzed. The recovery Figure 2. For most analytes, we found no chromatographic
was calculated by comparing the responses of the blank urine interferences and limited noise. However, for the earlier eluting
samples spiked before extraction to the average response of the compounds in the positive ionization mode, we observed interfer-
blank urine samples spiked after the extraction. ences that were not baseline-separated (Figure 2). Furthermore,
Precision. The precision of the method was determined by the pseudo-MS/MS transition used for TCPY generated a high
calculating the coefficient of variation of repeat measurements of noise background.
samples from the QC pools (QCH and QCL; see Quality Control The method validation data (extraction recovery, within- and
Materials for analyte concentration information). Five new samples between-day variation, and LOD) for the analytes are summarized
from each of the QCH and the QCL pools were prepared and in Table 4. Unfortunately, the depletion of native ALA during the
(29) Taylor, J. K. Quality Assurance of Chemical Measurements; Lewis Publisher: validation process precluded the report of any validation data for
Chelsea, MI, 1987. this compound. The recoveries of all analytes ranged from 81 to
Analytical Chemistry, Vol. 76, No. 9, May 1, 2004 2457
 more, the results of the same urine samples analyzed with both
TIS and APCI (n ) 115, 51 samples with detectable amounts) are
shown in Figure 3. The cross-instrument comparison shows a
good agreement between the two determinations.
The effects of coextracted matrix from the SPE on peak
intensity of the analytes are shown in Figure 4. Regarding
compounds analyzed with APCI, the first 200-µL elution from the
SPE cartridge negatively effected the peak intensity of IMPY and
CMHC, whereas compounds with a hydroxyl group on an
aromatic ring were generally not affected by coextracted matrix.
For the carboxylic acid compounds, the highest peak intensities
were observed in the first and second fractions (Figure 4). This
was most pronounced for MDA, which had no signal in the third,
fourth, or fifth fractions. Compounds analyzed with TIS had
suppressed ionization in the first 200-µL fraction, and 3-PBA and
3-F-4-PBA had reduced peak intensity in the later fractions, similar
to the carboxylic acid compounds analyzed with APCI.
A comparison of PNP and TCPY concentrations in 115 urine
samples measured using our method and a previously established
method11 is shown in Figure 5. We found detectable concentra-
tions of PNP and TCPY in 106 and 89 samples, respectively.
The Sciex API 4000 source was robust; an average of ∼2500
samples can be injected without any maintenance. Figure 6 shows
the trans-DCCA concentrations in QC samples from 60 consecutive
runs over ∼2.5 months, during which no source maintenance was
performed. Routine maintenance of the TSQ source was negligible;
the heated capillary was rinsed and wiped with methanol between
runs. With this limited maintenance, ∼2000 injections (or 1000
samples, positive and negative injection) can be made without
major maintenance.

DISCUSSION
Figure 2. Chromatograms of an extracted urine sample (2 mL) Comparison with Earlier Methods. This method is based
fortified with 1 ng/mL of the individual compounds. The upper and on three previous methods from our laboratory5,8,13 that measured
middle chromatograms illustrate the runs in negative and positive 17 different compounds. Of these, 14 were measured with the new
ionization modes, respectively, on the TSQ 7000, and the lower method. We have included five additional analytes in the new
chromatogram the run on the Sciex API 4000. Compound name
methods2,4,5-T, DEET, and three markers of exposure to
abbreviations are according to Table 1.
chloroacetanilide herbicides.
The LODs of the target analytes generally were comparable
114%, except for MDA, which had recoveries of 68-75%. The with LODs in the earlier methods.5,8,13 However, the LOD was
LODs for all analytes were below 0.5 ng/mL, except IMPY and markedly lower for TCPY and higher for IMPY. TCPY was
CIT, which had LODs of 0.7 and 1.5 ng/mL, respectively. analyzed by a pseudo-MS/MS transition, using the same product
The method precision of each analyte, expressed as the relative ion as the precursor ion. The use of this technique has been
standard deviation (RSD) of repeated analysis of the QCL and reported previously.8;9 The LOD for TCPY was improved to 0.4
QCH urine pools, is given in Table 4. We estimated both within- ng/mL of urine. Although the recoveries of ATZ and IMPY were
day and between-day RSDs for the individual analytes. The within- much better using SPE than the previous liquid-liquid extraction,
day variation for the analytes was 4-8%, except for CIT, which the LODs were not better, probably because of increased noise
had a RSD of >12%. The within-day RSDs were similar for both from coextracted compounds. The LODs for four of the five
QCL and QCH pools, even though the RSDs for the high pyrethroid metabolites were lower even though the final volume
concentration were generally slightly lower than for the low was almost 10-fold higher. This increase probably results from
concentration. In general, the between-day RSDs for analytes with improvement in instrumentation between the methods instead of
labeled internal standards were less than 10%, except MDA and improvements in sample preparation.
CIT, which had RSDs of approximately 14 and 20%, respectively. The product ion for the analysis of 3-PBA in the APCI method
The between-day RSDs of cis-DCCA and DBCA, which lacked was changed to the same ion as for the TIS to make the analysis
isotopically labeled internal standards, were higher (12-20%). more selective. This also made the APCI method more sensitive
The validation data for 3-PBA using both the APCI and the with a lower LOD than in the old APCI method5 and more similar
TIS sources are given in Table 4. The recovery, LOD, and to the TIS method.
between-day and within-day variation results were similar, even In general, the precision of the method was as good as or better
though slightly better results were obtained using TIS. Further- than those in the previous methods. The lack of labeled internal
2458 Analytical Chemistry, Vol. 76, No. 9, May 1, 2004
Table 4. Descriptive Measurements of the Method Expressed as Extraction Recovery at a High and a Low
Concentration, Precision of the Method Expressed as the Coefficient of Variation (CV) for the Different Analytes,
and Detection Limits (LOD)

extraction recovery (%) CV between-day (%) CV within-day (%)

5 ppb (n)19) 50 ppb (n)10) low pool high pool low pool high pool LOD,
compd mean SD mean SD n ) 96 n ) 93 n ) 25 n ) 25 ng/mL
MDA 75 5.8 68 4.5 12 14 5.7 5.0 0.3
PNP 95 3.4 93 4.4 6.2 5.6 5.3 5.0 0.1
CMHC 95 6.6 96 13 10 8.8 7.7 7.7 0.2
MET 91 8.6 n/a 9.5 9.1 n/a n/a 0.2
24D 96 8.6 87 3.8 7.4 5.9 4.3 5.3 0.2
TCPY 1 88 8.2 93 8.8 10 7.8 7.8 4.9 0.3
TCPY 2 93 9.6 9 7.8 8.9 9.4 8.5 6.3 0.4
245T 97 5.1 90 2.4 5.4 5.0 4.4 4.6 0.1
3PBA (APCI) 95 5.1 90 3.5 7. 5.6 4.9 4.5 0.2
IMPY 99 12 81 9.3 10 9.0 7.1 6.9 0.7
DEAMPY 98 6.5 95 3.4 10 7.3 5.5 4.8 0.2
CIT 98 20 90 11 20 15 14 12 1.5
ATZ 96 3.8 94 1.8 8.0 7.4 3.1 4.1 0.3
ACE 98 5.1 94 3.1 7.9 7.5 4.1 4.4 0.1
DEET 96 4.4 93 2.7 7.6 7.5 3.7 4.2 0.1
3PBA (TIS) 94 4.2 92 1.8 4.8 4.9 4.3 3.9 0.1
4F3PBA 106 13 104 9.1 7.0 6.5 4.8 5.0 0.2
cis-DCCA 108 15 101 15 15 12 4.6 5.6 0.2
trans-DCCA 95 3.6 92 2.1 5.6 4.9 4.3 5.5 0.4
DBCA 114 18 n/a 20 15 4.3 5.4 0.1

it necessary to divide the samples into two parts to be analyzed

on two different instruments. We measured all pyrethroid me-
tabolites in the TIS method so that they are in the same analytical
injection, even though 3-PBA and 4-F-3PBA perform equally well
on both APCI and TIS. However, because 3-PBA is a possible
metabolite of over half of the pyrethroid insecticides registered
for use in the United States, and thus can be considered a good
marker for general pyrethroid exposure, we also included the
measurement of 3-PBA with the APCI analysis. The agreement
in results was high when the same samples were analyzed with
the different ionization sources (Figure 3), even though TIS
Figure 3. 3-Phenoxybenzoic acid concentrations in urine samples yielded slightly better sensitivity. If specific pyrethroid metabolite
(n ) 115) determined on both the TSQ and the Sciex instruments.
data are not required, the samples can be analyzed using only
Fifty-one samples had concentrations above the detection limit for
the atmospheric pressure chemical ionization method. APCI; two fractions are not necessary.
The main advantages of this method over our previous
methods,5,8,13 aside from the addition of more target analytes, are
standard for three of the pyrethroid metabolites remained a the reduction of sample preparation and cleanup time and the
problem, especially for DBCA and cis-DCCA, which had much reduced urine volume required for analysis. A 2-mL aliquot of
higher RSDs than trans-DCCA. Despite the structural similarity
urine is used instead of the previous total volume of 17 mL, and
between the trans and cis isomers of DCCA, their chemical
each sample has to be extracted only once instead of three times.
properties are very different. This is evident by the almost 20%
Traditionally, a large sample volume has not been a problem when
difference in chromatographic retention time between cis-DCCA
urine is the sample matrix. However, with the increased number
and trans-DCCA (Figure 2). In fact, the cis isomers (i.e., DBCA
of children’s exposure studies for which only a limited amount of
and cis-DCCA) behaved more similarly chromatographically than
urine can be feasibly collected, sample volume has become an
the cis- and trans-DCCA, although the halogen substitution
differed. This emphasizes the need for labeled internal standards issue. Furthermore, the new cleanup procedure was optimized
for DBCA and cis-DCCA. to reduce the labor-intensiveness and analyst time required. The
Our method still employs two different analyses, an APCI SPE on 3-mL cartridges reduced the sample cleanup time
analysis for measuring multiple, chemically diverse pesticide dramatically, compared with the liquid-liquid extraction used in
metabolites and a TIS analysis for measuring metabolites of one of the previous methods5 and the larger cartridge used in
pyrethroid insecticides. Our attempts to combine all of the another of the methods.13 Another difference is that the samples
markers into a single mass spectral analysis were unsuccessful were evaporated to dryness and reconstituted in 50 µL of
because the pyrethroid metabolites DBCA and cis- and trans- acetonitrile compared with a labor-intensive reduction of the final
DCCA could not be measured with the APCI source. This made sample volume to 7 µL in one of the earlier methods.13 These
Analytical Chemistry, Vol. 76, No. 9, May 1, 2004 2459
Figure 4. Relative peak intensity of the analyzed compounds in five consecutive 200-µL methanol fractions after solid-phase extraction. The
analytes were spiked into the five fractions after extraction. The bars show the average peak area (n ) 5, brackets are standard deviation) for
each fraction normalized to the highest average area of the analyte in any of the five fractions. Compound name abbreviations are according
to Table 1.

Figure 5. Comparison of 3,5,6-trichloro-2-pyridinol (TCPY) and 4-nitrophenol (PNP) concentrations in 115 urine samples determined with the
presented liquid chromatography method and with a gas chromatography method.11 Dashed lines indicate a slope of 1.

changes shortened the sample preparation time without compro- from a total of 41 min with the three old methods to 28 min in
mising the sensitivity. the present method. Thus, we have reduced the sample prepara-
However, some compromises were necessary regarding the tion and cleanup time to 20% and the instrument time to 70% of
analytes measured. We were unable to achieve reliable measure- the time needed with the previous three methods. Consequently,
ments using the APCI source of the azinphos-methyl metabolite the throughput of the method is ∼50 samples (including standards
previously analyzed with positive electrospray.8 To measure this and QC samples) per TSQ 7000 per day. However, the TIS analysis
compound with TIS would have required an additional injection is much faster than the APCI analysis. Thus, if only pyrethroid
in positive ionization mode on the Sciex API4000. In addition, to data are of interest, the sample throughput can easily be increased
simplify the cleanup, the analytes acephate and methamidaphos to 100-150 samples/day. We routinely use two TSQ 7000 and
were excluded. These two analytes were previously measured by one Sciex API4000 instrument to run 100 samples/day.
a sorbent-immobilized liquid extraction of the combined break- Matrix Effects. Unretained or coextracted substances from
through and wash fraction from the SPE cartridge.8 the SPE, such as salts and biomolecules, can possibly interfere
The present method reduces sample preparation time for 50 with the ionization processes. We speculate that most of the
samples to only 4 h from a total of ∼20 h with the three old interfering compounds elute in the early subfractions of the 1.5-
methods. Furthermore, the total instrument time is shortened mL methanol fraction collected for analysis. We observed ion
2460 Analytical Chemistry, Vol. 76, No. 9, May 1, 2004
 the endogenous material essential for good sensitivity probably
is of organic origin. We do not know the mechanism for this
ionization enhancement, but the need of urine matrix to improve
sensitivity of mercapturic acids31 and the need to add humic acids
to water samples to enhance signals of acidic compounds by
electrospray ionization have been reported.32
In normal urine samples, this is not a problem because enough
endogenous material is eluted to enhance the signal. However, it
becomes a problem when dilute urine samples or artificial matrix
is analyzed. Because artificial urine is sometimes used to make
proficiency testing samples or field QC samples, this potential
effect should be documented for each target analyte to allow
interpretation of the quality assurance data.
Comparison between GC-MS/MS and LC-MS/MS of
TCPY and PNP. We validated the method by comparing PNP
and TCPY results from samples analyzed with our present method
Figure 6. Concentrations of trans-DCCA in QCH and QCL samples
analyzed on the Sciex API4000. The data set consists of 60 to those results obtained with a method using liquid-liquid
consecutive runs without any maintenance of the source. Lines extraction for cleanup and GC-MS/MS for analysis.11 The two
represent average, 95, and 99% confidence limits based on the methods give comparable results for both TCPY and PNP (Figure
standard deviation for the method. The method relative standard 5). The agreement is good even though the samples were analyzed
deviation is 5.6% for QCL and 4.9% for QCH.
separated in time and with different standard solutions for the
creation of calibration curves for quantification.
suppression in the first subfraction of all analytes analyzed with
CONCLUSIONS
TIS (Figure 4). Furthermore, a tendency of matrix-induced This method provides a high-throughput tool for measuring
suppression was observed using APCI for the early-eluting urinary markers of human exposure to commonly used pesticides.
compounds IMPY and CMHC. Ion suppression is a common Compared with methods previously used in our laboratory, this
problem in LC-MS applications using electrospray interfaces, but method decreased the total preparation time by 75%. The method
in APCI applications, this is generally not regarded as a major is robust, and the limited instrument maintenance is necessary
problem. For IMPY, the ion suppression is a problem and a likely to successfully run a large number of samples (Figure 6).
cause for the relatively high LOD. However, matrix effects for some analytes still need to be
We observed a different problem for all the carboxylic acid addressed.
compounds, except trans- and cis-DCCA, in both TIS and APCI
modes. In the later subfractions, the peak signal decreases (Figure
Received for review December 29, 2003. Accepted
4). The most dramatic effect was for MDA, where the compound February 24, 2004.
could be detected only in the first two subfractions. We tried to
measure the carboxylic acids in artificial urine made of a mixture AC0355404
of salts, urea, and creatinine30 with limited success. Therefore, (31) Barr, D. B.; Ashley, D. L. J. Anal. Toxicol. 1998, 22, 96-104.
(32) Dijkman, E.; Mooibroek, D.; Hoogerbrugge, R.; Hogendoorn, E.; Sancho,
(30) Gustafsson, J. E.; Uzqueda, H. R. Clin. Chim. Acta 1978, 90, 249-257. J. V.; Pozo, O.; Hernandez, F. J. Chromatogr., A 2001, 926, 113-125.

Analytical Chemistry, Vol. 76, No. 9, May 1, 2004 2461