Clinical and biochemical effects during
treatment of depression with
nortriptyline: The role of
1 0 - hydroxynortriptyline
Plasma concentrations of nortriptyline (NT) and its major metabolite 10-hydroxy-NT (10-0H-NT) were
measured in 30 patients with depression, treated with NT for 3 weeks. Nine patients who recovered
completely had plasma concentrations of NT and 10-0H-NT ranging from 358 to 728 nmol/L and from
428 to 688 nmol/L, respectively. Of the 21 patients who did not recover completely, only four had plasma
concentrations within the window limited by these two plasma concentration ranges. A correlation was
found between the degree of amelioration and the plasma concentration of NT (r, = 0.469; P < 0.01).
Lumbar punctures were performed in 26 patients before and after 3 weeks of NT treatment. Dur-
ing treatment there was a 30.9% mean decrease in the noradrenaline metabolite 4-hydroxy-3-me-
thoxyphenylglycol (HIVIPG) in cerebrospinal fluid (CSF). We could not evaluate the extent to which this
decrease was caused by NT or 10-0H-NT, respectively, because both are strong inhibitors of noradrenaline
uptake. The ratio between the concentration of NT and 10-0H-NT in CSF correlated to the reduction
of HA/PG in CSF (r = 0.397; P < 0.05) and to the amelioration of depression (r, = 0.623; P < 0.001).
This might indicate that NT and 10-0H-NT interact on the noradrenaline system in a nonadditive way.
During treatment there was a 15.2% decrease in CSF concentration of the serotonin metabolite
5-hydroxyindoleacetic acid. The reduction was significantly correlated to the CSF concentration of NT
but not to that of 10-0H-NT. This is in accordance with the fact that NT is a more potent inhibitor of
serotonin uptake than is 10-0H-NT. The dopamine metabolite homovanillic acid in CSF decreased
significantly by 10.0%. The biochemical data indicate that noradrenergic, serotoninergic, and dopamin-
ergic systems are affected by NT treatment and that 10-0H-NT might be more selective on noradrenergic
neurons than the parent drug. (CLIN PHARMACOL THER 1987;42:10-9.)
Conny Nordin, M.D., Leif Bertilsson, Ph.D., and Bo Siwers, M.D. Huddinge, Sweden
Interpretation of the plasma concentration-effect re-
lationship of a drug is complicated by the occurrence
of active metabolites.' The major metabolite of nor-
triptyline (NT), 10-hydroxy-NT (10-OH-NT) inhibits
From the Departments of Psychiatry and Clinical Pharmacology at
the Karolinska Institute, Huddinge University Hospital.
Supported by grants from the Swedish Medical Research Council
(Nos. 3902 and 5454), the Karolinska Institute, and the Soder-
Received for publication Sept. 19, 1986; accepted Dec. 26, 1986.
Reprint requests: Dr. C. Nordin, Department of Psychiatry, Huddinge
Hospital, S-141 86 Huddinge, Sweden.
noradrenaline (NA) uptake with a potency about 50%
that of NT.2 Because 10-0H-NT concentrations in
plasma and cerebrospinal fluid (CSF) usually are higher
than those of the parent drug,' 10-0H-NT probably
contributes to the effects of NT on the noradrenergic
system and may thereby influence NT's antidepressant
Treatment of depression with an NA uptake in-
hibitor like NT decreases the CSF concentration of
the NA metabolite 4-hydroxy-3-methoxyphenyl glycol
(HMPG), probably as a result of feedback inhibition of
transmitter biosynthesis .4 Analogously, treatment with
inhibitors of serotonin uptake such as chlorimipramine
Table I. Data on patients
and zimeldine results in a decreased level of the sero-
tonin metabolite 5-hydroxyindoleacetic acid (5-HIAA)
in CSF.' NT also slightly decreases 5-HIAA in CSF,
suggesting that it inhibits serotonin uptake.' A recent
study' shows that treatment with NT inhibits serotonin
uptake by platelets and that this effect is much stronger
for NT than for 10-0H-NT.
It has been hypothesized that CSF concentrations of
5-HIAA7" of the dopamine metabolite homovanillic
acid (HVA)12'13 and of HMPG-14 may predict the outcome
of treatment with various antidepressant drugs. Thus
endogenously depressed patients with high CSF con-
centrations of 5-HIAA (>15 ng/ml) have been reported
to respond better to NT than do patients with a lower
(<15 ng/ml) concentration.' However, relationships are
difficult to evaluate because of the many factors influ-
encing CSF concentrations of transmitter metabolites. '5
We have now examined indirectly the clinical and
biochemical roles of 10-0H-NT during treatment of
depression with NT.
The study included 39 nondelusional, depressed in-
patients who came to our psychiatric clinic on their own
initiative and who fulfilled the Feighner criteria for pri-
mary depression.' The subjects had a median depres-
sion score of 17, which reflects a depression of mod-
erate severity (see Methods) (Table I). All were judged
by two experienced psychiatrists to require inpatient
care. All were physically healthy according to history,
routine laboratory tests, and physical examination.
None of the patients had received treatment with an-
tidepressants or neuroleptics for 1 month before inclu-
sion. Some patients had been given occasional doses
of benzodiazepines during this period.
During an initial placebo period five patients were
excluded because of spontaneous recovery. Four pa-
tients left the study during the first week of NT treat-
ment. Of these, two recovered, one discontinued med-
ication because of a severe orthostatic reaction, and one
was excluded after a reconsideration of the Feighner
criteria. Data on the subjects are given in Table I. Of
the 30 patients with depression who completed 3 weeks
of treatment, 27 had unipolar disorder, two had bipolar
disorder with histories of manic episodes, and one had
bipolar disorder with previous hypomanic episodes.
After a placebo period of 3 to 7 days, NT (Sensaval,
Pharmacia Laboratories) treatment was begun at 50 mg
t.i.d. (8 AM, 2 PM, and 10 Pm) in 31 patients, 50 mg
b.i.d. (8 AM and 10 Pm) and 25 + 50 mg (8 AM and
10 Pm) in two patients with histories of side effects,
and 25 mg b.i.d. (8 AM and 10 Pm) in a 75-year-old
patient. These doses were maintained for at least 3
weeks except for the four patients excluded who were
treated with 50 mg t.i.d. Occasional daily doses of
nitrazepam (5 mg) or oxazepam (15 to 30 mg) were
Plasma samples were drawn at 8 AM, before the
morning dose of NT, once or twice a week from the
second week of treatment. Blood was collected in hep-
arinized glass tubes and centrifuged within 30 minutes.
Plasma was separated and stored at
ysis. All samples from each patient were analyzed si-
multaneously by mass fragmentography.19-" The re-
ported values of NT and unconjugated 10-0H-NT (sum
of E and Z isomers) are the means of the concentrations
measured during the second and the third weeks of
Lumbar punctures were performed by a standardized
technique at the level L 4-L 5 before and after 3 weeks
of NT treatment. Twelve milliliters of CSF was col-
lected in the early morning with the patient in a sitting
position after at least 8 hours of fasting and strict bed
rest. CSF samples were centrifuged immediately,
frozen in 3 ml aliquots in silanized glass tubes, and
stored at 20° C until analysis. CSF samples from
each patient were analyzed simultaneously. Concentra-
tions of NT, unconjugated 10-OH-NT (sum of E and Z
20° C until anal-
(mean ± SD)
46 ± 8
43 ± 13
45 ± 8
44 ± 13
45 ± 8
44 ±- 12
After the placebo period
After 3 weeks of NT treatment
Depression and 10-hydroxynortriptyline
12 Nordin et al.
Placebo Weeks of NT treatment
Fig. 1. Median depression scores in patients with low (<15 ng/m1 CSF) (n = 10) and high (>15
ng/ml CSF) (n = 19) concentration of 5-HIAA before treatment (x, difference between the two
groups at the end of placebo period [P < 0.05]; xx, difference between depression scores after 3
weeks of NT treatment compared with after placebo period; this was significant in both groups
[P < 0.01]).
<1 5ng 5-HIAA/m1 CSF
>1 5ng 5-HIAA/m1 CSF
(n=1 9 1xx
isomers), and amine metabolites 5-HIAA, HVA, and
HMPG were analyzed by mass fragmentography.17,19,20
At three lumbar punctures only 6 ml CSF was obtained.
The concentrations of 5-HIAA and HVA in these 6 ml
samples were adjusted to correspond to a 12 ml sample
of CSF according to Bertilsson et al.'
Severity of depression was assessed by using 10 items
described in the Montgomery-Asberg Depression Rat-
ing Scale (MADRS)2' at the time of inclusion, at the
end of the placebo period, and then once a week. Scale
steps 0, 1, 2, and 3 with half steps were used. The sum
of the rating scores was used as a measure of severity.
These scores are half of the corresponding MADRS
scores. The amelioration score was defined as the dif-
ference between the score at baseline (week 0) and
during treatment (week 3). Most ratings were performed
by two of the authors in a joint interview. The ratings
of C. N. were used in the statistical analysis except for
some interviews when B. S. rated the patient alone.
The added scores were used to calculate the Spearman
rank correlation coefficient. The interrater reliability
was 0.92 (n = 25) at inclusion and 0.95 (n = 17) after
3 weeks of NT treatment.
Side effects were rated simultaneously with depres-
sion with a specially designed scale as described by
Asberg et al 22 The summed score of the 11 items was
used as the "raw side effect score." By subtracting the
pretreatment raw side effect score from that after 3
weeks of treatment, the "corrected side effect score"
was obtained. Raw and corrected scores after 3 weeks
of treatment were used in the statistical analysis.
CLIN PHARMACOL THER
All statistical analyses were performed with programs
from the MIMER/ST system on a computer at the Ka-
Rating data were analyzed by nonparametric statistics
according to Siegel." Parametric statistics were used
according to Snedecor and Cochran.'
The study was approved by the Ethics Committee at
the Huddinge Hospital.
Clinical effects. Thirty patients completed 3 weeks
of NT treatment. Lumbar puncture was performed twice
(before and after 3 weeks of treatment) in 26 patients
and once (before treatment) in three patients.
The 10 patients with a low pretreatment concentration
of 5-HIAA (<15 ng/ml CSF) tended to have higher
depression scores than the 19 patients with a high con-
centration (>15 ng/ml CSF) from the time of inclusion
until week 3 (significant only at week 0; Mann-Whitney
U test, U = 51.5; P < 0.05) (Fig. 1). In both groups
of patients there was a highly significant reduction of
median depression score after 3 weeks of treatment
(Wilcoxon matched-pairs signed-ranks test; T = 0.00
and T = 6.50, respectively; P < 0.01) (Fig. 1).
The steady-state NT concentrations ranged from 186
to 960 nmol/L in plasma and from 13 to 98 nmol/L in
CSF. For 10-0H-NT the corresponding ranges were 313
to 1072 nmol/L and 24 to 131 nmol/L.
In the total group of patients there was a correla-
tion between the amelioration score and the plasma
(n = 30; I., = 0.469; P < 0.01), as well as CSF
*P <0.01; TP < 0.05
P < 0.001.
(n = 25; r, = 0.515; P < 0.01) concentration of NT
(Fig. 2; Table II). In patients with a high 5-HIAA
concentration, a correlation was found between ame-
lioration score and the plasma NT (n = 19; rs =
0.594; P < 0.05), as well as CSF NT (n = 17;
r, = 0.561;P <0.05). In patients with a low 5-HIAA
concentration, the relationships between amelioration
score and the plasma NT (n = 10; r, = 0.304), as well
as CSF NT (n = 8; r, = 0.448), were positive but did
not reach statistical significance.
Rank correlations were also calculated for relation-
ships between amelioration score and the plasma and
CSF concentration:NT (nM)
Fig. 2. Relationships between amelioration score and NT concentrations in plasma (left) and CSF
(right) in patients with a low (0) (<15 ng/ml CSF) and a high () (>15 ng/ml CSF) pretreatment
5-HIAA concentration. (In one patient () no CSF was obtained.) Coefficients of Spearman rank
correlation are given in Table II.
Table II. Relationships between amelioration score and concentrations of NT and 10-0H-NT in plasma
CSF concentrations of (a) 10-0H-NT, (b) NT +
(10-0H-NT), and (c) NT + 0.57 (10-0H-NT). (The
factor 0.57 is derived from in vitro data2; Table II.) No
correlation was stronger than that with NT alone.
Correlations were also found between the amelio-
ration score and the ratio NT/10-0H-NT in plasma
(n = 30; r, = 0.420; P < 0.05) and CSF (n = 25;
r, = 0.623; P < 0.001) (Table II). In patients with a
high CSF 5-HIAA concentration, corresponding cor-
relations were found for plasma (n = 19; rs = 0.415;
P < 0.05) and CSF (n = 17; r, = 0.636; P <0.01).
In patients with a low CSF 5-HIAA, a correlation
Pretreatment level of 5-HIAA in CSF
NT + 10-OH-NT
NT + 0.57 (10-OH-NT)
Plasma concentration:NT (nM)
Depression and 10-hydroxynortriptyline 13
14 Nordin et al.
*P < 0.01; tP < 0.05.
was found for CSF (n = 8; r, = 0.673; P < 0.05) but
not for plasma (n = 10; r, = 0.403; P not signif-
Before starting further statistical analysis, the pa-
tients were grouped as follows according to the final
depression score at week 3: responders:
partial responders: 5.5 to 10 (n = 9); and nonrespond-
ers: >10 (n = 12).
The nine responders had steady-state plasma NT con-
centrations within a range of 358 to 728 nmol/L with
a simultaneous plasma 10-0H-NT range of 428 to 688
nmol/L. In CSF the corresponding ranges were 23 to
60 nmol/L for NT and 47 to 81 nmol/L for 10-0H-NT
(Fig. 3). Only four patients who had partial or no
5 (n = 9);
Table IV. Coefficients of correlation between
pretreatment CSF concentrations of amine
metabolites and body height and age (n = 26)
CLIN PHARMACOL THER
response had plasma or CSF concentrations within the
windows limited by these two concentration ranges
No correlation was found between the plasma and
CSF concentrations of NT and 10-0H-NT and the raw
or corrected side effect score nor any specific side
Biochemical effects. As expected from a strong in-
hibitor of NA uptake, CSF concentrations of HMPG
decreased significantly (30.9%; P < 0.001) during
NT treatment (Table III). Concentrations of 5-HIAA
(P < 0.001) and HVA (P < 0.05) also decreased but
less markedly (Table III).
The pretreatment CSF concentrations of 5-HIAA and
HVA were correlated to body height (Table IV), and
CSF HMPG concentration was correlated to age (Table
IV). A correlation was found between the concentra-
tions of 5-HIAA and HVA before treatment (Table V).
There was also a significant, but weaker, correlation
between the CSF 5-HIAA and HMPG (Table V). For
all three amine metabolites, CSF concentrations before
and during NT treatment were strongly correlated
We studied the effects of NT and 10-0H-NT during
94.1 ± 37.7
200.4 -± 108.2
49.4 ± 10.3
Plasma concentration:NT (nM)
CSF concentration:NT (nM)
Fig. 3. Relationships between the concentrations of NT and 10-0H-NT in plasma (left) (n = 30)
and CSF (right) (n = 23) in patients grouped according to depression score after 3 weeks of NT
E , 5.5 to 10; and , >10).
Table III. Amine metabolite levels in CSF of 26 patients before and during NT treatment (mean ± SD)
(mean -± SD)
78.5 ± 33.0
176.2 ± 108.2
33.7 ± 6.0
-15.2 ± 20.2
-10.0 ± 23.2
-30.9 ± 10.3
Table V. Coefficients of correlation for
pretreatment concentrations of amine
metabolites (n = 26)
y = 66X, - 0.53X2 + 45
y = 73X, - 39X2 + 42
y = 101X, + 0.72X2 - 0.28X3 + 34X4
y = 101X, + 45X2 - 21
NT treatment on monoamine metabolite concentrations
by multiple stepwise regression analysis. The concen-
trations of 5-HIAA, HVA, and HMPG during treatment
as percent of pretreatment values (below indicated by
p5-HIAA, pHVA, and pHMPG) were in turn used as
dependent variables. The independent variables were
the percent of pretreatment values of the two other
monoamine metabolites and the CSF concentrations of
NT and 10-0H-NT. The results were as follows.
As seen in Table VII (upper panel), 42.9% of the
variance in the p5-HIAA was accounted for by the pHVA
(P < 0.01). The CSF NT concentration increased the
fraction of explained variance by 35.6% (P < 0.01).
The CSF concentration of 10-0H-NT and pHMPG
made almost no additional contribution. The explained
variance in pHVA was accounted for by the p5-HIAA
(42.9%) and the concentration of NT (30.1%). The CSF
concentration of 10-0H-NT increased the explained
variance by 3.7%, which was significant (P < 0.05)
but only when the pHMPG was taken into account. The
variance in the pHMPG was not explained significantly
by any of the four independent variables.
Replacing the CSF NT and 10-0H-NT concentra-
tions by the ratio NT/10-0H-NT as independent vari-
able (Table VII, lower panel) gave results for p5-HIAA
Depression and 10-hydroxynortriptyline 15
Table VI. Coefficients of correlation for
concentrations of amine metabolites before and
during NT treatment (n = 26)
*P <0.001; -1P <0.05.
Table VII. Multiple stepwise regression of monoamine metabolite concentrations in CSF as percent of
pretreatment values (p5-HIAA, pHVA, and pHMPG) and CSF NT and 10-0H-NT concentrations (n = 25)
*P < 0.01; t significance (P < 0.05) was reached when pHMPG was taken into account; R is the multiple correlation coefficient; t P < 0.05.
and pHVA that were very similar to the previous anal-
ysis (Table VII, upper panel). By contrast, 15.9% of
the variance in the pHMPG was accounted for by the
ratio (P < 0.05) (Table VII, lower panel).
Multiple stepwise regression analysis was performed
in the same way with plasma instead of CSF concen-
trations of NT and 10-0H-NT. Similar results were
obtained. This was expected because the CSF and
plasma levels are closely correlated for both NT
(r = 0.92; P < 0.001) and 10-0H-NT (r -=- 0.77;
P < 0.001) as previously reported.3
With a linear regression analysis, a relationship was
found between p5-HIAA and the CSF NT concentration
(r - 0.475; n = 25;P < 0.05;y = 101 - 0.42x)
whereas no correlation was found between the CSF NT
concentration and the pHVA (r = 0.172) or pHMPG
(r = -0.315). However, a linear relationship was
found between the pHMPG and the ratio NT/10-0H-
NT (r = -0.399; P < 0.05). As shown in Table VII,
there was a strong influence of the pHVA on the
p5-HIAA (and vice versa). This is in line with the
relationship between the percentage changes in HVA
and 5-HIAA concentrations (linear regression analysis)
shown in Fig. 4 (r = 0.610; P < 0.001).
Relationship between clinical and biochemical ef-
y = 77 - 13X,
Fig. 4. Relationship between the change in CSF HVA and
5-HIAA concentrations (percent of pretreatment level) after
3 weeks of NT treatment (n = 26; y = 0.0068 + 0.70x).
fects. Pretreatment concentrations of HVA and 5-HIAA
were strongly correlated (Table V) (regression line
y = 2.40 x 3.78). In the three subgroups of pa-
tients, no difference was found between the pretreat-
ment regression coefficients (analysis of covariance;
F2.20 = 0.137; P not significant) (Fig. 5, left). How-
ever, during treatment a significant difference was found
between the regression coefficients of responders, par-
tial responders, and nonresponders (analysis of covar-
iance; F220 = 4.983; P < 0.05) (Fig. 5, right).
We found significant correlations between the con-
centrations of NT in plasma and CSF and antidepressant
effect ( Fig. 2; Table II). Previous studies have reported
a curvilinear relationship between plasma NT and ther-
apeutic outcome.25-27 To demonstrate less response at
high plasma concentrations of NT, several such patients
need to be studied. This was the case in the study by
Montgomery et al.,' in which most of the patients had
plasma NT concentrations exceeding 800 nmol/L. Only
two of our 30 patients had NT concentrations above
800 nmol/L; therefore our results do not conflict with
the concept of a decreased therapeutic response to NT
at concentrations above the so-called "therapeutic win-
dow." Another factor contributing to the absence of a
curvilinear relationship might be the duration of treat-
ment. In the studies by Kragh-SOrensen et al. ,26,27 Mont_
gomery et al.,' and Ziegler et al.," NT was given
during 4 or 6 weeks. After that period of treatment
patients with high NT concentrations (>140 ng/ml or
532 nmol/L) deteriorated compared with patients hav-
ing a lower plasma NT concentration.'
Our principal aim was to explore the role of
10-0H-NT in the therapeutic outcome of patients
with primary depression, treated with NT. The
strong correlation (r, = 0.623) between the ratio of
NT/10-0H-NT in CSF and amelioration might indicate
that NT and 10-0H-NT interact in a nonadditive way.
The ranges described for plasma and CSF concen-
trations of NT and 10-0H-NT illustrated in Fig. 3 do
not fulfill the prerequisites for hypothesis testing be-
cause they were derived from inspection of the data.
Hypothetically there might exist a window, as outlined
in Fig. 3, within which an optimal balance between NT
and 10-0H-NT concentrations is reached. This balance
might offer a pharmacologic basis for a favorable an-
tidepressant effect. Interpretation is complex and dif-
ficult because both NT and 10-0H-NT are potent in-
hibitors of NA uptake2 and both compounds might have
postsynaptic receptor-blocking properties. In very early
studies, such properties were demonstrated for several
tricyclics including NT." We recently described the
absence of an antidepressant effect in a patient with a
"therapeutic" NT plasma concentration.' This patient
was an extremely rapid hydroxylator of NT and was
therefore treated with high doses of the drug. The
plasma concentration of 10-0H-NT was very high
(2730 nmol/L) and might have decreased the antide-
pressant effect of NT by receptor blockade.
The CSF concentration of 5-HIAA has a bimodal
distribution in depressed patients, and a biochemical
heterogeneity of endogenous depression has been sug-
gested.7'32-" Asberg et al.' found a significant antide-
pressant effect of NT treatment in patients with a high
(>15 ng/ml) but not with a low (<15 ng/ml) pretreat-
ment CSF concentration of 5-HIAA. Because NT is an
NA uptake inhibitor,' they interpreted this as support
for decreased noradrenergic activity in patients with a
high 5-HIAA in CSF. Our study in a slightly larger
patient cohort shows a significant and similar antide-
pressant effect of NT treatment in both groups of pa-
tients. Because NT influences both the noradrenergic2
and the serotoninergic neurons,' it is not an ideal tool
to demonstrate biochemical subtypes of patients with
depression. By contrast, 10-0H-NT is a more specific
inhibitor of NA uptake2'8 and in the future might better
serve this purpose.
NT treatment markedly reduced HMPG concentra-
tions in CSF, in keeping with results of an earlier study.'
The absence of a correlation between the degree of
reduction and the concentrations of NT and 10-0H-NT
(Table VII, upper panel) is surprising, because both
these compounds are potent inhibitors of NA uptake.2
CLIN PHARMACOL THER
16 Nordin et al.
5-HIAA in CSF (nM) before
A weak correlation (r = 0.42; P < 0.05) has been
found previously between the decrease of CSF HMPG
and the plasma concentration of demethylchlorimipra-
mine, a metabolite of chlorimipramine, which strongly
inhibits NA uptake.' The use of changes in CSF HMPG
as a measure of effects of drugs on central noradrenergic
mechanisms is complicated by the fact that 25% to 50%
of the HMPG in CSF is derived from the periphery.'
This peripheral contribution might explain why CSF
HMPG decreases by only about 30% (with very little
variation) during treatment with different potent NA
The significant decrease in CSF 5-HIAA concentra-
tion (Table III) was related to the CSF NT concentration
(Table VII), whereas CSF 10-0H-NT had no influence.
This is in accordance with an in vitro study showing
that the parent drug NT is a stronger serotonin uptake
inhibitor than its hydroxymetabolite in brain slices.' A
recent study' has also shown that NT treatment in-
creases the rate constant for serotonin uptake by plate-
lets, by competitive uptake inhibition. These results
show that uptake inhibition in serotonin neurons during
NT treatment is significant and caused by the parent
drug but not its hydroxymetabolite. Since the decrease
in HMPG concentration did not covary with that of
5-HIAA (Table VII), we cannot confirm the hypothesis
that the reduction of CSF 5-HIAA concentration is me-
diated via a noradrenergic mechanism.'
5-HIAA in CSF (nM) after
3 weeks of NT treatment
Fig. 5. Relationships between HVA and 5-HIAA in CSF. Left (before treatment): 1, responders
(n = 9) (y = 2.50 x 21.96); 2, partial responders (n = 8) (y = 2.59 x 43.63); and 3,
nonresponders (n = 9) (y = 2.20 x 20.71; analysis of covariance: F220 -= 0.137; NS). Right
(after 3 weeks of NT treatment): 1, responders (n = 9) (y
sponders (n = 8) (y = 2.37 x +2.21); and 3, nonresponders (n = 9) (y = 1.70 x + 4.52;
analysis of covariance: F220 = 4.983; P < 0.05).
5.42 x 243.7); 2, partial re-
The decrease in CSF HVA is less than that of
5-HIAA (Table III) but depends on NT concentration
to about the same degree (Table VII). For HVA the
10-0H-NT concentration seems to be important but
only when the change in HMPG is taken into account.
This might indicate that part of the effect on dopamine
neurons during NT treatment is mediated indirectly via
the NA system.
There is a pronounced gradient for 5-HIAA and HVA
along the spinal canal.' The concentrations of these
amine metabolites are different in CSF obtained from
two adjacent lumbar interspaces .39 Unpublished results
from our group indicate that the direction of the needle
within an interspace during lumbar puncture may also
influence 5-HIAA and HVA concentrations. We always
perform lumbar punctures in the interspace L4-L5, but
it is difficult to standardize for the inclination of the
needle within the interspace. This may explain in part
why pretreatment concentrations of these two amine
metabolites are strongly related (r = 0.867; Table V)
and why the effects of NT treatment on the two me-
tabolites covary (Table VII; Fig. 4). A similar covar-
iation between the effect of NT treatment on 5-HIAA
and HVA in CSF has previously been shown in a study
of obsessive-compulsive disorder."
This study confirms that NT treatment causes an in-
hibition of NA uptake, probably mediated by both the
parent drug and its 10-OH-metabolite. There is also an
Depression and 10-hydroxynortriptyline 17
18 Nordin et al.
effect on the serotonin system that seems to be mediated
entirely by NT itself. The effect on the dopaminergic
system is probably mediated principally by NT but also
by 10-0H-NT. A shift of dopaminergic-serotoninergic
balance might be important for clinical improvement.
Therefore the clinical and biochemical effects of NT
and its 10-OH-metabolite need to be analyzed further
in terms of interaction between transmitters. We do not
know which of these neuronal effects is most important
for the antidepressant effect of NT treatment. Because
E-10-0H-NT is a potent in vitro inhibitor of NA uptake
with little effect on serotonin, this drug is a more spe-
cific tool for elucidating the role of these three neuronal
systems in antidepressant drug action.
We thank Elisabeth Berg, Per Nasman, and John ohrvik
for statistical advice, Christina Lynga for professional care
of the patients, and Jolanta Widen for skillful technical as-
Atkinson AJ Jr, Strong JM. Effect of active drug me-
tabolites on plasma level-response correlations. J Phar-
macokinet Biopharm 1977;5:95-109.
Bertilsson L, Mellstrom B, Sjoqvist F. Pronounced in-
hibition of noradrenaline uptake by 10-hydroxymetabo-
lite of nortriptyline. Life Sci 1979;25:1285-92.
Nordin C, Bertilsson L, Siwers B. CSF and plasma levels
of nortriptyline and its 10-hydroxy metabolite. Br J Clin
Bertilsson L, Asberg M, Thor& P. Differential effect of
chlorimipramine and nortriptyline on cerebrospinal fluid
metabolites of serotonin and noradrenaline in depression.
Eur J Clin Pharmacol 1974;7:365-8.
Siwers B, Ringberger V-A, Tuck JR, SjOqvist F. Initial
clinical trial based on biochemical methodology of zi-
melidine (a serotonin uptake inhibitor) in depressed pa-
tients. CLIN PHARMACOL THER 1977;21:194-200.
Traskman L, Asberg M, Bertilsson L, etal. Plasma levels
of chlorimipramine and its demethyl metabolite during
treatment of depression. CLIN PHARMACOL THER 1979;
Asberg M, Bertilsson L, Tuck D, Cronholm B, Sjoq-
vist F. Indoleamine metabolites in the cerebrospinal fluid
of depressed patients before and during treatment with
nortriptyline. CLIN PHARMACOL THER 1973;14:277-86.
Malmgren R, Aberg-Wistedt A, Bertilsson L. Sero-
tonin uptake inhibition during treatment of depression
with nortriptyline caused by parent drug and not by
10-hydroxymetabolites. Psychopharmacology (Berlin)
Sjoqvist F. A pharmacokinetic approach to the treatment
of depression. Int Pharmacopsychiatry 1971;6:147-69.
Muscettola G, Goodwin FK, Potter WZ, Claeys MM,
Markey SP. Imipramine and desipramine in plasma and
spinal fluid. Arch Gen Psychiatry 1978;35:621-5.
Maas JW, Koslow SH, Katz MM, et al. Pretreatment
CLIN PHARMACOL THER
neurotransmitter metabolite levels and response to tri-
cyclic antidepressant drugs. Am J Psychiatry 1984;141:
Post RM, Gerner RH, Carman JS, et al. Effects of a
dopamine agonist piribedil in depressed patients. Arch
Gen Psychiatry 1978;35:609-14.
Gerner RH, Post RM, Gillin JC, Bunney WE. Biological
and behavioural effects of one night's sleep deprivation
in depressed patients and normals. J Psychiatr Res
Nordin C, Siwers B, Bertilsson L. Bromocriptine treat-
ment of depressive disorders. Acta Psychiatr Scand
Bertilsson L, Asberg M. Amine metabolites in the ce-
rebrospinal fluid as a measure of central neurotrans-
mitter function: methodological aspects. In: Usdin E,
Asberg M, Bertilsson L, Sjoqvist F, eds. Frontiers in
biochemical and pharmacological research in depression.
New York: Raven Press, 1984:27-34.
Feighner JP, Robins E, Guze SB, Woodruff RA, Wino-
cur G, Munoz R. Diagnostic criteria for use in psychiatric
research. Arch Gen Psychiatry 1972;26:57-63.
Bertilsson L. Quantitative mass fragmentography-a
valuable tool in clinical pharmacology. In: Usdin E, ed.
Clinical pharmacology in psychiatry. New York: Elsevier
Science Publishing Co Inc, 1981:59-71.
Bertilsson L, Asberg M, Lantto 0, Scalia-Tomba G-P,
Traskman-Bendz L, Tybring G. Gradients of monoamine
metabolites and cortisol in cerebrospinal fluid of psy-
chiatric patients and healthy controls. Psychiatry Res
Borga 0, Palmer L, Sjoqvist F, Holmstedt B. Mass frag-
mentography used in quantitative analysis of drugs and
endogenous compounds in biological fluids. In: Phar-
macology and the future of man. Proceedings of the Fifth
International Congress of Pharmacology, San Francisco,
Calif., 1972. Basel: Karger, 1973;3:56-68.
Bertilsson L, Nordin C, Otani K, et al. Disposition of
single oral doses of E-10-hydroxynortriptyline in healthy
subjects with some observations on pharmacodynamic
effects. CLIN PHARMACOL THER 1986;40:261-7.
Montgomery SA, Asberg M. A new depression scale
designed to be sensitive to change. Br J Psychiatry
Asberg M, Cronholm B, Sjoqvist F, Tuck D. Correlation
of subjective side effects with plasma concentrations of
nortriptyline. Br Med J 1970;4:18-21.
Siegel S. Nonparametric statistics for the behavioral sci-
ences. New York: McGraw-Hill Inc, 1956.
Snedecor GW, Cochran WG. Statistical methods, ed 7.
Ames: Iowa State University press, 1982.
Asberg M, Cronholm B, Sjoqvist F, Tuck D. Relation-
ship between plasma level and therapeutic effect of nor-
triptyline. Br Med J 1971;3:331-4.
Krogh-Sorensen P. Asberg M, Hansen CE. Plasma-nor-
triptyline levels in endogenous depression. Lancet 1973;
Krogh-Sorensen P, Hansen CE, Baastrup PC, Hvidberg
EF. Self-inhibiting action of nortriptyline's antidepres-
sive effect at high plasma levels. A randomized, double-
blind study controlled by plasma concentrations in pa-
tients with endogenous depression. Psychopharmacology
Montgomery S, Braithwaite R, Dawling S, McAuley R.
High plasma nortriptyline levels in the treatment of
depression. I. CLIN PHARMACOL THER 1978;23:309-14.
Ziegler VE, Clayton PJ, Taylor JR, Co BT, Biggs JT.
Nortriptyline plasma levels and therapeutic response.
CLIN PHARMACOL THER 1976;20:458-63.
Haefely W, Hfirlimann A, Thoenen E. Scheinbar para-
doxe Beeinflussung von peripherer Noradrenalinwirkun-
gen durch einige Thymoleptika. Hely Psycho! Acta
Bertilsson L, Aberg-Wistedt A, Gustafsson LL, Nor-
din C. Extremely rapid hydroxylation of debrisoquine: a
case report with implication for treatment with nortrip-
tyline and other tricyclic antidepressants. Ther Drug
van Praag HM, Korf J. Endogenous depressions with and
without disturbances in the 5-hydroxytryptamine metab-
olism: a biochemical classification? Psychopharmacology
Asberg M, Thoren P, Traskman L, Bertilsson L, Ring-
berger V. "Serotonin depression"a biochemical sub-
group within the affective disorders? Science 1976;
Depression and 10-hydroxynortriptyline 19
van Praag HM. Significance of biochemical parameters
in the diagnosis, treatment and prevention of depressive
disorders. Biol Psychiatry 1977;12:101-34.
Gibbons RD. Consistent evidence for a biological subtype
of depression characterized by low CSF monoamine
levels. Acta Psychiatr Scand 1986;74:8-12.
Carlsson A, Corrodi H, Fuxe K. Hokfelt T. Effects of
some antidepressant drugs on the depletion of intraneu-
ronal brain catecholamine stores caused by 4,a-dimethyl-
metatyramine. Eur J Pharmacol 1969;5:367-73.
Kopin IJ, Blomberg P. Ebert MH, et al. Disposition and
metabolism of MHPG-CD, in humans: plasma MHPG
as the peripheral pathway of norepinephrine metabolism
and as an important determinant of CSF levels of MHPG.
In: Usdin E, Asberg M, Bertilsson L, Sjoqvist F, eds.
Frontiers in biochemical and pharmacological research
in depression. New York: Raven Press, 1984:57-68.
Moir ATB, Ashcroft GW, Crawford TBB, Eccleston D,
Guldberg HC. Cerebral metabolites in cerebrospinal fluid
as a biochemical approach to the brain. Brain 1970;
Nordin C, Bertilsson L, Siwers B. Site of lumbar punc-
ture influences levels of monoamine metabolites. Arch
Gen Psychiatry 1982;39:1445.
Thor& P, Asberg M, Bertilsson L, Mellstrom B, Sjoq-
vist F, Traskman L. Clomipramine treatment of obses-
sive-compulsive disorder. Arch Gen Psychiatry 1980;37: