Determination of phenolic metabolites of polycyclic aromatic hydrocarbons in human urine as their pentafluorobenzyl ether derivatives using liquid chromatography-tandem mass spectrometry

Article abstract of DOI:10.1021/ac060920l

Polycyclic aromatic hydrocarbons (PAHs) are ubiquitous environmental contaminants, and a number of them are carcinogenic. One approach for measuring exposure to them is to determine the concentrations of metabolites in urine. The pyrene metabolite 1-hydroxypyrene has been used as a biomarker for exposure in numerous studies. However, determination of exposure to several PAHs may be advantageous, since the relative amounts may vary depending upon the exposure source. We developed a liquid chromatography-tandem mass spectrometty method for the determination of phenolic metabolites of naphthalene, fluorene, phenanthrene, and pyrene in human urine. Following enzymatic cleavage of the glucuronide and sulfate conjugates, the phenolic metabolites are extracted from urine and converted to pentafluorobenzyl ethers. These derivatives greatly enhance the sensitivity of detection by atmospheric pressure chemical ionization in the negative ion mode. Lower limits of quantitation range from 0.01 to 0.5 ng/mL. Stable isotope-labeled internal standards were synthesized or obtained commercially. Data on urinary excretion of several PAH metabolites in urine of smokers and nonsmokers are presented.

Full text of DOI:10.1021/ac060920l

Anal. Chem. 2007, 79, 587-598
Determination of Phenolic Metabolites of
Polycyclic Aromatic Hydrocarbons in Human Urine
as Their Pentafluorobenzyl Ether Derivatives Using
Liquid Chromatography-Tandem Mass
Spectrometry
Peyton Jacob, III,* Margaret Wilson, and Neal L. Benowitz
Division of Clinical Pharmacology and Experimental Therapeutics, Medical Service, San Francisco General Hospital Medical
Center, the Departments of Medicine, Psychiatry, and Biopharmaceutical Sciences, University of California, San Francisco,
San Francisco, California 94110
PAHs are carcinogenic, there has been a great deal of interest in
developing methods for determining human exposure. Generally
the higher molecular weight PAHs, such as benzo[a]pyrene, are
the most carcinogenic, but recent studies have concluded that
naphthalene is a substance that contributes considerably to human
cancer risk.4 A widely used approach for assessing exposure is
to measure concentrations of PAH metabolites in urine.6 Since
urinary excretion of metabolites of highly carcinogenic PAHs such
as benzo[a]pyrene is very low and variable due to extensive
elimination in the feces, and therefore difficult to quantify,6
“surrogate markers” are generally used. The pyrene metabolite
1-hydroxypyrene (1-HP) has been used as a biomarker for PAH
exposure in numerous studies.7 However, since ratios of various
PAHs may vary depending upon the source,8-11 measuring
concentrations of metabolites of several different PAHs may be
advantageous in studies of exposure.12-16 Liver enzyme induction
can affect metabolite ratios, so measuring concentrations of several
metabolites might be used to evaluate induction of cytochrome
Polycyclic aromatic hydrocarbons (PAHs) are ubiquitous
environmental contaminants, and a number of them are
carcinogenic. One approach for measuring exposure to
them is to determine the concentrations of metabolites
in urine. The pyrene metabolite 1-hydroxypyrene has been
used as a biomarker for exposure in numerous studies.
However, determination of exposure to several PAHs may
be advantageous, since the relative amounts may vary
depending upon the exposure source. We developed a
liquid chromatography-tandem mass spectrometry method
for the determination of phenolic metabolites of naphtha-
lene, fluorene, phenanthrene, and pyrene in human urine.
Following enzymatic cleavage of the glucuronide and
sulfate conjugates, the phenolic metabolites are extracted
from urine and converted to pentafluorobenzyl ethers.
These derivatives greatly enhance the sensitivity of detec-
tion by atmospheric pressure chemical ionization in the
negative ion mode. Lower limits of quantitation range from
0.01 to 0.5 ng/mL. Stable isotope-labeled internal stan-
dards were synthesized or obtained commercially. Data
on urinary excretion of several PAH metabolites in urine
of smokers and nonsmokers are presented.
(4) Preuss, R.; Angerer, J.; Drexler, H. Int. Arch. Occup. Environ. Health 2003,
76, 556-576.
(5) Guillen, M. D.; Sopelana, P. J. Dairy Sci. 2004, 87, 556-564.
(6) Jacob, J.; Seidel, A. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2002,
778, 31-47.
(7) Jongeneelen, F. J. Ann. Occup. Hyg. 2001, 45, 3-13.
(8) Schauer, J. J.; Kleeman, M. J.; Cass, G. R.; Simoneit, B. R. Environ. Sci.
Technol. 1999, 33, 1578-1587.
Polycyclic aromatic hydrocarbons (PAHs) are ubiquitous
environmental contaminants, arising from numerous combustion
sources, including motor vehicles, fireplaces, tobacco smoke,
power plants, and various other industrial sources.1,2 In addition,
diet is a major source of human exposure to these substances,
including the relatively volatile naphthalene, fluorene, and phenan-
threne, as well as less volatile PAHs such as pyrene and benzo-
[a]pyrene, with four or more fused rings.3-5 Since a number of
(9) Schauer, J. J.; Kleeman, M. J.; Cass, G. R.; Simoneit, B. R. Environ. Sci.
Technol. 2001, 35, 1716-1728.
(10) Schauer, J. J.; Kleeman, M. J.; Cass, G. R.; Simoneit, B. R. Environ. Sci.
Technol. 2002, 36, 1169-1180.
(11) Ding, Y. S.; Trommel, J. S.; Yan, X. J.; Ashley, D.; Watson, C. H. Environ.
Sci. Technol. 2005, 39, 471-478.
(12) Hollender, J.; Koch, B.; Dott, W. J. Chromatogr.. B: Biomed. Sci. Appl. 2000,
739, 225-229.
* Corresponding author. E-mail: peyton.jacob@.ucsf.edu. Phone: (415) 282-
9495. Fax: (415) 206-5080.
(13) Kang, J. W.; Cho, S. H.; Kim, H.; Lee, C. H. Arch. Environ. Health 2002,
57, 377-382.
(1) Bostrom, C. E.; Gerde, P.; Hanberg, A.; Jernstrom, B.; Johansson, C.;
Kyrklund, T.; Rannug, A.; Tornqvist, M.; Victorin, K.; Westerholm, R.
Environ. Health Perspect. 2002, 110 (Suppl 3), 451-488.
(2) Li, C. T.; Lin, Y. C.; Lee, W. J.; Tsai, P. J. Environ. Health Perspect. 2003,
111, 483-487.
(3) Wang, G.; Lee, A. S.; Lewis, M.; Kamath, B.; Archer, R. K. J. Agric. Food
Chem. 1999, 47, 1062-1066.
(14) Gmeiner, G.; Gartner, P.; Krassnig, C.; Tausch, H. J. Chromatogr., B: Anal.
Technol. Biomed. Life Sci. 2002, 766, 209-218.
(15) Toriba, A.; Chetiyanukornkul, T.; Kizu, R.; Hayakawa, K. Analyst 2003, 128,
605-610.
(16) Grainger, J.; Huang, W.; Patterson, D. G., Jr.; Turner, W. E.; Pirkle, J.; Caudill,
S. P.; Wang, R. Y.; Needham, L. L.; Sampson, E. J. Environ. Res. 2006,
100, 394-423.
10.1021/ac060920l CCC: $37.00 © 2007 American Chemical Society
Published on Web 12/09/2006
Analytical Chemistry, Vol. 79, No. 2, January 15, 2007 587

P450 (CYP) enzymes.17-21 For example, it has been reported that
the ratio of 1- + 2-hydroxyphenanthrene/3- + 4-hydroxyphenan-
threne is lower in smokers than in nonsmokers, as a result of
induction of 3,4-oxidation catalyzed by CYP1A2.17,18 Hydroxylated
PAH metabolites are excreted mainly as their glucuronide
conjugates, but varying amounts of sulfate conjugates and small
amounts of free phenols may also be excreted.6,22
Most of the methods that have been used to determine
phenolic metabolites of PAHs, in particular 1-HP, have employed
high-performance liquid chromatography (HPLC) with fluores-
cence detection.6,7,15,18,23 An advantage of these methods is that
HPLC instruments are available in many laboratories and that
fluorescence detection is highly sensitive and can provide fairly
good specificity for determining PAH metabolites. Disadvantages
are that often fairly large urine specimens (10 mL or more) are
needed, long HPLC run times may be required, and most methods
do not employ an internal standard, which, if used, would be
expected to result in better precision and accuracy.15,23,24
Figure 1. Ring-hydroxylated PAH metabolites. The subscripts of
the R groups indicate the ring positions of the hydroxy groups for the
monophenolic metabolites, e.g., for 2-hydroxyfluorene R2 ) OH and
all other R groups ) H.
graphic run times.29,30 Despite this, relatively few LC/MS methods
have been reported for determination of hydroxylated metabolites
of PAHs. Determination of 1-HP in urine using LC- time-of-flight
MS with electrospray ionization (ESI),31 and by LC/MS/MS with
ESI,32,33 have been reported, and an LC/MS method that utilized
a single quadrupole instrument with ESI has been reported for
determination of phenolic metabolites of PAHs from in vitro
metabolism studies.34 LC/MS methods using atmospheric pres-
sure chemical ionization (APCI) or ESI for determination of
various hydroxylated PAHs have been reported but not applied
to biological samples.35-38 Recently, an LC-ESI-MS/MS method
for selective detection and preliminary quantification of certain
monohydroxy PAH metabolites in human urine was reported,
including data on excretion of metabolites of the potent carcino-
gens benzo[a]pyrene and benz[a]anthracene.39 LC-APCI-MS/MS
determination of 1-HP and 3-hydroxybenzo[a]pyrene has been
reported,40 but the sensitivity was adequate only for urine from
persons with relatively high levels of PAH exposure.16
Gas chromatography (GC) and gas chromatography/mass
spectrometry (GC/MS) methods for determination of phenolic
PAH metabolites have also been reported. An advantage of these
methods is that capillary GC columns can provide better resolution
of isomeric metabolites than HPLC columns.25 Recently, GC-high-
resolution MS methods for simultaneous determination of several
PAH metabolites was reported; an advantage is the high sensitivity
and specificity attainable on a high-resolution instrument.26-28
A
disadvantage of GC/MS methods compared to HPLC-fluorescence
is that derivatization of the analytes is necessary and that more
expensive instruments are required, especially in the case of high-
resolution mass spectrometers, which are not widely available.
In recent years, liquid chromatography/mass spectrometry
(LC/MS) instruments have become available in many laboratories,
a major driving force being their widespread use in drug discovery
and development.29 With LC coupled to triple-stage quadrupole
tandem mass spectrometers (LC/MS/MS), a wide range of
substances in complex biological matrixes can be quantitated at
low levels and with high specificity, often with short chromato-
As part of our studies of toxic substance exposure in people
smoking cigarettes of different composition, we needed a method
for the determination of several PAH metabolites in human urine
that would be suitable for large numbers of samples generated in
clinical studies. In this paper, we describe an LC/MS/MS method
for determination of monophenolic metabolites of naphthalene,
fluorene, phenanthrene, and pyrene (Figure 1). The method
(17) Jacob, J.; Grimmer, G.; Dettbarn, G. Biomarkers 1999, 4, 319-327.
(18) Heudorf, U.; Angerer, J. Int. Arch. Occup. Environ. Health 2001, 74, 177-
183.
(19) Nan, H. M.; Kim, H.; Lim, H. S.; Choi, J. K.; Kawamoto, T.; Kang, J. W.;
Lee, C. H.; Kim, Y. D.; Kwon, E. H. Carcinogenesis 2001, 22, 787-793.
(20) Seidel, A.; Dahmann, D.; Krekeler, H.; Jacob, J. Int. J. Hyg. Environ. Health
2002, 204, 333-338.
(30) Maurer, H. H. Clin. Biochem. 2005, 38, 310-318.
(21) Carmella, S. G.; Chen, M.; Yagi, H.; Jerina, D. M.; Hecht, S. S. Cancer
(31) Holm, A.; Molander, P.; Lundanes, E.; Ovrebo, S.; Greibrokk, T. J.
Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2003, 794, 175-183.
(32) Pigini, D.; Cialdella, A. M.; Faranda, P.; Tranfo, G. Rapid Commun. Mass
Spectrom. 2006, 20, 1013-1018.
Epidemiol. Biomarkers Prev. 2004, 13, 2167-2174.
(22) Singh, R.; Tucek, M.; Maxa, K.; Tenglerova, J.; Weyand, E. H. Carcinogenesis
1995, 16, 2909-2915.
(23) Carmella, S. G.; Le, K. A.; Hecht, S. S. Cancer Epidemiol. Biomarkers Prev.
(33) Li, Y.; Li, A. C.; Shi, H.; Zhou, S.; Shou, W. Z.; Jiang, X.; Naidong, W.;
Lauterbach, J. H. Rapid Commun. Mass Spectrom. 2005, 19, 3331-3338.
(34) Van, de Wiele, T. R.; Peru, K. M.; Verstraete, W.; Siciliano, S. D.; Headley,
J. V. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2004, 806, 245-
253.
(35) Galceran, M. T.; Moyano, E. J. Chromatogr., A 1994, 683, 9-19.
(36) Galceran, M. T.; Moyano, E. J. Chromatogr., A 1996, 731, 75-84.
(37) Letzel, T.; Poschl, U.; Rosenberg, E.; Grasserbauer, M.; Niessner, R. Rapid
Commun. Mass Spectrom. 1999, 13, 2456-2468.
2004, 13, 1261-1264.
(24) Chetiyanukornkul, T.; Toriba, A.; Kameda, T.; Tang, N.; Hayakawa, K. Anal.
Bioanal. Chem. 2006, 386, 712-718.
(25) Jacob, J.; Grimmer, G. Rev. Ann. Chem. 1987, 9, 49-89.
(26) Romanoff, L. C.; Li, Z.; Young, K. J.; Blakely, N. C., 3rd; Patterson, D. G.,
Jr.; Sandau, C. D. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2006,
835, 47-54.
(27) Smith, C. J.; Walcott, C. J.; Huang, W.; Maggio, V.; Grainger, J.; Patterson,
D. G., Jr. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2002, 778,
157-164.
(38) Letzel, T.; Poschl, U.; Wissiack, R.; Rosenberg, E.; Grasserbauer, M.;
Niessner, R. Anal. Chem. 2001, 73, 1634-1645.
(28) Li, Z.; Romanoff, L. C.; Trinidad, D. A.; Hussain, N.; Jones, R. S.; Porter,
E. N.; Patterson, D. G., Jr.; Sjodin, A. Anal. Chem. 2006, 78, 5744-5751.
(29) Tiller, P. R.; Romanyshyn, L. A.; Neue, U. D. Anal. Bioanal. Chem. 2003,
377, 788-802.
(39) Xu, X.; Zhang, J.; Zhang, L.; Liu, W.; Weisel, C. P. Rapid Commun. Mass
Spectrom. 2004, 18, 2299-2308.
(40) Fan, R.; Dong, Y.; Zhang, W.; Wang, Y.; Yu, Z.; Sheng, G.; Fu, J. J.
Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2006.
588 Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

Figure 2. Synthesis of 2-hydroxyfluorene-d9. A mixture of hydroxyphenanthrene isomers was synthesized analogously. MCPBA )
m-chloroperbenzoic acid.
Table 1. Parent Ions, Product Ions, and Collision
Energies for Analytes and Internal Standards
analyte or internal
standard
parent
mass
product
mass
collision
energy, eV
2-naphthol
143
149
181
181
181
190
190
193
202
217
223
115
121
153
180
153
162
188
165
174
189
195
30
30
28
31
28
28
31
38
38
22
22
2-naphthol-d6
1-hydroxyfluorene
2-hydroxyfluorene
3-hydroxyfluorene
2-hydroxyfluorene-d9
2-hydroxyfluorene-d9
hydroxyphenanthrenes
hydroxyphenanthrenes-d9
1-hydroxypyrene
Figure 3. Conversion of phenolic metabolites to PFB derivatives
and proposed fragmentation pathways. Illustrated for 2-hydroxyfluo-
rene pentafluorobenzyl ether.
1-hydroxypyrene-13C6
2-, 3-, 4-, and 9-hydroxyphenanthrenes were purchased from
Crescent Chemical Co. (Islandia, NY). 1-Hydroxypyrene-13C6 was
obtained from the National Cancer Institute Chemical Carcinogen
Reference Standard Repository. Pentafluorobenzyl bromide was
obtained from Acros Organics/Fisher Chemical Co. 2-Naphthol-
d6, 2-hydroxyfluorene-d9, and hydroxyphenanthrene-d9 isomers
were synthesized in our laboratory (Figures 2, S-1, and S-2,
Supporting Information). The syntheses are described in the
Supporting Information. HPLC grade methanol and water from
Burdick and Jackson (Muskegon, MI) were used to prepare the
LC mobile phase. HPLC grade pentane and ethyl acetate from
Fisher were used for extractions.
involves conversion of the metabolites to the pentafluorobenzyl
ether derivatives, in order to enhance sensitivity by making use
of the recently developed electron capture atmospheric pressure
chemical ionization (ECAPCI) technique.41 Advantages of this
method include simultaneous determination of several analytes
with good precision and accuracy, very high sensitivity, an
extraction/derivatization procedure suitable for processing large
batches of samples, and a relatively short (15 min) instrument
run time.
EXPERIMENTAL SECTION
Instrumentation. LC/MS and LC/MS/MS analyses were
carried out with a Surveyor HPLC interfaced to a TSQ Quantum
Ultra triple-stage quadrupole mass spectrometer (Thermo-Finni-
gan, San Jose, CA). A PAH Green column (4.6 × 150 mm, Thermo-
Hypersil-Keystone (Bellefonte, PA) was used for the chromatog-
raphy. For GC/MS and GC/MS/MS analyses, a Thermo Trace
2000 GC (Milan, Italy) with a 0.25 mm i.d. × 25 m DB-5 column
(J & W Scientific, Folsom, CA) interfaced to a Finnigan TSQ 7000
mass spectrometer was used. Solvent evaporation was carried out
using a Savant Automatic Environmental SpeedVac model AES
2000 (Thermo-Savant, Marietta, OH)
Chemicals. 1- and 2-naphthols and 1-hydroxypyrene were
obtained from Acros Organics/Fisher Chemical Co. (Pittsburgh,
PA). 1-Hydroxyfluorene was obtained from ChemBridge Corp.
(San Diego, CA), and 2- and 3-hydroxyfluorene were from Aldrich
Chemical Co. (Milwaukee, WI). 1- and 2-naphthols and 2-hydroxy-
fluorene were sublimed under reduced pressure prior to use. 1-,
Preparation of Standards and Controls. An “artificial urine”
was used as a matrix for the standards, prepared from major
components reported in human urine.42 Concentrations in grams
per liter of deionized water were as follows: human albumin, 0.1;
ammonium sulfate, 1.0; andosterone, 0.02; ascorbic acid, 0.02;
bilirubin, 0.005; citric acid, 0.5; creatine, 0.5; creatinine, 1.5; cystine,
0.1; glucuronic acid, 0.5; hippuric acid, 1.0; histidine, 0.5; magne-
sium sulfate, 1.0; phenol, 0.5; potassium phosphate, monobasic,
1.0; sodium chloride, 10; urea, 15; uric acid, 0.5. The pH of the
artificial urine solution was adjusted to between 5 and 6, and it
was stored at -20 °C. Standards (including blanks) were prepared
on the day of use by diluting eight stock methanolic solutions of
the analyte standards 30-fold with the artificial urine. Controls were
prepared by spiking pooled nonsmokers’ urine with analyte stock
solutions in methanol. A stock solution of internal standards was
prepared by adding 100 µL of 1 mg/mL 2-naphthol-d6, 50 µL of 1
(41) Singh, G.; Gutierrez, A.; Xu, K.; Blair, I. A. Anal. Chem. 2000, 72, 3007-
(42) Putnam, D. F. In McDonnell Douglas Astronautics Company, Report # NASA
CR-1082; National Information Service: Springfield, VA, 1971.
3013.
Analytical Chemistry, Vol. 79, No. 2, January 15, 2007 589

Figure 4. APCI spectra of 2-hydroxyfluorene and 1-hydroxypyrene pentafluorobenzyl ethers.
mg/mL 2-hydroxyfluorene-d9, 50 µL of 1 mg/mL hydroxyphenan-
threne-d9 isomers (method A, Supporting Information), 60 µL of
1 mg/mL hydroxyphenanthrene-d9 isomers (method B, Support-
ing Information), and 1 mL of 10 µg/mL 1-hydroxypyrene-13C6 to
125 mL of methanol. This composition was chosen to give MS
responses comparable to those of the natural isotopomers in
smokers and to give similar peak areas of the isomeric deuterium-
labeled hydroxyphenanthrenes.
PFBBr could be reduced to 20 µL of 5%, and the ammonia and
sulfuric acid steps could be omitted, with about 20-50% improve-
ment in sensitivity.) The residues were dissolved in 120 µL of
methanol, and 20 µL was injected into the LC/MS/MS system.
Extraction efficiency was determined by analyzing 10 urine
specimens with the labeled internal standards added prior to
enzyme treatment and extraction, as well as aliquots of the same
specimens to which internal standards were added after extraction.
The differences in mean peak areas for samples with internal
standards added prior to extraction and those for which internal
standards were added after extraction were used to calculate
extraction recoveries. These were as follows: 2-naphthol-d6, 72%;
1-hydroxyphenanthrene-d9, 73%, 2-hydroxyphenanthrene-d9, 82%;
3- + 4-hydroxyphenanthrene-d9, 80%; 2-hydroxyfluorene-d9, 81%;
1-hydroxypyrene-13C6, 83%. Note:pentafluorobenzyl bromide is a
potent lachrymator, and it should be handled in an efficient fume
hood.
Liquid Chromatography. The analytes were separated using
a gradient of water and methanol, 0.7 mL/min, with the following
program: initial 80% methanol, 0.5 min 90% methanol, 5 min 94%
methanol, 5.5-9 min 100% methanol, and 9.1-15 min (end of run)
80% methanol.
Mass Spectrometry. The mass spectrometer was operated
using APCI, in the negative ion mode. The ion source parameters
were optimized by infusing a methanol solution of the pentafluo-
robenzyl (PFB) derivative of 2-hydroxyfluorene via syringe pump.
The vaporizer temperature was 317 °C, the heated capillary
temperature was 251 °C, and the corona discharge current was
set at 50 µA. The PFB derivatives of the individual analytes were
infused to determine appropriate ion transitions and the optimum
Sample Preparation. The internal standard solution (50 µL)
was added to urine specimens, standards, blanks, or controls (2.7
mL), which were buffered to pH 7 with 0.3 mL of 1 M phosphate
buffer and incubated overnight at 37 °C with â-glucuronidase (3000
units, type 1XA from Eschericheria coli, Sigma, St. Louis, MO)
and sulfatase (0.6 unit, type VI from Aerobacter aerogenes, Sigma).
The samples were extracted with a 90:10 mixture (v/v) of
pentane/ethyl acetate (4 mL) by vortex mixing, and the phases
were separated by centrifuging (∼1000g) and freezing the aqueous
layers in a dry ice/acetone bath. The organic phases were poured
into tubes containing 150 µg of gallic acid in 30 µL of methanol,
and the solvent was removed using a centrifugal vacuum evapora-
tor at ambient temperature. To the residues were added pen-
tafluorobenzyl bromide (100 µL of 5% in methylene chloride),
aqueous tetrabutylammonium bromide (50 µL of 5%), and aqueous
tripotassium phosphate (50 µL of 20%). The tubes were capped,
vortex-mixed for 30 min, and then vortexed with 100 µL of
ammonium hydroxide (10% in 40/60 water/methanol) to destroy
excess pentafluorobenzyl bromide. After adding 1 mL of 4 M
sulfuric acid, the derivatives were extracted with pentane (3 mL,
vortex, centrifuge, freeze/pour) and evaporated to dryness.
(Subsequent to method validation, we found that the amount of
590 Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

Figure 5. Effect of Q 1 mass resolution (Q 3 0.7 amu fwhm). LC/MS SRM chromatograms, hydroxyfluorenes and hydroxyphenanthrenes in
extract of nonsmoker’s urine.
Data Analysis. The Finnigan XCalibur/LC Quan software was
used to generate calibration curves (linear regression, 1/X
weighting) and calculate concentrations using peak area ratios of
analyte/internal standard. The ISTD for 2-naphthol was 2-naphthol-
d6, the ISTD for 1- and 3-hydroxyfluorenes was 2-hydroxyfluorene-
d9 (m/z 190 > m/z 162), the ISTD for 2-hydroxyfluorene was
2-hydroxyfluorene-d9 (m/z 190 > m/z 188), the ISTD for 3- and
4-hydroxyphenanthrene was 3-hydroxyphenanthrene-d9 (first elut-
ing), the ISTD for 1-hydroxyphenanthrene was 1-hydroxyphenan-
threne-d9 (and/or 9- hydroxyphenanthrene-d9, second eluting), the
ISTD for 2-hydroxyphenanthrene was 2-hydroxyphenanthrene-d9
(third eluting), and the ISTD for 1-HP was 1-HP-13C6 (Table 1).
Figure 6. Relative recovery of 2-hydroxyfluorene-d9 and 1-hydroxy-
pyrene-13C6 from different sample matrixes. Urine and artificial urine
Eight standards were used, at concentrations spanning the
expected range, 0.25-50 ng/mL for 2-naphthol and from 0.01 to
were treated with enzyme â-glucuronidase/arylsulfatase.
5 ng/mL for all other analytes. “Artificial urine” blanks were
included in all runs. The following analyte pairs were not resolved
collision energies for collision-induced dissociation (CID). The
and were reported as sums: 1- and 2-naphthols; 1- and 9-hydroxy-
collision gas (argon) pressure was 1.5 mTorr. The parent and
phenanthrenes; 3- and 4-hydroxyphenanthrenes. Correlation coef-
product ions and collision energies used for quantitation are in
Table 1. Three acquisition segments were used. The PFB
derivatives of 2-naphthol and 2-naphthol-d6 eluted in the first; the
ficients (r2) for typical calibration curves are as follows: 2-naphthol,
0.9960; 1-hydroxyfluorene, 0.9989; 2-hydroxyfluorene, 0.9974;
3-hydroxyfluorene, 0.9739; 1-hydroxyphenanthrene, 0.9995; 2-hy-
PFB derivatives of the hydroxyfluorenes, hydroxyphenanthrenes,
and their deuterated isotopomers eluted in the second; 1-hydroxy-
pyrene and 1-hydroxypyrene-13C6 PFB derivatives eluted in the
third. Source CID collision energy was 33 eV for 1-hydroxypyrene
and 1-hydroxypyrene-13C6. Source CID was not used for the other
analytes. For all analytes, Q1 peak width was 0.1 amu fwhm, and
Q3 peak width was 0.7 amu fwhm.
droxyphenanthrene, 0.9988; 3- + 4-hydroxyphenanthrenes, 0.9998;
1-hydroxypyrene, 0.9996.
Clinical Samples. Urine specimens were obtained from
smokers and nonsmokers participating in research studies, which
were approved by the Institutional Review Board at the University
of California, San Francisco, and were stored frozen at -20 °C
prior to analysis. Stability of stored urine samples was evaluated
Analytical Chemistry, Vol. 79, No. 2, January 15, 2007 591

Figure 7. LC/MS SRM chromatograms of urine extracts. Q 1 resolution 0.1 amu fwhm, Q3 resolution 0.7 amu fwhm. (A) SRM transitions for
analytes; (B) SRM transitions for internal standards.
by reanalyzing samples 1 year after the original analysis. The
means (six nonsmokers and seven smokers) of the percent
differences were within (15% of the original analytical results
(Table S-1, Supporting Information).
592 Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

Figure 8. GC/MS chromatograms, methane NICI.
RESULTS AND DISCUSSION
and specificity would not be expected to be as good as that
attainable using SRM. Indeed, considerable background ion
current and extraneous peaks were encountered in SIM analysis
of both standards and urine extracts. Therefore, we turned our
attention to converting the PAH metabolites to PFB derivatives
in order to use the highly sensitive ECAPCI technique.41 1-HP
and the other phenolic PAH metabolites were readily converted
to the PFB derivatives with pentafluorobenzyl bromide and a base.
Conversion to derivatives was indicated by HPLC, and in the case
of 2-naphthol, the structure was verified by GC/MS with electron
ionization, which produced the molecular ion m/z 324 (17%), as
well as fragments m/z 181 (C6F5CH2+, 78%) and 115 (C10H17O+,
100%).
ECAPCI produced aryloxy anions by loss of the PFB radical
from the initially generated radical anion. In ECAPCI, as well as
negative ion chemical ionization (NICI) in GC/MS of PFB
derivatives, the initially formed radical anions generally lose the
PFB radical if a relatively stable anion can be formed.41 Interest-
ingly, applying source CID (also called in-source fragmentation)
dramatically increased the ion current for some analytes, by 1
order of magnitude or more. Perhaps this aids in desolvation, or
reduces the extent of protonation of the initially formed radical
anion by methanol or water in the vaporized mobile phase, but
we have no experimental evidence for these hypotheses. However,
with the exception of 1-HP, source CID significantly increased
API Mass Spectrometry. Initially, we attempted to develop
a method for determining the phenolic metabolites without
conversion to a derivative. Using APCI or ESI ionization modes
and single-stage MS, 1-HP was converted to the expected (M -
H)- ion, m/z 217. Operating the instrument in the MS/MS mode,
CID converted the (M - H)- ion to the major product ion m/z
189, presumably via loss of CO. Using the transition m/z 217 to
189, we attempted to develop a selected reaction monitoring
(SRM) method for 1-HP, but sensitivity with APCI appeared
inadequate for determination of the subnanogram per milliliter
levels that would be required. ESI was found to be more sensitive
than APCI, but we were concerned that matrix suppression of
ionization that often occurs with ESI might be a problem39,43-47
and we did not evaluate this ionization mode with urine extracts.
Using selected ion monitoring (SIM) of m/z 217, sensitivity for
pure standards was better than when using SRM by 1 order of
magnitude or more, but sensitivity still appeared to be marginal,
(43) Matuszewski, B. K.; Constanzer, M. L.; Chavez-Eng, C. M. Anal. Chem.
1998, 70, 882-889.
(44) Liang, H. R.; Foltz, R. L.; Meng, M.; Bennett, P. Rapid Commun. Mass
Spectrom. 2003, 17, 2815-2821.
(45) Niessen, W. M. J. Chromatogr., A 1999, 856, 179-197.
(46) Mallet, C. R.; Lu, Z.; Mazzeo, J. R. Rapid Commun. Mass Spectrom. 2004,
18, 49-58.
(47) Jacob, P., 3rd; Wilson, M.; Yu, L.; Mendelson, J.; Jones, R. T. Anal. Chem.
2002, 74, 5290-5296.
Analytical Chemistry, Vol. 79, No. 2, January 15, 2007 593

Table 2. Precision and Accuracy, Spiked Pooled Nonsmokers’ Urine, Intraday (N ) 7)
added
amount
(ng/mL)
expected
amounta
(ng/mL)
measd
mean
accuracy
(% of
precision
CV (%)
analyte
(ng/mL)
expected)
2-naphthol
0
1.43
8.0
11
0.25
1
1.68
2.43
11.4
51.4
0.505
1.71
102
109
108
101
103
2.64
4.2
5.1
6.0
7.6
19
10
50
0.500
0
12.3
52.1
LOQ (artificial urine)
1-hydroxyfluorene
0.518
0.038
0.051
0.128
1.01
0.025
0.1
1
0.063
0.138
1.04
5.04
0.105
80.9
92.9
98.0
95.7
101
17
8.4
5.8
9.4
19.5
4.9
5.1
3.8
4.3
4.1
13
5
4.83
LOQ (artificial urine)
2-hydrolxyfluorene
0.100
0
0.106
0.142
0.177
0.273
1.20
0.025
0.1
1
0.167
0.242
1.14
5.14
0.028
106
113
105
108
84
5
5.56
LOQ (artificial urine)
3-hydroxyfluorene
0.025
0
0.024
0.045
0.056
0.146
1.22
17
0.025
0.1
1
0.070
0.145
1.05
81.0
101
116
106
29
25
10
5
0
5.05
5.34
22
1-hydroxyphenanthrene
0.124
0.148
0.227
1.117
5.20
12
0.025
0.1
1
0.149
0.224
1.12
5.12
0.025
99.1
101
99.3
102
86
9.2
3.9
5.9
4.6
16
5
LOQ (artificial urine)
2-hydroxyphenanthrene
0.025
0
0.022
0.032
0.057
0.144
1.05
12
0.025
0.1
1
0.057
0.132
1.03
5.03
0.010
99
7.4
7.0
5.9
6.2
12
109
101
99.2
100
5
4.99
LOQ (artificial urine)
0.010
0
0.010
0.096
0.144
0.292
2.00
3- + 4-hydroxyphenanthreneb
7.5
6.8
3.8
3.5
3.2
10
7.0
2.0
1.7
3.4
2.8
5.2
0.05
0.2
2
0.146
0.296
2.10
5.10
0.020
98.6
98.6
95.2
95.5
96
5
4.87
LOQ (artificial urine)
1-hydroxypyrene
0.020
0
0.019
0.057
0.082
0.159
0.97
0.025
0.1
1
0.082
0.157
1.06
5.06
0.025
100
101
91.8
94.5
104
5
4.78
0.026
LOQ (artificial urine)
0.025
b
a Added amount + mean amount measured in blank urine (pool of three nonsmokers). Spiked with equal amounts of 3- and 4-hydroxyphenan-
threne. Relative response of 4-hydroxyphenanthrene was ∼25% of the response of 3-hydroxyphenanthrene.
chemical noise observed in chromatograms derived from urine
samples, so except for 1-HP, source CID was not used.
of the PFB derivatives of 1-hydroxypyrene and 2-hydroxyfluorene
are shown in Figure 4. In addition to the major (M - PFB)- ions,
(M - H)- ions were observed in low abundance, m/z 361 and
397, respectively. The ion transitions used for quantitation and
collision energies are listed in Table 1. Sensitivity in the SRM
mode was sufficient to detect a few picograms of derivatized
standards.
Effect of Mass Resolution. The TSQ Quantum is capable of
enhanced mass resolution, to 0.1 amu full width half-maximum
(fwhm), without drastic loss in sensitivity. This can minimize
interference from extraneous substances. For some analytes,
operating the mass spectrometer at enhanced resolution (0.1 amu
fwhm) greatly improved the specificity when applied to urine
extracts. This was particularly evident for the hydroxyfluorenes
and hydroxyphenanthrenes, for which high background ion
MS/MS spectra of the PFB derivatives of all analytes were
recorded. With the exception of 2-hydroxyfluorene, the major
product ions were (M - PFB - 28)- presumably formed by loss
of CO, as was observed with the underivatized analytes. These
transitions provided cleaner chromatograms of urine extracts than
did the only reasonable alternatives, namely, the parent (M -
PFB)- ions (in the SIM mode) or in the case of 1- and
3-hydroxyfluorene, the (M - PFB - H)- ion, m/z 180, or (M -
PFB - 28 - H)- ion, m/z 152. For 2-hydroxyfluorene, the (M -
PFB - H)- product ion provided much better sensitivity than the
(M - PFB - 28)- ion, as well as excellent specificity. The
ionization process and proposed structures for the product ions,
illustrated for 2-hydroxyfluorene, are depicted in Figure 3. Spectra
594 Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

Table 3. Urine Concentrations of PAH Metabolites in Eighteen Smokers and Eight Nonsmokers Determined by
LC/MS/MS and GC/MS/MSa
LC/MS/MS
GC/MS/MS
correlation
coefficient
analyte
2-naphthol
ng/mL mean (range)
ng/mL mean (range)
regression equationb
10.5
10.8
y ) 1.08x - 0.547
r2 ) 0.965
r2 ) 0.968
r2 ) 0.993
r2 ) 0.976
(BLQ-51.1)
0.65
(BLQ -56)
0.61
1-hydroxyfluorene
y ) 0.758x + 0.116
y ) 0.936x + 0.0265
y ) 0.746x + 0.00448
(BLQ -4.57)
1.12
(BLQ -3.49)
1.08
2-hydroxyfluorene
(BLQ -6.62)
0.13
(0.047-6.29)
0.10
2-hydroxyphenanthrene
(BLQ -0.70)
(BLQ -0.57)
a BLQ, below the lower limit of quantitation. b Equation y ) Mx + B, where y is the concentration by GC/MS/MS and x is the concentration by
LC/MS/MS.
current and extraneous peaks made low-level quantitation difficult
when the instrument was operated at unit resolution (0.7 amu
fwhm). Much cleaner SRM chromatograms were obtained when
Q1 was operated at 0.1 amu fwhm, thus minimizing interfering
peaks and facilitating integration (Figure 5). For 2-hydroxyfluo-
rene, the appearance of the chromatograms was further improved
by operating Q3 at 0.1 amu resolution as well, in that the size of
extraneous peaks were reduced (Figure S-3, Supporting Informa-
tion). However, this did not appear to have a significant effect on
analytical results and caused some loss in sensitivity, so in the
analysis of biofluid samples Q1 resolution was set at 0.1 amu and
Q3 resolution was set at 0.7 amu for all analytes.
Standards and Internal Standards. Our goal was to quan-
titate monophenolic metabolites of naphthalene, fluorene, phenan-
threne, and pyrene (Figure 1). Most of these metabolites were
commercially available, the exceptions being 4-hydroxyfluorene,
2-hydroxypyrene, and 3-hydroxypyrene, which fortunately have
been reported to be minor metabolites.14,48 Consequently, we used
the commercially available standards in our method development.
Using mass spectrometry as the detection method makes it
possible to use stable isotope-labeled internal standards in order
to maximize precision and accuracy. Since acquiring stable
isotope-labeled analogues of all analytes did not appear to be
feasible for our studies, we decided to purchase or synthesize at
least one labeled monophenol derived from each of the four parent
hydrocarbons. Perdeutero-1- and 2-naphthol are commercially
available. An analogue of 1-hydroxypyrene labeled with six carbon-
13 atoms is available from the National Cancer Institute. 2-Hy-
droxyfluorene-d9 and a mixture of hydroxyphenanthrene-d9 iso-
mers were synthesized in our laboratory. The synthetic route is
shown in Figure 2 using 2-hydroxyfluorene-d9 as an example.
Completely deuterated parent hydrocarbons are commercially
available. These were acylated with acetic anhydride and stannic
chloride (Friedel-Crafts, method A) or formylated with R,R-
dichloromethyl methyl ether (method B) to give aryl methyl
ketones or aldehydes, which were then treated with m-chloro-
peroxybenzoic acid (Baeyer-Villiger oxidation) to give the cor-
responding acetoxy- or formyloxyarene. Base hydrolysis of the
esters provided the deuterated phenols. This method yielded
2-hydroxyfluorene-d9, since electrophilic substitution of fluorene
occurs predominantly at this position.49 Electrophilic substitution
of phenanthrene is more complex, and generally a mixture of
isomers is obtained, the composition depending upon the reaction
conditions.50 Mixtures that contained 1-, (and/or 9-), 2-, and
3-hydroxyphenanthrene-d9 as the major components were ob-
tained, as determined by the relative retention times of the PFB
derivatives compared to nondeuterated standards (Figure S-2,
Supporting Information). The amounts of d0-isotopomers in the
hydroxyfluorene-d9 and hydroxyphenanthrene-d9 preparations
were less than 0.1%.
A potential problem with the use of deuterium-labeled phenols
as internal standards is the possibility of deuterium loss by enol-
keto tautomerization during extraction or derivatization. Although
carbon-13-labeled internal standards would eliminate this problem,
obtaining these for naphthols, hydroxyfluorenes, and hydrox-
yphenanthrenes was not feasible at the time our studies were
initiated. Therefore, in order to investigate the possibility of
deuterium exchange, we dissolved 2-naphthol-d7 and a mixture
of hydroxyphenanthrene-d9 isomers in pH 7 buffer, let the solution
stand for 3 days at room temperature, extracted the mixture, and
then subjected the extract to the derivatization procedure. In
another experiment, we incubated the deuterium-labeled metabo-
lites with â-glucuronidase and arylsulfatase at pH 7 for 4 days at
37 °C. These procedures were similar to the extraction/derivati-
zation of biological samples (see Experimental Section), with the
exception that the incubation with buffer (3-4 days) was longer
than the corresponding period (overnight) used for enzymatic
deconjugation of the urine samples. This was done so that these
conditions for potential deuterium exchange would be more
extreme than the conditions used for the biological samples. The
derivatized extracts were analyzed by LC/MS for isotopomers that
could result from deuteron-proton exchange. Interestingly,
deuterium loss occurred to a large extent only with 2-naphthol-
d7, and it was converted to the d6-isotopomer, with further
exchange occurring only at a very slow rate (Table S-2, Supporting
Information). These results suggested a possible solution to the
problem: it should be possible to convert 2-naphthol-d7 to
2-naphthol-d6 and use this as the internal standard. For preparative
purposes, triethylamine proved to be an effective catalyst in
refluxing ethanol/acetonitrile solvent (Figure S-1, Supporting
(49) Keumi, T.; Taniguchi, R.; Kitajima, H. Synthesis 1980, 2, 139-141.
(50) Girdler, R. B.; Gore, P. H.; Thadani, C. K. J. Chem. Soc. [Sect.] C: Org.
1967, 24, 2619-2624.
(48) Gmeiner, G.; Krassnig, C.; Schmid, E.; Tausch, H. J. Chromatogr., B: Biomed.
Sci. Appl. 1998, 705, 132-138.
Analytical Chemistry, Vol. 79, No. 2, January 15, 2007 595

Table 4. Urinary Excretion of PAH Metabolites in Smokers and Nonsmokersa
ng/mL mean
(range)
pmol/mg creatinined
mean (range)
smokers
nonsmokers
smokers
nonsmokers
analyte
(N ) 21)
(N ) 22)
(N ) 21)
(N ) 21)
2-naphtholb
14.3
2.44
p < 0.0001
p ) 0.0002
p < 0.0001
110
19.0
p < 0.0001
p < 0.0001
p < 0.0001
p < 0.0001
p ) 0.0007
p < 0.0001
p < 0.0001
p < 0.0001
p < 0.0001
(1.7-51.1)
0.96
(BLQ-17.5)
all BLQ
(41-190)
5.41
(3.5-88.5)
0.97
1-hydroxyfluorene
(BLQ-4.57)
1.60
(1.54-14.7)
9.50
(0.17-2.75)
0.90
2-hydroxyfluorene
0.11
(0.20-6.62)
(BLQ-0.33)
(3.22-24.4)
14.9
(0.36-2.18)
1.86
sum of 1- and 2-hydroxyfluorene
1-hydroxyphenanthrene
2-hydroxyphenanthrene
3- + 4-hydroxyphenanthrenec
sum of hydroxyphenanthrenes
1-hydroxypyrene
(5.29-39.1)
1.54
(0.53-3.75)
0.69
0.28
0.095
p ) 0.0067
p ) 0.0019
p ) 0.0025
(0.029-1.34)
0.16
(BLQ-0.46)
0.04
(0.36-4.86)
0.85
(0.16-1.62)
0.29
(0.017-0.70)
0.450
(BLQ-0.15)
0.063
(0.35-1.91)
2.31
(0.14-0.76)
0.52
(0.066-2.13)
(BLQ-0.23)
(0.86-6.48)
4.69
(0.24-1.92)
1.49
(1.78-11.5)
1.59
(0.62-4.17)
0.39
0.33
0.061
p ) 0.0044
(0.029-2.00)
(BLQ-0.23)
(0.32-4.04)
(0.093-0.77)
a Comparison of smokers and nonsmokers by unpaired t-test. For determining means and ranges of creatinine-normalized values, concentrations
below the lower limit of quantitation (BLQ) were assigned the value LOQ divided by the square root of 2. b The isomeric metabolite 1-naphthol
coeluted. Since the MS response was ∼5% of that of 2-naphthol, and human urine concentrations of the two metabolites are generally similar, the
analytical results should be fairly close to those of 2-naphthol for most samples. See text, Table 3, and Figure S-5. c The relative response of
4-hydroxyphenanthrene was ∼25% that of 3-hydroxyphenanthrene. Concentrations of 3-hydroxyphenanthrene generally exceed those of
4-hydroxyphenanthrene in human urine by 5-10-fold18,21 so this sum should be a good estimate of the concentration of 3-hydroxyphenanthrene
d
for most samples. Concentrations normalized to creatinine to adjust for differences in urinary flow.
Information). Loss of one deuterium atom was complete after a
2-h reflux period, with only a few percent loss of a second
deuterium. The relative abundance of m/z 143, corresponding to
the unlabeled isotopomer, was 0.04%. Syntheses of the internal
standards and the experiments on deuterium exchange are
described in the Supporting Information. Subsequent to the
development of our method, 13C-labeled 1-, 3-, and 4-hydroxy-
phenanthrenes and 2-naphthol have become available com-
mercially.
Sample Preparation. The phenolic metabolites of PAHs are
excreted in urine largely as the glucuronide conjugates, although
small amounts of the sulfates and free phenols may also be
excreted.6,22 Consequently, the first step in nearly all reported
methods for PAH metabolite determination involves incubation
with â-glucuronidase and arylsulfatase to convert conjugates to
the free phenols. Most published methods have used an enzyme
mixture isolated from Helix pomatia.15,17,23,26 We found that using
â-glucuronidase from E. coli and sulfatase from Aerobacter aero-
genes gave cleaner extracts (fewer emulsions) and better recovery
of analytes than with samples incubated with the Helix enzymes.
Following deconjugation, the free phenols are extracted with a
90:10 mixture of pentane/ethyl acetate. The extracts are then
evaporated and converted to the PFB derivatives by a phase-
transfer procedure using pentafluorobenzyl bromide and tetrabu-
tylammonium bromide as the catalyst. The derivatives are
extracted into pentane, the pentane extracts are evaporated, and
reconstituted in methanol for LC/MS/MS analysis.
assays.51 In addition, some investigators have added surfactants
to samples in order to minimize losses.52 Initially we utilized
aqueous instead of urine calibration standards, since all urine
samples contain PAH metabolites. Surprisingly, recoveries (de-
termined from stable isotope-labeled internal standard peak areas)
were greater from spiked urine than from spiked water! Further-
more, recoveries from an “artificial urine” (see Experimental
Section) developed as a blank matrix for preparation of standards
were greater than from water and were generally similar to
recoveries from real urine specimens (Figure 6). On the assump-
tion that the losses were due to adsorption on surfaces of
extraction glassware, inhibited by surfactants in the urine samples,
Triton X-100 was added to aqueous standards, but this had little
effect. However, addition of antioxidants (R-tocopherol, BHA,
quercetin, gallic acid) did improve recoveries. The hydroxyfluo-
renes and hydroxyphenanthrenes were most susceptible to
oxidation. It was found that losses were occurring both during
the evaporation and derivatization steps, and addition of a small
amount of gallic acid to the extracts prior to evaporation and
derivatization was effective for inhibiting losses. (Figure S-4,
Supporting Information). Recoveries of labeled internal standards
ranged from 72 to 83%. Final extracts of derivatives in methanol
were stable for at least 24 h at room temperature in autosampler
trays and for at least 5 days at -20 °C.
Chromatography. Our goal was to separate as many of the
isomeric metabolites as possible with a reasonably short chro-
matographic run time. Phenols derived from different hydrocar-
bons could be distinguished by mass, but isomeric compounds
Analyte Stability and Extraction Recovery. Phenolic com-
pounds are susceptible to oxidation, and some investigators have
added antioxidants to prevent losses from occurring during
extraction procedures or chromatography in PAH metabolite
(51) Grimmer, G.; Jacob, J.; Dettbarn, G.; Naujack, K. W. Int. Arch. Occup.
Environ. Health 1997, 69, 231-239.
(52) Simon, P.; Morele, Y.; Delsaut, P.; Nicot, T. J. Chromatogr., B: Biomed. Sci.
Appl. 1999, 732, 91-101.
596 Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

could not, since MS/MS produced the same product ions. Each
parent hydrocarbon may be ring-hydroxylated to isomeric monophe-
nolic metabolites: two for naphthalene, four for fluorene, five for
phenanthrene, and three for pyrene (Figure 1). All are reported
to be human metabolites. We evaluated several columns, and of
these, the best separation of the isomeric PFB derivatives was
achieved with a Thermo-Hypersil PAH Green column (Figure 7).
The three commercially available monophenolic hydroxyfluorene
isomers (1-, 2-, and 3-hydroxyfluorene) were all separated. 4-Hy-
droxyfluorene has been detected in human urine, but of the four
isomers, it appears to be the least abundant.14 9-Hydroxyfluorene
appears to be a minor metabolite,14 and being an aliphatic alcohol,
it predictably did not derivatize under our conditions. The PFB
derivatives of 1- and 2-naphthol were not separated. Sensitivity
for 1-naphthol was much lower than that for 2-naphthol by a factor
of ∼20, as determined by integration of SRM chromatograms of
derivatized pure standards injected separately, apparently due
largely to inefficient fragmentation of 1-naphthol anion under the
CID conditions. Since the isomeric metabolites are generally
excreted in comparable amounts in human urine,53 the concentra-
tions determined by our method should be fairly close to those
of 2-naphthol for most samples. In support of this, 2-naphthol
concentrations in 18 urine samples determined by LC/MS/MS
were similar to those determined by GC/MS/MS, for which the
isomeric naphthols were well separated (see below, Figure 8,
Table 3, and Figure S-5). Of course, if determination of both
metabolites were important, this would be a limitation of our
method. The PFB derivatives of 1- and 9-hydroxyphenanthrene
had identical retention times, but since the 9-hydroxy isomer has
been reported to be a minor metabolite in humans,26,51 and the
sensitivity for 9-hydroxyphenanthrene was only 30% of 1-hydroxy-
phenanthrene, the analytical results can be considered to be
mainly 1-hydroxyphenanthrene concentrations. The PFB deriva-
tives of 3- and 4-hydroxyphenanthrene were partially resolved and
determined as a sum. Since concentrations of 3-hydroxyphenan-
threne in human urine have been reported to generally exceed
those of 4-hydroxyphenanthrene by 5-10-fold,18,21 and since the
relative response of 4-hydroxyphenanthrene was 25% of 3-hydrox-
yphenanthrene, analytical results should be a good estimate of
3-hydroxyphenanthrene concentrations for most samples. As with
2-naphthol, if separation of all isomeric metabolites were important,
this would be a limitation of our method. 1-Hydroxypyrene was
the only pyrene metabolite available to us; it has been reported
to be the major metabolite.7
Artificial urine blanks, treated with â-glucuronidase and arylsul-
fatase and derivatized in the same manner as standards and urine
samples were free from interfering substances (Figure 7A).
Precision and accuracy, evaluated by analyzing nonsmokers’ urine
spiked with the analytes at concentrations spanning the expected
ranges, were very good for all analytes except 3-hydroxyfluorene,
which was erratic and for which reliable quantitation was not
possible (Table 2). Fortunately, 2-hydroxyfluorene is the major
fluorene metabolite,14,15 and precision, accuracy, and sensitivity
for it were excellent. Limits of quantitation (LOQ) were deter-
mined by analyzing spiked artificial urine, because the sensitivity
of the method was good enough to determine concentrations
below those found in urine of nonexposed persons, for most
analytes (Table 2). We used the criteria of Shah et al.54 to validate
the method for precision and accuracy and to determine the LOQ.
These are precision (CV) of (15% and accuracy within (15% of
the expected amount, except at the lower limit of quantitation,
for which (20% is considered acceptable.
To further validate the method, derivatized extracts of urine
samples were analyzed by both LC/MS/MS and GC/MS with
NICI. Without putting much effort into optimizing chromato-
graphic separations or mass spectrometric parameters, we were
able to quantitate some metabolites using GC/MS. 1- and
2-naphthol PFB derivatives were easily separated, as were the
derivatives of all three available hydroxyfluorene isomers. 2-Hy-
droxyphenanthrene was well separated from the other isomers;
1- and 3-hydroxyphenanthrene coeluted; sensitivity was poor for
4- and 9-hydroxyphenanthrene. We were unable to detect the
1-hydroxypyrene PFB derivative even at column temperatures
considerably above those satisfactory for the other analytes. We
evaluated both SIM and SRM, and as expected, SRM gave cleaner
chromatograms of urine extracts. Using SIM, the hydroxyphenan-
threne PFB derivative peaks were lost in baseline noise, but it
was possible to quantitate 2-hydroxyphenanthrene using SRM
(Figure 8). A fourth peak was present in the m/z 181-153 SRM
chromatograms of urine extracts, but not in standards. We
speculate that this is due to the presence of 4-hydroxyfluorene,
for which a standard was not available. For those analytes that
could be quantitated by both methods, there was generally a good
correlation between LC/MS/MS and GC/MS/MS results. (Table
3, Figure S-5).
Determination of PAH Metabolites in Urine of Smokers
and Nonsmokers. For the metabolites of all four PAHs, con-
centrations were significantly higher in smokers compared to
nonsmokers (Table 4). Interestingly, the differences were most
pronounced for hydroxyfluorenes, and for these two groups of
subjects, there was no overlap in creatinine-normalized excretion
of 2-hydroxyfluorene. In contrast, there was some overlap in
concentrations between the two groups for the widely used PAH
biomarker 1-hydroxypyrene, and for the other PAH metabolites,
as has been previously reported.55 The relatively high specificity
of fluorene for tobacco smoke compared to other PAHs was
suggested by data from the literature on emissions from various
sources8-11 and supports the value of measuring metabolites of
several PAHs in studies of exposure, as suggested by other
Performance of the Method. Since PAH metabolites are
present in all human urine specimens due to environmental and
dietary exposure, working standards were prepared in “artificial
urine” (Experimental Section), by spiking with a stock solution
prepared in methanol. To validate this approach, controls were
prepared by spiking nonsmokers’ urine with the analytes, and the
calculated amounts were compared to the amounts expected based
on the difference between spiked and nonspiked results. Calibra-
tion curves were prepared by linear regression with 1/X weight-
ing. These were linear over the ranges 0.25-50 ng/mL for
2-naphthol and 0.01-5 ng/mL for the other analytes. Correlation
coefficients for standard curves are in the Experimental Section.
(54) Shah, V. P.; Midha, K. K.; Findlay, J. W.; Hill, H. M.; Hulse, J. D.; McGilveray,
I. J.; McKay, G.; Miller, K. J.; Patnaik, R. N.; Powell, M. L.; Tonelli, A.;
Viswanathan, C. T.; Yacobi, A. Pharm. Res. 2000, 17, 1551-1557.
(55) Hecht, S. S. Carcinogenesis 2002, 23, 907-922.
(53) Yang, M.; Koga, M.; Katoh, T.; Kawamoto, T. Arch. Environ. Contam. Toxicol.
1999, 36, 99-108.
Analytical Chemistry, Vol. 79, No. 2, January 15, 2007 597

investigators.6,16-20 There were also large differences in 2-naphthol
concentrations between the two groups. It has been proposed that
urinary naphthols may be particularly appropriate biomarkers for
inhalation exposure to PAHs and that 2-naphthol is a better
biomarker than 1-naphthol for inhalation exposure.19
In summary, a method for the simultaneous determination of
ring-hydroxylated metabolites of four PAHs using liquid chroma-
tography with electron capture atmospheric pressure chemical
ionization mass spectrometry has been developed. Applicability
of the method to studies of PAH exposure in smokers has been
demonstrated, and sensitivity is adequate for determination of
several PAH metabolites in urine of persons without unusually
high exposures.
Medical Research Institute. 1-Hydroxypyrene-13C6 was obtained
from the National Cancer Institute Chemical Carcinogen Refer-
ence Standard Repository. The urine samples were obtained from
studies carried out in part at the General Clinical Research Center
at San Francisco General Hospital Medical Center with support
of the Division of Research Resources, National Institutes of
Health (RR-00083). The authors thank Lisa Yu for assistance with
LC/MS/MS operation and data analysis, Dr. Faith Allen for
Statistical analyses, Dr. Alexander Shulgin for advice, and Marc
Olmsted for editorial assistance.
SUPPORTING INFORMATION AVAILABLE
Additional information as noted in text. This material is
available free of charge via the Internet at http://pubs.acs.org.
ACKNOWLEDGMENT
This study was supported by the State of California Tobacco-
Related Disease Research Program Grant 10RT-0215, Grant
CA78603 from the National Cancer Institute, and Grants DA02277
and DA12393 from the National Institute on Drug Abuse, National
Institutes of Health, and a grant from the Flight Attendant’s
Received for review May 18, 2006. Accepted October 30,
2006.
AC060920L
598 Analytical Chemistry, Vol. 79, No. 2, January 15, 2007