Journal of Affective Disorders 44 (1997) 101–109

Research report

Serotonergic autoreceptor blockade in the reduction of

antidepressant latency: personality variables and response to
paroxetine and pindolol
a b a a,
Maria B. Tome , C. Robert Cloninger , James P. Watson , Michael T. Isaac *
a
Department of Psychological Medicine, UMDS ( Guy’ s Campus), London UK
b
Department of Psychiatry, Washington University, St Louis, MO, USA

Received 17 December 1996; accepted 26 February 1997

Abstract

No antidepressant currently in use exerts a significant antidepressant effect for at least two to three weeks after the patient
starts taking it. Open studies suggest that, for selective serotonergic re-uptake inhibitor (SSRI) antidepressants, this latency
may be reduced when the drug is taken with the 5HT 1A receptor blocker pindolol. We have undertaken a randomised,
placebo controlled, double blind trial of augmentation of the selective SSRI antidepressant paroxetine in combination with
pindolol. All our patients (n 5 54; mean age 36 [range 19–65]) met criteria for major depression and received a standard
dose (20 mg o.d.) of paroxetine plus, randomly, either pindolol (2.5 mg t.d.s.) or placebo for six weeks. We examined
personality variables in 48 consecutive subjects according to a short version (TCI-125) of Cloninger et al’s self-rated
Temperament and Character Inventory (Cloninger et al., 1994) and correlated the results with clinical responses in the trial.
The results suggest that personality can influence clinical outcome. After the double blind period patients were offered
paroxetine 20 mg or 40 mg for up to 6 months. Twenty-six patients took this up. The results suggest that high scores in the
temperament dimension of Reward Dependence and low scores in the temperament dimension of Harm Avoidance had a
better outcome at 6 weeks. Patients who had received paroxetine and pindolol during the trial and who reported high Novelty
Seeking and low Harm Avoidance scores had a better outcome at 6 weeks and 6 months. We suggest that temperament
factors may influence outcome of antidepressant treatment.  1997 Elsevier Science B.V.

Keywords: Major depression; Paroxetine; Pindolol; Randomised controlled trial; Temperament and character inventory;
Compliance

1. Introduction
*Corresponding author. Suite 6, Lewisham Hospital, Lewisham
High St, London SE13 6LH, UK. Tel.: 1 44 181 3333000 x6338; This paper examines the predictive value of
fax: 1 44 181 3333333; e-mail: 100736.3070@compuserve.com personality variables in the response to treatment

0165-0327 / 97 / $17.00  1997 Elsevier Science B.V. All rights reserved

PII S0165-0327( 97 )00030-X
102 M.B. Tome et al. / Journal of Affective Disorders 44 (1997) 101 – 109
with paroxetine and pindolol compared with parox-
etine and placebo. The addition of pindolol, a 5HT 1A
presynaptic autoreceptor blocker, to paroxetine, a
selective serotonin receptor inhibitor, has been
shown to increase serotonergic transmission in ani-
mal studies (Artigas et al., 1996). The combination
has been used in humans to accelerate the antidepres-
sant effect of SSRIs (Artigas et al., 1994; Blier and
Bergeron, 1995; Isaac and Tome, 1997). However,
the mechanism of action is still unclear.
In a psychobiological model of temperament and
character, Cloninger (1986) identified four dimen-
sions of temperament, (Novelty Seeking, Harm
Avoidance, Reward Dependence and Persistence),
which represent the more ‘hard wired’ aspects of
personality development, with a hereditary compo-
nent and a putative basis in what may be called the
neurochemical tone of the main aminergic neuro-
transmitter systems. Cloninger also described three
dimensions of character (Self Directedness, Co-
operativeness and Self Transcendence), that represent
the more ‘soft wired’ elements of personality involv-
ing the view of self. The character dimensions are
susceptible to change through experience and educa-
tion, for example, to a degree that the temperament
dimensions are not. Unsurprisingly, neurochemical
correlates of character are more elusive than in the
case of temperament.
One feels intuitively that personality variables
should influence the outcome of treatment. Joyce and
Paykel (1989) argued that the depressed patient with
a good premorbid personality, an insidious onset of
depression, psychomotor retardation, and an inter-
mediate level of severity and ‘endogeneity’, without
psychotic features, may be the patient who responds
best to treatment with tricyclic antidepressants. The
presence of personality disorder is believed to predict
a poor outcome of antidepressant treatment (Kerr et
al., 1970; Weissman et al., 1978; Duggan et al.,
1990; Andrews et al., 1990).
Joyce et al. (1994), explored the possible role of
personality in the short term clinical outcome of a
randomised comparative trial of clomipramine and
desipramine in major depression using a self rated,
tridimensional personality questionnaire (TPQ).
Their results in 45 severely depressed women
showed that low Harm Avoidance and high Reward
Dependence scores in the group of patients receiving
clomipramine (which has a potent serotonergic reup-
take blockade effect) had a better outcome at 6
weeks.
The present study employs a shorter version (TCI
125) of the self-rated Temperament and Character
Inventory (Cloninger et al., 1994) to try and throw
light upon the question of whether personality vari-
ables can influence outcome, in patients treated with
antidepressants, and also to elucidate possible neuro-
chemical bases of the clinical response to the parox-
etine and pindolol combination. We were also most
interested to see if it was possible, from considering
personality variables, to predict which patients would
not be compliant with medication for the further six
months. Compliance with medication for at least six
months has been regarded as fundamental for the
treatment of an episode of depression. An earlier
study (Wingerson et al., 1993) using the Tempera-
ment and Character Inventory showed that patients
with Generalised Anxiety Disorder who scored high-
ly in the temperament dimension of Novelty Seeking
tended to comply less well with treatment.
2. Method
A double blind randomised trial was designed in
order to assess the efficacy of paroxetine (20 mg
o.d.) and pindolol (2.5 mg t.d.s) compared with
paroxetine (20 mg o.d.) and placebo (2.5 mg t.d.s.).
One way ANOVA was performed with scores in the
˚
Montgomery-Asberg Depression Rating Scale
(MADRS: Montgomery and Asberg, ˚ 1979) at days
0, 4, 7, 10, 14, 21, 28 and 42, and at week 24 (6
months) of the trial (Tome et al., 1997). The first part
of the study was a six week, double blind, placebo
controlled study. After six weeks, patients were able
to continue taking paroxetine 20 mg or 40 mg alone
for a further 6 months or more. Written informed
consent was obtained from each patient. Twenty-six
patients elected to continue taking medication, 10
who had been pre-treated with paroxetine and pin-
dolol and 16 who had received paroxetine and
placebo. The clinical characteristics of the study
population are summarised in Table 1 and Table 2.
The data were analysed for the pooled population;
for patients who received paroxetine and pindolol
(the paroxetine and pindolol group); for patients who
 M.B. Tome et al. / Journal of Affective Disorders 44 (1997) 101 – 109 103

Table 1
Clinical characteristics of patients by treatment group
Paroxetine 1 placebo (n 5 27) Paroxetine 1 pindolol (n 5 21)
Male / female 11 / 16 8 / 13
Age (mean6S.D.) 41.00610.52 34.3869.62
MADRS day 0 (mean6S.D.) 33.0365.53 30.7165.35
MADRS day 42 (mean6S.D.) 15.22611.79 9.80610.68
MADRS week 24 (mean6S.D.) 19.26615.65 10.8168.62
Past psychiatric history 15 9
Past treatment with antidepressants 11 5
50% response at day 42 15 15
Work 8 11
Compliant 16 10
Reward dependence
Higher than median score 9 7
Lower than median score 18 14
Novelty seeking
Higher than median score 10 10
Lower than median score 17 11
Harm avoidance
Higher than median score 14 8
Lower than median score 13 13
Value represents number of patients in each group unless otherwise specified.

Table 2
Clinical characteristics of patients by compliance with further medication for 6 months
Non-compliant (n 5 22) Compliant (n 5 26)
Male / female 6 / 16 13 / 13
Age (mean6S.D.) 36.87611.67 39.52610.25
MADRS day 0 (mean6S.D.) 32.4067.37 31.7063.90
MADRS day 42 (mean6S.D.) 14.80613.31 10.8169.67
MADRS week 24 (mean6S.D.) 21.05611.26 9.3367.41
Past psychiatric history 11 13
Past treatment with antidepressants 5 11
50% response at day 42 9 8
Work 4 15
Pindolol 11 10
Reward Dependence
Higher than median score 6 10
Lower than median score 16 16
Novelty seeking
Higher than median score 9 11
Lower than median score 13 15
Harm avoidance
Higher than median score 9 13
Lower than median score 13 13
Value represents number of patients in each group unless otherwise specified.
104 M.B. Tome et al. / Journal of Affective Disorders 44 (1997) 101 – 109

received paroxetine and placebo (the paroxetine and
placebo group); for patients who complied with
medication for the whole 6 months of the study (the
‘compliant’ group); and for those patients who took
medication only for the first 6 weeks of the study,
but elected not to take medication beyond that (the
‘non-compliant’ group).
Fifty-four consecutive subjects were invited to
complete the TCI-125 during the first week of the
trial and the results were correlated with subsequent
clinical responses. Patients completed the ques-
tionnaires at home. Forty eight questionnaires (89%)
were properly completed and entered into the analy-
sis.
Our analysis of the TCI scores followed the
methods of Joyce et al. (1994), namely Pearson
correlation analysis between questionnaire scores and
mood scales and regression analysis to predict the
treatment outcome. Pearson correlation analysis be-
tween MADRS scores at baseline (day 0), day 42,
and week 24 has been used to predict the best
response to medication, reflected by lower MADRS
scores.
Multiple regression analyses of TCI scores were
used as predictors of the MADRS score at 42 days (6
weeks) and 24 weeks (6 months) from start of
medication.
The software package SPSS for Windows (v.6)
was used for all calculations.
3. Results
3.1. Total patient population (n 5 48)
The results show overall improvement in depres-
sive symptoms, as expected after treatment with
SSRIs. The effect of pindolol is also observed at 6
months (Table 3).
Means and standard deviations (S.D.) of the TCI-
125 scores are shown in Table 4, as is the Pearson
correlation analysis of relationships between TCI-

Table 3
MADRS scores all patients
MADRS Placebo 1 paroxetine (n 5 27) Pindolol 1 paroxetine (n 5 21) P
Day 0 33.0365.53 30.7165.35 0.1502
Day 4 28.1466.9 19.2168.18 0.0002
Day 7 26.0468.90 16.10610.72 0.0019
Day 10 23.0069.85 13.22610.40 0.0035
Day 14 21.4269.46 13.10611.70 0.0106
Day 21 21.13611.09 12.05610.17 0.0092
Day 28 15.91610.14 12.2469.57 0.2608
Day 42 15.22611.79 9.80610.68 0.1073

24 weeks 19.26615.65 10.8068.62 0.0317

Table 4
TCI-125 scores and correlation coefficients (r) between MADRS scores and dimensions of TCI
Dimension mean S.D. r (MADRS 0) r (% change MADRS 42) r (MADRS 24w)
Harm avoidance 15.18 4.83 0.10 2 0.30* 0.17
Novelty seeking 8.27 3.73 0.01 0.21 2 0.19
Reward dependence 9.31 2.77 2 0.05 0.30* 2 0.08
Persistence 1.90 1.58 2 0.26 0.22 2 0.03
Self directedness 10.71 5.73 2 0.05 0.32* 2 0.15
Co-operativeness 17.84 5.11 0.06 0.08 0.01
Self transcendence 5.43 3.65 2 0.09 0.13 0.09
* Significance P , 0.05; MADRS 0 5 MADRS score, day 0; % change MADRS 42 5 (MADRS 0–MADRS 42 / MADRS 0)*100; MADRS
24w 5 MADRS score at 24 weeks.
 M.B. Tome et al. / Journal of Affective Disorders 44 (1997) 101 – 109
125 scores, baseline MADRS scores and percentage
change in MADRS scores at day 42 and 6 months. It
will be seen that there was no correlation between
basal MADRS and TCI scores. At 6 weeks, though,
patients with high scores in the dimensions of
Reward Dependence and Self Directedness had a
lower MADRS score, while those with high Harm
Avoidance scores had high MADRS scores.
Multiple regression analysis with percentage
change in MADRS at day 42 as the dependent
variable against Reward Dependence and Harm
Avoidance scores is shown in Table 5. The sample
was divided evenly at the median value of the
Reward Dependence Score (median 5 10), into high
and low scores. High scores in Reward Dependence
were associated with a 60% fall in MADRS score.
Four patients with a high score did not show this
magnitude of improvement at day 42, compared with
12 patients who did. However, among those with low
scores in Reward Dependence, 18 did not experience
a 60% fall in MADRS at day 42, compared with 14
subjects who did (x 2 5 0.03). Harm Avoidance was
also divided into higher and lower scores (median 5
17). Lower scores in Harm Avoidance were associ-
ated with greater improvement in MADRS (18 of 26
displayed a fall in MADRS of more than 60%;
x 2 5 0.04).
We did the same in relation to the temperament
dimension of Novelty Seeking (median score 5 8).
Table 5
Regression analysis: MADRS against dimensions of TCI
B S.E. of B
a
All patients (% change in MADRS at day 42)
Reward dependence 3.94 1.57
Harm avoidance 2 2.23 0.90
Constant 58.13 19.74
Paroxetine 1 pindolol patients ( MADRS at day 42) b
Novelty seeking 2 1.36 0.64
Harm avoidance 0.93 0.39
Constant 9.58 6.92
Paroxetine 1 pindolol patients ( MADRS at week 24) c
Novelty seeking 2 1.38 0.50
Harm avoidance 0.69 0.30
Constant 14.03 5.37
a
n 5 49; R 2 5 0.20; df 5 2.46; P , 0.0057; F 5 5.78.
b
R 2 5 0.30; df 5 2.18; F 5 3.88; P , 0.0395.
b
R 2 5 0.35; df 5 2.18; F 5 4.93; P , 0.02.
105
High scores in Novelty Seeking were associated with
a 60% fall in MADRS score by day 42. Fourteen
patients with high scores showed this magnitude of
improvement, compared with 6 patients who did not.
Among those with low scores in Novelty Seeking, 11
subjects experienced a 60% fall in MADRS at day
42, compared with 17 who did not (x 2 5 0.05).
3.2. Paroxetine and pindolol patients (n 5 21)
TCI-125 scores for this group are shown in Table
6, which also shows correlation analyses used as
above to examine the relationship between TCI-125
scores and MADRS score at day 0 and percentage
change in MADRS at day 42 and at week 24. There
was a statistically significant difference at 24 weeks
between patients treated with pindolol during the 6
weeks of trial, regardless of subsequent compliance
between weeks 6 and 24. At day 0, those with high
scores in Self Transcendence are seen to have a
lower score in MADRS.
At 6 weeks, patients with a high score in Harm
Avoidance have higher MADRS scores, while pa-
tients with high scores in Novelty Seeking show
lower MADRS scores. Patients with high scores in
the character dimensions of Self Transcendence
show an improvement in mood as indicated by their
lower MADRS scores. At 6 months, high scores in
(temperament) Novelty Seeking, and in (character)
T Significance of T
2.508 0.01
2 2.48 0.01
2.94 0.01
2 2.01 0.05
2.38 0.02
1.38 0.18
2 2.75 0.01
2.27 0.03
2.61 0.01
106 M.B. Tome et al. / Journal of Affective Disorders 44 (1997) 101 – 109

Table 6
TCI-125 scores of patients and correlation coefficients (r) between MADRS scores and dimensions of TCI by treatment group.
Dimension Mean S.D. r (MADRS 0) r (%change MADRS 42) r (MADRS 24w)
Paroxetine and pindolol
Harm Avoidance 13.43 5.61 0.19 2 0.38 0.29
Novelty Seeking 9.05 3.40 2 0.32 0.23 2 0.42*
Reward Dependence 9.67 2.31 2 0.01 0.13 2 0.01
Persistence 2.10 1.61 2 0.03 0.32 2 0.26
Self Directedness 11.10 6.57 0.06 0.30 2 0.27
Co-operativeness 16.43 4.90 2 0.14 0.14 2 0.29
Self Transcendence 4.71 2.67 2 0.60* 0.59* 2 0.55*
Paroxetine and placebo
Harm Avoidance 16.48 3.80 2 0.08 2 0.09 2 0.01
Novelty Seeking 7.85 3.88 0.22 0.11 2 0.05
Reward Dependence 9.14 3.08 2 0.10 0.35 2 0.08
Persistence 1.63 1.47 2 0.34 0.20 0.13
Self Directedness 10.14 4.98 2 0.06 0.40* 2 0.09
Co-operativeness 18.81 5.17 0.16 0.17 0.02
Self Transcendence 6.11 4.20 0.03 2 0.01 0.33
*significance P , 0.05

Self Transcendence are associated with lower scores
in MADRS.
There is a trend (P , 0.10) that suggests that
Novelty Seeking scores can predict response at days
14 and 21 and week 24, when the differences
between treatments are higher. Higher scores in Self
Transcendence were correlated with MADRS scores
at basal visit (day 0), at 42 days and 6 months. A
stepwise multiple regression shows Self Transcend-
ence as a predictor of MADRS at 42 days.
(Constant 5 21.57, P , 0.001; Beta 5 2 2.49, P ,
0.002; R 2 5 0.38, F 5 12.05; sig. F , .002; df 5
1.19).
Table 5 also shows the regression results at day
42 and at week 24. High scores in Novelty Seeking
and low scores in Harm Avoidance predict reduction
in MADRS at day 42. This prediction is replicated at
24 weeks although half the pindolol patients were
not compliant with the medication offered at the
conclusion of the study period (day 42).
3.3. Patients receiving placebo and paroxetine.
(n 5 27)
TCI-125 scores and correlation analyses were used
as above to examine the relationship between TCI-
125 scores and MADRS at day 0 and percentage
change in MADRS at day 42 (Table 6). At day 0,
there are no significant correlations. At 6 weeks,
patients with high scores in Self Directedness
showed greater clinical improvement; but at 6
months, there were no correlations.
A simultaneous multiple regression analysis was
performed using the terms with the Self Directedness
score at 6 weeks (Constant 5 27.12, p , 0.05;
Beta 5 2.69, p , 0.03; R 2 5 0.16; df 5 1.25; F 5
4.89; sig. F , 0.03).
3.4. Compliant with medication: weeks 6 to 24.
(n 5 26)
Twenty six patients continued with paroxetine (20
mg or 40 mg per day: mean 5 26.34 mg per day;
sd 5 9.65) for up to 6 months after they finished the
double blind period of the trial. Ten of these patients
had received paroxetine and pindolol during the
initial six week trial phase, and 16 had received
paroxetine and placebo.
Pearson Correlation Analysis was performed to
examine relationships between TCI-125 scores and
MADRS scores at basal (day 0), day 42 and week 24
(Table 7). At day 0, subjects who scored highly in
the dimension of Harm Avoidance had in general
higher MADRS scores (i.e., they were more de-
pressed). Subjects with high scores in Persistence
were in general less depressed, with lower MADRS
scores. At 6 weeks, high Harm Avoidance scores
were associated with poorer outcome, with higher
 M.B. Tome et al. / Journal of Affective Disorders 44 (1997) 101 – 109 107

Table 7
TCI-125 scores of patients and correlation coefficients (r) between MADRS scores and dimensions of TCI by compliance with further
medication for up to 6 months.
Dimension Mean S.D. r (MADRS 0) r (%change MADRS 42) r (MADRS 24w)
Compliant with medication, weeks 6 to 24 (n 5 26)
Harm Avoidance 15.00 4.84 0.45* 2 0.42* 0.11
Novelty Seeking 7.81 4.10 2 0.01 0.33 2 0.34*
Reward Dependence 9.54 2.97 2 0.18 0.06 2 0.12
Persistence 2.08 1.52 2 0.57* 0.33 0.10
Self Directedness 12.15 6.29 2 0.21 0.49* 2 0.01
Co-operativeness 18.00 4.37 2 0.06 0.19 2 0.08
Self Transcendence 4.85 3.73 2 0.33 0.14 0.16
Non compliant patients, weeks 6 to 24 (n 5 22)
Harm Avoidance 15.00 5.16 2 0.05 2 0.19 0.32
Novelty Seeking 9.25 3.19 2 0.06 0.16 2 0.17
Reward Dependence 9.35 2.56 2 0.08 0.67* 2 0.02
Persistence 1.60 1.57 2 0.01 0.13 2 0.11
Self Directedness 8.90 4.45 0.15 0.16 2 0.19
Co-operativeness 17.65 5.90 0.09 0.18 0.13
Self Transcendence 6.30 3.53 2 0.04 0.30 0.06
*Significance P , 0.05.

MADRS scores. On the other hand, high scores in
Self Directedness had lower MADRS scores, having
a better treatment outcome at this stage. At 6 months,
subjects with high Novelty Seeking scores sustain
this outcome, although the character dimensions of
Self Directedness no longer predict the MADRS at
this point. Self Directedness and Harm Avoidance
scores predicted reduction in MADRS at 42 weeks.
Novelty Seeking scores predicted decrease in
MADRS at 16 weeks (Constant 5 24.77, P , 0.01;
Beta 5 2 1.69, P , 0.01; R 2 5 0.25; df 5 1.23; F 5
7.62; P , 0.01) and a trend at week 24 of treatment
(Constant 5 20.44, P , 0.01; Beta 5 2 1.17, P ,
0.09; R 2 5 0.12; df 5 1.24; F 5 3.15; P , 0.09).
3.5. Non compliant patients, weeks 6 to 24 (n 5
22)
Twenty-two patients from the double blind ran-
domised trial elected not to continue with further
treatment after 6 weeks. Half of these patients had
been pre-treated with paroxetine and pindolol and the
other half with paroxetine and placebo. TCI 125
scores and correlation results are shown in Table 7.
There were no correlations at day 0, but at 6 weeks,
those who had scored highly in Reward Dependence
had a lower MADRS score. It is notable that in this
group, scores in Reward Dependence predict
MADRS scores at six weeks (Constant 5 16.82, P ,
0.01; Beta 5 3.42, P , 0.01; R 2 5 0.43; df 5 1.20;
F 5 13.79; P , 0.01), but not at 6 months.
4. Discussion
Comparison of the mean TCI-125 scores with
previous studies shows that, in the present UK study,
Harm Avoidance scores are higher and Reward
Dependence, Novelty Seeking, Persistence, Self Di-
rectedness, Co-operativeness and Self Transcendence
scores are lower than those previously reported in
inpatients (USA: Cloninger et al., 1994).
Earlier studies of the paroxetine pindolol combina-
tion (reviewed by Artigas et al., 1996), postulated
that any effect of pindolol was almost exclusively
due to its effect on serotonergic neurotransmission,
with little if any effect on either the noradrenergic or
dopaminergic systems. According to Cloninger’s
biosocial model of personality (Cloninger, 1987),
scores in the temperament dimension of Harm
Avoidance correlate positively with mesolimbic
serotonergic activity. The suggestion is that frequent
releases of serotonin at high levels from presynaptic
neuron down-regulation of postsynaptic serotonergic
receptors. Scores in Harm Avoidance positively
correlate with increased prolactin secretion in re-
108 M.B. Tome et al. / Journal of Affective Disorders 44 (1997) 101 – 109
sponse to serotonin challenge, and have been re-
ported in previous studies of depression. The correla-
tion between high Harm Avoidance scores and
higher MADRS scores in the present study is
consistent with this general picture. One may extend
this idea and suggest that high Harm Avoidance
scores may be associated with poorer response to
antidepressants. This seems to apply to the group in
our study who received pindolol and paroxetine, who
also demonstrate the converse, namely that subjects
who have low scores in Harm Avoidance have a
better clinical outcome at six months.
High scores in the temperament dimension of
Novelty Seeking correlate with better outcome at 6
weeks and 6 months in patients who had received
paroxetine and pindolol in the initial study, but not in
patients who had received paroxetine and placebo.
Novelty Seeking is said to modulate the response to
novel stimuli, and high Novelty Seeking scores have
been related to a lower threshold of sedation (Wing-
erson et al., 1993), and a reduced level of dopamine
secretion and activity, especially in the corpus
striatum, reflected in greater sensitivity of postsynap-
tic dopamine receptors and higher prolactin levels
(Cloninger et al., 1994). Such ideas fit with our data,
but we did not duplicate the observation that com-
pliance with medication is poorer in subjects with
high scores in Novelty Seeking (Wingerson et al.,
1993).
In terms of the other dimensions of personality,
we found that, for the undivided study population,
high Reward Dependence scores correlate with better
outcome at 6 weeks. Self Directedness seems to be
important in the response to medication, possibly
indicating the behaviour of a more mature personal-
ity.
While our population size may be too small to
allow substantive conclusions to be drawn, we have
replicated the earlier results of Joyce et al. (1994),
who looked at outcome in treatment with tricyclic
antidepressants. These authors, using a similar num-
ber of subjects to us, found that the personality
variables of Reward Dependence and Harm Avoid-
ance accounted for 45% of the variance in per-
centage change in Hamilton Depression Score in
their subjects. Our results indicate that high scores in
Reward Dependence and low scores in Harm Avoid-
ance predict outcome in treatment with SSRIs in the
subjects of this study taken as a whole.
However, a combination of a high score in Novel-
ty Seeking and a low score in Harm Avoidance was
a better predictor of response in the paroxetine and
pindolol treatment group at both 6 weeks and six
months, regardless of subsequent compliance with
antidepressant treatment after 6 weeks. Moreover, we
detected a trend that high scores in Novelty Seeking
can predict MADRS scores at 16 and 24 weeks in
patients who had received pindolol, and suggest that,
where pindolol is used, it better predicts MADRS
score at week 16 than at week 24, indicating an
attenuation of the pindolol effect with time. The
prominence of Novelty Seeking, and, by extension,
the possible role of the dopamine system, marks the
divergence of the present study from the results of
Joyce et al. (1994) and conflicts with the expecta-
tions of Artigas et al. (1996).
According to the latter authors, pindolol serves
only to block 5HT 1A somatodendritic autoreceptors,
so increasing the firing of serotonergic neurons and
improving serotonergic transmission when combined
with SSRIs such as paroxetine. However, this model
relies chiefly on data from rats, and there is compara-
tively little direct evidence in humans. A recent
immunohistochemical study in primates (Azmitia et
al., 1996) has indicated that 5HT 1A somatodentritic
autoreceptor receptors are widely distributed in not
only in the dorsal raphe´ nucleus, but also in the
(noradrenaline-rich) locus coeruleus and large mul-
tipolar neurons in the pontine reticular formation.
The somatodendritic location may influence norad-
renergic, cholinergic and GABA-ergic neurons, as
well as serotonergic. To this milieu we can perhaps
now add dopaminergic neurons.
The patients that best respond to the pindolol
paroxetine combination have high scores in Novelty
Seeking and low scores in Harm Avoidance, sug-
gesting both low basal dopamine levels and low
serotonin levels. The former has been well docu-
mented in melancholic patients (Willner, 1995) and
low serotonin levels have also been observed in
depressed subjects (Maes and Meltzer, 1995; Isaac
and Tome, 1997). Low serotonin levels may be
associated with up regulation of serotonin receptors
in the mesolimbic area, which can explain the
acceleration effect of pindolol in these patients.
Interestingly, pindolol has been shown in humans to
increase prolactin levels, supposedly under the in-
fluence of the serotonergic system (Meltzer and
 M.B. Tome et al. / Journal of Affective Disorders 44 (1997) 101 – 109
Maes, 1994). The present results in the pindolol
group may support the notion that the pindolol
paroxetine combination alters brain dopamine levels
via the serotonergic system.
Among the dimensions of character, high scores in
Self-Transcendence are associated with lower
MADRS scores from day 0, perhaps reflecting more
‘positive’ emotional states (Svrakic et al., 1996).
Similarly, high, ‘mature’, Self-Directedness scores
may be supplementary predictors of compliance with
medication.
Augmentation of SSRIs with pindolol represents a
novel treatment of depression, likely perhaps to have
its greatest impact in the primary care treatment of
this disorder. Such treatment may be more closely
tailored to patient need by considering the personali-
ty of the patient. The possibility of correlating
neurochemical findings with clinical outcome and
individual personality variables represents an excit-
ing departure in the quest for understanding the
biological substrate for depression.
References
Andrews, G., Neilson, M., Hunt, C., Stewart, G. and Kiloh, L.G.
(1990) Diagnosis, personality and the long term outcome of
depression. Br. J. Psychiatry 157, 13–18.
Artigas, F., Perez, V. and Alvarez, E. (1994) Pindolol induces a
rapid improvement of depressed patients treated with serotonin
reuptake inhibitors. Arch. Gen. Psychiatry, 51; 248–251.
Artigas, F., Romero, L., de Montigny, C. and Blier, P. (1996).
Acceleration of the effect of selected antidepressant drugs in
major depression by 5-HT 1A antagonists. Trends Neurosci. 19,
378–383.
Azmitia, E.C., Gannon, P.J., Kheck, N.M. and Whitaker-Azmitia,
P.M. (1996) Cellular localisation of the 5HT 1A receptor in
primate brain neurons and glial cells. Neuropsychopharmacolo-
gy 14, 35–46.
Blier, P. and Bergeron, R. (1995) Effectiveness of pindolol with
selected antidepressant drugs in the treatment of major depres-
sion. J. Clin. Psychopharmacol. 15; 217–222.
Cloninger, C.R. (1986) A unified biosocial theory of personality
and its role in the development on anxiety states. Psychiatry
Dev. 3, 167–226.
109
Cloninger, C.R. (1987) A systematic method for clinical descrip-
tion and classification of personality variants. Arch. Gen.
Psychiatry 44, 573–588.
Cloninger, C.R., Przybeck, T.R. and Svakic, D.M.(1994) Wetzel
Reward Dependence. TCI: The Temperament and Character
Inventory (TCI): A Guide to its Development and Use. Center
for Psychobiology of Personality. Washington University, St.
Louis, MO.
Duggan, C.F., Lee, A.S. and Murray, R.M. (1990) Does personali-
ty predict long term outcome in depression? Br. J. Psychiatry
157, 19–24.
Isaac, M.T. and Tome, M.B. (1997) Augmentation of antidepres-
sants with pindolol. Primary Psychiatry. In press.
Joyce, P.R. and Paykel, E.S. (1989) Predictors of drug response in
depression. Arch. Gen. Psychiatry 46, 89–99.
Joyce, P.R., Mulder, T.R. and Cloninger, C.R. (1994). Tempera-
ment predicts clomipramine and desipramine response in major
depression. J. Affect. Disord. 30, 35–46.
Kerr, T.A., Schapiro, K., Roth, M. and Garside, R.F. (1970) The
relationship between the Mausdley Personality Inventory and
the course of affective illness. Br. J. Psychiatry 116, 11–19.
Maes, M. and Meltzer, H.Y. (1995) The serotonin hypothesis of
major depression. In: Bloom, F.E. and Kupfer, D.J. (Eds.),
Psychopharmacology: the Fourth Generation of Progress.
Raven Press, New York, pp 933–944.
Meltzer, H.Y. and Maes, M. (1994) Effect of pindolol on the
L-5HTP-induced increase in plasma prolactin and cortisol
concentrations in man. Psychopharmacology 114, 635–643.
˚
Montgomery, S.A. and Asberg, M. (1979). A new depression
scale designed to be sensitive to change. Br. J. Psychiatry 134,
382–389.
Svrakic, N.M., Svrakic, D.M. and Cloninger, C.R. (1996) A
general quantitative theory of personality development: fun-
damentals of a self-organising psychobiological complex. Dev.
Psychopathol. 8, 247–272.
Tome, M.B., Isaac, M.T., Harte, R. and Holland, C. (1997)
Serotonergic autoreceptor blockade in the reduction of antide-
pressant latency: a controlled trial. Int. Clin. Psychopharmacol.
in press.
Weissman, M.M., Prusoff, B.A. and Klerman, G.L. (1978)
Personality and the prediction of long term outcome of
depression. Am. J. Psychiatry 135, 797–800.
Willner, P. (1995). Dopaminergic mechanisms in depression and
mania. In: Bloom, F.E. and Kupfer, D.J. (Eds.), Psycho-
pharmacology: The Fourth Generation of Progress. Raven
Press, New York, pp. 921–931.
Wingerson, D., Sullivan, M., Dager, S., Flick, S., Dunner. D. and
Roy-Birne, P., (1993). Personality traits and early discontinua-
tion from clinical trials in anxious patients. J. Clin. Psycho-
pharmacol., 13 (3), 194–197.