A neuroendocrine study of serotonin func

Journal of Affective Disorders 52 (1999) 121–133

Research report

A neuroendocrine study of serotonin function in depressed stroke
patients compared to non depressed stroke patients and healthy
controls
Rajamannar Ramasubbu a , *, Alastair Flint b , Gregory Brown c , George Awad c ,
Sidney Kennedy c
b

a
The Department of Psychiatry, University of Ottawa, Royal Ottawa Hospital, Ottawa, Canada
The Department of Psychiatry, University of Toronto, The Queen Elizabeth and The Toronto Hospital, Toronto, Canada
c
The Clarke Institute of Psychiatry, University of Toronto, Toronto, Canada

Received 5 February 1998; received in revised form 6 March 1998; accepted 6 March 1998

Abstract
Objectives: We employed a neuroendocrine challenge paradigm to study serotonergic abnormalities associated with

poststroke depression. Method: Twelve depressed stroke patients (major depression N 5 5, minor depression N 5 7), 8
nondepressed stroke patients and 12 healthy volunteers completed a single-blind, placebo-controlled, challenge tests.
Baseline cortisol (CORT) and prolactin (PRL) values, and these hormonal responses to 30 mg of oral d-FEN and placebo
over a 4 hour period were measured in the three groups. Results: There were intergroup differences for baseline adjusted
PRL responses (change scores from baseline) to d-FEN (group effect F 5 4.38, df 5 2,29, p 5 0.02) while these responses to
placebo were comparable between groups (group effect F 5 1.82, df 5 2,29, p 5 0.18). Peak PRL responses (post d-FEN
maximal PRL change from baseline scores) in depressed stroke patients were significantly greater than in nondepressed
patients ( p 5 0.005) but comparable to healthy normals ( p 5 0.47). However, these responses between major and minor
depression were not significant ( p 5 0.34). There was a trend suggesting a negative correlation between peak PRL response
and severity of depression ( p 5 0.056). Depressed patients were younger than the controls ( p 5 0.054). Also, the depressed
group was more functionally impaired ( p 5 0.04) and more likely to have right-sided lesions ( p 5 0.009) compared with the
nondepressed group. Differences in baseline adjusted PRL changes between depressed and nondepressed groups became non
significant when the influence of laterality of lesions was covaried, whereas covariation of functional scores and age did not
alter the significance. CORT responses did not show intergroup differences. Limitations: The study group was small and was
heterogenous in lesion characteristics, time since stroke and type of depression. A fixed-order design was used in the
challenge test paradigm. Conclusions: When laterality of stroke lesion was taken into account, depressed and nondepressed
stroke patients did not differ in PRL responses to d-FEN.  1999 Elsevier Science B.V. All rights reserved.
Keywords: Serotonin; Poststroke depression; d-Fenfluramine; Prolactin; Cortisol

*Corresponding author.

0165-0327 / 99 / $ – see front matter  1999 Elsevier Science B.V. All rights reserved.
PII: S0165-0327( 98 )00050-0

122

R. Ramasubbu et al. / Journal of Affective Disorders 52 (1999) 121 – 133

1. Introduction
Depression is a common mood disturbance in
stroke patients with a prevalence of 30–50% at the
initial evaluation (Robinson and Starkstein, 1990).
Despite growing recognition of its clinical importance, the nature and etiology of depression associated with stroke remains uncertain. There is a
controversy regarding the etiology of poststroke
depression (PSD). Some have described the depression as an understandable emotional response to
multiple deficits associated with stroke (Charatan
and Fisk, 1978). However, investigators who
attempted to relate severity of depression to functional, cognitive and social impairment have found
only weak positive correlations in some studies
(Ebrahim et al., 1987; Robinson et al., 1984a) and no
correlation in other studies (Feibel and Springer,

1982; Robinson and Szetela, 1981). Further, there
have been suggestions that depression may contribute to functional (Ramasubbu et al., in press) and
cognitive impairments (Robinson et al., 1986).
Moreover, trials involving psychosocial interventions
have failed to show any significant improvement in
poststroke depression compared to routine care
(Friedland and McColl, 1992).
The evidence in support of an organic hypothesis
comes from studies reporting a significant association between major depression and left frontal cortical or left basal ganglia lesions (Robinson et al.,
1984b; Astrom et al., 1993; Hermann et al., 1993).
Focal vascular brain lesions cause direct damage to
specific brain structures, altering not only local
neurochemical and physiological systems but also
those of distant brain regions (Andrews, 1991)
involved in neural organisations that subserve emotions, resulting in mood disturbances (Robinson and
Bloom, 1977). Hence, the pathogenesis of depression
after stroke is better understood in terms of disruption to neurotransmitter systems rather than by
traditional clinical pathological correlations. In addition, since clinical approaches are inadequate in
recognising the presence of depression in stroke
patients with severe communication and comprehension deficits, identification of biological markers

might improve diagnosis (Ramasubbu and Kennedy,
1994a,b). A growing body of experimental and
clinical evidence suggests that ischaemic brain le-

sions are associated with disturbances in the
serotonergic system (Ferrarise et al., 1986; Mrsulja
et al., 1976). Further, the demonstration of a negative
correlation between severity of depression and 5HT 2
receptor binding in the left temporal cortex in left
hemispheric stroke (Mayberg et al., 1988), low
levels of CSF 5H1AA, the principal metabolite of
5HT in PSD (Bryer et al., 1992) and proven efficacy
of serotonergic antidepressants such as citalopram
and fluoxetine in treatment of PSD (Anderson et al.,
1994; Stamenkovic et al., 1996) provide evidence
linking the serotonergic system and PSD. However,
specific 5HT receptor binding studies and CSF
studies may not be reliable indicators of net physiological responsiveness of 5HT systems.
Among several strategies that have been employed
to examine brain 5HT, neuroendocrine challenge

tests provide a reliable measure of physiological
responsiveness of the central 5HT systems. These
tests have been designed on the basis of abundant
evidence substantiating the stimulatory role of 5HT
in the release of PRL, ACTH, and CORT in animals
and humans (Van de Kar, 1991). Hormonal responsivity to fenfluramine (FEN) challenge has been
widely used as a measure of central 5HT function to
study serotonergic dysfunction in depressive disorders. FEN releases 5HT, inhibits 5HT reuptake and
stimulates post-synaptic receptors directly and indirectly (Costa et al., 1971; Fuex et al., 1975). This
provides a measure of net pre-synaptic and postsynaptic serotonergic activity. The racemic compound of fenfluramine d,l-FEN is a mixture of disomer (d-FEN) and l-isomer (l-FEN). The d-isomer
has been reported to be a more specific 5HT
releasing agent than the d,l-isomer since the l-isomer
has additional influences on the catecholomine system (Invernizzi et al., 1989). Hence, hormonal
responses induced by d-FEN might be assumed to be
mediated solely by stimulation of the 5HT system.
d-FEN induces a dose dependent increase in plasma
PRL levels which has been shown to be diminished
by pretreatment with the 5HT 2A / 2C receptor antagonist Ritanserin (Goodall et al., 1993) or the 5HT 1A
receptor antagonist pindolol (Palazidou et al., 1995),
suggesting that this response is mediated by a

serotonergic mechanism. The cortisolemic effect of
d-FEN in humans has also been reported to be
specifically mediated via a serotonergic mechanism

R. Ramasubbu et al. / Journal of Affective Disorders 52 (1999) 121 – 133

(Feeney et al., 1993). Furthermore, studies indicate
that d-FEN and its main metabolite d-nor-FEN are
more potent than racemic d,l-FEN and d,l-nor-FEN
(Garattini et al., 1986; Invernizzi et al., 1986). And
also 30 mg of oral d-FEN is well tolerated in healthy
subjects without any spontaneously reported adverse
effects (O’Keane and Dinan, 1991) as opposed to 60
mg of oral d,l-FEN. Thus d-FEN appears to be a
more specific, potent and safe serotonergic releasing
agent than d,l-FEN. Further, given the growing
evidence that the primary 5HT abnormality in depression may be presynaptic (Anand et al., 1994),
d-FEN which exerts a specific excitatory influence
primarily on presynaptic neurotransmission is a
preferred challenge agent to study serotonergic abnormalities of depression.

To our knowledge this is the first study to
investigate the central 5HT system in depressed
stroke patients using d-FEN in a neuroendocrine
challenge paradigm. The purpose of this study was to
test the hypothesis that there will be a significant
decrease in PRL and CORT responses to d-FEN
among depressed stroke patients compared to nondepressed stroke patients and healthy controls.

2. Methods

2.1. Subject selection
A group of twelve depressed patients (8 women, 4
men) with a mean age of 61.75615.37 years (S.D), 8
nondepressed stroke patients (4 women, 4 men) with
a mean age of 7367.55 years and 12 healthy
volunteers (9 women, 3 men) with a mean age of
71.0866.77 years participated in this study. Depressed and nondepressed stroke subjects were recruited from the Stroke Rehabilitation Unit of The
Queen Elizabeth Hospital, Toronto and also through
local distribution of posters and media advertisements.
For inclusion in the study, the subjects had to have

a confirmed diagnosis of stroke on the basis of a
clinical examination by a qualified neurologist and
by CT scan findings. Stroke included thrombotic or
embolic infarct or cerebral hemorrhage. Since alterations in cerebral blood flow to both hemispheres
occurs during the acute poststroke (Cordes et al.,

123

1989) period, stroke subjects were allowed to participate in the study only after one or more months had
elapsed poststroke. Stroke subjects who scored 16 or
higher on The Center for Epidemiological Studies
Depression (CES-D) Scale, a self-report instrument,
were considered to be depressed (Radloff, 1977).
Using a score of 16 or higher, CES-D has been
found to have a sensitivity of 0.86, specificity of
0.90, and a positive predictive validity of 0.80 in the
stroke population (Parikh et al., 1988). Further,
severity of depression was evaluated in depressed
stroke subjects using the Hamilton Rating Scale for
Depression (HRSD) (Hamilton, 1960). This instrument has been shown to be useful in assessing the

severity of depression in previous studies of stroke
subjects (Robinson et al., 1984a). The Hamilton
Rating Scale for Anxiety (HRSA) was used to assess
the presence of co-existing anxiety symptoms
(Hamilton, 1959). The Schedule for Affective Disorders and Schizophrenia, a diagnostic interview,
was used to qualify the symptoms of depression and
to derive affective disorder diagnosis according to
the Research Diagnostic Criteria (RDC, Spitzer and
Endicott, 1979). The Barthel Index was administered
to evaluate the functional abilities of stroke patients
(Mahoney and Barthel, 1965). CT scan findings were
obtained from chart analysis.
The exclusion criteria were as follows: (a) patients
who were unable to provide informed consent in
English and who had poor communications skills in
English; (b) subjects with severe cognitive deficits as
determined by the Mini Mental State Examination
(score , 15) (Folstein et al., 1975); (c) severe
impairment in comprehension and expressive language; (d) severe essential hypertension (diastolic
blood pressure $ 120 mm Hg) (Williams, 1994); (e)

uncontrolled diabetes mellitus (fasting blood sugar $
8 mmol); (f) patients with myocardial infarction
within the last 2 months; (g) hypothyroidism; (h)
neurological illnesses other than stroke; (i) history of
alcohol or substance abuse in the last six months; (j)
schizophrenia and other psychoses; (k) past history
of clinical depression and family history of depression; (l) subjects who were taking drugs that are
known to produce changes in the 5HT system or in
PRL concentration. These drugs included ketanserin
(5HT2 antagonist used in the treatment of hypertension), methyldopa, oral contraceptives, estrogen,

124

R. Ramasubbu et al. / Journal of Affective Disorders 52 (1999) 121 – 133

thyroxine, cimetidine, metaclopropamide, antidepressants and antipsychotics.
Healthy normals were recruited by media advertisement and word of mouth. Subjects were free of
present or past psychiatric illnesses as determined by
a clinical psychiatric interview. These healthy normals were further screened for current severe medical illnesses, substance abuse and also for a family
history of mental disorders. Subjects were excluded

if they had medical disorders such as diabetes
mellitus requiring insulin or hypoglycemic agents,
moderate and severe hypertension (diastolic blood
pressure $ 105 mm Hg), (Williams, 1994) coronary
heart disease, hypothyroidism, serious liver diseases,
cancer, stroke, epilepsy, or any other severe medical
or neurological conditions. Subjects receiving any of
the aforementioned medications that could alter
serotonin function or PRL secretion were also excluded. Physical examination and appropriate laboratory tests were performed to rule out any of the
above mentioned conditions. Healthy normals who
participated in the study were ambulatory and well
functioning with normal independent existence in the
community.

inserted into the anterior cubital vein and a physiological saline drip was commenced. Baseline blood
samples were drawn through a threeway stop cock in
the intravenous line 30 minutes after insertion, and
then 30 minutes after that (Time 0) before the oral
administration of 30 mg of d-FEN or placebo
capsule. Thereafter, blood samples were drawn at 60,
120, 180, and 240 minutes post-challenge for hormonal assays. Vital signs (B.P/ Pulse) and self rated
behavioral responses on a Visual Analogue Scale
were also recorded at these time points. The Visual
Analouge Scale (10 cm scale) was used to measure
changes in 4 parameters (drowsy, anxiety, sad, high).
The sampling period was limited to 4 hrs because the
peak RPL response seemed to occur at 4 hrs after
d-FEN administration (Quattrone et al., 1983) and
also to reduce the inconvenience of prolonged testing. A fixed dose of 30 mg of oral d-FEN was
chosen as this might be equivalent to the dose of
d-isomer in 60 mg of oral d,l-FEN (Silverstone et al.,
1987). An additional blood sample was taken at 3 hrs
after administration of d-FEN for assays of FEN and
nor-FEN levels. Maximal plasma levels of d-FEN
and d-nor-FEN metabolites were found at 2–4 hrs
post d-FEN (Campbell, 1991).

2.2. d-FEN challenge test
2.3. Assays
The challenge tests were conducted at the Clinical
Investigation Unit of The Toronto Hospital. Subjects
attended the test centre at 08:00 hours after an
overnight fast. On each test day, they were served a
low tryptophan breakfast (apple sauce, jello, orange
juice) to prevent hormonal changes secondary to
hypoglycemia which could occur as a consequence
of prolonged testing or d-FEN intake. They were not
allowed to consume caffeine or nicotine and they
remained in a resting state but awake during the test
session. All subjects participated in two test sessions
on two successive days. As oral d-FEN has long
carry over effects on neuroendocrine parameters it
was administered in the second session. Study subjects, the research nurse who administered clinical
scales, medication and drew blood samples, and the
laboratory technicians who performed hormonal
assays were blind to the sequence of placebo and
d-FEN presentation. An intravenous cannula was

Each blood sample was centrifuged and plasma
was stored at 2 258C before assays. Each sample
was assayed for PRL and CORT by quantitative
enzyme immuno assay using transferable solid phase
technology. The kits were supplied by Sychron
Enzyme Linked Immunosorbent Assay (SYNELISA–USA). The mean intra and inter assay coefficients of variation for plasma PRL were 5.0% and
7.8% respectively and sensitivity was 2 mg / ml. The
intra and inter assay co-efficients of variation for
plasma CORT were 5.2% and 7.1% respectively and
sensitivity was 0.5 mg / ml. Levels of d-FEN and
d-nor-FEN were determined by gas liquid chromatography using a nitrogen selective detector using the
method of Kerbs et al. (1984) with minor modifications. The intra assay and inter assay of coefficient of variation for d-FEN at 15 mg / ml was
2.7% and 2.7% respectively and for nor-d-FEN at 15

R. Ramasubbu et al. / Journal of Affective Disorders 52 (1999) 121 – 133

125

mg / ml was 3.7% and 5.4% respectively. The lower
detection limit was 2 mg / ml.

significance of all tests was set at p 5 0.05 (two
tailed).

2.4. Data analysis

3. Results

Hormonal responses were calculated for 3 outcome measures: (a) Mean baseline (mean of 2 30, 0
time) adjusted net changes (change scores from
baseline), (b) peak hormonal response concentration
(baseline values subtracted from the maximum increase in post d-FEN), (c) area under the response
curve (AUC), calculated using the trapezoid rule, for
both PRL and CORT responses to d-FEN and
placebo from time 0 until time 240 minutes. Mean
baseline values for each of the hormones were
compared between placebo and d-FEN sessions by
using student t-tests and between depressed and
comparison groups using ANOVA. Baseline adjusted
net changes in placebo and drug conditions were
compared between groups using a threeway analysis
of variance for repeated measures (ANOVA). A
comparison between groups for peak hormonal responses was performed using ANOVA and student t
tests as appropriate. Tests of differences in characteristics between groups for continuous and categorical variables were performed with student t tests and
x 2 tests, respectively. AUC values were compared
using ANOVA and student t tests where significant.
All results are expressed as means and standard
deviations unless stated otherwise. Data were analysed using the computer program SPSS. Statistical

3.1. Sample characteristics
Among the twelve depressed stroke patients, 7 had
minor depression and 5 had major depression according to RDC criteria. Since these two groups did not
significantly differ in hormonal response to d-FEN or
in demographic statistics, they were combined into
one group for analyses. Their mean scores on CESD, HRSD and HRSA were 22.6, 23.1 and 14
respectively. As shown in Table 1, the depressed
group did not significantly differ from control groups
in sex ( x 2 5 1.33, df 5 2, p 5 0.51), body weight
(F 5 1.38, df 5 2,29, p 5 .27) or plasma concentration of d-FEN and d-nor-FEN (t 5 2.15, df 5 2,26,
p 5 .13). The d-FEN and d-nor-FEN levels of three
subjects were not available due to technical difficulties. Further, there were no differences in time since
stroke (t 5 2 .27, df 5 18, p 5 .79) or in Mini
Mental Status Examination Scores (t 5 2 1.26, df 5
18, p 5 .22) between the depressed and nondepressed stroke group. However, the depressed group was
younger than the control groups (F 5 3.24, df 5 2,29,
p 5 .054) and differed significantly from the nondepressed group in laterality of lesions and in functional
impairment. The majority of depressed patients with
stroke had right-sided lesions (81.8%) while non-

Table 1
Sample characteristics
Variables

Depressed a
N 5 12

Non-depressed
N58

Healthy normals
N 5 12

P-value

Sex (MF)
Age (years)
Weight (Kg)
Plasma concentration of d-FEN 1 d-nor-FEN (ng / ml)
MMSE
Barthel Scores
Laterality of lesions
Time since stroke

4/8
61.75615.37
68.20612.49
24.63613.32
26.2562.73
58.75622.17
right sided 81.8%
18.79617.08

4/4
73.0067.56
75.14616.19
19.0067.64
24.7562.38
80.63620.43
left sided 83.3%
22.14614.88

3/9
71.0866.78
68.0168.70
24.64610.85
2
2
2
2

P 5 0.51
P 5 0.054
P 5 0.27
P 5 0.13
P 5 0.22
P 5 0.04
P 5 0.009
P 5 0.79

a

Major depression (N 5 5) and Minor depression (N 5 7) are combined.

126

R. Ramasubbu et al. / Journal of Affective Disorders 52 (1999) 121 – 133

depressed stroke patients had left-sided lesions ( p 5
83.3%) ( x 2 5 6.80, df 5 1, p 5 .009). Depressed
patients expressed greater functional impairment than
nondepressed patients (t 5 2.23, df 5 18, p 5 .04).

3.2. Prolactin responses
The baseline PRL values in drug and placebo
conditions did not differ between groups. A three
way analysis of variance for repeated measures
(ANOVA) yielded a main effect of group (F 5 3.84,
df 5 2,29, p 5 .033), drug (F 5 10.99, df 5 1,29,
p 5 .002), time (F 5 8.47, df 5 3,87, p 5 .00),
group 3 time interaction (F 5 2.74, df 5 6,87, p 5
.017), and drug 3 time interaction (F 5 8.39, df 5
3,87, p 5 .00) on baseline adjusted PRL responses
for drug and placebo condition indicating intergroup
differences in PRL responses. In order to determine
that the observed group effect and group 3 time
interaction were not due to different PRL responses

to placebo, a two way ANOVA with repeated
measures was performed (Fig. 1). There were no
differences in baseline adjusted PRL responses to
placebo in these three groups (group effect F 5 1.82,
df 5 2,29, p 5 0.18: group 3 time F 5 1.06, df 5
6,87, p 5 .39) while baseline adjusted PRL changes
to d-FEN were significant between groups (group
effect F 5 4.38, df 5 2,29, p 5 .02: group 3 time
F 5 2.52, df 5 6,87, p 5 .03). Further, peak PRL
responses of depressed stroke patients were greater
than nondepressed patients (t 5 3.38, df 5 11.98, p 5
.005) while the differences in these responses between depressed stroke and healthy normals were not
statistically significant (t 5 2 .73, df 5 33, p 5 .47).
Similarly, AUC 60–240 (area under the curve response from time point 60–240 minutes) PRL response post d-FEN was significantly greater in the
depressed stroke group than the nondepressed stroke
(t 5 3.09, df 5 17.18, p 5 .007) and comparable with
healthy normals (t 5 1.75, df 5 18.45, p 5 0.1).

Fig. 1. Comparison of baseline adjusted prolactin responses in depresed stroke patients with non-depressed stroke and healthy normals.
Baseline adjusted prolactin changes to d-fenfluramine (d-fen) were significant between groups while these responses to placebo were not
significant (ANOVA).

R. Ramasubbu et al. / Journal of Affective Disorders 52 (1999) 121 – 133

127

When laterality of lesion was entered into an analysis
of co-variance, the differences in baseline adjusted
PRL responses between depressed and nondepressed
patients became non significant (group effect F 5
.82, df 5 1,14, p 5 .381; group 3 time interaction
F 5 2.14, df 5 3,45, p 5 .11). However, covariation
of Barthel scores did not alter the significance of
main effect of group (depressed Vs nondepressed)
(F 5 4.66, df 5 1,17 p 5 .045) and group 3 time
interaction (F 5 2.75, df 5 3,54, p 5 .05). When age,
was added as a co-variate, there was no change in
the significance of group effect (F 5 4.54, df 5 2,29,
p 5 .020) or group 3 time interaction (F 5 2.74, df 5
6,87, p 5 .017) in net PRL responses. The differences in peak PRL responses between patients with
major and minor depression did not reach statistical
significance (t 5 1.01, df 5 10, p 5 .34). However,
the level of significance of the differences in peak
PRL responses between stroke patients with depression and nondepressed patients (t 5 2.72, df 5 6.22,
p 5 .033) was greater than the level of significance
in PRL differences between patients with and nondepressed patients (t 5 2.44, df 5 4.45, p 5 .064).
And also there was a trend suggesting a negative
correlation between peak PRL response and severity
of depression (HRSD) (r 5 2 .56, p 5 .056). A
significant negative correlation was found between
peak PRL response and anxiety symptoms (HRSA)
(r 5 2 .57, p 5 .05). There was no correlation between peak PRL response and MMSE scores (r 5
2 .26, p 5 .41), or Barthel scores (r 5 2 .28,
p 5 .23).

(r 5 2 .33, p 5 .29), and Barthel scores (r 5 .07,
p 5 .81).

The main finding of this study is that PRL
responses to acute administration of 30 mg of oral
d-FEN in depressed stroke patients who had a
preponderance of right sided lesions were comparable with healthy normals while these responses
were attenuated in nondepressed stroke patients who
were predominantly affected by left sided lesions.
Our findings indicate that differences in lateralised
lesions could account for differential PRL responsivity between depressed and nondepressed stroke
patients. Furthermore, there was a trend indicating a
negative correlation between peak PRL responses
and severity of depression.

3.3. Cortisol responses

4.1. Methodological limitations

The baseline CORT values on the two challenge
days did not differ between and within groups.
Although there was a main effect of drug (F 5 18.82,
df 5 1,29, p 5 .00) for baseline adjusted CORT
changes, there was no group effect (F 5 .94, df 5
2,29, p 5 .40) or group 3 drug (F 5 1.06, df 5 2,29,
p 5 .36), or group 3 time (F 5 .58, df 5 6,87, p 5
.74), or group 3 drug 3 time interaction (F 5 .60,
df 5 6,87, p 5 .73) indicating an absence of intergroup differences in d-FEN induced CORT responses (Fig. 2). There was no significant correlation
between CORT responses and scores of HRSD, (r 5
2 .25, p 5 .43), HRSA (r 5 2 .47, p 5 .12), MMSE

Before discussing these findings further, several
methodological shortcomings concerning patient
selection and the neuroendocrine challenge paradigm
need to be addressed. Despite aggressive recruitment
efforts, it was extremely difficult to identify poststroke depressed subjects who met the rigorous entry
criteria. Similar difficulties in recruitment have been
reported in a previous challenge study involving
depressed stroke patients (Barry and Dinan, 1990a).
As a result of small sample size we were not able to
control for important variables such as laterality, size
and location of lesion and medication. Furthermore,
some severely depressed stroke patients with left-

3.4. Behavioral and cardiovascular responses
None of the subjects reported unpleasant side
effects or showed any change in self rated anxiety,
sad affect, mood elevation or alertness following
d-FEN. There were no significant differences in
diastolic blood pressure (group 3 time 3 drug interaction; f 5 1.15, df 5 8,100, p 5 .33) or heart rate
(group 3 time 3 drug interaction f 5 .77, df 5 8,100,
p 5 0.63). However, ANOVA yielded significant
differences in systolic blood pressure (group 3
time 3 drug; f 5 2.25, df 5 8,100, p 5 .03).

4. Discussion

128

R. Ramasubbu et al. / Journal of Affective Disorders 52 (1999) 121 – 133

Fig. 2. Comparison of baseline adjusted cortisol responses in depressed stroke patients with non-depressed stroke and healthy normals. There
was a main effect of drug for baseline adjusted cortisol responses whereas group effect or group 3 drug or group 3 time or group 3 drug 3
time were not significant (ANOVA).

sided lesions who fulfilled entry criteria were unable
to take part in the challenge tests due to agitation and
catastrophic reactions in response to communication
deficits. This results in a bias towards inclusion of
moderately depressed stroke patients with a preponderance of right-sided vascular lesions. Another
major drawback is that the stroke patients participating in this study were heterogenous in age and
gender distribution, time since stroke and type of
depression. With respect to the challenge paradigm,
in a fixed-order design, the differences in stress
response to the procedure between first and second
challenge tests were not controlled. Another limitation that needs mentioning is the omission of quantitative evaluation of cognitive functioning and activities of daily living in healthy normals. These
factors may have influenced some or all of our

results and hence caution should be exercised before
drawing any clear conclusions. Future research
should address these methodological difficulties
more adequately.

4.2. Prolactin responses between depressed and
non-depressed stroke patients
Given that PRL response to d-FEN is a relative
index of central 5HT function, the observed increased PRL response in depressed stroke subjects
compared to nondepressed seem to be inconsistent
with the findings of Bryer et al. (1992) who reported
a diminished serotonergic functioning in poststroke
depression as evidenced by a lower concentration of
5HIAA in the CSF of depressed stroke compared to
nondepressed stroke patients matched for age, gender

R. Ramasubbu et al. / Journal of Affective Disorders 52 (1999) 121 – 133

and hemispheric lesion location within 5 days following stroke. There are several possible explanations for the discrepancies in these findings. In our
study, the majority of depressed stroke patients had
right-sided lesions (9 of 12) and suffered from minor
depression (7 of 12) in contrast to those of the CSF
study who were predominantly affected by left sided
lesions and had major depression (3 of 4). Furthermore, 5HIAA concentration in CSF was measured
during the acute phase of infarction (5 days after
stroke) while stroke patients participated in this
challenge test at least one month following stroke.
Therefore, it is likely that laterality of lesions, time
since stroke and type of depression might influence
the nature of serotonergic abnormalities associated
with poststroke depression. There were no differences in time since stroke between depressed and
nondepressed groups and hence the observed differences in PRL responsivity between groups could not
be accounted for by this variable. Due to the small
sample size, we were unable to match hemispheric
lesions in order to examine whether the type of
depression may influence PRL responses. To further
explore the possibility that the relative increase in
PRL response in the depressed stroke group could be
a stress response to limited physical functioning, the
relationship between Barthel scores and PRL were
examined. There was no correlation between Barthel
scores and peak PRL response and covarying Barthel
scores did not alter the PRL responses between
groups.
Our results suggest that preponderance of rightsided lesions in the depressed group might have
accounted for the increased PRL responses compared
to nondepressed patients who were predominantly
affected by left-sided lesions. This is partly in
agreement with previous studies reporting right-left
asymmetry in serotonergic functioning. Arato et al.
(1987) reported a significantly higher number of
imipramine binding sites in right compared with left
frontal cortices of normal controls suggesting a
relative increase in net serotonergic transmission in
the right hemisphere. Mayberg et al. (1988) also
demonstrated that patients with right-sided lesions,
16 months poststroke, had greater cortical 5HT 2
receptor binding than a similar group of patients with
left-sided lesions. Similarly, suction lesions involving right frontal cortex produced bilateral increases

129

in 5HT 2 receptors compared to left-sided lesions in
rats (Mayberg et al., 1990). In contrast, 5HT uptake
into platelets did not differ between those patients
with left and right sided lesions (Barry et al., 1990b).
However, the interpretation of platelets studies is
complicated by lack of normative data concerning
the influence of potential physiological factors on
platelet 5HT uptake (Hrdina, 1994). In summary, the
findings of our study, and the above mentioned
studies, suggest that there may be a relative increase
in 5HT function in the right hemisphere compared to
the left hemisphere and left-sided lesions may cause
a greater depletion of 5HT than right-sided lesions.

4.3. Prolactin responses between depressed stroke
and healthy normal
Another important finding is that PRL responsivity
to d-FEN was unable to distinguish stroke depressed
patients from healthy controls. Even taking into
account the negative findings (Park et al., 1996;
Maes et al., 1991) most of the studies to date seem to
point to the presence of a diminished serotonergic
responsivity in functional major depression (Siever
et al., 1984; O’Keane and Dinan, 1991; Cleare et al.,
1996; Mann et al., 1995). Although we do not have
an exact answer to explain our negative findings,
there are several possibilities. First, it is possible that
a type II error occurred owing to small sample size.
Second, the preponderance of minor depression (7 of
12) among the depressed stroke patients might have
neutralised the intensity of reduced PRL responsivity
associated with major depression. In support of this
notion, our study and several other studies suggest
that the intensity of blunted PRL response may be
proportional to severity or type of depressive
syndrome. Lopez–ibor et al. (1989) reported that
PRL responses to 60 mg of oral d,l-FEN were most
blunted in melancholic depression while these responses were least blunted in dysthymia. Mitchell
and Smythe (1990) similarly reported that depressed
patients with only endogenous features were most
likely to have blunted PRL responses while there
were no overall differences in these responses between the depressed group and matched controls.
However, the association between melancholic or
endogenous depression and blunted PRL responses is
not supported by studies by O’Keane and Dinan

130

R. Ramasubbu et al. / Journal of Affective Disorders 52 (1999) 121 – 133

(1991); Park et al. (1996); Maes et al. (1991). Third,
since PRL responsivity is inversely correlated with
age (McBride et al., 1990) and there is an age-related
decline in post-synaptic 5HT receptors, (Marcusson
et al., 1984a,b) it could be argued that the low
therapeutic dose used in this study (normal range
30–60 mg) might have masked the subsensitivity of
5HT receptors in depressed stroke patients compared
to healthy normals. However, mean body weight and
plasma concentration of d-FEN and d-nor-FEN were
comparable between groups and the weight adjusted
dose used in this study (0.44 mg / kg) was higher
than the recommended therapeutic dose (0.2–0.3
mg / kg) (Guy–Grand et al., 1989), suggesting that
the findings concerning PRL responses could not be
attributed to pharmacokinetic aspects of d-FEN but
might be related to receptor mechanisms. Thus, the
absence of deductible differences between depressed
and healthy controls may be due to an age-related
decline in post-synaptic receptor responsivity as
reported in previous studies (Mann et al., 1995;
McBride et al., 1990). Fourth, the validity of blunted
PRL responses to d-FEN reported in previous studies
is questionable as these studies failed to examine
pharmacokinetic aspects of d-FEN and also omitted a
placebo-controlled condition (O’Keane and Dinan,
1991; Cleare et al., 1996; Lopez–ibor et al., 1989).
Fifth, the association of blunted PRL responses to
d-FEN and major depression may depend on personality variables such as aggression and impulsivity
(Coccaro et al., 1989). We did not measure personality in this study and so we were not able to examine
its impact on hormonal responses.

4.4. Cortisol responses
The observed negative results on CORT responses
are consistent with other studies reporting neither
blunted nor enhanced CORT responses to d,l-FEN or
d-FEN in patients with major depression (Lopez–
ibor et al., 1989; Mitchell and Smythe, 1990). Prior
to validating our results on CORT responses a
number of issues need examination. First, compared
to PRL responses, increases in plasma CORT after
d-FEN challenge have not been consistently replicated, raising the possibility that FEN induced
CORT responses may not be a sensitive index of
central 5HT function. Second, as ACTH responses

were not examined in this study it is not possible to
rule out the peripheral mechanisms in the mediation
of d-FEN induced CORT release. Third, hypercortisolism has been shown to be associated with stroke
lesions (Olsson et al., 1989). Hence, due to heightened HPA axis in our elderly stroke sample, subtle
differences in CORT responses to d-FEN between
depressed and controls might have gone undetected.

4.5. Conclusions
The influence of lateralised lesions on PRL responsivity and a probable negative correlation of
PRL responses with severity of depression seem to
fit nicely with the hypothetical model proposed by
Mayberg et al. (1988). According to this model,
depletion of 5HT following right-sided lesions might
lead to upregulation of 5HT receptors while there
might be a failure in upregulation following leftsided lesions. This might explain why major depression is commonly associated with left anterior lesions while right anterior lesions are frequently
associated with undue cheerfulness or emotional
indifference (Robinson et al., 1984a). In addition,
different mechanisms might be responsible for depressive illnesses associated with right-sided lesions and
left-sided lesions (Starkstein et al., 1989). Further a
recent PET study has shown an increase in metabolic
response in left prefrontal, and temporo-parietal
cortical areas and decrease in the right prefrontal
cortex to 60 mg of oral d,l-FEN challenge in healthy
subjects compared to patients with major depression.
This suggests that, unlike patients with major depression, inhibitory effects of the right hemisphere are
balanced by excitatory effects of the left hemisphere
in normals (Mann et al., 1996). Similarly, the
hemispheric regulation of 5HT function also might
be lateralised. Thus, the occurrence of emotional
indifference or pleasant mood states in right sided
lesions might be accounted for by an unbalanced
release of excitatory effects of the left hemisphere on
serotonergic function and the reverse phenomena
could explain the frequent association of depression
with left-sided lesions. In conclusion, the observed
effect of lateralised lesions on PRL responsivity to
5HT activation highlights the importance of stroke
lesion-induced biochemical changes in determining

R. Ramasubbu et al. / Journal of Affective Disorders 52 (1999) 121 – 133

the emergence of depressive disorders in stroke
patients.

Acknowledgements
Part of this study was presented at the International Congress in Neuropsychiatry at Seville 1996 and
the Biological Psychiatry Conference at San Diego,
1997. This research is supported by a grant from the
Canadian Psychiatric Research Foundation. Authors
thank Dr. Hajek and Dr. Ruderman and the staff of
the stroke unit at the Queen Elizabeth Hospital, staff
of the Clinical Investigation Unit, the Toronto Hospital, Toronto for their co-operation and assistance in
recruitment and performing challenge tests. We
would also like to thank Servier for supplying dFenfluramine and placebo capsules and Mr. Thomas
Cooper, Nathan Kline Institute of Psychiatric Research, New York State for measuring d-fenfluramine
and d-nor-fenfluramine metabolites. Dr. David
Streiner provided statistical consultation and Mr.
Paul Miceli assisted with statistical analysis. We also
acknowledge Mary Kinckle for technical assistance
and Sangeetha Ramasubbu for administrative support.

References
Anand, A., Charney, D.S., Delgado, P.L., Mcdongle, C.J., Heninger, G.R., Price, L.H., 1994. Neuroendocrine and behavioural
responses to intravenous m-chlorophenyl piperazine (mcpp) in
depressed patients and healthy subjects. Am. J. Psychiatry 151,
1626–1630.
Andrews, R.J., 1991. Transhemispheric diaschisis: a review and
comment. Stroke 22, 943–949.
Anderson, G., Vestergaard, K., Lauritzen, L., 1994. Effective
treatment of poststroke depression with the selective serotonin
reuptake inhibitor citalopram. Stroke 25, 1009–1104.
Arato, M., Tekes, K., Palkovits, M., Demeter, E., Falues, A.,
1987. Serotonergic split brain and suicide. Psychiatry Res. 21,
355–356.
Astrom, M., Adolfsoson, R., Asplund, K., 1993. Major depression
in stroke patients: a 3 year longitudinal study. Stroke 24,
976–982.
Barry, S., Dinan, T.G., 1990. Alpha-2 adrenergic receptor function
in post-stroke depression. Psychol. Med. 20, 305–356.
Barry, S., Phillips, O.M., Williams, D.C., Dinan, T.G., 1990.
Platelet 5-HT uptake in post stroke depression. Acta. Psychiatr.
Scand. 82, 88–89.

131

Bryer, J.B., Starkstein, S.E., Votypka, V., Parikh, R.M., Price, T.R.,
Robinson, R.G., 1992. Reduction of CSF monoamine metabolites in poststroke depression: a preliminary report. J. Neuropsychiatry Clin. Neurosci. 4, 440–442.
Campbell, D.B., 1991. Dexfenfluramine: an overview of its
mechanisms of action. Rev. Contemp. Pharmacother. 2, 93–
113.
Charatan, F.B., Fisk, A., 1978. The mental and emotional results
of stroke. New York State Journal of Medicine 78, 1403–1405.
Cleare, A.J., Murray, R.M., O’Keane, V., 1996. Reduced prolactin
and cortisol response to d-fenfluramine in depressed compared
to healthy matched control subjects. Neuropsychopharmacology 14, 349–354.
Coccaro, E.F., Siever, L.J., Klar, H.M., Maurer, G., Cochrane, K.,
Cooper, T.B., Mohs, R.C., Davis, K.L., 1989. Serotonergic
studies in patients with affective and personality disorders.
Arch. Gen. Psychiatry 46, 587–599.
Cordes, M., Henkes, H., Roll, D., Eichstadt, H., Christe, W.,
Langer, M., Felix, R., 1989. Subacute and chronic cerebral
infarctions: SPECT and gadolinium-DAPT enhanced MR
imaging. J. Comput. Assist. Tomogr. 13, 567–571.
Costa, E.A., Groppetti, A., Refuelta, A., 1971. Action of fenfluramine on monoamine stores of rat tissue. Br. J. Pharmacol.
41, 57–64.
Ebrahim, S., Barer, K.D., Nouri, F., 1987. Affective illness after
stroke. Br. J. Psychiatry 151, 52–56.
Feeney, S., Goodall, E., Silverstone, T., 1993. The effects of d-and
l-Fenfluramine (and their interactions with d-amphetamine) on
cortisol secretion. Int. Clin. Psychopharmacol. 8, 139–142.
Feibel, J.H., Springer, C.J., 1982. Depression and failure to
resume social activities after stroke. Arch. Phys. Med. Rehabil.
63, 276–278.
Ferrarise, C., Bassie, S., Frattola, L., Locatelli, P., Piolti, R.,
Trabucchi, M., 1986. Different patterns of CSF neurotransmitter metabolism in patients with left or right hemispheric
stroke. Acta. Neurol. Scand. 73, 581–585.
Folstein, M.F., Folstein, S.E., McHugh, P.R., 1975. Mini mental
state: a practical method for grading the cognitive state of
patients for the clinician. J. Psychiatr. Res. 12, 189–198.
Friedland, J.F., McColl, M., 1992. Social support intervention
after stroke: results of a randomized trial. Arch. Phys. Med.
Rehabil. 73, 573–581.
Fuex, K., Farnebo, L.O., Hamberger, B., Ogren, S.O., 1975. On
the in vivo and in vitro actions of fenfluramine and its
derivatives on central monoamine neurons, especially 5 hydroxytryptamine neurons and their relation to the anorectic
activity of fenfluramine. Postgrad. Med. J. 1, 35–40.
Garattini, S., Mennini, T., Bendotti, C., Invernizzi, R., 1986.
Neurochemical mechanism of action of drugs which modify
feeding via the serotonergic system. Appetite 7, 15–38.
Goodall, E.M., Cowen, P.J., Franklin, M., Silverstone, T., 1993.
Ritanserin attentuates anorectic, endocrine and thermic response to d-fenfluramine in human volunteers. Psychopharmacology 112, 461–466.
Guy–Grand, B.J.P., Crepaldi, G., Lefebvre, P., Apfelbaum, M.,
Gries, A., Turner, P., 1989. International trial of long-term
dexfenfluramine in obesity. Lancet ii, 1142–1145.

132

R. Ramasubbu et al. / Journal of Affective Disorders 52 (1999) 121 – 133

Hamilton, M.A., 1960. A rating scale for depression. J. Neurol.
Neurosurg. Psychiatry 23, 56.
Hamilton, M.A., 1959. The assessment of anxiety states by rating.
Br. J. Med. Psychol. 32, 50.
Hermann, M., Bartles, C., Wallesh, C.W., 1993. Depression in
acute and chronic aphasia. J. Neurol. Neurosurg. Psychiatry 56,
672–678.
Hrdina, P.D., 1994. Platelet serotonergic markers in psychiatric
disorders: use, abuse and limitations. J. Psychiatr. Neurosci. 19,
87–90.
Invernizzi, R., Berettera, C., Garattini, S., Samanin, R., 1986. D
and L-isomer of fenfluramine differ markedly in their interaction with brain serotonin and catecholomines in the rat. Euro.
J. Pharmacol. 120, 9–15.
Invernizzi, R., Bertorelli, R., Consolo, S., Garattini, S., Samanin,
R., 1989. The effects of the l-isomer of fenfluramine and
dopamine mechanism in rat brain. Eur. J. Pharmacol. 164,
241–248.
Kerbs, H.A., Cheng, L.K., Wright, G.S., 1984. Determination of
fenfluramine and nor-fenfluramine in plasma using a nitrogen
sensitive detector. J. Chromato. Biomed. Appl. 310, 412–413.
Lopez–ibor, J.J., Saiz–Ruiz, J., Iglesias, L.M., 1989. Neuroendocrine challanges in the diagnosis of depressive disorders. Br.
J. Psychiatry 154, 73–76.
Maes, M., D’Hondt, P., Suy, E., Minner, B., Vandavost, C., Raus,
J., 1991. HPA axis hormones and prolactin responses to dextro
fenfluramine in depressed patients and healthy controls. Prog.
Neuropsychopharmacol. Biol. Psychiat. 15, 781–790.
Mann, J.J., McBride, P.A., Malone, K.M., DeMeo, M., Keilp, J.,
1995. Blunted serotonergic responsivity in depressed inpatients. Neuropsychopharmacology 13, 53–64.
Mann, J.J., Malone, K.M., Dienl, D.J., Cooper, T.B., Mintum,
M.A., 1996. Demonstration in vivo of reduced serotonin
responsivity in brain of untreated patients. Am. J. Psychiatry
153, 174–182.
Mahoney, F.I., Barthel, D.W., 1965. Functional evaluation: the
Barthel index. Md. Med. J. 14, 61–65.
Marcusson, J., Oreland, L., Winbald, B., 1984. Effect of age on
human brain serotonin (SI) binding sites. J. Neurochem. 43,
1699–1705.
Marcusson, J., Morgan, D.G., Winbaild, B., Finch, C.E., 1984.
Serotonin-2 binding sites in human frontal cortex and hippocampus: selective loss of S-2A sites with age. Brain Res.
311, 51–56.
McBride, P.A., Tierney, H., Demco, M., Chen, J.S., Mann, J.,
1990. Effect of age and gender on CSN serotonergic responsivity in normal adults. Biol. Psychiatry 27, 1143–1155.
Mayberg, H.S., Robinson, R.G., Wong, D.F., Parikh, R., Bolduc,
P., Starkstein, S.E., Price, T., Dannals, R.F., Links, J.M.,
Wilson, A.A., Ravert, H.T., Wagner, H.N., 1988. PET imaging
of cortical: S2 serotonin receptors after stroke: laterlized
changes and relationship to depression. Am.J. Psychiatry. 145,
937–943.
Mayberg, H.S., Moran, T.H., Robinson, R.G., 1990. Remote
lateralised changes in cortical (3H) spiperone binding following focal frontal cortex lesions in the rat. Brain Res. 515,
127–131.

Mitchell, P., Smythe, G., 1990. Hormonal responses to fenfluramine in depressed and control subjects. J. Affect. Disord.
19, 43–51.
Mrsulja, B.B., Mrsulja, B.J., Spatz, M., Klatzo, I., 1976. Brain
serotonin alter experimental vascular occlusion. Neurology 26,
785–787.
O’Keane, V., Dinan, T.G., 1991. Prolactin and cortisol responses to
d-fenfluramine in major depression: evidence for diminished
responsivity of central serotonergic function. Am. J. Psychiatry
148, 1009–1015.
Olsson, T., Astrom, M., Ericksson, S., Forssell, A., 1989. Hypercortisolism revealed by the Dexamethasone Supression Test
with acute ischaemic stroke. Stroke 20, 1685–1690.
Palazidou, E., Stephenson, T., Butler, J., Coskeran, P., Chambers,
S., McGregor, A., 1995. Evidence for 5HT1A receptor involvement in the control of prolactin in man. Psychopharmacology 119, 311–314.
Park, S.B.G., Williamson, D.J., Cowen, P.J., 1996. 5-HT neuroendocrine function in major depression: prolactin and cortisol
response to D-fenfluramine. Psychol. Med. 26, 1191–1196.
Parikh, R.M., Eden, D.T., Robinson, R.G., 1988. The sensitivity
and specificity of the centre of the centre for epidemiological
studies depression scale in screening for poststroke depression.
Int. J. Psychiatry. Med. 18, 169–181.
Quattrone, A., Tedeschi, G., Aguglia, U., Scopacasa, F., Di Renzo,
G.F., Annunziato, L., 1983. Prolactin secretion in man: a useful
tool to evaluate the activity of drugs on central 5-hydroxytryptaminergic neurones. Studies with fenfluramine. Br. J. Clin.
Pharmac. 16, 471–475.
Radloff, L.S., 1977. The CES-D scale: a self report depression
scale for research in the general population. Applied Psychological Measurement 1, 385–401.
Ramasubbu, R., Robinson, R.G., Flint, A.J., Kosier, T., Price,
T.R., in press. Functional impairment associated with acute
post stroke depression: the stroke data bank study. J. Neuropsychiatry. Clin. Neurosci.
Ramasubbu, R., Kennedy, S.H., 1994. Factors complicating the
diagnosis of depression in cerebrovascular disease, Part 1 –
Phenomenological and nosological issues. Can. J. Psychiatry
39, 596–600.
Ramasubbu, R., Kennedy, S.H., 1994. Factors complicating the
diagnosis of depression in cerebrovascular disease, Part 11 –
Neurological deficits and various assessment methods. Can. J.
Psychiatry 39, 601–607.
Robinson, R.G., Bloom, F.E., 1977. Pharmacologic treatment
following experimental cerebral infarction: implications for
understanding psychological symptoms of human stroke. Biol.
Psychiatry 12, 669–680.
Robinson, R.G., Szetela, B., 1981. Mood change following left
hemispheric brain injury. Ann. Neurol. 9, 447–453.
Robinson, R.G., Starr, L., Lipsey, J.R., Rao, K., Price, T.R., 1984.
A two year longitudinal study of post-stroke mood disorders:
dynamic changes in associated variables over the first six
months of follow-up. Stroke 15, 510–517.
Robinson, R.G., Kubos, K.L., Starr, L.B., Rao, K., Price, T.R.,
1984. Mood disorders in stroke patients: Importance of location
in brain. Brain 107, 81–93.

R. Ramasubbu et al. / Journal of Affective Disorders 52 (1999) 121 – 133
Robinson, R.G., Bolla, Wilson, K., Kaplan, E., Lipsey, J.R., Price,
T.R., 1986. Depression influence intellectual impairment in
stroke patients. Br. J. Psychiatry. 148, 541–547.
Robinson, R.G., Starkstein, S.E., 1990. Current research in
affective disorders following stroke. J. Neuropsychiatry Clin.
Neurosci. 2, 1–14.
Siever, L.J., Murphy, D.L., Slater, S., de la Vega, E., L