Short-Term Augmentation of Fluoxetine With Clonazepam

in the Treatment of Depression: A Double-Blind Study

Ward T. Smith, M.D., Peter D. Londborg, M.D.,

Vincent Glaudin, Ph.D., and John R. Painter, Ph.D.

Objective: Because selective serotonin reuptake inhibitors (SSRIs) require 2–4 weeks
to reach efficacy, the authors determined whether clonazepam augmentation of fluoxe-
tine is superior to fluoxetine alone at the beginning of treatment for major depression.
Method: Eighty adult outpatients with major depression who were rated as “moderately ill”
or “markedly ill” on the Clinical Global Impression of Severity underwent 8 weeks of double-
blind, randomized treatment with fluoxetine, 20 mg/day for all patients initially and 40 mg/
day if needed after 6 weeks. One-half of these patients received clonazepam, 0.5 mg h.s.
adjusted to two tablets by day 10 if needed, and the remainder received placebo, likewise
adjusted. Clonazepam/placebo was gradually discontinued during days 21–33. Efficacy
was evaluated by means of the Hamilton Depression Rating Scale, the Clinical Global Im-
pression of Improvement, and a patient rating of global improvement. Results: The pa-
tients taking clonazepam improved significantly more during the first 3 weeks of treatment
according to ratings on the Hamilton scale (≥50% improvement) and the clinician- and pa-
tient-rated global improvement measures (“much” or “very much” improved). Analysis of
variance confirmed a significant effect of clonazepam for average Hamilton depression
scores. No serious adverse events were found in either treatment group. Taper effects ap-
peared modest and transitory. Conclusions: Clonazepam augmentation of fluoxetine was
superior to fluoxetine alone in the first 3 weeks of treatment. This strategy may reduce suf-
fering during early SSRI treatment, may partially suppress SSRI side effects, may increase
compliance, and could possibly reduce the risk of suicide.
(Am J Psychiatry 1998; 155:1339–1345)

T he early phase of treatment for major depression

poses a significant clinical challenge because antide-
pressive agents typically take 2–4 weeks to provide
substantial benefit, during which time the patient con-
tinues to suffer, may stop the treatment regimen, and
may become even more discouraged, with an attendant
risk of suicide. The problem of slow response may be
compounded by the necessity for as many as one-third
of patients (1) to change medications in search of one
that is both well tolerated and effective.
A number of controlled studies have combined ben-
zodiazepines with tricyclic antidepressive compounds,
demonstrating positive results (2–5). Several double-
Received Dec. 2, 1997; revision received April 15, 1998;
accepted May 21, 1998. From the Summit Research Network,
Portland, Ore., and Seattle. Address reprint requests to Dr. Smith,
Summit Research Network, Suite 201, 1849 N.W. Kearney, Port-
land, OR 97209; wsmith@summitnetwork.com (e-mail).
Summit Research Network developed the study protocol, man-
aged the study, and prepared the final text, supported by a grant
from Hoffman LaRoche, Inc.
blind investigations have also suggested that some ben-
zodiazepines alone may have an antidepressive effect
beyond initial anxiety reduction (3, 6–8). The practice
of augmenting antidepressive medication with benzo-
diazepines has especially gained favor recently since se-
lective serotonin reuptake inhibitors (SSRIs) (9) have
replaced tricyclic antidepressants as the first line of
treatment of major depression. Whereas the tricyclic
antidepressants often produce sedation early in treat-
ment, the SSRIs more commonly have side effects of
stimulation and sleep impairment (10–12). The high-
potency benzodiazepines (clonazepam, alprazolam,
lorazepam) are frequently used in clinical practice con-
currently with SSRIs in treating major depression, but
reports of enhanced efficacy tend to be anecdotal, and
a search including the electronic MEDLINE and Psyc-
Info databases for the last 10 years failed to locate any
controlled trials of such combined therapy in the pub-
lished literature.
The purpose of the present double-blind study was
to determine whether augmenting fluoxetine with

Am J Psychiatry 155:10, October 1998 1339

CLONAZEPAM AUGMENTATION OF FLUOXETINE
clonazepam is both safe and more efficacious than flu-
oxetine alone in the early phase of treatment for major
depression.
METHOD
Study Drugs
Fluoxetine was the first SSRI approved by the Food and Drug Ad-
ministration (FDA) and is the most prescribed medication for major
depression worldwide. Numerous studies (13) have established its
efficacy. Clonazepam is a high-potency, long-acting benzodiazepine
originally approved by the FDA as an anticonvulsant, and more re-
cently it was approved for the treatment of panic disorder (14–17).
Clonazepam’s long half-life of 20 to 80 hours (18) might render
this compound especially promising for augmentation therapy in
major depression, because interdose fluctuation in mood state
should be reduced and because it might diminish the likelihood or
severity of rebound worsening of symptoms following the discontin-
uation of augmentation (19, 20).
Subjects
Adult outpatients (aged 18 or older) from Seattle and Portland,
Ore., were recruited through solicitations in the public media; those
meeting the DSM-IV criteria for nonpsychotic major depressive dis-
order of at least 1 month’s duration underwent a screening examina-
tion. Those selected also met the study entry criteria of “moderate”
or “marked” symptom severity and a score of at least 18 on the 17-
item Hamilton Depression Rating Scale (21). Comprehensive psychi-
atric and medical examinations, including laboratory tests and ECG,
screened out patients with symptoms or histories that might cloud
the diagnosis of unipolar major depression. Patients with a history of
alcohol or drug abuse within the previous 12 months were excluded.
All patients were required to abstain from alcohol and mood-alter-
ing substances, including concomitant medications such as most an-
tihistamines, during the course of the study. Women of child-bearing
potential were accepted under the treatment protocol if they prac-
ticed contraception. After complete description of the study to the
subjects, written informed consent was obtained. All study proce-
dures were in accordance with the Helsinki Declaration of 1975
(22), as revised in 1983.
Research Design
A random permuted block scheme balanced assignment to study
groups in each site. Only senior Summit Research Network person-
nel knew the assignment code, which was inaccessible to the princi-
pal investigators or other study personnel. The code could be broken
in individual cases in emergencies. The patients were reexamined 3
to 7 days after the initial psychiatric and medical screening. Those
who continued to meet the entrance criteria were randomly assigned
to treatment with fluoxetine plus clonazepam or with fluoxetine plus
placebo. A low dose of clonazepam at bedtime was judged to be the
best regimen for outpatients with depression, in contrast to the
higher, divided doses prescribed for other disorders. Double-blind
treatment with clonazepam or placebo was started in the evening of
day 0. One 0.5-mg tablet of clonazepam at bedtime was prescribed
randomly to one-half of the patients, and one tablet of inert placebo,
identical in appearance to the clonazepam tablet, was given to the
other half of the study group. At day 1 all patients started open-label
treatment with the standard dose of fluoxetine, one 20-mg capsule
each morning (12, 23). At day 4 the research clinicians had the op-
tion of increasing clonazepam/placebo to two tablets at bedtime if
one tablet did not appear to be effective, provided the patients were
tolerating the combination of drugs. At day 7 and day 10 clinicians
could likewise increase the patient’s dose to two tablets of clon-
azepam/placebo, reduce it from two to one, or maintain the prior
dose. No adjustments were made after day 10 until gradual discon-
tinuation of clonazepam/placebo from day 21 through day 33, per
the following schedule: patients taking two tablets decreased to one
1340
from day 21 through day 27 and all patients decreased to one tablet
every other night from day 28 through day 33. At day 42, more than
a week after discontinuation of clonazepam/placebo, the research
clinician had the option of increasing the dose of fluoxetine to 40 mg
each day for the last 2 weeks of the study.
Measurement of Efficacy and Safety
The primary measurement of treatment efficacy was the change in
the total score on the Hamilton depression scale from baseline. Sec-
ondary estimates of efficacy were based on declaring each patient as
a “responder” or “nonresponder” to treatment, operationally de-
fined as follows: 1) reduction of the total score on the Hamilton de-
pression scale (at least 50% from the patient’s baseline score); 2) a
clinician’s rating of “much improved” or “very much improved” on
the Clinical Global Impression of Improvement (24); and 3) the pa-
tient’s report of “much improved” or “very much improved” as a
global impression of improvement on a 1–7 scale paralleling the de-
scriptors used by clinicians.
The safety of the treatment program was assessed by monitoring
vital signs, repeating laboratory tests and ECG at endpoint, and sys-
tematically recording all adverse events in text and by coding ac-
cording to a body-system classification. Plasma drug concentrations
were not determined in this study.
Statistical Analysis
A sample size of 40 subjects per treatment group (N=80) was es-
tablished to provide a power of approximately 0.80 to detect a dif-
ference in the Hamilton depression score of 3 points at day 14, with
α set at 0.05 and an estimated standard deviation of 6.0 (25). Group
(clonazepam versus placebo) and time were the factors in a repeated-
measures analysis of variance (ANOVA) used as the test for the pri-
mary efficacy variable, Hamilton depression score; the most recent
observation was carried forward for missing data (26). Fisher’s least
significant difference statistic (26) determined the significance of dif-
ferences between treatment groups in mean Hamilton depression
scores at each week. For the latter, the rate of experimentwise
type 1 error, the risk of overreporting positive results due to multiple
tests of significance, was maintained at or below 0.05 by using a
Bonferroni technique setting the standard at 0.005. Differences in
categorical frequency scores, such as treatment response rates by
group, were tested by using the Mantel-Haenszel chi-square statistic
(27, 28). All tests were produced by using SYSTAT for Windows,
version 5 (29).
RESULTS
Disposition of Patients
A total of 85 patients completed the psychiatric and
medical examinations with findings that met the
screening criteria; however, four of these did not con-
tinue to meet all of the entry criteria at day 0 (baseline)
and were dropped from the study. Eighty patients were
originally randomly assigned to treatment and started
taking the medication, but one patient dropped out at
day 4 and provided safety data but no efficacy data.
Another patient who met all of the entry criteria re-
placed this subject. Therefore, the overall study group
was 80 patients for testing efficacy and 81 patients for
safety assessment (intent to treat). Of the 80 patients
included in the efficacy evaluation, all 40 of the sub-
jects in the augmentation group completed the first 3
weeks and 37 subjects in the placebo group completed
this phase of treatment. Sixty-five patients continued
treatment for 8 weeks, for an 81% completion rate.
Am J Psychiatry 155:10, October 1998

FIGURE 1. Scores on the Hamilton Depression Rating Scale Over 8 Weeks for Patients Taking Clonazepam Plus Fluoxetine or Pla-
cebo Plus Fluoxetine
The 15 patients who terminated treatment prema-
turely were almost equally divided between clon-
azepam plus fluoxetine (N=8) and placebo plus fluox-
etine (N=7), and they showed similar patterns of
reasons for discontinuing, with the exception of a sta-
tistically nonsignificant trend for the placebo patients
to discontinue treatment because of adverse events
(five patients versus one). Of the 40 efficacy patients in
the clonazepam group, 33 completed the study (83%).
One dropped on day 49 because of adverse events, two
dropped because of insufficient clinical response, one
was lost to follow-up, one left town, one used a pro-
scribed drug for cough, and one missed study drug
dosing for more than 3 days. Thirty-two of the patients
treated with placebo completed treatment (80%): five
dropped out because of adverse events, two were lost
to follow-up, and one was dropped after taking pro-
hibited pain medications for an ankle injury.
Subject Characteristics at Baseline
The efficacy group ranged in age from 20 to 73 years
and averaged 41.5 years. More than 90% identified
themselves as Caucasians, and the study group con-
sisted of almost equal numbers of men and women
(52% female). The research subjects had an average of
14 years of education (SD=2). No significant differ-
ences were found on any of these demographic vari-
ables. At baseline, the two 40-subject treatment groups
had identical mean ratings of 4.4 (SD=0.5) (between
“moderately ill” and “markedly ill”) on the Clinical
Global Impression of Severity (24) and similar mean
scores on the Hamilton depression scale: 21.7 (SD=
2.6) for the clonazepam-plus-fluoxetine treatment
group and 22.6 (SD=3.1) for the placebo-plus-fluoxet-
ine group.
Am J Psychiatry 155:10, October 1998
SMITH, LONDBORG, GLAUDIN, ET AL.
Dosing Pattern
By day 10, 24 (60%) of the 40 patients receiving
clonazepam had stabilized at two tablets at bedtime
(1.0 mg) while 27 (68%) of those in the placebo group
had also increased to two tablets. The patients treated
with 0.5 mg and those treated with 1.0 mg of clon-
azepam were similar at baseline on measures of sever-
ity; further, the measures of improvement showed no
significant differences between these subgroups. Of the
35 patients receiving clonazepam who remained in
treatment through day 42, somewhat more than one-
half (57%, N=20) were increased to 40 mg/day of flu-
oxetine (the percentage was essentially the same for
both doses of clonazepam). Thirty-two patients
treated with fluoxetine alone were still in treatment at
day 42, and 59% of them (N=19) had the fluoxetine
dose raised to 40 mg.
Efficacy
Patients in both treatment groups improved over the
course of the study, as reflected by changes in mean
scores on the Hamilton depression scale (figure 1).
Two-group repeated-measures ANOVA revealed a sig-
nificant treatment-by-time interaction (F=2.55, df=9,
702, p=0.03, with the Greenhouse-Geisser correction).
Pairwise comparison of Hamilton depression means
scores by Fisher’s least significant difference first
showed that the augmentation group had improved
significantly more than the group receiving fluoxetine
plus placebo at day 7 (p=0.002). The clonazepam pa-
tients also had significantly lower Hamilton depression
scores at day 10 and again at day 21 (p<0.001). By day
28, 1 week after the clonazepam discontinuation
schedule was initiated, there was no significant differ-
ence between clonazepam augmentation and placebo.
1341
CLONAZEPAM AUGMENTATION OF FLUOXETINE

TABLE 1. Percent of Patients Taking Clonazepam Plus Fluoxetine (N=40) or Placebo Plus Fluoxetine (N=40) Who Were Considered
Responders According to Three Measuresa
Clinical Global Impression Patient Global Impression
Hamilton Depression Scale of Improvement (“much” or “very of Improvement (“much” or “very
(≥50% decrease in score)b much” improved)c much” improved)d
Clonazepam Placebo Clonazepam Placebo Clonazepam Placebo
Plus Plus Plus Plus Plus Plus
Fluoxetine Fluoxetine Fluoxetine Fluoxetine Fluoxetine Fluoxetine
Visit N % N % N % N % N % N %
Day 4 2 5 3 8 5 13 1 3 8 20 4 10
Day 7 12 30 5 13 14 35 5 13 15 38 9 23
Day 10 17 43 8 20 18 45 13 33 17 43 8 20
Day 14 19 48 10 25 22 55 13 33 21 53 12 30
Day 21 18 45 11 28 24 60 13 33 19 48 13 33
Day 28 23 58 15 38 26 65 18 45 18 45 16 40
Day 35 21 53 15 38 27 68 18 45 24 60 17 43
Day 42 15 38 16 40 24 60 20 50 21 53 21 53
Day 56 24 60 21 53 28 70 25 63 25 63 27 68
a Last observation carried forward.
b Significant difference between groups (Mantel-Haenszel chi-square test: χ2=14.48, df=1, p<0.001).
c Significant difference between groups (χ2=23.51, df=1, p<0.001).
d Significant difference between groups (χ2=9.40, df=1, p<0.003).

The graphic display reveals that the Hamilton depres-
sion scores for the clonazepam group increased slightly
during the taper (day 35) and the first week after dis-
continuation (day 42). Figure 1 also shows that the pa-
tients receiving placebo continued to improve, so that
the mean Hamilton depression scores were similar for
the two treatment groups by day 42, and they re-
mained so at day 56, when both groups reached the
point of greatest reduction in depression symptoms.
The patients were declared to be “responders” or
“nonresponders” to treatment by three different stan-
dard measures: the Hamilton depression scale, a clini-
cian’s overall judgment of improvement (Clinical Glo-
bal Impression of Improvement), and the patient’s own
estimate of at least “much improved.” The rates of re-
sponse from day 4 to day 56 study are shown in table
1. Significantly more of the patients treated with clon-
azepam augmentation had an improvement of at least
50% on the Hamilton depression scale, gained ratings
of “much” or “very much” improved on the Clinical
Global Impression of Improvement, and reported
themselves “much” or “very much” improved. The
overall difference in frequencies of positive response
over the nine data points was significant for each of
these measures according to the Mantel-Haenszel chi-
square test (χ2=10.35, df=1, p<0.0001). An exception
to the pattern of results was the decrease in the per-
centage of responders in the augmentation group on all
three measures at day 42, 9 days after clonazepam dis-
continuation, before reaching the highest rates 2 weeks
later (day 56). This result confirms the temporary re-
versal of improvement observed at day 42 in mean
scores on the Hamilton depression scale (figure 1).
Safety
Discontinuation because of adverse events. No seri-
ous adverse events were reported in the experimental
or control group. Of the 81 patients who took medi-
cine, six terminated treatment prematurely because of
adverse events—one who was treated with clonazepam
plus fluoxetine and five who were treated with placebo
plus fluoxetine (Fisher’s exact test, p>0.10). The single
early termination in the augmentation group, occur-
ring on day 49, was attributed to decreased appetite,
nervousness, nausea, poor concentration, sleepiness,
fatigue, decreased libido, and headache. The remaining
early terminators, all treated with fluoxetine only,
stopped treatment during the first 3 weeks of treat-
ment, as follows: one patient at day 4 with headache
and agitation; one patient at day 7 with lightheaded-
ness, nausea, and anxiety, which resolved with discon-
tinuation of placebo; and three patients at day 21, one
with stomach cramps, diarrhea, diaphoresis, increased
anxiety and anxiety attack, jitteriness, shaky voice,
restlessness, insomnia, and nightmares, one patient
with lightheadedness, headaches, diarrhea/loose stool,
and nausea, and one patient with drowsiness/sedation.
Frequency of adverse events. The patients receiving
clonazepam experienced less anxiety and sleep distur-
bance during the 8 weeks of treatment, although the
difference was not statistically significant. Not surpris-
ingly, these patients were more likely to experience se-
dation. The only difference that was statistically signif-
icant by Fisher’s exact test was dry mouth (table 2).
More patients in the placebo group (N=10) com-
plained of new or worsened side effects during the
clonazepam/placebo taper and immediately after dis-
continuation (day 22 to day 42) than did patients in
the clonazepam augmentation group (N=4). The ma-
jority of these patients in both groups were tapering
from two clonazepam/placebo tablets over 2 weeks
(seven of the 10 placebo patients, and three of the four
clonazepam patients). A total of nine adverse events
were reported by the four clonazepam patients com-
plaining of new or worsened side effects during taper:
1) headache, nightmares, insomnia, and diaphoresis
(day 30); 2) delayed ejaculation (day 32); 3) restless-

1342 Am J Psychiatry 155:10, October 1998

 SMITH, LONDBORG, GLAUDIN, ET AL.

TABLE 2. Percent of Patients Taking Clonazepam Plus Fluox- gesting significant levels of insecticide in vegetables
etine or Placebo Plus Fluoxetine Who Reported Adverse from a neighbor’s garden.
Events Experienced by at Least 10% of Either Group
Clonazepam Placebo
Plus Plus
Fluoxetine Fluoxetine
DISCUSSION
Fisher’s
(N=41) (N=40) Exact
Adverse Event N % N % Test (p) This investigation demonstrated that clonazepam
Drowsiness/ augmentation of fluoxetine was superior to fluoxetine
sedation 18 44 13 33 >0.20 alone in relieving the symptoms of moderate to
Nausea 7 17 10 25 >0.20 marked major depression during the first phase of
Gastrointestinal treatment, especially the first 3 weeks. The rapidity of
disturbance 7 17 8 20 >0.20
Headache 6 15 8 20 >0.20
response was most striking: patients receiving clon-
Nervousness or azepam achieved in 10 days the same improvement on
anxiety 6 15 13 33 0.07 the Hamilton depression scale that fluoxetine alone
Sleep distur- reached by day 56. Similarly, 43% of the patients re-
bance 3 7 8 20 0.12 ceiving clonazepam rated themselves as “much” or
Diarrhea 9 22 6 15 >0.20
Weight loss 5 12 5 13 >0.20 “very much” Improved by day 10, a proportion not
Diaphoresis 4 10 5 13 >0.20 matched by fluoxetine alone until 5 weeks into treat-
Delayed ejacu- ment. The findings were consistently significant over a
lation 3 20a 5 22b >0.20 variety of outcome measures based on clinician judg-
Decreased ap-
petite 5 12 3 8 >0.20
ments and reports from patients.
Dry mouth 9 22 2 5 0.05 These results are apparently the first to confirm, in a
a Three of 15 men. controlled trial, the efficacy of augmenting an SSRI
b Five of 23 men. with a benzodiazepine. This evidence of the enhanced
efficacy of augmentation therapy is compatible with
anecdotal reports from clinical practice (2, 10) and
ness (day 25), weight loss (day 28), and indigestion earlier studies of combining benzodiazepines effica-
(day 30); and 4) heartburn (day 23). The 10 patients ciously with tricyclic antidepressants (2–5). Our find-
tapering from placebo reported a total of 15 adverse ings tend to support the typical rationale for aug-
events: 1) increased dreaming (day 22), constipation mentation with benzodiazepines, that is, reduction of
(day 22), and tinnitus (day 30); 2) agitation (day 31); the anxiety and insomnia components of the illness
3) restlessness (day 22); 4) sedation (day 30); 5) panic and suppression of the stimulating side effects of the
attack (day 30) and delayed ejaculation (day 33); 6) SSRI (9, 11, 12). The findings do not specifically sup-
CNS stimulation (day 29); 7) constipation (day 28); 8) port but are consistent with suggestions that high-po-
weight loss (day 29); 9) nightmares (day 22), heart- tency benzodiazepines, such as alprazolam (3, 6) or
burn (day 28), and decreased libido (day 32); and 10) adinazolam (7), may have a direct antidepressive effect
dry mouth (day 29). It should be recalled, however, (8).
that the clonazepam group’s efficacy scores deterio- Our study did not replicate the 60% rate of positive
rated temporarily after the taper was completed, espe- response to fluoxetine after 3 weeks that has been re-
cially at day 42, 9 days posttaper. After the day-42 rat- ported by some investigators (12), but our result is
ings were recorded, the majority in each group consistent with reports that one-third or more of pa-
increased the fluoxetine dose to two 20-mg capsules tients may be unresponsive to an adequate trial of
each day. Of the nine patients who complained of late- SSRIs (1). Further, almost 60% of both study groups
appearing adverse events, starting day 43 through day required an increase in fluoxetine dose at week 6,
49, eight were taking the increased dose of fluoxetine. whereas in an earlier study (31) only 37% of patients
Adverse events that were first recorded or that wors- required a dose increase after 4 weeks. The difference
ened after the increase in fluoxetine dose were as fol- in these results may in part be related to the timing of
lows: diaphoresis (N=3), dry mouth (N=2), decreased the dose adjustment, i.e., 4 versus 6 weeks of treat-
libido (N=2), sedation (N=2), anxiety (N=1), initial in- ment. The longer a patient has been treated without a
somnia (N=1), muscle twitching (N=1), and weight satisfactory response, the more likely a clinician is to
loss (N=1). increase the dose or change medications. In the present
There were no clinically significant changes in labo- group of depressed outpatients, fluoxetine appeared to
ratory test results or ECGs from screen to endpoint produce a slow, steady positive response that was still
with one exception. Patient 4, in the clonazepam increasing at the end of 8 weeks.
group, terminated treatment at day 21 because she There was suggestive evidence of a temporary in-
used an analgesic for a concomitant illness, and she crease in symptoms after clonazepam was discon-
had high levels of SGOT (154 U/liter), SGPT (167 U/li- tinued, possibly reflecting a mild rebound, loss of
ter), and GOT (142 U/liter). Her case was followed for suppression of some SSRI side effects, or both. Discon-
several weeks until her laboratory values returned to tinuation of augmentation confounds at least three po-
normal, after it was discovered that she had been in- tential factors: 1) elimination of directly additive and/

Am J Psychiatry 155:10, October 1998 1343

CLONAZEPAM AUGMENTATION OF FLUOXETINE
or synergistic treatment effects on the symptoms of de-
pression, 2) removal of possible suppression of fluox-
etine side effects, 3) a withdrawal syndrome or lesser
physiologic adjustment. This complex mixture of
symptom return and emergence of medication side ef-
fects requires cautious interpretation (for example, if a
patient experienced insomnia during the clonazepam
taper period, it would be difficult to determine to what
extent the change was due to adjustment to medication
withdrawal per se versus reduction in antidepressant
effect versus unmasking of fluoxetine side effects). A
specific point at issue is the temporary reversal of im-
provement in the augmentation patients at day 42, 9
days after discontinuation of clonazepam. This regres-
sion in response may reflect all of the factors just listed,
especially in the context of an increase in new adverse
events most characteristic of fluoxetine side effects. At
this point, it may be prudent to expect that some pa-
tients will temporarily display a reversal of treatment
response due to discontinuation of clonazepam aug-
mentation, even from a low dose.
Overall, clonazepam augmentation therapy was
shown to be a safe and effective treatment modality for
major depression in the low dose range (0.5 mg or 1.0
mg) prescribed in this research protocol. The results in-
cluded no serious adverse events, and the group receiv-
ing clonazepam plus fluoxetine had fewer premature
terminations for side effects than fluoxetine alone. The
long half-life of clonazepam (18) probably minimized
rebound anxiety (19, 20) without the use of a long
taper period, although there was some evidence of a
temporary reduction of benefit during taper. Only one
patient treated with clonazepam (0.5 mg) discontinued
treatment because of an adverse experience. In this
case, the adverse events appeared during the taper
schedule. The overall pattern of results suggested that
some other patients experienced a plateau in improve-
ment, a mild, transitory worsening of depression
symptoms, or an increase in SSRI side effects when
clonazepam was discontinued. However, it should be
kept in mind that augmentation was associated with
fewer adverse events and terminated cases because of
side effects, although the difference in rates was not
statistically significant.
Generalization of the clonazepam augmentation strat-
egy to other SSRIs might involve different dosing, based
on the pattern of inhibition the specific SSRI demon-
strates for the cytochrome isoenzyme CYP3A4, which is
central in the metabolism of clonazepam (32–34).
The results of this study indicate that augmentation
of fluoxetine with low-dose clonazepam during the
first 3 weeks of treatment is safe and more efficacious
than fluoxetine alone for moderate-to-marked depres-
sion. Augmentation therapy has the advantage of re-
ducing suffering during the period it takes an antide-
pressive drug to become effective. The addition of the
high-potency benzodiazepine seems to help suppress
the anxiety and insomnia often experienced in adjust-
ing to an SSRI and may be associated with a lower rate
of adverse events. Clonazepam augmentation has the
1344
potential value of increasing compliance, especially if
there is a need to change the SSRI in search of one that
is better tolerated, and faster relief may reduce the risk
of suicide at the start of treatment. The discontinua-
tion of augmentation appeared to be associated with
mild, transitory worsening of symptoms.
REFERENCES
1. Joffe RT, Levitt AJ, Sokolov ST: Augmentation strategies: fo-
cus on anxiolytics. J Clin Psychiatry 1996; 57(suppl 7):25–31;
discussion 57(suppl 7):32–33
2. Feighner JP, Brauzer B, Gelenberg AJ, Gomez E, Kiev A,
Kurland ML, Weiss BL: A placebo-controlled multicenter trial
of Limbitrol versus its components (amitriptyline and chlor-
diazepoxide) in the symptomatic treatment of depressive ill-
ness. Psychopharmacology (Berl) 1979; 61:217–225
3. Fawcett J, Edwards JH, Kravitz HM, Jeffriess H: Alprazolam:
an antidepressant? alprazolam, desipramine, and an alpra-
zolam-desipramine combination in the treatment of adult de-
pressed outpatients. J Clin Psychopharmacol 1987; 7:295–310
4. Calcedo-Ordonez A, Arosemene X, Otero-Perez FJ, Hernan-
dez-Herror C, Garcia A, Moral L, Baeza F, Fernandez Benitez
J, Velasco J, Garrido J: Clomipramine/bentazepam combina-
tion in the treatment of major depressive disorder. Hum Psy-
chopharmacol 1992; 7:115–122
5. Nolen WA, Haffmans JP, Bouvy PF, Duivenvooren HJ: Hyp-
notics as concurrent medication in depression: a placebo-
controlled, double-blind comparison of flunitrazepam and
lormetazepam in patients with major depression, treated with
a tricyclic antidepressant. J Affect Disord 1993; 28:179–188
6. Feighner JP, Aden GC, Fabre LF, Rickels K, Smith WT: Com-
parison of alprazolam, imipramine, and placebo in the treat-
ment of depression. JAMA 1983; 249:3057–3064
7. Smith WT, Glaudin V: Double-blind efficacy and safety study
comparing adinazolam mesylate and placebo in depression.
Acta Psychiatr Scand 1986; 74:238–245
8. Petty F, Trivedi MH, Fulton M, Rush AJ: Benzodiazepines as
antidepressants: does GABA play a role in depression? Biol
Psychiatry 1995; 38:578–591
9. Feighner JP, Boyer WF: Selective Serotonin Re-Uptake Inhib-
itors. Chichester, England, John Wiley & Sons, 1991
10. Nelson JC: Combined treatment strategies in psychiatry. J
Clin Psychiatry 1993; 55(Sept suppl):42–49
11. Pollack MH: Innovative uses of benzodiazepines in psychia-
try. Can J Psychiatry 1993; 38(Nov suppl):122–126
12. Arky R: Physicians’ Desk Reference. Montvale, NJ, Medical
Economics Data Production Co, 1994
13. Stark P, Hardison DC: A review of multicenter controlled stud-
ies of fluoxetine, imipramine and placebo in outpatients with
major depressive disorder. J Clin Psychiatry 1985; 46:53–58
14. Pollack MH, Rosenbaum JF, Tesar GE, Herman JB, Sachs
GS: Clonazepam in the treatment of panic disorder and ago-
raphobia. Psychopharmacol Bull 1987; 23:141–144
15. Tesar GE, Rosenbaum JF, Pollack MH, Otto MW, Sacks GS,
Herman JB, Cohen LS, Speir SA: Double-blind, placebo-con-
trolled comparison of clonazepam and alprazolam for the
treatment of panic disorder. J Clin Psychiatry 1991; 52:69–76
16. Pollack MH, Otto MW, Tesar GE, Cohen LS, Meltzer-Brody S;
Rosenbaum JF: Long-term outcome after acute treatment of
panic disorder with alprazolam or clonazepam. J Clin Psy-
chopharmacol 1993; 13:257–263
17. Beauclair L, Fountain R, Annable L, Holobow N, Chouinard G:
Clonazepam in the treatment of panic disorder: a double-
blind, placebo-controlled trial investigating the correlation be-
tween clonazepam concentrations in plasma and clinical re-
sponse. J Clin Psychopharmacol 1994; 27:111–118
18. Devane LC, Ware RM, Lydiard BR: Pharmacokinetics, phar-
macodynamics, and treatment issues of benzodiazepines: al-
prazolam, adinazolam, and clonazepam. Psychopharmacol
Bull 1991; 27:463–473
Am J Psychiatry 155:10, October 1998

19. Herman JB, Brotman AW, Rosenbaum JF: Rebound anxiety
in panic disorder patients treated with shorter-acting benzodi-
azepines. J Clin Psychiatry 1987; 48(Oct suppl):22–28
20. Pecknold JC: Discontinuation reactions to alprazolam in panic
disorder. J Psychiatr Res 1990; 27:155–170
21. Hamilton M: Development of a rating scale for primary de-
pressive illness. Br J Soc Clin Psychol 1967; 6:278–296
22. Declaration of Helsinki. Bulletin of the Pan American Health
Organization 1990; 24:606–609
23. Rickels K, Smith WT, Glaudin V, Weiss C, Settle GP: Compar-
ison of two dosage regimens of fluoxetine in major depres-
sion. J Clin Psychiatry 1985; 46:53–58
24. Guy W (ed): ECDEU Assessment Manual for Psychopharma-
cology: Publication ADM 76-338. Rockville, Md, US Depart-
ment of Health, Education, and Welfare, 1976, pp 217–222
25. Kraemer HC, Thiemann S: How Many Subjects? Newbury
Park, Calif, Sage Publications, 1987
26. Winer BJ: Statistical Principles in Experimental Design. New
York, McGraw-Hill, 1971
27. Mantel N, Haenszel W: Statistical aspects of the analysis of
data from retrospective studies of disease. J Natl Cancer Inst
1959; 22:719–748
Am J Psychiatry 155:10, October 1998
SMITH, LONDBORG, GLAUDIN, ET AL.
28. Pocock SJ: Clinical Trials: A Practical Approach. Chichester,
England, John Wiley & Sons, 1983
29. SYSTAT for Windows: Statistics, Version 5. Evanston, Ill,
SYSTAT, 1992
30. Smith WT, Glaudin V, Sheridan AQ: Long-term open-label
study of adinazolam in depressed outpatients (abstract). Con-
gress Internationale Psychopharmacologie Proceedings
(suppl) 1990; 1:84
31. Van Moffaert M, Bartholome F, Cosyns P, De Nayer, AR: A
controlled comparison of sertraline and fluoxetine in acute
and continuation treatment of major depression. Hum Psy-
chopharmacol 1995; 10:393–405
32. Shen WW: Cytochrome P450 monooxygenases and interac-
tions of psychotropic drugs: a five-year update. Int J Psychia-
try Med 1995; 25:277–290
33. Sproule BA, Naranjo CA, Brenmer KE, Hassan PC: Selective
serotonin reuptake inhibitors and CNS drug interactions: a
critical review of the evidence. Clin Pharmacokinet 1997; 33:
454–471
34. Nemeroff CB, DeVane CL, Pollock BG: Newer antidepres-
sants and the cytochrome P450 system. Am J Psychiatry
1996; 153:311–320
1345
 Effect of Pindolol on Onset of Action of Paroxetine
in the Treatment of Major Depression: Intermediate
Analysis of a Double-Blind, Placebo-Controlled Trial

Régis Bordet, M.D., Pierre Thomas, M.D., and Bernard Dupuis, M.D., Ph.D.,
on Behalf of the Réseau de Recherche et d’Expérimentation Psychopharmacologique

Objective: The purpose of this study was to investigate the effect of pindolol to accelerate
the onset of action of paroxetine in patients suffering from major depression. Method: Pa-
tients who met DSM-IV criteria for a nonpsychotic disorder, who had no previously treated
episode of major depression episode, and who had a score of at least 18 on the 17-item
Hamilton Depression Rating Scale were randomly assigned, for the first 21 days, to treat-
ment with paroxetine (20 mg/day) and either pindolol (5 mg t.i.d.) or placebo. Patients were
evaluated with the Hamilton depression scale, the Montgomery-Åsberg Depression Rating
Scale, and Global Clinical Impression (CGI) on days 0 (baseline), 5, 10, 15, 21, 25, 31, 60,
120, and 180. Results: Intermediate analysis of the first month’s results for the first 100 pa-
tients (pindolol, N=50; placebo, N=50) was performed. At day 10 there were more im-
proved patients (defined as patients with a maximum score of 10 on the Hamilton depres-
sion scale) in the pindolol plus paroxetine group (N=24; 48%) than in the placebo plus
paroxetine group (N=13; 26%). At day 5 there was no statistically significant difference,
and at day 15 and thereafter, the differences between the two groups disappeared. Hamil-
ton depression scale scores were significantly lower on days 5 and 10 for the pindolol plus
paroxetine group (mean=15.7, SD=5.3, and mean=11.7, SD=6.4, respectively) than for the
placebo plus paroxetine group (mean=19, SD=5.9, and mean=14.7, SD=6.8); this was also
true for Montgomery-Åsberg depression scale and CGI scores. Conclusions: The addi-
tion of pindolol to paroxetine treatment significantly accelerates the onset of therapeutic re-
sponse in patients suffering from major depression. Nevertheless, the mechanism (phar-
macodynamic or pharmacokinetic) of this beneficial effect remains unclear.
(Am J Psychiatry 1998; 155:1346–1351)

D espite advances in major depression treatment

(1), the problem of the delay of onset of therapeutic
benefits remains approximately the same for all avail-
Received March 12, 1997; revisions received Dec. 17, 1997, and
Jan. 22 and March 19, 1998; accepted April 23, 1998. From the
Réseau de Recherche et d’Expérimentation Psychopharma-
cologique (REPP) and Unité d’Evaluation, Centre Hospitalier et
Universitaire, Lille, France. Members of the REPP are P. Thomas
(general secretary), J.Y. Alexandre, R. Bordet, J. Catteau, C. Cyran,
T. Danel, J. Debieve, J.P. Dumon, P. Dumont, B. Dupuis, D.
Duthoit, D. Dutoit, C. Geraud, N. Lalaux, F. Lanvin, F. Lebert, J.
Louvrier, J.P. Meaux, P.J. Parquet, C. Plumecocq, B. Scottez, D.
Servant, C.E. Thomas, E. Trinh, G. Vaiva, J. Vignau, and A. Vin-
cent. Address reprint requests to Dr. Bordet, Service de Pharma-
cologie Hospitalière, Centre Hospitalier Regional Universitaire de
Lille, 59045 Lille, France; bordet@pop.univ-lille2.fr (e-mail).
Supported by grants from the Delegation à la Recherche du
Centre Hospitalier et Universitaire de Lille (France).
Pindolol was provided by Sandoz Pharmaceuticals.
As principal investigator, Dr. Dupuis assumes full responsibility
for the overall content of this article.
able antidepressants, i.e., 2 to 3 weeks (2). Most anti-
depressant drugs, in particular the selective serotonin
reuptake inhibitors (SSRIs), increase the concentration
of 5-hydroxytryptamine (5-HT; serotonin) present in
the synaptic cleft (3, 4). The increased levels of 5-HT
will act on postsynaptic receptors as well as presynap-
tic receptors including 5-HT1A somatodendritic recep-
tors, which exert negative feedback on the axon, re-
ducing the release of new 5-HT into the cleft, thereby
reducing the effect of SSRI antidepressant medication
(5, 6). There is experimental evidence, with use of both
microdialysis and electrophysiological techniques,
which indicates that during treatment with SSRI medi-
cation, synaptic concentrations of 5-HT do not in-
crease until after the 5-HT1A receptors are functionally
desensitized (7–10). The 2 weeks that are thought to be
necessary for desensitization to occur seem to coincide
with the onset of antidepressant effects (6, 11, 12).

1346 Am J Psychiatry 155:10, October 1998


Pindolol is a nonselective liposoluble β-adrenergic
antagonist that exerts weak sympathomimetic activi-
ties and acts also as a 5-HT1A receptor antagonist. It
therefore follows that pindolol should interfere with
the feedback effect and allow SSRI medication to in-
crease synaptic concentrations of 5-HT more rapidly
(6, 13). Two open therapeutic trials of limited size have
shown that the association of pindolol with SSRI in the
treatment of major depression produced a more rapid
therapeutic response than administration of the SSRI
alone in patients who were naive to SSRI treatment, as
well as in patients who had been previously judged to
be resistant to antidepressant SSRI treatment (14, 15).
Nevertheless, these preliminary results were not ini-
tially confirmed by a double-blind, placebo-controlled
trial conducted with 43 patients receiving fluoxetine
associated with pindolol (2.5 mg t.i.d.) or placebo
(16). Recently, two large controlled trials have con-
cluded that the addition of pindolol to antidepressant
treatment increases, in a prolonged manner, the effec-
tiveness of two SSRIs (17, 18). This effect of pindolol
was explained by a higher response rate in the pindolol
group rather than by a hastening response to SSRIs in
contrast with the experimental hypothesis supporting
the initial use of pindolol in depression treatment (19).
The goal of this study was to test the effect of pin-
dolol on the onset of action of paroxetine in patients
suffering from major depression in a multicenter ran-
domized, double-blind, placebo-controlled study. To
study its hastening effect, pindolol has been prescribed
only for a short period in the early phase of a depres-
sive episode when the effect of SSRIs is delayed. While
the study is still ongoing, analysis of the first 100 pa-
tients has been performed as planned in the protocol.
These partial results are presented in this report be-
cause they confirmed the promising effect of pindolol,
but they are in contrast with previous studies on the
nature of this effect (16, 18).
METHOD
Patients
The study group consisted of patients recruited by a group of 20
psychiatrists of the Réseau de Recherche et d’Expérimentation Psy-
chopharmacologique (northern part of France). The investigators
were psychiatrists in private, public, and institutional practice and
university hospital practice. The study was centralized by the Unité
d’Evaluation of the Lille University Hospital (Lille, France), which
was responsible for logistics, statistical analysis, and assessment of
quality of the study.
Potential participants were male and female inpatients and outpa-
tients between 18 and 65 years of age. We included patients who met
DSM-IV criteria for a current unipolar, major depressive episode,
nonpsychotic subtype, with a score of at least 18 on the 17-item
Hamilton Depression Rating Scale (20). Exclusion criteria for the
study were serious or unstable medical illness (in particular, cardio-
vascular disorders); a history of organic mental disorders or brain
disorders, bipolar disorders, any psychotic disorders, active sub-
stance use disorders (including alcohol) within the last 12 months;
history of adverse reactions to or intolerance of or nonresponse to
paroxetine or pindolol; history of use of a monoamine oxidase in-
Am J Psychiatry 155:10, October 1998
BORDET, THOMAS, AND DUPUIS
hibitor (MAOI) within the past 15 days or lithium within the previ-
ous 6 months; pregnancy; or lactation. We also excluded patients
currently taking any type of cardiac antiarrhythmic treatment, vera-
pamil, diltiazem, bepridil, or alprazolam. If the patient had been tak-
ing an anxiolytic drug other than alprazolam before entering the
study, he or she was allowed to continue at the same dose that was
taken before the patient received paroxetine; if a new prescription of
an anxiolytic was necessary, oxazepam (up to 100 mg/day) was pre-
scribed.
The study was approved by the Ethics Committee of the Lille Uni-
versity Hospital. Before inclusion in the study, all patients completed
a written informed consent form after the study protocol had been
fully explained, in accordance with French law.
Procedure
In a double-blind, placebo-controlled, parallel-design, 6-month
study, we compared pindolol to placebo, in association with parox-
etine, for the first 3 weeks of the 6-month treatment period of a ma-
jor depressive episode. All eligible patients received paroxetine in
one daily dose (20 mg/day) for the entire 6 months of the study pe-
riod. On the same day that they started paroxetine treatment, pa-
tients were randomly assigned to one of two parallel treatment
groups receiving placebo or pindolol. Randomization was per-
formed by an independent center, generated by tables of random
numbers, and stratified by participating investigators to allocate
treatment by blocks of four. Pindolol, as well as placebo, was given
three times a day for the first 21 days of treatment (15 mg/day) and
was then tapered to twice a day for 4 days (10 mg/day) and once a
day for 3 days (5 mg/day) and was then stopped. The total length of
pindolol treatment was 28 days.
Clinical Assessment
Before inclusion in the study, patients underwent standard clinical
assessment comprising a psychiatric evaluation, structured diagnos-
tic interview, and medical history to verify inclusion and exclusion
criteria. The structured diagnostic interview used was the Structured
Clinical Interview for DSM-IV (21). Sociodemographic data col-
lected included gender, age, inpatient or outpatient status, past his-
tory of depressive episode, response to previous antidepressant treat-
ment, current anxiety disorder, and use of anxiolytic treatment.
During the study, depressive symptoms were assessed at baseline
(day 0) with the 17-item Hamilton Depression Rating Scale (mini-
mum score=0; maximum score=52) and with the 10-item Montgom-
ery-Åsberg depression scale (minimum score=0; maximum score=
60) (22). Items on the Hamilton depression scale were grouped in
different subscales to analyze the nature of symptoms improved by
the addition of pindolol, as follows: depression (items 1, 2, 7, 8, 10,
and 13), suicide (item 3), and anxiety (items 9, 11, 15, and 16) (23).
Global improvement was assessed with the Clinical Global Impres-
sion (CGI) (24), a 7-point scale (1=very much improved; 4=no
change; 7=very much worse). The different scales were administered
by the blind evaluators on days 5, 10, 15, 21, 25, 31, 60, 120, and
180. The repetitive evaluations for the first 2 weeks were performed
to assess the specific effect of pindolol on the onset of action of par-
oxetine. Evaluations at days 21, 25, and 31 were performed to verify
the lack of withdrawal effect or maintenance of improvement; eval-
uations at days 60, 120, and 180 were performed to assess long-term
evolution of the depressive episode. Each evaluation was performed
at the foreseen date with a window of 2 days during the first month
and of 4 days at the subsequent dates. In addition, adverse experi-
ences were recorded at each visit. The Hamilton depression scale and
Montgomery-Åsberg depression scale interviews were all conducted
by the 20 psychiatrists of the Réseau de Recherche et d’Expérimen-
tation Psychopharmacologique, who were fully trained in the use of
these instruments through live and videotaped interviews. For one
patient, each evaluation was performed by the same investigator. Al-
though interrater reliability in the use of the two scales was not for-
mally assessed, participating clinicians established consensus
through repeated interrater reliability sessions.
1347
PINDOLOL AND ONSET OF ACTION OF PAROXETINE
FIGURE 1. Percent of Depressed Patients Receiving Paroxet-
ine Plus Either Pindolol or Placebo Who Improved From Day 5
to Day 31
a Significant
difference between the groups receiving paroxetine
plus pindolol and paroxetine plus placebo (intent-to-treat, χ2=4.3,
df=1, p=0.04).
Statistical Analysis
The comparability of the patients in the two groups (pindolol plus
paroxetine and placebo plus paroxetine) was tested by chi-square
analysis on qualitative data and by Student’s t test for continuous
measures. All analyses have been performed in intention-to-treat,
i.e., in all the patients randomly assigned to either treatment condi-
tion, including those who failed to complete the study. Intermediate
analysis of the first month’s results for the first 100 patients was
planned to screen the pindolol effect on the onset of action of parox-
etine. Experimental and clinical rationales for the use of pindolol
have suggested that the effect of pindolol is to hasten the SSRI effect.
Thus, the hypothesis-driven analysis was that the expected date of
acceleration would be during the first 2 weeks of treatment, which
explained the choice of frequent evaluations during this period. Be-
yond the third week of treatment, the interest of analysis was only to
test whether differences in improvement were maintained or disap-
peared, despite the lack of a hypothesis supporting such an effect.
The aim of analysis was to evaluate the difference between pindolol
and placebo in onset of action of paroxetine, rather than evaluate
the long-term effect of the two drug associations—pindolol plus par-
oxetine and placebo plus paroxetine—on depression because all pa-
tients received paroxetine, a treatment known to be effective in de-
pression, with a delay of 2 or 3 weeks. These hypotheses concerning
the effect of pindolol support the use of individual tests at each
point.
The main endpoint of the study was remission, and the depression
of the patients with a maximum score of 10 on the Hamilton depres-
sion scale was classified as remitted (25). Among the patients who
did not complete the study until day 31, we considered as remitted
the depression of those patients who dropped out the study with a
maximum score of 10 on the Hamilton depression scale, without ad-
verse effect. The other patients were classified as unsuccessful pa-
tients. This comparison was performed through use of a chi-square
analysis, with Yates’s correction, at each point until day 31.
To determine precisely the effect of pindolol during the first 2
weeks of paroxetine treatment, we compared the severity of depres-
sion (evaluated by scores on the Hamilton depression scale and its
subscales and on the Montgomery-Åsberg depression scale) and glo-
bal improvement (evaluated by CGI scores) (expressed as means and
standard deviations) in the pindolol plus paroxetine and placebo
plus paroxetine groups at each point in the first 2 weeks. All the
comparisons were performed at days 5, 10, and 15 by using Stu-
dent’s t test for continuous measures. All statistical tests were per-
formed (P. Devos, Unité d’Evaluation) with SAS software (Cary,
N.C.), and statistical significance was set at p≤0.05.
1348
RESULTS
Patient Characteristics
Between June 1, 1995, and May 31, 1996, 112 pa-
tients were recruited, and intermediate analysis was
performed on the first 100 patients for whom results
were available until day 31. These 100 patients receiv-
ing paroxetine (20 mg/day) were randomly assigned to
pindolol (N=50) or to placebo (N=50). Patients ran-
domly assigned, respectively, to pindolol or placebo
were comparable for several demographic and clinical
characteristics: age (mean=40 years, SD=11, versus
mean=44, SD=10), gender (33 women and 17 men ver-
sus 37 women and 13 men), past depressive episode
(55% versus 58%), past unsuccessful treatment of de-
pression (18% versus 18%), current anxiety disorder
(43% versus 43%), current anxiolytic treatment
(65% versus 62%), hospitalization for current de-
pressive episode (45% versus 44%), baseline Hamil-
ton depression scores (mean=24.2, SD=3.3, versus
mean=24.1, SD=3.5), and baseline Montgomery-Ås-
berg depression scores (mean=30.4, SD=4.5, versus
mean=30.6, SD=4.8).
Among the 100 patients, 20 (pindolol=10; placebo=
10) had dropped out of the study before day 31. Five
patients were withdrawn from the study because of
side effects (pindolol=two; placebo=three). Adverse ef-
fects were nausea (one patient), drowsiness (one pa-
tient), hypotension (one patient), and hepatitis (two
patients). Of the 15 remaining patients, eight dropped
out of the study because of lack of response to treat-
ment (pindolol=three; placebo=five), two because of
early relapse of depressive symptoms (one in each
group), and five because of improvement and patient
unwillingness to continue treatment (pindolol=four;
placebo=one). Among the five improved patients, the
depression of four was classified as remitted (pindolol=
three; placebo=one).
Depression Remission
Figure 1 shows the effect of pindolol treatment on
remission of depressive episode over time for the first
month of paroxetine treatment, for all patients ran-
domly assigned to pindolol (N=50) or placebo (N=50).
At day 5, there was no significant difference in the
percent of patients with remitted depression in the
pindolol plus paroxetine group (12%) and placebo
plus paroxetine group (6%). At day 10, the propor-
tion of patients with remitted depression (N=24 of
50; 48%) in the pindolol plus paroxetine group was
significantly higher than that (N=13 of 50; 26%) in
the placebo plus paroxetine group. At day 15 and
thereafter, there were no significant differences be-
tween the two groups. At day 31 the proportion of
patients with remitted depression was 76% in the pin-
dolol plus paroxetine group and 70% in the placebo
plus paroxetine group; the difference between the
groups was not significant.
Am J Psychiatry 155:10, October 1998
 BORDET, THOMAS, AND DUPUIS

TABLE 1. Psychometric Test Scores for Days 0 (Baseline) to TABLE 2. Hamilton Depression Subscale Scores for Days 0
15 of Depressed Patients Receiving Paroxetine Plus Either (Baseline) to 15 of Depressed Patients Receiving Paroxetine
Placebo or Pindolol Plus Either Placebo or Pindolol
Score Score
Placebo Pindolol Placebo Pindolol
Plus Plus Two-Tailed Plus Plus Two-Tailed Stu-
Paroxetine Paroxetine Student’s t Paroxetine Paroxetine dent’s t Test
(N=50) (N=50) Test (df=98) Subscale (N=50) (N=50) (df=98)
Test and Day Mean SD Mean SD t p and Day Mean SD Mean SD t p
17-item Hamilton De- Suicide
pression Rating 0 1.4 1.3 1.5 1.1 0.25 0.80
Scale 5 1.0 1.0 0.6 0.7 –2.21 0.03
0 24.1 3.5 24.2 3.3 0.06 0.95 10 0.6 0.9 0.3 0.6 –2.31 0.03
5 19.0 5.9 15.7 5.3 –2.96 0.004 15 0.4 0.8 0.3 0.6 –1.15 0.25
10 14.7 6.8 11.7 6.4 –2.22 0.03 Depression
15 11.5 7.7 9.7 6.4 –1.33 0.19 0 12.4 2.0 12.5 1.9 0.15 0.88
10-item Montgomery- 5 10.5 3.0 8.8 2.8 –2.93 0.005
Åsberg Depression 10 8.6 3.4 6.8 3.3 –2.65 0.01
Rating Scale 15 6.9 3.8 5.7 3.5 –1.70 0.17
0 30.6 4.8 30.4 4.5 –0.22 0.83 Anxiety
5 24.3 7.1 20.6 7.1 –2.66 0.01 0 4.2 2.0 3.9 1.8 –1.00 0.32
10 19.2 8.8 14.2 8.2 –2.91 0.005 5 3.1 1.8 2.5 1.5 –1.73 0.09
15 14.7 9.4 12.0 8.0 –1.53 0.13 10 2.1 1.5 1.9 1.5 –0.54 0.59
CGI 15 1.6 1.5 1.5 1.4 –0.07 0.95
0a Sleep
5 3.2 0.9 2.6 1.0 –3.15 0.003 0 3.8 1.3 3.9 1.2 0.66 0.51
10 2.8 1.0 2.1 1.0 –3.15 0.003 5 2.6 1.7 2.3 1.5 –0.83 0.41
15 2.2 1.2 2.0 1.0 –1.10 0.31 10 1.8 1.8 1.6 1.6 –0.65 0.52
a Not evaluated at baseline. 15 1.4 1.7 1.2 1.5 –0.70 0.50

Analysis of global improvement with the CGI
pointed out the effect of pindolol in the early phase of
paroxetine treatment (table 1). At days 5 and 10 CGI
scores were significantly lower in the pindolol plus
paroxetine group than in the placebo plus paroxetine
group. At day 15 CGI scores were not significantly dif-
ferent in the two groups.
Depression Severity
Changes in the level of severity of depression, as
measured from days 0 to 15 by both the Hamilton and
Montgomery-Åsberg depression scales, are presented
in table 1. Hamilton depression scale scores were sig-
nificantly lower in the pindolol plus paroxetine group
than in the placebo plus paroxetine group at both days
5 and 10. At day 15 there was no significant difference
between the two groups. The same pattern was seen
for the Montgomery-Åsberg depression scale for the
pindolol plus paroxetine and placebo plus paroxetine
groups at both days 5 and 10. At day 15 there was no
significant difference between the two groups.
Table 2 shows the mean scores for the subscales of
the Hamilton depression scale from days 0 to 15. Sig-
nificant differences between the pindolol plus paroxet-
ine and placebo plus paroxetine groups were found at
both days 5 and 10 for the depression and suicide sub-
scales but not for the anxiety and sleep subscales. De-
pression subscale scores were significantly lower with
pindolol plus paroxetine than with placebo plus par-
oxetine at both days 5 and 10. Scores on the suicide
subscale were significantly lower in the pindolol plus
paroxetine group than in the placebo plus paroxetine
group at day 5 and at day 10. At day 15 there was no
significant difference between pindolol plus paroxetine
and placebo plus paroxetine for all subscales.
DISCUSSION
Intermediate analysis at the end of the first month of
treatment of the first 100 patients included in this
study indicates that the addition of pindolol signifi-
cantly accelerates the onset of response to paroxetine
treatment in patients suffering from major depression.
By day 10 of treatment, the percent of patients with
clinical improvement significantly increased, by almost
twofold, in the pindolol plus paroxetine group, while
scores on the Hamilton depression scale, Montgom-
ery-Åsberg depression scale, and CGI were signifi-
cantly lower on days 5 and 10 in the pindolol plus par-
oxetine group. The number of patients with remitted
depression in the two groups merged on day 15 and
thereafter followed the same improvement course until
day 31. The placebo group responded to treatment
within the usual 2-week delay reported in standard
SSRI studies, indicating that there was no particular
bias in the recruitment of patients. The proportion of
successful therapeutic results in this study is slightly
higher than the 60% to 70% usually observed in stud-
ies of antidepressant drugs, which might in part be a
result of the frequency of visits at the beginning of the
study (days 0, 5, 10, 15, 21, 25, and 31), during which
a supportive psychotherapeutic effect might have come
into play.

Am J Psychiatry 155:10, October 1998 1349

PINDOLOL AND ONSET OF ACTION OF PAROXETINE
Chemical blocking of β-receptors is widely used in
psychiatry for problems as diverse as antipsychotic
drug-induced akathisia and social phobia anxiety dis-
order (26, 27). Although β-blockers can partially
mimic a small antidepressant effect through their anti-
anxiety properties (27), our data suggest that this ef-
fect alone is not responsible for pindolol-induced im-
provement, because anxiety was not predominantly
improved by pindolol plus paroxetine while depressive
symptoms were clearly improved. The drug associa-
tion of pindolol plus paroxetine improved psychic anx-
iety, as shown by changes in the score of this item in
our study. Nonetheless, the significance of this item re-
mains controversial because it is considered by some
authors as a depressive symptom and by others as an
anxiety symptom (23). Although many antidepressants
down-regulate β-adrenergic receptors, there is no evi-
dence that β-adrenergic antagonists induce any type of
antidepressant effect by themselves, but, rather, can in-
duce depressive symptoms (28). The lack of antide-
pressant activity in this class of drugs was also pointed
out clearly in Blier and Bergeron’s work testing propra-
nolol (15). There is, however, a certain heterogeneity
concerning pharmacological properties within the β-
adrenergic antagonist class, and, as a result, pindolol is
known to have a higher affinity for 5-HT1A receptors
that propranolol (29).
A more rapid onset of action of SSRIs by pindolol
addition has been reported in two clinical trials (17,
18), while in another one (16), the authors failed to
find any difference because of a lack of sufficiently fre-
quent assessment in the first 2 weeks of treatment.
Tome et al. (18) reported a significant acceleration of
onset of action of paroxetine observable from day 4 in
a group receiving pindolol (7.5 mg t.i.d. for 6 weeks).
In their study of fluoxetine and pindolol (7.5 mg t.i.d.
for 6 weeks), Pérez et al. (17) reported that the number
of days to reach sustained response was lower in the
fluoxetine and pindolol group than in the fluoxetine
and placebo group (median=19 versus 29 days). The
results concerning onset of action in our study are ap-
parently in accordance with those of these and previ-
ous studies, while we used a higher dose and a shorter
period of pindolol treatment. However, there is a dis-
crepancy in the long-term effect of the addition of pin-
dolol. In the study by Tome et al. (18), patients origi-
nally treated with pindolol showed better outcome at
week 24 than patients taking paroxetine alone. In the
study by Pérez et al. (17), the difference between pin-
dolol and placebo in the cumulative percentage of pa-
tients with a sustained response was maintained until
the end of the 6-week period. Moreover, there was no
difference in time-to-onset of clinical improvement
when only patients with sustained response were con-
sidered. These results might suggest that the more
rapid onset of action of SSRIs would be related to a su-
perior outcome in depression when pindolol is added.
The differences in results could be explained by differ-
ences in length of pindolol treatment period. Neverthe-
1350
less, if it is true that prolonged pindolol treatment
leads also to prolonged enhancement of the effect of
SSRIs, the mechanism of action of pindolol remains
unclear and must be discussed.
From a pharmacodynamic point of view, our data
emphasize the hypothesis of a presynaptic mechanism,
e.g., an inhibition of 5-HT1A somatodendritic recep-
tors by pindolol. Experimental evidence, with use of
microdialysis techniques, suggests that pindolol acts
preferentially on somatodendrite 5-HT 1A autorecep-
tors to inhibit the negative feedback effect of 5-HT fol-
lowing the increase of 5-HT levels in the synaptic cleft
as a result of local SSRI activity (30, 31). While it is
clear that pindolol causes a functional blockade of 5-
HT1A receptors (therefore including pre- or postsynap-
tic receptors), the waning of the feedback effect of 5-
HT1A receptor stimulation after several days of SSRI
treatment is not yet clearly understood, in part because
it has not been consistently observed. There is still
much discussion as to whether the density or the sensi-
tivity of these receptors is affected by chronic SSRI ad-
ministration (32). It has been suggested that SSRIs may
also induce a decrease in the number of neuronal 5-HT
carriers, leading to a long-term decrease in 5-HT re-
uptake (33). Postsynaptic adaptive mechanisms could
also be possible, as suggested by prolonged enhance-
ment of SSRI effects by pindolol when it is adminis-
tered for 6 weeks (17, 18), as well as by results ob-
tained in some treatment-resistant patients after
pindolol addition (15). It remains unknown whether
there is an enhanced activation of certain 5-HT
postsynaptic receptors or whether long-term adaptive
changes in limbic and cortical areas are also required.
The mechanism of action of pindolol could be more
complex than initially thought, with both presynaptic
action explaining acceleration of onset of action and
postsynaptic action explaining the higher rate of re-
sponse when pindolol treatment is prolonged. More-
over, pindolol is often considered a “dirty drug,”
meaning that the regulation of 5-HT receptors that it
induces is complex and multifaceted. A pharmaco-
kinetic interaction (i.e., increase in the plasma levels of
paroxetine induced by pindolol) cannot be excluded,
although it has not been confirmed in another study
using fluoxetine (17). Nevertheless, these aspects were
not investigated in this clinical study.
Whereas much has yet to be done to explain the
mechanisms (pharmacodynamic or pharmacokinetic)
involved, the results of this preliminary report strongly
suggest that the association of pindolol with an SSRI
will accelerate the therapeutic response to antidepres-
sants, by more rapid improvement in depressive symp-
toms. The choice of the SSRI used remains to be pre-
cisely determined in relationship with differences in
pharmacological profiles of SSRIs. Moreover, in view
of two recent studies, it will also be necessary to pre-
cisely determine if this acceleration cannot be associ-
ated with an enhancement of response to SSRIs when
pindolol treatment is prolonged.
Am J Psychiatry 155:10, October 1998

REFERENCES
1. Boyer WF, McFadden GA, Feighner JP: The efficacy of selec-
tive serotonin reuptake inhibitors in depression, in Selective
Serotonin Reuptake Inhibitors. Edited by Feighner JP, Boyer
WF. Chichester, England, John Wiley & Sons, 1991, pp 89–
108
2. Blier P, de Montigny C: Current advances and trends in the
treatment of depression. Trends Pharmacol Sci 1994; 15:
220–226
3. Blier P, De Montigny C, Chaput Y: Modifications of the seroto-
nin system by antidepressant treatments: implications for the
therapeutic response in major depression. J Clin Psychophar-
macol 1987; 7:24S–35S
4. Hyttel J: Pharmalogical characterization of selective serotonin
reuptake inhibitors (SSRIs). Int Clin Psychopharmacol 1994;
9(suppl 1):19–26
5. Hoyer D, Clarke ED, Fozard J, Hartig P, Graeme M, Mylecha-
rane E, Saxena P, Humphrey P: International Union of Phar-
macology classification of receptors for 5-hydroxytryptamine
(serotonin). Pharmacol Rev 1994; 46:157–203
6. Artigas F, Romero L, de Montigny C, Blier P: Acceleration of
the effect of selected antidepressant drugs in major depres-
sion by 5-HT1A antagonists. Trends Neurosci 1996; 19:378–
383
7. Chaput Y, de Montigny C, Blier P: Effects of a selective 5-HT
reuptake blocker, citalopram, on the sensitivity of 5-HT auto-
receptors: electrophysiological studies in the rat brain. Nau-
nyn Schmiedebergs Arch Pharmacol 1986; 333:342–348
8. Bel N, Artigas F: Chronic treatment with fluvoxamine in-
creases extracellular serotonin in frontal cortex but not in
raphe nuclei. Synapse 1993; 15:243–245
9. Fuller RW: Uptake inhibitors increase extracellular serotonin
concentration measured by brain microdialysis. Life Sci 1994;
55:163–167
10. Rutter JJ, Gundlah C, Auerbach SB: Increase in extracellular
serotonin produced by uptake inhibitors is enhanced after
chronic treatment with fluoxetine. Neurosci Lett 1994; 171:
183–186
11. Heninger GR, Charney DS: Mechanism of action of antide-
pressant treatment: implication for the etiology and treatment
of depressive disorders, in Psychopharmacology: The Third
Generation of Progress. Edited by Meltzer HY. New York,
Raven Press, 1987, pp 535–544
12. De Montigny C, Chaput Y, Blier P: Classical and novel targets
for antidepressant drugs: new pharmacological approaches to
the therapy of depressive disorders. Int Acad Biomed Drug
Res 1993; 5:8–17
13. Bel N, Romero L, Celada P, de Montigny C, Blier P, Artigas F:
Neurobiological basis for the potentiation of the antidepres-
sant effect of 5-HT reuptake inhibitors by the 5-HT1A antago-
nist pindolol. Monit Mol Neurosci 1994; 6:209–210
14. Artigas F, Perez V, Alvarez E: Pindolol induces a rapid im-
provement of depressed patients treated with serotonin re-
uptake inhibitors (letter). Arch Gen Psychiatry 1994; 51:248–
251
Am J Psychiatry 155:10, October 1998
BORDET, THOMAS, AND DUPUIS
15. Blier P, Bergeron R: Effectiveness of pindolol with selected
antidepressant drugs in the treatment of major depression. J
Clin Psychopharmacol 1995; 15:217–222
16. Berman RM, Darnell AM, Miller HL, Anand A, Charney DS: Ef-
fect of pindolol in hastening response to fluoxetine in the treat-
ment of major depression: a double-blind, placebo-controlled
trial. Am J Psychiatry 1997; 154:37–43
17. Pérez V, Gilaberte I, Faries D, Alvarez E, Artigas F: Ran-
domised, double-blind, placebo-controlled trial of pindolol in
combination with fluoxetine antidepressant treatment. Lancet
1997; 349:1594–1597
18. Tome MB, Isaac MT, Harte R, Holland C: Paroxetine and pin-
dolol: a randomised trial of serotoninergic autoreceptor block-
ade in the reduction of antidepressant latency. Int Clin Psy-
chopharmacol 1997; 12:81–89
19. Bordet R, Thomas P, Dupuis B: Selective serotonin reuptake
inhibitors plus pindolol (letter). Lancet 1997; 350:289
20. Hamilton M: A rating scale for depression. J Neurol Neurosurg
Psychiatry 1960; 23:56–62
21. Spitzer RL, Williams JBW, Gibbon M, First MB: User’s Guide
for the Structured Clinical Interview for DSM-III-R (SCID).
Washington, DC, American Psychiatric Press, 1990
22. Montgomery SA, Åsberg M: A new depression scale designed
to be sensitive to change. Br J Psychiatry 1979; 134:382–389
23. Bech P, Allerup P, Gram LF, Reisby N, Rosenberg R, Jacob-
sen O, Nagy A: The Hamilton depression scale. Evaluation of
objectivity using logistic models. Acta Psychiatr Scand 1981;
63:290–299
24. Conners CK, Barkley RA: Rating scales and checklists for
child psychopharmacology. Psychopharmacol Bull 1985; 21:
809–843
25. Hirschfeld RMA: Guidelines for the long-term treatment of de-
pression. J Clin Psychiatry 1994; 55(Dec suppl):61–69
26. Dumon J-P, Catteau J, Lanvin F, Dupuis BA: Randomized,
double-blind, crossover, placebo-controlled comparison of
propranolol and betaxolol in the treatment of neuroleptic-in-
duced akathisia. Am J Psychiatry 1992; 149:647–650
27. Bailly D: The role of beta-adrenoreceptor blockers in the treat-
ment of psychiatric disorders. CNS Drugs 1996; 5:115–136
28. Avorn J, Everitt DE, Weiss S: Increased antidepressant use in
patients prescribed β-blockers. JAMA 1986; 255:357–360
29. Tricklebank MD, Forler C, Fogard JR: The involvement of
subtypes of the 5-NT1 receptor and of catecholaminergic sys-
tems in the behavioral response to 8-hydroxy-2-(di-n-propy-
lamino) tetralin in the rat. Eur J Pharmacol 1984; 106:271–
282
30. Artigas F: 5HT and antidepressants: new views from micro-
dialysis studies. Trends Pharmacol Sci 1993; 15:262–263
31. Hjorth S: Serotonin 5HT1A autoreceptor blockade potentiates
the ability of the 5-HT reuptake inhibitor citalopram to increase
nerve terminals output of 5-HT in vivo: a microdialysis study. J
Neurochem 1993; 60:776–779
32. Kreiss DS, Lucki I: Desensitization of 5-HT1A autoreceptors
by chronic administration of 8-OH-DPAT. Neuropharmacology
1992; 31:1073–1076
33. Pineyro G, Blier P, Dennis T, De Montigny C: Desensitization
of the neuronal 5-HT carrier following its long-term blockade.
J Neurosci 1994; 14:3036–3047
1351