Studies on the stereoselective metabolism of citalopram by human liver microsomes and cDNA-expressed cytochrome P450 enzymes

Article abstract of DOI:10.1159/000028333

The involvement of CYP enzymes in the metabolism of citalopram was studied, inclusive the conversion of demethylcitalopram to didemethylcitalopram and the formation of citalopram N-oxide, which both have not been considered previously. Using human mixed liver microsomes and cDNA-expressed CYP enzymes, we confirmed that CYP3A4, 2C19 and 2D6 are involved in the first demethylation step of citalopram, all favouring conversion of the biologically active S-enantiomer. Inhibitor studies indicated that at therapeutic citalopram concentrations CYP3A4 was responsible for 40-50% of demethylcitalopram formation, while the contribution of CYP2C19 increased and that of CYP2D6 tended to decrease with increasing drug concentration, CYP2D6 exclusively mediated the second demethylation step, and citalopram N-oxide was also exclusively formed by CYP2D6. None of the studied CYP enzymes mediated deamination to the propionic acid derivative.

Full text of DOI:10.1159/000028333

Original Paper
Received: February 8, 1999
Accepted after revision: May 31, 1999
Pharmacology 1999;59:298–309
StudiesontheStereoselectiveMetabolism
ofCitaloprambyHumanLiverMicrosomes
and cDNA-Expressed Cytochrome P450
Enzymes
Ole V. Olesen Kristian Linnet
Institute for Basic Psychiatric Research, Department of Biological Psychiatry,
Psychiatric Hospital in Aarhus, Aarhus University Hospital, Risskov, Denmark
KeyWords
second demethylation step, and citalopram
Citalopram W Stereoselectivity W Metabolism W
Cytochrome P450
N-oxide was also exclusively formed by
CYP2D6. None of the studied CYP enzymes
mediated deamination to the propionic acid
Abstract
derivative.
Copyright © 1999 S. Karger AG, Basel
The involvement of CYP enzymes in the me-
tabolism of citalopram was studied, inclusive
the conversion of demethylcitalopram to di-
demethylcitalopram and the formation of ci-
talopram N-oxide, which both have not been
considered previously. Using human mixed
liver microsomes and cDNA-expressed CYP
enzymes, we confirmed that CYP3A4, 2C19
and 2D6 are involved in the first demethyl-
ation step of citalopram, all favouring con-
version of the biologically active S-enantio-
mer. Inhibitor studies indicated that at thera-
peutic citalopram concentrations CYP3A4
was responsible for 40–50% of demethylcita-
lopram formation, while the contribution of
CYP2C19 increased and that of CYP2D6
tended to decrease with increasing drug con-
centration. CYP2D6 exclusively mediated the
Citalopram (CITA) belongs to the group
of selective serotonin reuptake inhibitors
(SSRIs), used in the treatment of depression
[1, 2]. CITA is marketed as a racemic mixture
of the S-(+)-CITA and R-(–)-CITA enantio-
mers. The S-(+) enantiomer is most effective
as inhibitor of serotonin reuptake and is con-
sidered to be of main importance for the clini-
cal effects [3].
CITA is partly excreted unchanged or con-
jugated to glucuronic acid in the urine and
partly metabolized in the liver [4]. The main
metabolites are N-demethylcitalopram (DCI-
TA), citalopram-N-oxide (CITA-NO), N-di-
demethylcitalopram (DDCITA) and citalo-
© 1999 S. Karger AG, Basel
ABC
Ole V. Olesen, PhD, Institute for Basic Psychiatric Research
Department of Biological Psychiatry, Psychiatric University Hospital
Skovagervej 2, DK–8240 Risskov (Denmark)
0031–7012/99/0596–0298$17.50/0
Fax + 41 61 306 12 34
E-Mail karger@karger.ch
www.karger.com
Accessible online at:
www.karger.com/journals/pha
Tel. +45 8617777 ext. 3522, Fax +45 86170778
E-Mail viol1907@post3.tele.dk

pram propionic acid derivative (CITA-
PROP) (fig. 1) [5, 6]. Two independent in
vitro studies have dealt with the identification
of the cytochrome P450 (CYP) enzymes in-
volved in the first demethylation step of
CITA to DCITA [7, 8]. The two studies
agreed so far as both suggested that CYP3A4
and CYP2C19 play a role for the demethyl-
ation of citalopram but disagreed with respect
to the relative importance of the two enzymes
and a possible contribution of CYP2D6. Ro-
chat et al. [8] found that CYP2D6 had a high
affinity but a low capacity for DCITA forma-
tion, while Kobayashi et al. [7] was unable to
demonstrate a significant contribution of the
enzyme. Rochat et al. [8] also compared the
enzyme kinetic constants for demethylation
of the two enantiomers of CITA and found a
pronounced stereoselectivity with respect to
reaction velocity. The Vmax of the CYPs 3A4
and 2C19 catalyzed DCITA formation of the
S-enantiomer was about 1.5 times the Vmax of
the R-enantiomer. CYP2D6 showed an oppo-
site stereoselectivity, the capacity for demeth-
ylation of the S-enantiomer being 0.7 times
that of R-DCITA formation. In the above-
Fig. 1. Molecular structures of citalopram and metab-
olites.
mentioned studies the formation of CITA- pressed CYP enzymes. As a part of the stud-
NO, DDCITA and CITA-PROP was not con- ies, the selectivities of traditionally used CYP
sidered, but recent investigations have shown inhibitors were considered.
that monoamine oxidases (MAO) and alde-
hyde oxidases presumably are of importance
for CITA-PROP formation [9].
Materials and Methods
The purpose of the present study was to
Drugs and Chemicals
clarify the discrepancies of the above-men-
tioned studies and extend the knowledge on
citalopram disposition. The focus was on the
stereoselectivity of DCITA formation. Addi-
tionally, the second demethylation step of
CITA to DD-CITA was examined in detail
and finally the formation of CITA-NO, which
both have not been studied previously. The
present investigations were based both on a
mixed human liver microsome preparation
(HLM) and individual human cDNA-ex-
Lundbeck A/S (Copenhagen, Denmark) kindly
supplied racemates and individual enantiomers of
CITA, DCITA, DDCITA, CITA-NO and CITA-
PROP. LU-10-2029, which corresponds to citalopram
with chlorine substituted for fluor in the phenyl ring,
amitriptyline, nortriptyline and E-10-OH-nortripty-
line were also gifts from Lundbeck A/S. Fluvoxamine
and ketoconazole were donated by Solvay Duphar
B.V. (Weesp, the Netherlands) and Janssen Pharma-
ceutica N.V. (Beerse, Belgium), respectively. Caffeine
and 1,7-dimethylxanthine (paraxanthine) were from
Sigma Chemical Co. (St Louis, Mo., USA). (S)-mephe-
Pharmacology 1999;59:298–309
Citalopram Metabolism
299

nytoin, 4)-hydroxymephenytoin, 4)hydroxydiclofenac,
chlorzoxazone and 6-hydroxychlorzoxazone were pur-
chased from Gentest Co. (Woburn, Mass., USA). Oth-
er chemicals were of analytical grade. An HLM, pre-
pared as a mixture of microsomes from six different
donors, and cDNA-expressed human CPYs 1A2, 3A4,
2C9, 2C19, 2E1 and 2D6, prepared from baculovirus-
infected insect cells, were also obtained from Gentest
Co. The microsomal preparations were received frozen
Inhibitor Studies
In order to determine the quantitative contribution
of the individual CYP enzymes to the total HLM cata-
lyzed demethylation of CITA, inhibitors were added to
the reaction medium. Ketoconazole in the concentra-
tion range 0.25–2.5 Ìmol/l was used to inhibit
CYP3A4 activity, fluvoxamine in the concentration
range 0.5–5 Ìmol/l served as inhibitor of CYP2C19,
and quinidine in the concentration range 0.125–
5 Ìmol/l was used as inhibitor of CYP2D6. The opti-
mal concentration of an inhibitor was finally selected
as a compromise between an acceptable high inhibito-
ry effect on the target enzyme and a reasonably low
inhibitory activity towards the other enzymes involved
in DCITA formation as described in detail later.
°
and were stored at –80 C until use. According to the
supplier, the contents of the individual CYP enzymes
in HLM have been determined by immunoquantita-
tion. The amounts of the CYP enzymes investigated,
expressed per mg protein, were 45 pmol CYP1A2, 108
pmol CYP3A4, 96 pmol CYP2C9, 19 pmol CYP2C19,
49 pmol 2E1 and 10 pmol CYP2D6. The activity of
flavine containing monooxygenase (FMO) was given
as 2.53 nmol/mg protein/min methyl p-tolyl sulfide
oxidation.
Assay of DCITA, DDCITA, CITA-NO and
CITA-PROP
DCITA and DDCITA in the incubate were deter-
mined essentially as previously described [17]. In
short, after addition of 0.5 ml 1.6 mol/l sodium car-
bonate/bicarbonate buffer, pH 10 and internal stan-
dard (LU-10-2029), the compounds were extracted
into 7.5 ml heptane-isoamylalcohol (98:2, v/v) by
shaking. After centrifugation the analytes were back-
extracted from the organic phase into 125 Ìl phospho-
ric acid 250 mmol/l, of which 65 Ìl was injected into
the HPLC apparatus. After extraction of DCITA and
DDCITA from the incubate, amitriptyline was added
as a new internal standard, and the more polar CITA-
NO was extracted into 7.5 ml heptane-isoamylalcohol
(90:10, v/v) and back-extracted into 125 Ìl phosphoric
acid as described above. The acid CITA-PROP was
extracted directly from the incubate after addition of
0.5 ml 1 mol/l acetic acid also using 10% isoamylalco-
hol in heptane. The organic phase was evaporated to
dryness, and the residue was dissolved in 125 Ìl mo-
bile phase, of which 65 Ìl was injected into the HPLC
apparatus. The compounds were separated with an
acetonitrile-44 mmol/l phosphate buffer, pH 2.5-mo-
bile phase containing 10 mmol/l TEA on a Hypersil
BDS C18 column and measured by UV detection at
240 nm. The lower levels of quantification were 1.8,
2.5, 3.3 and 3.8 pmol for DCITA, DDCITA, CITA-
NO and CITA-PROP, respectively. The interday CV
for DCITA and DDCITA determination was !8%,
and for CITA-NO and CITA-PROP the CV was
!10%.
Experimental Procedure
The enzymatic reactions were carried out in poly-
°
propylene tubes at 37 C in a total volume of 300 Ìl of
0.1 mol/l phosphate buffer, at pH 7.4 as previously
described [10]. To avoid unspecific inhibition no or-
ganic solvents were used for dissolving any of the com-
pounds. The NADPH generating system consisted
of 1 U isocitrate dehydrogenase, 5 mmol/l NADP,
5 mmol/l isocitrate and 5 mmol/l MgCl2. The reaction
was started by addition of 50–100 Ìg HLM protein or
2.5–5 pmol of the individual CYP enzymes and
stopped after 7.5 min by addition of 0.5 ml ice-cold
1 mol/l acetic acid. Blank values were obtained by add-
ing the stop reagent before enzymes. In order to quali-
tatively assess the CYP enzymes involved in the me-
tabolism of CITA, incubation times of 60 min were
used with substrate concentrations of 2, 10 and 200
Ìmol/l. Qualitative checks of the activities and identi-
ties of the individual CYP enzymes were carried out
using established marker reactions. For CYP1A2 the
ability to mediate the 3-demethylation of caffeine to
paraxanthine was used [11]. Amitriptyline N-demeth-
ylation to nortriptyline served to demonstrate the ac-
tivity of CYP3A4 [12]. CYP2C9 activity was checked
by formation of 4)-hydroxydiclofenac [13]. CYP2C19
activity was assessed by the conversion of S-mepheny-
toin to 4)-hydroxy-mephenytoin [14]. The activity of
CYP2D6 was evaluated by conversion of nortriptyline
to E-10-hydroxynortriptyline [15]. Finally, the conver-
sion of chlorzoxazone to 6-hydroxychlorzoxazone
served as a check of CYP2E1 activity [16]. For all pro-
cesses proportionality between the amount of enzyme
added and the products formed was found.
Data Analysis
Reaction velocities (V) were expressed as nmol of
product formed per hour per mg protein, when HLM
was used, and as mol of product formed per hour per
Pharmacology 1999;59:298–309
300
Olesen/Linnet

Table 1. Kinetic parameter values for the metabolism of citalopram enantiomers
Reaction
Parameters
Microsomes
(+)S
CYP2C19
(+)S
CYP2D6
(+)S
(–)R
(–)R
(–)R
N-demethyl-
ation
Km, Ìmol/l
95 (12)
76 (11)
18 (0.7)
148 (25) 135 (18)
1490 (88)b 890 (42)
7.3 (0.6)b 24 (1.6)
Vmax, nmol h–1 mg–1 protein 26 (1.0)b
Vmax, mol h–1 mol–1 CYP
626 (11)b
360 (7.2)
N-oxidation
Km, Ìmol/l
13 (3.9)
33 (11)
7.0 (1.5)a 18 (3.7)
Vmax, nmol h–1 mg–1 protein
Vmax, mol h–1 mol–1 CYP
0.63 (0.05)b 0.16 (0.02)
110 (5.0)a
131 (7.5)
N-didemethyl- Km, Ìmol/l
ation
Vmax, mol h–1 mol–1 CYP
7.3 (0.9)
82 (2.4)b
6.0 (0.6)1
99 (2.0)1
Values are means with SEs in parentheses based on 2–3 replicates at 6–8 different substrate concentrations.
S-enantiomer versus R-enantiomer: a p ! 0.05; b p ! 0.01.
Km and Vmax calculated using one-binding site.
1
mol of CYP enzyme, when the CYP enzyme activities
were measured. V as a function of increasing substrate
concentrations (S) was fitted by nonlinear regression
according to the Michaelis-Menten equation, using
GraphPad Prism software (San Diego, Calif., USA).
The software allows a statistical comparison between a
one-site binding fit and a two-site binding fit by means
of an F test. Vmax and Km were determined on the basis
of the mean of two or three replicates of V measure-
ments using 6–8 levels of substrate concentration.
Standard errors for the parameter estimates Vmax and
Km were derived from the scatter of the individual
points around the estimated curve on the basis of a
least-squares computation principle [18]. In addition
to the saturation curves, Eadie-Hofstee plots were also
displayed. In this linearisation of the Michaelis-Men-
ten equation, V is plotted against V/S, and the slope
and intercept at the y-axis, determined by linear regres-
sion, correspond to –Km and Vmax, respectively [19].
DCITA and CITA-NO formation were signif-
icantly highest for the S-enantiomer. The ap-
parent Km values of the two enantiomers were
not significantly different for the two pro-
cesses (table 1).
Initial screening of the CYP enzymes 1A2,
3A4, 2C9, 2C19, 2D6 and 2E1 for activity
towards citalopram showed that CYPs 3A4,
2C19 and 2D6 displayed significant N-de-
methylation activity. The formations of S-
and R-DCITA as a function of incubation
time are shown in figure 2. The figure shows
that the reaction rate generally decreased with
time, and for CYP3A4 the DCITA formation
ceased completely after about 10 min. There-
fore, an incubation time of 7.5 min was used
throughout in the quantitative studies. Only
CYP2D6 mediated formation of measurable
amounts of CITA-NO, and CYP2D6 was also
the only of the CYP enzymes tested, which
Results
The estimated Km and Vmax values for the mediated the second demethylation of CITA,
HLM-mediated N-demethylation and N-oxi- the conversion of DCITA to DDCITA. Satu-
dation of S- and R-CITA are shown in table 1. ration plots for CYP2C19- and CYP2D6-
It appears that the rate of N-oxidation was mediated reactions and the corresponding
small compared to the rate for N-demethyl- Eadie Hofstee plots are shown in figure 3, and
ation. The apparent Vmax values for both the measured Km and Vmax values are given in
Pharmacology 1999;59:298–309
Citalopram Metabolism
301

table 1. CYP2D6 exhibited the highest affini- were obtained with respect to DCITA forma-
ty, while CYP2C19 had the highest capacity tion. However, an almost linear relationship
for conversion of CITA to DCITA.
between substrate concentrations and DCITA
With regard to CYP3A4, it was not possi- formation was found, when the CITA concen-
ble to determine Km and Vmax, because at high tration was kept below 20 Ìmol/l. The rela-
substrate concentration irreproducible results tions (mean and SE) between the reaction rate
Fig. 2. Demethylation of citalo-
pram enantiomers mediated by
CYP2C19, CYP2D6 and CYP3A4
as a function of time.
Fig. 3. Saturation curves and the corresponding Eadie-Hofstee plots for demethylation of
citalopram enantiomers by CYP2C19 (
enantiomers by CYP2D6 ( ). The slope and intercept of the dotted lines in figure 5f repre-
a, b), CYP2D6 (c, d) and demethylation of DCITA
e, f
sent the calculated –Km and Vmax values using a two-binding-site equation. Each point repre-
sents the mean of 2 or 3 determinations of V.
Pharmacology 1999;59:298–309
302
Olesen/Linnet

3
Pharmacology 1999;59:298–309
Citalopram Metabolism
303

Fig. 4. Inhibition of CYP3A4,
2C19 or 2D6 catalyzed demethyl-
ation of racemic citalopram. Bars
represent means of triplicate deter-
minations of the percentage inhibi-
tion of control reactions without
inhibitors.
Pharmacology 1999;59:298–309
304
Olesen/Linnet

Fig. 5. Inhibition of HLM-me-
diated demethylation of racemic
citalopram by ketoconazole, flu-
voxamine and quinidine. Bars re-
present mean of triplicate determi-
nations of percentage inhibition
of control reaction without inhibi-
tors.
(V) expressed as mol h–1 mol CYP and the respectively (fig. 4f). The recorded Km and
substrate concentration (S) expressed in
Vmax values, provided that a one-binding-site
Ìmol/l were V = S ! 0.82 B 0.017 (n = 10) equation was used for the CYP2D6-mediated
and V = S ! 0.55 B 0.015 (n = 8) for S-CITA DDCITA formation, are given in table 1.
and R-CITA, respectively. The difference be-
The individual contribution of CYPs 3A4,
tween the slopes was significant (p ! 0.01), 2C19 and 2D6 to the demethylation of ra-
indicating that for all the three enzymes in- cemic CITA mediated by the mixed human
volved in the demethylation of CITA, the liver microsome preparation was determined
velocity for demethylation of the S-enantio- by addition of inhibitors to the incubation
mer was higher than for the R-enantiomer.
medium. Reasonable compromises with re-
The second demethylation of CITA to spect to inhibitor concentrations are shown in
DDCITA was exclusively mediated by figure 4. Ketoconazole (0.5 Ìmol/l) reduced
CYP2D6 (fig. 3e, f). A significantly better fit the mean CYP3A4 control (uninhibited) ac-
(p ! 0.001) was obtained when a two-binding- tivity by 80 B 7% (SD), while the activities of
site model was used for S-DDCITA forma- CYP2C19 and CYP2D6 were reduced by 17
tion. Km was 3.0 Ìmol/l and Km2 was 44 B 3 and 4 B 3%, respectively (fig. 4a). Flu-
Ìmol/l, and the corresponding Vmax values voxamine (1 Ìmol/l) reduced the CYP2C19
were 51 and 40 mol h–1 (mol CYP2D6)–1, activity by 82 B 2% and the activities of
Pharmacology 1999;59:298–309
Citalopram Metabolism
305

CYP3A4 and CYP2D6 by 12 B 9 and 11
Vmax for CYP2D6 and also the same stereose-
B 4%, respectively (fig. 4b). Finally, the lective pattern for this enzyme as for the other
CYP2D6 catalyzed control activity was re- two. According to our study, all enzymes
duced by 94 B 2% by quinidine (0.5 Ìmol/l), favour a more rapid demethylation of the S-
while CYP2C19 and CYP3A4 activities were enantiomer, which agrees with the fact that
diminished by 6 B 2% and 4 B 1%, respec- the mean plasma concentration of the S-enan-
tively (fig. 4c). Addition of ketoconazole in- tiomer in patients in steady state only
hibited HLM catalyzed DCITA formation by amounts to 35–37% of the total concentration
36–46% independent of the substrate concen- of CITA [20, 21].
tration, indicating that CYP3A4 could be
To judge about the relative contribution in
held responsible for 40–50% of DCITA vivo of the three enzymes from in vitro stud-
formation (fig. 5). The contribution of ies, estimates of drug concentrations in hepat-
CYP2C19, measured by fluvoxamine inhibi- ic tissue are necessary. The plasma concentra-
tion, increased significantly (p ! 0.05) from tions of racemic CITA in patients treated with
13 to 30% with increasing substrate concen- standard doses of 20–40 mg/24 h are in the
tration, while the contribution of CYP2D6, range 30–500 nmol/l, and about 80% is pro-
measured by quinidine inhibition, tended to tein bound [6]. Rochat et al. [8] used the
decrease when the substrate concentration unbound fraction multiplied by a liver-water
was increased from 1 to 50 Ìmol/l.
partition coefficient of 20 as a measure of a
The ability of HLM to catalyze CITA- likely concentration of CITA at the CYP re-
PROP formation was tested with CITA, DCI- ceptor site. Using this calculation principle,
TA, DDCITA and CITA-NO, respectively, as the concentrations of CITA in liver tissue will
substrates. Using an incubation time of be in the range 0.1–2 Ìmol/l. If protein bind-
60 min and a substrate concentration of 100 ing is not taken into account, the CITA con-
Ìmol/l, DCITA was the only one of the com- centration in liver would be in the range 0.6–
pounds tested which gave rise to a measurable 10 Ìmol/l. Therefore, it seems reasonable to
amount of CITA-PROP formation. However, suggest that during human steady-state treat-
only about 0.1 nmol/mg protein of CITA- ment with CITA the concentration in liver tis-
PROP was formed during 1 h, and at lower sue hardly exceeds 10 Ìmol/l. According to
substrate concentrations the amount of me- the measured kinetic parameters of the pure
tabolite formed was below the level of quan- CYPs 2C19 and 2D6 and an activity of
tification.
CYP3A4, which is proportional to the sub-
strate concentration when kept below 20
Ìmol/l, each of the three CYP enzymes should
significantly contribute to the metabolism of
CITA. The content of the CYPs 3A4, 2C19
Discussion
Concerning the first demethylation step of and 2D6 in the HLM was, according to the
citalopram, our results support the findings of supplier, determined to be 108, 19 and 10
Rochat et al. [8]. Both inhibition studies of pmol per mg protein. Using the kinetic pa-
HLM-mediated demethylation and cDNA- rameters obtained from the experiments with
expressed CYP-mediated demethylation of pure CYP enzymes, the simulated contribu-
citalopram show that CYP3A4, 2C19 and tion of the individual CYP enzymes to the
2D6 are involved. Compared to the study by demethylation of CITA can be calculated. For
Rochat et al. [8], we found a somewhat higher CYP2C19 and 2D6 the DCITA formation
Pharmacology 1999;59:298–309
306
Olesen/Linnet

(V), expressed as pmol h–1 (mg protein)–1 is solely played a role for the second demethyl-
V = Vmax !S/(Km + S) multiplied by the num- ation step and 2C19 was partially responsible
ber of pmol CYP per mg HLM. For CYP3A4, for the first demethylation step [25]. The lat-
which showed atypical kinetic properties cor- ter is in agreement with our in vitro findings.
responding to a linear relationship between Contradictory results have been obtained
substrate concentration and DCITA forma- with respect to the role of CYP2D6 for
tion at substrate concentrations below 20 DCITA formation in humans. Administra-
Ìmol/l, the corresponding DCITA formation tion of a single dose of methotrimeprazine, a
is V = 0.69 ! S multiplied by 108. The simu- phenothiazine derivative known as inhibitor
lated percentage contribution of CYP3A4 to of CYP2D6, to human volunteers given
DCITA formation using substrate concentra- CITA increased the plasma concentration of
tions in the range 1–50 Ìmol/l increases from DCITA but not the half-life of CITA [26].
13 to 28%, the contribution of CYP2C19 However, long-term coadministration of phe-
increases from 26 to 43% and the contribu- nothiazines to patients in steady-state treat-
tion of CYP2D6 decreases from 61 to 29%. ment with CITA led to a 30% increase of the
Thus, compared to the inhibitor study of the mean plasma CITA concentration [27]. Con-
HLM-mediated DCITA formation, a simula- cerning a role of CYP3A4 for the DCITA for-
tion study highly overestimates the contribu- mation in humans, it was found that among
tion of CYP2D6 and underestimates the con- benzodiazepines only alprazolam, known as a
tribution of CYP3A4 to the total DCITA pro- rather specific substrate for CYP3A4, in-
duction. As both quinidine and ketoconazole creased the plasma CITA concentration [28].
in the used concentrations can be regarded as The finding that concomitant treatment with
relatively specific inhibitors of CYP2D6 and carbamazepine lowered the plasma CITA
3A4, respectively, it seems reasonable to sug- concentration may also suggest a role for
gest that the contributions found by the inhib- CYP3A4 [29]. Finally, comedication with the
itor study is more close to in vivo conditions. multiple CYP enzyme inhibitors cimetidine
We used fluvoxamine as inhibitor of and fluvoxamine has led to increase of CITA
CYP2C19 knowing that fluvoxamine is an in plasma [30, 31].
even stronger inhibitor of CYP1A2 and there-
Only CYP2D6 mediated a measurable N-
fore only can be used as inhibitor of oxidation of CITA, and both our experiments
CYP2C19 in the absence of CYP1A2 activity using pure enzyme and HLM suggest a minor
[22, 23]. At higher fluvoxamine concentra- role of this pathway for the metabolism of
tions, the activities of CYP2D6 and CYP3A4 CITA. The concentration of CITA-NO in
are also inhibited [24]. This explains the find- plasma is negligible, and the excretion of the
ing of Rochat et al. [8] that fluvoxamine in a metabolite in urine amounts to only 7% of the
concentration of 100 Ìmol/l in the incubation ingested dose [4, 5]. Finally, our results indi-
medium inhibited HLM catalyzed DCITA cate that CYP enzymes play an insignificant
formation by 64%.
role for the deamination of CITA, DCITA,
CYP2D6 was found to be exclusively re- DDCITA or CITA-NO to CITA-PROP. A
sponsible for the second demethylation of recent study indicates that deamination of
CITA to DDCITA and it seems in agreement CITA, DCITA, and DDCITA are mediated
with in vivo findings. Studies on panels of by monoamine oxidases (MAO) and aldehyde
poor and extensive metabolizers with regard oxidases [9].
to CYP2D6 and 2C19 suggested that 2D6
Pharmacology 1999;59:298–309
Citalopram Metabolism
307

Fig. 6. Schematic drawing of the metabolic pathways and excretion of citalopram.
Acknowledgements
The main metabolic pathways and the to-
tal elimination of CITA in humans seem
almost fully elucidated (fig. 6). About 38% of
an ingested dose is excreted in the urine either
unchanged or conjugated to glucuronic acid
and about 19% as free DCITA in the urine [4].
About 15% of a dose is excreted as free and
conjugated DDCITA and about 7% as CITA-
NO. Finally, CITA, DCITA, DCCITA and
CITA-NO may be deaminated to and ex-
creted as CITA-PROP in amounts corre-
sponding to about 8% of the dose.
We thank Pia Ploughmann and Birgit Poulsen for
excellent technical assistance. This work has been sup-
ported by the Psychiatric Research Foundation, the
Novo Nordisk Research Foundation, the Lundbeck
Foundation, Jacob Madsen Foundation, Geert-Jør-
gensen’s Foundation and Marshall’s Foundation.
References
1 Milne RJ, Goa KL: Citalopram: A
4 Dalgaard L, Jensen KG, Andersen
JB, Jensen B, Larsen CS: Metabo-
lism of citalopram in man. ISSX
Proc (ISSN 1061-3439), 1995.
5 Oyehaug E, Ostensen ET, Salvesen
B: High-performance liquid chro-
matographic determination of cita-
lopram and four of its metabolites in
plasma and urine samples from psy-
7 Kobayashi K, Chiba K, Yagi T, Shi-
mada N, Taniguchi T, Horie T, Tani
M, Yamamoto T, Ishizaki T, Kuroi-
wa Y: Identification of cytochrome
P450 isoforms involved in citalo-
pram N-demethylation by human
liver microsomes. J Pharmacol Exp
Ther 1997;280:927–933.
8 Rochat B, Amey M, Gillet M, Meyer
UA, Baumann P: Identification of
three cytochrome P450 isozymes in-
volves in N-demethylation of cita-
lopram enantiomers in human
liver microsomes. Pharmacogenet-
ics 1997;7:1–10.
review of its pharmacodynamic and
pharmacokinetic properties, and
therapeutic potential in depressive
illness. Drugs 1991;41:450–477.
2 Hyttel J: Pharmacological character-
ization of selective serotonin reup-
take inhibitors (SSRIs). Int Clin Psy-
chopharmacol 1994;9:19–26.
3 Hyttel J, Bogeso KP, Perregaard J,
Sanchez C: The pharmacological ef-
fect of citalopram residues in the (S)-
(+)-enantiomer. J Neural Transm
Gen Sect 1992;88:157–160.
chiatric patients.
1984;308:199–208.
J Chromatogr
6 Baumann P, Larsen F: The pharma-
cokinetics of citalopram. Rev Con-
temp Pharmacother 1995;6:287–
295.
Pharmacology 1999;59:298–309
308
Olesen/Linnet

9 Rochat B, Kosel M, Boss G, Testa B,
Gillet M, Baumann P: Stereoselec-
tive biotransformation of the se-
lective serotonin reuptake inhibitor
citalopram and its demethylated
metabolites by monoamine oxidases
in human liver. Biochem Pharmacol
1998;56:15–23.
10 Linnet K, Olesen OV: Metabolism
of clozapine by cDNA-expressed hu-
man cytochrome P450 enzymes.
Drug Metab Dispos 1997;25:1379–
1382.
11 Fuhr U, Doehmer J, Battula N, Wol-
fel C, Kudla C, Keita Y, Staib AH:
Biotransformation of caffeine and
theophylline in mammalian cell
lines genetically engineered for ex-
pression of single cytochrome P450
isoforms. Biochem Pharmacol 1992;
43:225–235.
12 Olesen OV, Linnet K: Metabolism
of the tricyclic antidepressant ami-
triptyline by cDNA-expressed hu-
man cytochrome P450 enzymes.
Pharmacology 1997;55:235–243.
13 Leemann T, Transon C, Dayer P:
Cytochrome P450TB (CYP2C): A
major monooxygenase catalyzing di-
clofenac 4)-hydroxylation in human
liver. Life Sci 1993;52:29–34.
14 Hall SD, Guengerich FP, Branch
RAm, Wilkinson GR: Characteriza-
tion and inhibition of mephenytoin
4-hydroxylase activity in human liv-
er microsomes. J Pharmacol Exp
Ther 1987;240:216–222.
15 Dahl ML, Nordin C, Bertilsson L:
Enantioselective hydroxylation of
nortriptyline in human liver micro-
somes, intestinal homogenate, and
patients treated with nortriptyline.
Ther Drug Monit 1991;13:189–
194.
16 Kim RB, O’Shea D: Interindividual
variability of chlorzoxazone 6-hy-
droxylation in men and women and
its relationship to CYPE2E1 genetic
polymorphisms. Clin Pharmacol
Ther 1995;67:645–655.
17 Olesen OV, Linnet K: Simplified
high-performance liquid chromato-
graphic method for the determina-
tion of citalopram and desmethylci-
talopram in serum without interfer-
ence from commonly used psycho-
tropic drugs and their metabolites. J
Chromatogr [B] 1996;675:83–88.
18 Motulsky H: Intuitive Biostatistics.
London, Oxford University Press,
1995, pp 279–282.
19 Hofstee BHJ: On the evaluation of
the constants Vm and Km in enzyme
reactions. Science 1952;116:329–
331.
20 Rochat B, Amey M, Baumann P:
Analysis of enantiomers of citalo-
pram and its demethylated metabo-
lites in plasma of depressive patients
using chiral reverse-phase liquid
chromatography. Ther Drug Monit
1995;17:273–279.
21 Sidhu J, Priskorn M, Poulsen M,
Segonzac A, Grollier G, Larsen F:
Steady-state pharmacokinetics of
the enantiomers of citalopram and
its metabolites in humans. Chirality
1997;9:686–692.
22 Jeppesen U, Gram LF, Vistisen K,
Loft S, Poulsen HE, Brosen K: Dose-
dependent inhibition of CYP1A2,
CYP2C19 and CYP2D6 by citalo-
pram, fluoxetine, fluvoxamine and
paroxetine. Eur J Clin Pharmacol
1996;51:73–78.
23 Kashuba AD, Nafziger AN, Kearns
GL, Leeder JS, Gotschall R, Rocci
ML Jr, Kulawy RW, Beck DJ, Berti-
no JS Jr: Effect of fluvoxamine ther-
apy on the activities of CYP1A2,
CYP2D6, and CYP3A as deter-
mined by phenotyping. Clin Phar-
macol Ther 1998;64:257–268.
24 Olesen OV, Linnet K: Fluvoxamine-
clozapine drug interaction: Inhibi-
tion in vitro of five cytochrome
P450 isoforms involved in clozapine
metabolism. J Clin Psychopharma-
col 1999;in press.
25 Sindrup SH, Brosen K, Hansen MG,
Aaes-Jorgensen T, Overo KF, Gram
LF: Pharmacokinetics of citalopram
in relation to the sparteine and the
mephenytoin oxidation polymor-
phisms. Ther Drug Monit 1993;15:
11–17.
26 Gram LF, Hansen MG, Sindrup
SH, Brosen K, Poulsen JH, Aaes-
Jorgensen T, Overo KF: Citalo-
pram: Interaction studies with le-
vomepromazine, imipramine, and
lithium. Ther Drug Monit 1993;15:
18–24.
27 Øyehaug E, Eide E, Salvesen B: Ef-
fect of phenothiazines on citalopram
steady-state kinetics in psychiatric
patients. Nord Pharm Acta 1984;46:
37–46.
28 Venkatakrishnan K, Greenblatt DJ,
von Moltke LL, Shader RI: Alprazo-
lam is another substrate for human
cytochrome P450-3A isoforms. J
Clin Psychopharmacol 1998;18:
256.
29 Leinonen E, Lepola U, Koponen H:
Substituting carbamazepine with
oxcarbazepine increases citalopram
levels: A report on two cases. Phar-
macopsychiatry 1996;29:156–158.
30 Priskorn M, Larsen F, Segonzac A,
Moulin M: Pharmacokinetic inter-
action study of citalopram and ci-
metidine in healthy subjects. Eur J
Clin Pharmacol 1997;52:241–242.
31 Bondolfi G, Chautems C, Rochat B,
Bertschy G, Baumann P: Non-re-
sponse to citalopram in depressive
patients: Pharmacokinetic and clini-
cal consequences of a fluvoxamine
augmentation. Psychopharmacolo-
gy (Berl) 1996;128:421–425.
Pharmacology 1999;59:298–309
Citalopram Metabolism
309