IUIIPlII J i l l i i t IP

PHARMACEUTICAL
ELSEVIER European Journal of Pharmaceutical Sciences. 4 1996) 133-138 SCIENCES

The impact of stereoisomerism in a bioequivalence study on two

formulations of doxepin
K.K. Midha ~'b'*, J.W. H u b b a r d a. M. R a w s o n ~, R. S c h w e d e ~
'College ~t Pharmacy and Nutrition. Unit'ersitv ~!f"Saskatchewan. I I0 Science Place. Sa.vkatoon. SK. S Z~I 5C0. Canada
hCollcee of Medicine University of Saskatchewan. I lO Science Place. Saskat~on. SK. ST,V fiCg. Canada
~Geneva Pharmaceuticals 2555 W Midway Blvd. P.O. Box 446. Broomfield, Co 800.;8-0.146. USA
Received If) April 1995; accepted 22 August 1995

Abstract

There are few archival data on the impact of stercoisomerism on bioequivalence studies. A hioequivalcncc study wax carried out on
two fortnulations of doxepin containitlg 15% of the active (-i.v isomer and 85% of the less active trans isomer. Plasma concentrations of
cis N-dcsmcthyldoxepin remained at approxim:ttcly 15% of total in only 2 of 30 healthy voltmtcers who completed the study. In the
retnaining 28 sulRjects, however, the proportion of cis N-desmcthyldoxepitl steadily increased with time such that by 72 h post dose. cis
N-desmethykh~xcpin comprised approximately 90% of the metabolite present in plasma. Nevertheless. the 90% conlidence intervals (In
AUCh, ,. In C ..... ;.llld In C ...... /AUC~.,,,) for N-desntcthyldoxcpin fell entirely within bit~equivalenee limits of 8()-125%, irrespective of
whether stereoselective or tlon-stereosclective data were analyzed. Thus the results of this study did not support arguments in favor of the
use of stercoselcctive methods in bioequivalence studies.

Keywords." Bioeqtfivalcnce; Stcreoisomerism; Geometric isomerism; Metabolites: Doxepin; N-Desmethyldoxepin

1. Introduction systemic clearance. It was felt that if the two enantio-

Many important and widely used drugs are mar-
keted as mixtures of optical isomers and/or geomet-
ric isomers that often differ in their pharmacological,
toxicological and pharmacokinetic properties (Wainer
and Draycr, 1988). It is clear that stereoisomerism
has an important impact on pharmacokinetics (Ariens,
1984) and therapeutic monitoring (Hubbard et al.,
1986) but there are few archival data on the relevance
of stcreochcmistry in bioequivalence studies. A con-
fcrence report from "'Bio-lnternational "92"' (Blurne
and Midha, 1993) commented subjectively on the
bioequivalence issue in the following terms: "'...The
use of specific analytical methods was briefly dis-
cussed. Two aspects must be taken into account (!)
systemic availability of the two enantiomers and (2)
Corresponding author.
mers are documented to show no difference in these
two pharmacokinetic characteristics, a non-
stereoselective method may be used..," This state-
ment was made in the absence of substantive data
since most bioequivalence studies are conducted
without recourse to intravenous dosing so that sys-
temic availability (F) and clearance (CL) usually
remain unknown, and the closest approximation that
can be made is the quotient CL/F.
One published study on this topic (Jamali et al.,
1991) compared a stereoselective approach with a
more conventional non-stereoselective method in a
bioequivalence study on two formulations of flurbip-
rofen. The authors' conclusion that stereospecilic
approaches provide a more accurate assessment of
products in bioequivalency studies, however, was
based on a significant difference in a paired t-test.
The latter simply tests for differences between formu-

Elsevier Science B.k(

S S D I 0928-0987(95)00039-9
 134 K.K. Midha et al. I European Journal of Pharmaceutical Sciences 4 (1996) 133-138
lations based on between subject variability, whereas
the standard confidence interval approach to bio-
equivalence decision making (Schuirmann, 1987)
relies on the within subject variability estimated from
the residual error term in ANOVA (Midha et al.,
1993a). Upon calculation of confidence intervals from
the published data (Jamali et al., 1991) it was found
that the ratios of geometric means were close to unity
and the confidence intervals fell within the regulatory
interval of 80-125% for both isomers. Interpreted in
this way, these data did not support arguments in
favor of stereoselective methods for bioequivalence
studies.
Doxepin is marketed as an irrational mixture of
15% the cis-isomer and 85% of the trans-isomer.
Tests in animals suggest that doxepin exhibits antide-
pressant activities with pronounced sedation and
tranquillizing properties, together with anticholiner-
gic, antispasmodic and mild peripheral vasodilating
effects (Pinder et al., 1977). cis-Doxepin was more
active than the trans-isomer in virtually all these
animal tests. Doxepin is metabolized in humans to
form a variety of phase I and phase I! metabolites
(Luo et al., 1991 Shu et al., 1990a Shu et al., 1990b).
The major mctabolite, N-desmethyldoxepin was also
active in animal tests with more sedative properties
than the parent drug (Pinder et al., 1977). Animal
data are difficult to interpret in the context of clinical
depression in humans, and while it has been reported
that the total plasma concentrations of doxepin plus
N-desmethyldoxepin correlate better with antidepres-
sant activity than do concentrations of the parent drug
alone (Faulkner et al., 1983 Green, 1978), there are
no data on the relative extents to which the cis-and
trans isomers of doxepin and its N-desmethyl metab-
olite contribute to therapeutic activity in patients. It
is, however, known that plasma concentrations of cis
N-desmethyldoxepin were comparable to those of
trans N-desmethyldoxepin in depressed patients
(Hrdina et ai., 1990 Rosseei et al., 1978) and healthy
volunteers (Midha et al., 1992), although the mecha-
nism(s) responsible for the change in proportion of
plasma concentration of the N-desmethyl isomers
remain(s) unknown. Evidence for the involvement of
trans to cis conversion has been reported in humans
(Ghabrial et ai., 1991), and an unstable metabolite of
N-desmethyldoxepin hydroxylated at the exocyclic
double bond was detected in the urine of patients
medicated with doxepin (Shu et al., 1990a). The latter
may be an indication that the trans to cis conversion
process could involve successive hydration and dehy-
dration of the exocyclic double bond. The potential
involvement of cis N-desmethyldoxepin in the effica-
cy of doxepin therapy was taken as an indication that
stereoisomerism should be taken into account in a
study on the bioequivalence of two formulations of
the drug. Therefore a bioequivalence study on two
capsule formulations of doxepin was carried out with
a view to studying the impact of stereoisomerism on
bioequivalence decision making.
2. Experimental procedures
2.1. Subjects
The subjects (n=30) were recruited from a cohort
of healthy male Caucasian volunteers aged 18-45
years and deviating by no more than 15% from the
ideal weight for height according to the Metropolitan
Life Insurance Company Bulletin, 1983. Each subject
was required to be free from cardiovascular, hepatic,
renal, or gastrointestinal diseases, drug abuse or
alcohol dependence, as assessed by physical exami-
nation and review of medical history. In addition, the
blood pressure, electrocardiogram and results of
clinical laboratory tests were required to be within
normal ranges. The subjects were required to refrain
from use of all other drugs (including non-prescrip-
tion drugs), for at least one week before the study
until one week after its completion. The consumption
of alcohol or caffeinated beverages was prohibited
from 48 h prior to dosing until after collection of the
last blood sample.
2.2. Study design
The study protocol and consent form were ap-
proved by the local ethics review board. The subjects
were given test and reference capsules containing
doxepin ( 3 x 2 5 mg) in a two-period, two-sequence,
randomized cross-over design (Grizzle, 1965) with a
wash-out period of three weeks between the last
blood sample collection after the first phase and
dosing of the second phase. Blood samples (15 mi)
were harvested in heparinized evacuated tubes imme-
diately before dosing and at 0.5, 1.0, 1.5, 2.0, 2.5, 3,
4, 6, 8, 12, 24, 48, 72, 96, 120, and 168 h after
 K.K. Midha et al. I European Journal of Pharmaceutical Sciences 4 (1996) 133-138 135

dosing. Care was taken to ensure that the blood did
not contact the rubber stopper at any time during
sample collection. The blood samples were cen-
trifuged within 30 min of collection. The plasma
samples were divided into two aliquots and stored at
-20C until analysis by methods described previous-
ly (Midha et al.. 1992).
2.3. Calculations and statistics
Pharmacokinetic parameters were calculated (PC-
SAS) by standard methods. The maximum plasma
concentration (Cm,~) and the time (tm~,,) to maximum
concentration were obtained directly from the plasma
concentration versus time curves. The area under the
plasma concentration versus time curve (AUCI,~,) up
to the time (/last) of the last measurable concentration
(C~,~,) was estimated by the linear trapezoidal rule for
ascending portions and the log trapezoidal rule lor
descending portions of the curve. A rate constant (k)
was calculated by least squares regression of the
natural logs of the last 2-5 data points. AUCI,,~t was
extrapolated to inlinite time (AUC) by adding the
quotient of the predicted Ct~.,, and the appropriate
value of k. A half life was calculated as the quotient
of In 2 and k. Bioequivalence calculations (PC-SAS)
were calculated from log transformed data as de-
scribed previously (Midha et al., 1993b).
3. Results and discussion
In the present study, geometric mean ratios and
90% confidence intervals were calculated for both
doxepin and N-desmethyldoxepin. In view of the
likelihood that concentration-dependent parameters
are not normally distributed, Table I shows geomet-
ric means, minima and maxima of AUC~,~t and Cm,,
for total doxepin, total N-desmethyldoxepin. and cis
and trans N-desmethyldoxepin after administration of
the test and reference formulations. Table 2 shows
corresponding arithmetic means (__+standard error) for
the time dependent parameters tr.,,' and t t/'.
Unfortunately plasma concentrations of cis dox-
epin were so low that stereoselective monitoring was
possible in only three subjects in whom plasma

Table 1
GeolllCttic nle;.ins and ranges for concentration-dependent parameters

AUC,.., (nglh'ml ' ) C... (nglml)

Test Reference Test Reference
Total doxepin
Mean (n = 30) 143.9 134.8 14.0 11.8
Minimum 32.3 33.4 4.0 3.2
Maximum 619.8 670.3 43.0 41.5
Total N-desmethyldoxepin
Mean (n = 30) 155.7 158.5 4.8 4.6
Minimum 64.9 78.7 2.9 2.5
Maximum 956.9 970. I 10.6 9. I
cis N-Desmethyldoxepin
Mean 01 = 30) 95.6 99.9 2.2 2.2
Minimum 49.2 59.9 1.5 1.5
Maximum 201.9 218.8 3.7 3.6
trans N-Desmethyldoxepin
Mean (n = 30) 43.6 41.7 2.7 2.5
Minimum 1 I. I 4. I 1.3 0.7
Maximum 848.8 855.5 8.4 7.8
cis N-Desmethyldoxepin
Mean (n = 28. all except 6 and 18) 94.6 98.5 2.3 2.3
Mean (n=2. subjects 6 and 18) 106.2 116.7 1.8 1.8
trans N-Desmethyldoxepin
Mean (n =28, all except 6 and 18) 35,9 34.1 2.6 2.3
Mean (n =2. subjects 6 and 18) 700.7 667.8 7.2 6.9
 136 K.K. Midha et al. I European Journal of Pharmaceutical Sciences 4 (1996) 133-138

Table 2
Anthmetic means_+SE for time dependent parameters
illustrates that a bioequivalence study based on a
cross-over design is not compromised by the inclu-
Test Reference
sion of subjects from different populations.
Total Doxepin
Table 2 reveals that the mean t .... and t ~/2 values
t,,, (h) 2.2-+0.2 2.50.2
t " : (h) 13.5__. 1.0 14.0___ 1.0 of cis N-desmethyldoxepin were considerably greater
Total N-desmethyldoxepin
that those of the trans isomer after administration of
t (h) 6.6__.I.0 7.2'-I.I the test and reference formulations of doxepin, sug-
fl': (h) 31.9"4.2 31.8"3.8 gesting that the pharmacokinetics of cis N-desmeth-
cis N-Desmethyldoxepin yldoxepin may have been rate limited by formation.
t~,, (h) 11.5__.2.4 11.3_+1.7 Studies are presently under way in our laboratories to
t ~': (h) 23.8_+2.5 28.1 _+3.3
investigate the absolute bioavailabilities of doxepin
trans N-Desmethyldoxepin and N-desmethyldoxepin in healthy volunteers which
t (hi 4.6_+0.8 4.6 1.0
t " : (h) 16.0_+ 2.5 14.2-+2.1
should yield valuable data on clearance and elimina-
tion half lives in addition to information on possible
trans to cis conversion during the formation of cis
N-desmethyldoxepin.
Non-stereoselective data tbr doxepin (Table 3)
concentrations of the cis isomer were highest. In show 90% confidence intervals for In AUC~,,~ (rela-
these individuals, the cis isomer comprised approxi- tive extent of absorption) that fell entirely within the
mately 15% of the total doxepin in plasma, similar to bioequivalence limits of 80-125%. The 90% confi-
the proportion in the dosage forms administered. The dence intervals tbr In C ..... , however, violated the
mean AUCt.~, values for cis N-desmethyldoxepin, upper boundary, suggesting a possible difference in
however, were more than double those of the trans relative rates of absorption from the two formula-
isomer (Table I ). Two individuals (subjects 6 and i 8) tions. Table 3 also shows that C ...... was highly
had very high values for AUCj.~, of trans N-dcsmeth- variable in that the ANOVA CV>30% (Blume and
yldoxcpin (Table l), and their proportion of cis to Midha, 1993) indicating there was a high probability
total N-dcsmethyldoxepin renmincd close to 15% of failure in terms of in C,,,,~ even in simulations
(Fig. I), suggesting these two individuals were from
a different population than the rest of the subjects.
Nevertheless, Fig. I shows that test and reference
Table 3
formulation means of 28 subjects were simihtr to
90% Confidence intervals and within-subject variabilities (n =30) for
each other, as were those of subjects 6 and 18, which total doxepin, total N-desmethyldoxepin. and cis and tran.~ N-desmeth-
yldoxepin

90% CIs CV%"

100 Total Doxepin

~. 9O In AUC,,., 96.4- I 18.2 23.2
a0 In C ..... 102.2-13Z9 .t4.1
O
o 70 1~
N In C,,,,./AUC,,, 101.4-122.0 21.0
5 6O Total N-Desmethyldosepin
In AUC,,., 93.3- I03.3 I 1.6
40 L
3o /M.en, of 6 10 In C .... 99.4-112.3 13.9
In C,,,,,/AUC~,,, 100.9-114.8 14.8
~ ~ ...... , . . . . 15%
cis N-Desmethyldoxcpin
0 / - I - I I . I . I I . I
In AUC~,,, 91 .1-100.6 I 1.4
0 10 2O 30 40 50 60 70 80 In C .... 96.0-105.5 10.7
Time (hours) In Cm,,/AUC,,,, I(X).2-110.4 I 1.0
trans N-Desmethyldoxepin
Fig. I. Plots of the mean % cis N-desmethyldoxepin for 28 subjects
(diamonds) and for subjects 6 and 18 (circles) after administration of test In AUC,,,, 93. I - I 17.8 26.9
(solid lines) or reference (broken lines} formulations. The plot is cut off at In Cm," 99.8-120.4 21.4
72 h since there were fewer than 28 subjects contributing to each mean In Cm, IAUC,,,, 92.8-118.0 27.3
(diamonds) after this time. 'Calculated from the residual mean square in ANOVA.
 K.K. Midha et al. / European Journal of Pharmaceutical Sciences 4 (1996) 133-138
where the true ratio of geometric means was set at
unity (Midha et al., 1993a). The ratio Cm~/AUC is
theoretically less variable than Cm~X (Endrenyi and
Yan, 1993) and has the additional advantage over
Cm~~ that it is sensitive to the rate of absorption, but
is unaffected by the extent of absorption (Endrenyi et
al., 1991 Schall et al., 1994). The 90% confidence
interval for In C,,,~/AUCt~st of total doxepin fell
within the bioequivalence limits of 80-125% (Table
3) although Cm~x/AUC has not been accepted as a
bioequivalence parameter by drug regulatory agen-
cies.
Some stereoselective data on doxepin were made
available by Novapharm Ltd., Canada, who kindly
consented to its inclusion in this manuscript. The
Novapharm study was carded out in 36 healthy
young men each of whom received two 150-mg
formulations of doxepin (twice the dose in the
present study) in a two-treatment, two-period, cross-
over design. This study was completed at a time
when Canadian drug regulatory authorities required
calculation of symmetrical 95% contidence intervals
which are reported here as such. The results, how-
ever, were similar to those in the present study. The
stereoselective 95% conlidence intervals for cis and
trans doxepin fell entirely within the bioequivalence
limits for In AUCt~,,, (cis 86.9-107.6; trans 82.0-
101.9) whereas those for In C,,,,~ violated the lower
bound (cis 7 8 . 6 - 1 0 8 . 2 ; trans 7 4 . 6 - 1 0 9 . 1 ) . C ..... /
AUC data were not available from the Novapharm
study, although it would appear that the failure of In
C~,~ was also likely due to high within-subject
variability of this parameter. The key point to be
made in the present context is that stereoselective
data gave the same overall result as a non-
stereoselective procedure in similar studies.
The inlluence of data from subjects 6 and 18 can
be seen in Table 1 as wide ranges in AUCt,s, for both
trans and total N-desmethyldoxepin after administra-
tion of either test or reference formulations. The
lower two panels of Table ! illustrate that inclusion
of data from these two subjects had a marked effect
on the geometric means for AUCta~t of trans N-
desmethyidoxepin, but little effect on those for
AUC~,~, of cis N-desmethyldoxepin, or on C,,,~ for
either isomer. Removal of data from these subjects,
however, cannot be justified in a bioequivalence
study since each acts as his own control in the
cross-over design. Nevertheless, Table 3 shows that
137
the 90% confidence intervals for each phar-
macokinetic parameter of cis, transo and total N-
desmethyldoxepin all fell within the bioequivalence
interval of 80-125%.
This study has demonstrated the necessity of
stereoselective data to define the bioavailability of
doxepin and N-desmethyldoxepin because it appears
likely that cis N-desmethyldoxepin plays an impor-
tant role in doxepin therapy. The terms "bioavail-
ability" and "bioequivalence", however, are often
used as if they are interchangeable. This confusion
would lead one to reason that bioequivalence of two
formulations of doxepin must also be based on
stereoselective data on the parent drug and the N-
desmethyl metabolite. In fact, a bioequivalence study
based on the confidence interval method applied to a
randomized cross-over design examines the relative
bioavailability of the test and reference formulations,
a completely different phenomenon. For example, it
is necessary to estimate AUC in order to estimate
extent of absorption in a bioavailability study, but
experimental data (Midha et al., 1994) has shown
that the use of partial areas of AUC (e.g. AUC,,,,, )
in conjunction with tile conlidence interval method is
a very effective means of investigating relative
bioavailability in a bioequivalence study.
In conclusion, the present study was carried out on
a drug marketed as an irrational mixture of geometric
isomers in which the active cis isomer comprises
only 15% of the administered dose, but which
undergoes a substantial change of proportion to the
extent that, for most subjects, A U C ~ , of cis N-
desmethyl metabolite was greater than that of the
trans isomer. The use of metabolite data in bio-
equivalence decision making, however, remains an
open question. Theoretical arguments suggest that
bioequivalence can be based on data from parent drug
or metabolite, active or inactive, provided the choice
of analyte is not made'retrospectively (Tucker et al.,
1993). In the absence of acceptance criteria for
metabolite data, it could be argued that the bio-
equivalence could not be declared in the present
context because the confidence intervals for In C ....
of the parent drug failed to fall within bioequivalence
limits of 80-125%, based either on stereoselective
data in the Novapharm study, or on non-stereoselec-
tive data in the present study. In other words, there
was insufficient evidence to support arguments in
favor of the use of stereoselective methods in these
138 K.K. Midha et al. I European Journal of Pharmaceutical Sciences 4 (1996) 133-138
studies, whether the bioequivalence decision was
based on parent drug or major metabolite.
References
Ariens. E. { 1984) Stereochemistry. a basis for sophisticated nonsense in
pharmacokinetics and clinical pharmacology. Eur. J. Clin. Pharmacol.
26. 663-668.
Blume. H. and Midha. K. (1993) Bio-lntemational 92. Conference on
Bioavailability. Bioequivalence. and Pharmacokinetic Studies. J.
Pharm. Sci. 82. I 186- I 189.
Endrenyi. L.. Fritsch. S. and Yan. W {1991) C~,,/AUC is a clearer
measure than C,,, for absorption rates in investigations of bio-
equivalence. Int. J. Clin. Pharmacol. Ther. Toxicol. 29. 394-399.
Endrenyi. L. and Yah. W. (1993) Variation of C~o, and C,o,/AUC in
investigations of bioequivalence. Int. J, Clin. Pharmacol. Ther. Tox-
icol. 31. 184-189.
Faulkner. R.. Pitts. W.. Lee. C., Lewis, W. and Farm. W. (1983) Multiple-
dose doxepin kinetics in depressed patients. Clin. Pharmacol. Ther. 34.
509-515.
Gfiabrial. H.. Prakash. C.. Tacke. U.. Blair. 1. and Wilkinson. G. ( I~)1 )
Geometric isomerism of doxepin during its N-demethylation. Drug
Metab. Dispos. 19. 596-599.
Green. D. (1978) Clinical importance of doxepin plasma levels. J. Clin.
Psychiat. 39. 481-482.
Grizzle. J. (1965) The two peri~v,I change-over design and its use in
clinical trials, l~liometrics. 21. 467-480.
Hrdina. P., Bakish. D., Swenson. S. and Lapierre. Y. (1000) cis- and
trans-isomers of doxepin and N-desmetfiyldoxepin ill the plasnm of
depressed patients treated with doxepin. Therap. Drug Monitor. 12.
129-133.
tlubhard. J.. Ganes. D.. Lira. !1. and Midha. K. { 1986) Chiral ph:trma-
cology and its consequences for therapeutic nlonitoring. Clin. Bio-
chem. 19. 107-112.
Jamali. F.. Collins. D., Berry. B.o Molder. S.. Cheung, R.. McColl, K. and
Cheung. H. (1991) Comparative bioavailability of two llurbiprol~en
products: Stereospecific versus conventional approach. Biopharm.
Drug Dislx~s. 12. 435-445.
Luo, El.. Hawes, E.. McKay. G.. Korchinski, E. and Midha, K. (1991)
The quaternary ammonium-linked glucuronide of doxepin: A major
metabolite in depressed patients treated with doxepin. Drug Metab.
Dispos. 1991. 722-723.
Midha, K.. Hubbard, J., McKay. G., Hawes. E.. Korchinski. E.. Gurnsey.
T.. Cooper. J. and Schwede. R. (1992) Stereoselective phar-
macokinetics of doxepin immers. Eur. J. Clin. Pharmacol. 42. 539-
544.
Midha. K.. Hubbard. J., Rawson. M.. and Gavalas. L. (1994) The
application of partial areas in assessment of rate and extent of
absorption in bioequivalence studies of conventional release products:
Experimental evidence. Eur. J. Pharm. Sci. 2. 351-363.
Midha. K.. Hubbard. J.. Yeung. P.. Ormsby. E., McKay. G.. Hawes. E .
Korchinski. E.. Gurnsey. T.. Rawson. M. and Schwede. R. (1993a)
Application of replicate design, in K.M.H. Blume tEd.). Bio-lmema-
tional: bioavailability, bioequivalence and pharmacokinetics; Interna-
tional Conference of F.I.P. "'Bio-international "92"" held in Bad
Homburg. Germany. May 20-22. 1992. Medpharm Scientific Pub-
lishers, Stuttgart. pp. 53-68.
Midha. K.. Ormsby. E.. Hubbard. L. McKay. G.. Hawes, E.. Gavalas. L.
and McGilveray. I. (1993b) Logarithmic transformations in hit-equiv-
alence: Application with two formulations of perphenazine. J. Pharm.
Sci. 82. 138-144.
Pinder. R., Brogden. R.. Speight, T. and Avery. G. (1977) Doxepin
up-to-date: A review of its pharmacological properties and therapeutic
efficacy with particular reference to depression. Drugs 13. 161-218.
Rosseel. M., Bogaert. M. and Claeys. M. (1978) Quantitative GLC
detennination of cis-and trans, isomers of doxepin and desmeth-
yldoxepin. J. Pharm. Sci. 67. 802-805.
Schall, R., Hundt. HK.L. and Luus. H.G. (1904) Pharmacokinetic
characteristics h~r extent of absorption anti clearance in drug/drug
inter;lotion studies. Int. J. Clin. Pharmacol. Therapeut. 32(12). 633-
637.
Schuirmann. D. (1987) A comparison of the two one-sided texts pro-
cedure and the power approach for assessing the equivalence of
average bioavail:tbility. J. Pharmacokinet. Biopharul. 15, 657-680.
Shu. Y.-Z., Hubbard. J., Cooper. J., McKay. G., Korchinski. [-.. Kumar. R.
-'rod Midha, K. (1990a) The identilication of urinary metabaflites of
doxepin in patients, l)nlg Metab. Dispos. 18. 735-741.
Shu, Y.-Z.. Hubbard. J., McKay. G. and Midha. K. (1990b) Identification
of phenolic doxepm nletal~lites fronl p:ttient urine and rat bile. Drug
Metab. l)ispos. 18. I0q6-1008.
Tucker. G.. Rostami. A. and Jackson, P. { 1903) Metabolite measurement
in bioequiwdence studies: Theoretical considerations. In K. Midha and
tl. Blume (Eds.). Bio-luternational Bioavailability. Bioequivalence and
Pharmacokinetics. Medpharm Scientific Publishers. Stuttgart, pp. 163-
170.
Wainer. 1. and Drayer. D. (Eds.) (1988) Drug Stereochemistry. Analytical
Methods and Pharmacology. Marcel Dekker. New York.