Journal of Immunological Methods 251 (2001) 19

www.elsevier.nl / locate / jim

Validity of an ELISA for N-acetyltransferase-2 (NAT2)

phenotyping
P. Wong a , K. Banerjee a , J. Massengill b , S. Nowell b , N. Lang b ,
a,
B. Leyland-Jones *
a
Department of Oncology, McIntyre Medical Sciences Building, McGill University, Suite 701, 3655 Promenade Sir William Osler,
Montreal, Quebec, Canada H3 G 1 Y6
b
University of Arkansas for Medical Sciences and Central Arkansas Veterans Health Care System, 4300 W. 7 th (112 -LR), Little Rock,
AR 72205, USA

Received 17 July 2000; received in revised form 21 December 2000; accepted 23 December 2000

Abstract

A competitive antigen ELISA was previously developed for NAT2 phenotyping, using caffeine as the probe drug. The
ELISA phenotypes by measuring the ratio of 5-acetamido-6-amino-3-methyluracil (AAMU) and 1-methylxanthine (1X) after
transformation of 5-acetamido-6-formylamino-3-methyluracil (AFMU) to AAMU, in contrast to capillary electrophoresis
high-pressure liquid chromatography (HPLC) which phenotype by measuring the AFMU / 1X ratio. The ELISA phenotyping
was previously determined in 30 samples and correlated well with phenotypes determined by capillary electrophoresis
(29 / 30). The correlation was extended with the standard HPLC methodology by expanding the data set by 146 in order to
test the validity of the ELISA methodology. The correlation with HPLC in this larger sample size was 96%; whereas the
correlation between the two methods for determination of 1X was high (r 2 50.90), that for determination of AAMU by
ELISA and AFMU by HPLC was low (r 2 50.53). The poor correlation between the two methodologies could not be
attributed to the age of urine samples, nor to a significant decomposition of AFMU in the body prior to collection of the
urine sample. The addition of a simple caffeine metabolite extraction method, originally developed for HPLC analysis of
metabolites, to the ELISA phenotyping protocol produced a methodology with absolute correlation to the standard HPLC
method. 2001 Elsevier Science B.V. All rights reserved.

Keywords: Caffeine; Metabolic phenotyping; N-Acetyltransferase-2 (NAT2); ELISA

1. Introduction droxylated arylamines. NAT2 is polymorphic re-

N-Acetyltransferase-2 (NAT2) is a phase II meta-
bolic enzyme of xenobiotics and catalyzes N-acetyla-
tion of arylamines and N-acetoxylation of N-hy-
*Corresponding author. Tel.: 11-514-398-8986; fax: 11-514-
398-5071.
E-mail address: leylandj@med.mcgill.ca (B. Leyland-Jones).
sulting in two distinct phenotypes: slow and fast
acetylators. The relative proportion of slow and fast
acetylators varies widely between ethnic groups,
with slow acetylators representing 5% in Inuit, 10
20% in Japanese, 5060% in Caucasians, 74% in
Hmong residing in US and 90% in Northern Africans
(Le Marchand et al., 1996; Straka et al., 1996). The
NAT2 polymorphism is of clinical interest because

0022-1759 / 01 / $ see front matter 2001 Elsevier Science B.V. All rights reserved.
PII: S0022-1759( 01 )00310-6
2 P. Wong et al. / Journal of Immunological Methods 251 (2001) 1 9
of its association with both adverse drug effects and
cancer susceptibility (Clark, 1985; Evans, 1989;
Rothman et al., 1993; Ambrosone et al., 1996; Lang
et al., 1994; Vineis et al., 1994; Ratain et al., 1995;
ONeil et al., 1997; Cascorbi et al., 1999). The
ability to rapidly and cost effectively determine an
individuals NAT2 phenotype will be most valuable
in identifying individuals prone to adverse drug
effects and allowing the prescription of suitable drug
regimens. In addition, this information will allow
physicians to identify individuals with increased
risks of certain forms of cancer.
NAT2 phenotype can be determined using caffeine
as a probe drug (Grant et al., 1983; Lloyd et al.,
1992; Meacher et al., 1998). The current method
involves the extraction of caffeine metabolites from
urine and separation by high-pressure chromatog-
raphy (HPLC) or by capillary electrophoresis; the
ratio of AFMU and 1X is subsequently determined.
NAT2 phenotyping can also be extrapolated from
genotyping, as determined by allele-specific amplifi-
cation of peripheral leukocyte DNA using the poly-
merase chain reaction and restriction fragment length
polymorphism. However, we believe that the de-
termination of genotype as compared to phenotype is
limited (Table 1).
We have previously reported a competitive antigen
ELISA for NAT2 phenotyping using caffeine as a
probe drug (Wong et al., 1995). The ELISA method
has the advantage that it is more rapid and efficient
then the current HPLC method. Moreover, it can be
easily and immediately utilized in hospitals and
commercial laboratories in contrast with other tech-
niques which can only be performed by specialist
research laboratories. Additionally, we estimate the
cost for determination of NAT2 phenotype by HPLC
Table 1
Comparison of the characteristics of using phenotyping and genotyping for determining metabolic phenotype of drug metabolism enzymes
Phenotyping
Quantitative (actual rate of activity)
One measurement summates all
various polymorphisms
Direct measurement of function
Incorporates function, viz.:
disease state of the individual
exposure to suppressors and inducers
polypharmacy in the elderly
to be |$180; this is significantly higher than the cost
of the majority of standard ELISA testing currently
performed in hospital clinical chemistry laboratories.
Initial data obtained using non-extracted urine sam-
ples suggested that the ELISA methodology was
valid for NAT2 phenotyping, as it correctly iden-
tified the NAT2 phenotypes in 29 of the 30 in-
dividuals phenotyped by capillary electrophoresis.
This article reports the results of an expanded sample
set. The initial ELISA methodology produced an
excellent (96%), but not absolute correlation with the
HPLC method. We demonstrated that the lack of
absolute correlation is unrelated to the intrinsic
instability of AFMU but that the addition of a simple
extraction to isolate the caffeine metabolites prior to
ELISA analysis produced a 100% correlation with
HPLC phenotyping. In addition, we briefly comment
on the optimization of the ELISA configuration.
2. Materials and methods
2.1. Materials
Bovine serum albumin (BSA) (catalogue no. A-
3803); 1-ethyl-3-(3-dimethylaminopropyl) carbodi-
imide hydrochloride (EDAC); protein A-Sepharose
4B; 1-methylxanthine (1X); polyoxyethylene sor-
bitan monolaurate (Tween 20) were purchased from
SigmaAldrich (St Louis, MO, USA). Horseradish
peroxidase (HRP) was purchased from Boerhinger
Mannheim Canada (Montreal, Quebec). Keyhole
Limpet Hemocyanin (KLH) was purchased from
Pierce (Rockford, IL, USA); 5-acetamido-6-formyl-
3-methyluracil (AFMU) was a generous gift of Dr.
G. Villeneuve; 5-acetamido-6-amino-3-methyluracil
Genotyping
Non-quantitative (fast or slow)
Need to measure multiple polymorphisms
separately
Indirect measurement of function
Fails to take into account:
disease state of the individual
exposure to suppressors and inducers
polypharmacy in the elderly
 P. Wong et al. / Journal of Immunological Methods 251 (2001) 1 9
(AAMU), AAMU-hemisuccinic acid and 1X8-pro-
pionic acid were obtained by synthesis (Wong et al.,
1995); ELISA plates (Corning 96-well Easy WashE
modified flat bottom, high binding) and Costar 96-
well tissue culture plate no. 3595 were obtained from
Fisher (Montreal, Quebec, Canada). Other reagents
used were ACS grade.
2.2. Conjugates
Conjugates of AAMUBSA, 1XBSA were pre-
pared by a procedure similar to that of Rojo et al.
(1986). Fifteen milligrams of BSA were dissolved in
6 ml of an AAMUhemisuccinic acid derivative (or
1Xpropionic acid derivative) solution (1.25 mmol /
ml of water). After adding 1.43 ml of an EDAC
solution (10 mg / ml of water), the solution was
stirred overnight and dialyzed against 250 ml water
for 48 h with three changes. Conjugates of AAMU
KLH and 1XKLH were prepared as follows. To a
1.1 ml KLH solution (10 mg / ml dialyzed against a
0.9 M NaCl solution), was added 0.8 ml of the
AAMU or 1X derivative solution (2.5 mmol / ml of a
0.9 M NaCl solution), 2 ml of an EDAC solution (10
mg / ml of a 0.9 M NaCl solution), and 1.8 ml of a
0.9 M NaCl solution. The solution was stirred
overnight and dialyzed against 125 ml of a 0.9 M
NaCl solution for 48 h with three changes. Conju-
gates of AAMUHRP and 1XHRP were prepared
by the mixed anhydride method by a procedure
previously described (Wong et al., 1995).
2.3. Antibodies
Procedures for raising antibodies in rabbits against
AAMUKLH and 1XKLH conjugates and for
determining antiserum titers have been previously
described (Wong et al., 1995). Rabbit antisera were
stored at 2208C. Rabbit IgG antibodies against
Table 2
Characteristics of four sets of urine samples collected for determining NAT2 phenotype
Sample Number
group of samples
1 129
2 10
3 11
4 135
3
AAMUKLH and 1XKLH were purified by affini-
ty chromatography on a protein ASepharose 4B
column as follows. The affinity column (1 ml) was
equilibrated with 0.15 M NaCl, 0.01 M sodium
phosphate buffer, pH 8.0 (PBS), and then washed
with 34 ml of a 0.1 M trisodium citrate buffer, pH
3.0. The column was then re-equilibrated with PBS.
One milliliter of rabbit antiserum was diluted with 1
ml PBS and the resulting solution was slowly applied
to the column. The column was washed with 15 ml
PBS and eluted with a 0.1 M trisodium citrate buffer,
pH 3.0. Three fractions of 2.2 ml were collected in
graduate tubes containing 0.8 ml of 1 M TrisHCl
buffer, pH 8.5. The purified rabbit IgG antibodies
were stored at 4 8C in the presence at 0.01% sodium
azide.
2.4. Urine samples and extraction of the caffeine
metabolites
Four groups of urine samples were used for NAT2
phenotyping by ELISA (Table 2). Those samples
collected at the University of Arkansas were
phenotyped by HPLC and frozen aliquots were sent
blinded to the University of McGill (except group 2).
Caffeine metabolites were extracted from urine
samples according to the procedure of Butler et al.
(1992), with the modification that 20 ml of 0.5
mg / ml of the internal standard acetaminophen were
added per 200 ml of urine. The urine extract samples
were stored at 2208C.
2.5. Preparation of samples for determination of
AAMU and 1 X by ELISA
Urine samples (non-acidified and acidified to pH
3. 5 with HCl) were prepared as follows for de-
termination of AAMU by ELISA. One hundred ml of
the urine sample were mixed with 100 ml of a 0.1 N
Location of collection ELISA performed
and HPLC analysis blinded (Y/ N)
University of Arkansas Y
University of Arkansas N
McGill University N
University of Arkansas Y
4 P. Wong et al. / Journal of Immunological Methods 251 (2001) 1 9
NaOH solution for the transformation of AFMU into
AAMU. After 10 min, 100 ml of a 0.1 N HCl
solution and 700 ml of 310 mOsm sodium phosphate
buffer, pH 7.5 (PB), were sequentially added. The
solution was then appropriately diluted with PB (80
20003). Two hundred ml of the diluted sample were
mixed with 200 ml of PB containing 2% BSA, 0.1%
Tween 20, 10.7 mg / ml AAMUHRP. Urine samples
acidified with glacial acetic acid to pH 3.5 were
prepared as above with the addition of the following
preceding step. The pH of 5 ml of the sample was
adjusted to 67 with a 0.11 N NaOH solution and
the volume of the urine solution was completed to 10
ml with water. For determination of AAMU in urine
extracts, the samples were prepared as above, but
with the following modification. A sample of 25 ml
was lyophilized for 2030 min. The residue was
dissolved in 50 ml of water and 50 ml of a 0.1 N
NaOH solution were added. After 10 min, 50 ml 0.1
N HCl and 350 ml PB were sequentially added. The
dilution factor varied between 40 and 400 for the
urine sample extracts. For determination of 1X in
non-extracted and extracted urine samples, the sam-
ples were prepared as above, with the following
differences; the transformation step of AFMU to
AAMU was omitted; the non-extracted urine samples
were diluted with PB by a factor of 4200; 25100
ml of the urine sample extract were lyophilized and
the residue was dissolved with 200800 ml of PB;
the dilution factor varied between 4 and 64 for the
urine sample extracts.
2.6. Competitive antigen ELISA
The competitive antigen ELISA was performed as
described previously (Wong et al., 1995) with the
following modifications: (a) wells of the ELISA plate
were washed with a Nunc-Immuno wash 12 washer
and they were coated with antibodies at 4 8C for 20
h; (b) 150 ml were used per well in contrast to 100
ml used previously; (c) unoccupied sites were
blocked by incubation for 3 h with PB containing 1%
BSA solution, 0.05% Tween 20; (d) for determi-
nation of AAMU and 1X concentrations, samples of
400 ml were prepared (see Section 2.5); (e) 200 ml of
each sample were pipetted in duplicate in a Costar
96-well culture tissue plate according to the pattern
shown in Fig. 3; (f) 150 ml of samples were
transferred in the corresponding wells of a 96-well
ELISA microtiter plate coated with antibodies; (g)
after the addition of the samples, the microtiter plates
were covered and left standing for 2 h.
2.7. High pressure liquid chromatography
High-pressure liquid chromatography was per-
formed using a Spectra-Physics P100 pump model, a
Rheodyne model 7123 Syringe loading sample injec-
tor with a 20-ml loop, a Spectra-Physics detector
(UV 100) model and a Spectra-Physics DataJet
integrator model. Caffeine metabolites were sepa-
rated on a Luna 5m C18 (2) column (25034.6 mm)
(Phenomenex; Torrance, CA, USA). Samples of 20
mL were injected and the column eluted isocratically
at a flow rate of 1 ml / min with a 5% acetonitrile,
0.05% acetic acid solution (v / v) (Lloyd et al., 1992).
The eluate was monitored at 280 nm. Retention
times for AAMU, AFMU and 1X were 3.8, 5.7 and
11.5 min, respectively. Concentrations of AFMU and
1X in urine samples were determined by peak area
ratio using the internal standard acetaminophen,
which eluted with a retention time of 19.3 min.
3. Results and discussion
3.1. Validation of the ELISA
Urine samples were collected from patients 45 h
after ingestion of caffeine. The 146 samples were
then phenotyped by the standard HPLC protocol.
The samples were then phenotyped again using the
ELISA methodology with non extracted-diluted urine
samples. Using an antimode of 2.0, the ELISA
correctly identified the NAT2 phenotypes in 96% of
the 146 urine samples that had been initially
phenotyped by HPLC (the urine samples of groups
13, 66 fast acetylators (45.2%) and 80 slow
acetylators (54.8%) with six non-concordant sam-
ples). To determine the possible reasons for the
imperfect correlation, we examined the correlation
between the ELISA and HPLC measurements for the
individual metabolites. For 1X, the correlation was
high (r 2 50.90) and 1X concentrations determined by
ELISA were on average 1.3860.34 (n518) higher
than that determined by HPLC (urine samples of the
 P. Wong et al. / Journal of Immunological Methods 251 (2001) 1 9 5

Fig. 1. Correlation curves of HPLC versus ELISA for determination of 1X, AAMU and AFMU obtained with newly collected urine
samples. The same patterns were obtained for correlation curves of urine samples collected more than a year ago. NAT2 phenotypes
assigned by both HPLC and ELISA were the same. See text for additional details.

third group) (Fig. 1B). However, the correlation
between AAMU (transformed AFMU) as measured
by ELISA and AFMU as measured by HPLC was
low (r 2 50.53), and AAMU concentrations deter-
mined by ELISA were on average 4.6064.17 (n5
18) times higher than those of AFMU determined by
HPLC (urine samples of the third group) (Fig. 1A). It
was thought that the poor correlation between ELISA
for determination of AAMU and HPLC for de-
termination of AFMU could be due to a lack of
AFMU stability in older urine samples or the de-
composition of AFMU to AAMU in the body prior
to collection of the urine samples. The effect of urine
age as a factor was excluded as the poor correlation
was observed in fresh urine samples as well as in
urine samples stored for 2 years. Stability studies of
AFMU indicated that no more than 15% of AFMU
decomposes to AAMU under physiological condi-
tions within the 4-h period prior to urine sample
collection after caffeine intake (unpublished results).
Thus, AFMU decomposition could not account for
the lack of correlation for AFMU and AAMU levels
between the two methods.
Since our aim is to produce an ELISA based
phenotyping assay with absolute correlation to the
current HPLC methodology, and whereas the ELISA
methodology using non-extracted urine produced
only a 96% correlation with HPLC, we investigated
whether the method of extraction of the caffeine
metabolites used for HPLC analysis could be used
prior to ELISA to improve the degree of correlation.
We performed the extraction on 20 samples obtained
in our laboratory (pre- and post-caffeine ingestion)
and found that the use of urine extraction produced a
100% correlation between not only HPLC and
ELISA phenotypes, but also between AAMU levels
as measured by ELISA and AFMU levels as mea-
sured by HPLC. Next the six discordant samples (of
the 146 samples tested without extraction) were re-
phenotyped by ELISA using urine extracts. Using
the extracts, these six samples produced phenotypes
by ELISA that corresponded to the HPLC
phenotypes. Therefore, these results suggested that
the use of a simple extraction procedure on the urine
samples prior to ELISA analysis produced a 100%
correlation in NAT2 phenotyping between ELISA
and HPLC. Using the extraction methodology, corre-
lation coefficients between the two methods for
determining AAMU and AFMU is 0.96 (Fig. 2A),
while that for determining 1X is 0.90 (Fig. 2B).
Moreover, ELISA and HPLC were in agreement for
1X, AAMU and AFMU concentrations, as demon-
6 P. Wong et al. / Journal of Immunological Methods 251 (2001) 1 9

Fig. 2. Correlation curves of HPLC versus ELISA for determination of 1X, AAMU and AFMU obtained after extraction of caffeine
metabolites from newly collected urine samples. Extractions were performed with the same urine samples used for the determinations of the
correlation curves of Fig. 1. NAT2 phenotypes assigned by both HPLC and ELISA were the same. See text for additional details.

strated by a slope close to the value of 1, in the
correlation curves of HPLC versus ELISA for these
caffeine metabolites.
As the urine extraction procedure had only been
performed on 26 samples, we used the new protocol
to phenotype the group four urine samples (Table 2).
The ELISA phenotyping results matched the HPLC
phenotypes in all 135 samples, with the exception of
three samples that had borderline ratios. In these
three samples the metabolic ratios agreed, but due to
the nature of phenotyping, the choice of a cut off
value for slow versus rapid metabolizers caused a
discrepancy.
On the basis of the above results, NAT2 phenotyp-
ing by the ELISA has to be performed after ex-
traction of caffeine metabolites to avoid ambiguities
using urine samples in the assay.
3.2. ELISA configuration optimization
To ensure a consistent reproducibility of results
using our ELISA methodology we have examined
the impact of 96-well plate location on the ex-
perimental results. Fig. 3 shows the design used for
dispensing the standard solutions and samples in
wells of the ELISA plate. The choice of this design
was based on the following observations.
(a) While a sufficiently accurate calibration curve
can be defined with 15 standards over the con-
centration range of 10 24 to 10 29 M, we have
chosen to determine our calibration curve with 23
serially diluted standard solutions over the con-
centration range of 10 24 to 10 213 M since it
produces a better defined curve and allows the
removal of any absorbency values of standard
solutions that are clearly erroneous without com-
promising the calibration curves accuracy.
(b) An effect of well position on absorbency
values was observed as the absorbency value of a
given sample perceptibly decreased from column
1 to column 12. To ensure that the effect of well
position on absorbency values did not create an
inaccurate calibration curve, the curve was de-
termined in duplicate with standards being
aliquotted in wells at opposite ends of the 96 well
plate (Fig. 3).
(c) Under the conditions of the ELISA, the
calibration curves were smooth and reproducible
 P. Wong et al. / Journal of Immunological Methods 251 (2001) 1 9 7

Fig. 3. Dispositions of standards and samples in the wells of a 96-well microtiter plate. The standard curve of AAMU (or 1X) and
concentrations of AAMU (or 1X) in samples were determined in duplicate. Blank (Blk), Standards (STD) and samples (S) were disposed in
the wells of a 96-well microtiter plate, as shown in the figure. Standard concentrations varied between 10 24 and 10 213 M and decreased
from STD1 to STD23. See the text for the choice of this design of ELISA for NAT2 phenotyping.

(Fig. 4): the average and standard values of EC 50
derived from the calibration curves of AAMU and
1X were 1.1060.15310 27 M (n512) and
4.3560.26310 26 M (n510), respectively.
(d) Wells A1-H3 and A9-H12 were used for the
standard solutions as the concentrations of
AAMU and 1X were generally found to be
overestimated in wells A1-H1 and wells A12-
H12, as indicated by lower AAMU and 1X
concentration values obtained by HPLC.
(e) Wells H4-H8 were not used for determination
of AAMU and 1X in sample solutions as the
concentrations were generally underestimated in
these wells, as indicated by higher AAMU and
1X concentration values obtained by HPLC: an
underestimation of as much as 100% could be
observed when AAMU and 1X concentrations in
the extracts were very low.
(f) Wells A4A8 were also not used for the
determination of AAMU and 1X as the con-
centrations tended to be overestimated, as indi-
cated by lower values of AAMU and 1X con-
centrations obtained by HPLC.
(g) AAMU and 1X concentrations were deter-
mined in duplicate and were within 5% of each
other when the sample were pipetted in wells in
the same row but separated by three columns
(e.g., A4 and A7). Moreover, the inter-assay
coefficient of variation (CV) was less than 10%
when the samples were pipetted in the same wells
8 P. Wong et al. / Journal of Immunological Methods 251 (2001) 1 9
Fig. 4. Calibration curves for determination of 1X and AAMU by ELISA. Each calibration curve represents the average of two calibration
curves. Data points without standard deviation bars indicate that deviations of the absorbency values are equal or less than the size of the
symbols representing the data points. See text for additional details.
from one day to another and that the conditions of
ELISA were precisely followed.
Acknowledgements
This study supported in part by NCI grant number
CA55751.
References
Ambrosone, C.B., Freudenheim, J.L., Graham, S., Marshall, J.R.,
Vena, J.E., Brasure, J.R., Michalek, A.M., Laughlin, R.,
Nemoto, T., Gillenwater, K.A., Harrington, A.M., Shields,
P.G., 1996. Cigarette smoking, N-acetyltransferase 2 genetic
polymorphisms, and breast cancer risk. J. Am. Med. Assoc.
276, 1494.
Butler, M.A., Lang, N.P., Young, J.F., Caporaso, N.E., Vineis, P.,
Hayes, R.B., Teitel, C.H., Massengill, J.P., Lawsen, M.F.,
Kadlubar, F.F., 1992. Determination of CYP1A2 and NAT2
phenotypes in human populations by analysis of caffeine
urinary metabolites. Pharmacogenetics 2, 116.
Cascorbi, I., Brockmoller,
J., Mrozikiewicz, P.M., Muller, A.,
Roots, I., 1999. Arylamine N-acetyltransferase activity in man.
Drug Metab. Rev. 31, 489.
Clark, D.W., 1985. Genetically determined variability in acetyla-
tion and oxidation. Therapeutic implications. Drugs 29, 342.
Evans, D.A., 1989. N-Acetyltransferase. Pharmacol. Ther. 42, 157.
Grant, D.M., Tang, B.K., Kalow, W., 1983. Polymorphic N-
acetylation of a caffeine metabolite. Clin. Pharmacol. Ther. 33,
355.
Lang, N.P., Butler, M.A., Massengill, J., Lawson, M., Stotts, R.C.,
Hauer-Jensen, M., Kadlubar, F.F., 1994. Rapid metabolic
phenotypes for acetyltransferase and cytochrome P4501A2 and
putative exposure to food-borne heterocyclic amines increase
the risk for colorectal cancer or polyps. Cancer Epidemiol.
Biomarkers Prev. 3, 675.
Le Marchand, L., Sivaraman, L., Franke, A.A., Custer, L.J.,
Wilkens, L.R., Lau, A.F., Cooney, R.V., 1996. Predictors of
N-acetyltransferase activity: should caffeine phenotyping and
NAT2 genotyping be used interchangeably in epidemiological
studies? Cancer Epidemiol. Biomarkers Prev. 5, 449.
Lloyd, D.K., Fried, K., Wainer, I.W., 1992. Determination of
N-acetylator phenotype using caffeine as a probe compound: a
comparison of high-performance liquid chromatography and
capillary electrophoresis methods. J. Chromatogr. 578, 283.
Meacher, D.M., Rasmussen, R.E., Menzel, D.B., 1998. Analysis of
NAT and CYP1A2 phenotypes and NAT2* genotypes by
capillary electrophoresis. Biomarkers 3, 205.
ONeil, W.M., Gilfix, B.M., DiGirolamo, A., Tsoukas, C.M.,
Wainer, I.W., 1997. N-Acetylation among HIV-positive pa-
tients and patients with AIDS: when is fast, fast and slow,
slow? Clin. Pharmacol. Ther. 62, 261.
Ratain, M.J., Rosner, G., Allen, S.L., Costanza, M., Van Echo,
D.A., Henderson, I.C., Schilsky, R.L., 1995. Population phar-
macodynamic study of amonafide: a cancer and leukemia
group B study. J. Clin. Oncol. 13, 741.
Rojo, S., Lopez De Castro, J.A., Aparicio, P., Van Seventer, G.,
Bragado, R., 1986. HLA-B27 antigenicity: antibodies against
the chemically synthesized 6384 peptide from HLA-B27.1
 P. Wong et al. / Journal of Immunological Methods 251 (2001) 1 9
display alloantigenic specificity and discriminate among HLA-
B27 subtypes. J. Immunol. 137, 904.
Straka, R.J., Hansen, S.R., Benson, S.R., Walker, P.F., 1996.
Predominance of slow acetylators of N-acetyltransferase in a
Hmong population residing in the United States. J. Clin.
Pharmacol. 36, 740.
Rothman, N., Hayes, R.B., Bi, W., Caporaso, N., Broly, F.,
Woosley, R.L., Yin, S., Feng, P., You, X., Meyer, U.A., 1993.
Correlation between N-acetyltransferase activity and NAT2
genotype in Chinese males. Pharmacogenetics 3, 250.
9
Vineis, P., Bartsch, H., Caporaso, N., Harrington, A.M., Kadlubar,
F.F., Landi, M.T., Malaveille, C., Shields, P.G., Skipper, P.,
Talaska, G. et al., 1994. Genetically based N-acetyltransferase
metabolic polymorphism and low-level environmental expo-
sure to carcinogens. Nature 369, 154.
Wong, P., Leyland-Jones, B., Wainer, I.W., 1995. A competitive
enzyme linked immunosorbent assay for the determination of
N-acetyltransferase-2 (NAT2) phenotypes. J. Pharm. Biomed.
Anal. 13, 1079.