Professional Documents
Culture Documents
*Clinical Research Institute; yPfizer, New York, NY; zPfizer, Groton CT;
xWyeth, Collegeville, PA; and kPfizer, New London, CT.
Received January 2, 2008; accepted after revision July 28, 2008.
Pfizer Inc supported the conduct of this study.
Dr Preskorn and Mr Baker are employees of the Clinical Research Institute
which has received grant support from Pfizer. Drs Landren, Kolluri, and
Menniti are employees of Pfizer. Dr Krams is currently an employee of
Wyeth but was an employee of Pfizer during the course of this study.
Address correspondence and reprint requests to Sheldon H. Preskorn, MD,
Clinical Research Institute, 201 S. Hillside, Wichita, KS 67204. E-mail:
spreskorn@cri-research.net.
Copyright * 2008 by Lippincott Williams & Wilkins
ISSN: 0271-0749/08/2806-0631
DOI: 10.1097/JCP.0b013e31818a6cea
Copyright @ 2008 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
631
Preskorn et al
METHODS
Patient Selection
Subjects were recruited from referrals from physicians
and through an extensive media campaign and the Internet. Men
and women, aged 18 to 55 years, who were outpatients and had
a major depressive disorder recurrent without psychotic features
according to Diagnostic and Statistical Manual of Mental
Disorders, Fourth Edition criteria, were eligible to participate.
Patients with Diagnostic and Statistical Manual of Mental
Disorders, Fourth Edition diagnosis of bipolar disorder or who
had a history of antidepressant- or substance-induced hypo-
Study Design
This study used a design involving 2 periods (Fig. 1).
Period 1 consisted of 6 weeks of open-label treatment with
paroxetine, with a single-blind intravenous (IV) placebo infusion after the third week of paroxetine treatment. Paroxetine
nonresponders in period 1 were randomized in period 2 to
receive a single, double-blind, random-assignment IV infusion of either CP-101,606 or placebo and continued paroxetine treatment for an additional 4 weeks. All infusions
(periods 1 and 2) were identical in terms of volume infused
and duration. The schedule for infusions and outpatient visits
sberg Depression Ratincluding HDRS and Montgomery-A
ing Scale (MADRS) ratings and safety assessments are
shown in Table 1.
During period 1, subjects were started on 20 mg of
paroxetine in the morning (every morning) and then escalated
to 40 mg every morning within 1 week. At the end of week 3
632
Copyright @ 2008 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Screen
HDRS
MADRS
12-Lead ECG
Infusion 1
Period 2
D5
D8
D12
Wk 6
D15
D2
D5
D8
D12
D15
D30
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
PK samples
Safety labs
Infusion 2
X
X
X
X
Blood samples for drug analyses were obtained before drug infusion and then at 0, 0.5, 1.5, 4, 8, 10, 12, 24, and 96 hours after initiation of drug infusion.
PK indicates pharmacokinetic.
Infusions
The infusions in periods 1 and 2 were identical in
volume and duration. The first 7 subjects in the trial received
an 8-hour infusion. The CP-101,606 infusion during period 2
for these patients consisted of 0.75 mg/kg per hour for 1.5
hours followed by 0.15 mg/kg per hour for 6.5 hours. Four of
these first 7 subjects experienced a moderate to severe dissociative response after the double-blind infusion. The blind
was not broken at the site but was broken by sponsor to
ensure that the target concentration had not been exceeded.
All patients experiencing the dissociative response had
received active CP-101,606 treatment. Although target
plasma concentrations had not been exceeded, the dosage of
CP-101,606 was reduced for all subsequent patients by decreasing the infusion duration to 1.5 hours and the dose to 0.5
mg/kg per hour with the intent of accomplishing 3 goals: (a)
ensure patient safety, (b) maintain the study blind, and (c)
determine if antidepressant effect could be achieved at doses
that did not cause a dissociative response. The duration of the
placebo infusions was also correspondingly decreased to 1.5
hours in both periods 1 and 2.
Outcome Measures
The primary efficacy end point was change from
baseline in the MADRS total score 4 days after the period 2
infusion (approximately 96 hours after this infusion; period 2,
day 5). The baseline measurement for the MADRS total score
was defined as the mean of the measurements at the end of
week j1 (ie, after 6 weeks of paroxetine treatment) and at hour
0 (pre-dose) on day 1 of period 2. Secondary efficacy end point
was change from baseline in the HDRS total score. Safety
evaluations included the incidence and magnitude of ECG
changes, adverse events (AEs), and the rate of laboratory and
vital sign abnormalities in subjects treated with CP-101,606
versus those administered placebo. The schedule of depression
and safety assessments are given in Table 1.
Staff
The infusions and the efficacy ratings were done by
different research staff. Neither the infusion staff nor the patients
discussed with the efficacy rating staff how the patients did
during the infusions. The efficacy rating staff were on the unit
only for the ratings, leaving before the infusion started and
returning 24 hours later for the day 2 ratings.
Statistical Analysis
Sample Size
A sample size of 15 subjects per group (a total of 30
subjects) was selected to provide at least 80% power to detect a
true difference of 8 points in mean for change from baseline in
the total score of MADRS on day 5, based on a 1-sided, 2sample t test at ! = 10%. This calculation assumed a common
SD of 10 points for change from baseline in the total score
of MADRS.
Copyright @ 2008 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
633
Preskorn et al
Primary Analysis
The primary end point was analyzed as a continuous
measurement using a linear model including the following
terms:
& Treatment (CP-101,606 placebo)
& Baseline total scores of MADRS
Based on the above linear model, a 1-sided test was used
to compare the CP-101,606 group with the placebo group at a
10% level of significance. An 80% 2-sided confidence interval
(CI) was provided for the true mean difference between the 2
treatment groups. All analyses were performed using available
data only. No multiplicity adjustments were made to the P
values or the CIs.
The analysis methodology described above for the
primary end point was also used for the key secondary end
points which were as follows: (1) change from baseline in the
total score of MADRS at each planned time point in period 2
other than the primary time point (ie, day 5) and (2) change
from baseline in the total HDRS score at each planned time
point in period 2.
A post hoc statistical analysis following the same parameters as the prespecified statistical plan was performed
on the 23 participants who received the reduced IV infusion
of CP-101,606 or placebo. A similar analysis was not done for
the high-dose infusion patients because this group contained
only 7 subjects (4 receiving CP-101,606 and 3 receiving
placebo). This number is so small that it was not feasible to
apply the statistical model to obtain reliable estimates of the
treatment effects.
Descriptive statistics (n, arithmetic mean, SD) for all
efficacy end points were presented for the treatment group.
RESULTS
Patients
The recruitment campaign started with the 260,000
persons who were within the ages of 18 to 55 years living
within convenient driving distance to the research facility. An
estimated 4333 individuals in this population had SSRI nonresponsive major depression; 1272 went through a thorough
screening via telephone, and 192 went through a face-to-face
interview to yield a total of 47 patients who were entered to
period 1. A total of 30 subjects (15 subjects per group) were
randomized to period 2.
Subjects ranged in age from 21 to 54 years. The majority
of the subjects were white (90%) and female (73%). All
subjects had a primary diagnosis of major depressive disorder.
There was no difference between those treated with CP101,606 versus placebo in period 2 in terms of age, sex,
ethnicity, body mass index, severity of depressive symptoms as
measured by the 17-item HDRS scale, or number of previous
failed SSRI trials. There were no other presenting conditions or
medical histories that the investigator considered sufficient to
affect the conduct of the study or to represent potential risk to
the subject during study participation.
Efficacy
Table 2 presents change from baseline in MADRS and
17-item HDRS scores by time point for CP-101,606 versus
634
Adverse Events
There were no deaths or discontinuations due to AEs
or abnormal laboratory findings. Fifteen subjects from the
CP-101,606-plus-paroxetine group had 55 AEs versus 61
AEs for the 15 subjects from the placebo-plus-paroxetine
group. Six CP-101,606treated patients experienced a dissociative reaction (2 mild, 2 moderate, and 2 severe). Two
placebo-treated patients experienced a mild dissociative
reaction. Most AEs were mild and did not differ between
CP-101,606 and placebo-treated patients. These AEs included feeling abnormal, dizziness, paresthesia, somnolence,
dry mouth, and abnormal urine odor.
A dissociative reaction was defined as the symptoms
caused in healthy humans by other NMDA antagonists such as
ketamine or phencyclidine.24,25 These include difficulty initiating and sustaining attention; memory impairment; disturbance in time, body, and/or environmental perception; stilted
speech; emotional withdrawal; impaired coordination; psychomotor retardation; dizziness; a sense of intoxication; emotional
blunting; intense emotional reactions; anxiety; feelings of
unreality and loss of control over thought processes; concrete
thinking; bizarre reasoning; illusory experiences in all sensory
* 2008 Lippincott Williams & Wilkins
Copyright @ 2008 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Placebo
LS Means*
Difference From Placebo (80% CI)
t Statistic (df )
P (1-Sided)
Change from baseline in MADRS total score by time point, CP-101,606 vs placebo
Period 2, wk 0 (day 2)
n
Arithmetic mean (SD)
LS means
15
15
j8.0 (9.5)
j7.7
j7.6 (8.2)
j7.8
0.02 (27)
0.491
Period 2, wk 0 (day 5)
n
Arithmetic mean (SD)
15
15
j14.1 (8.1)
j5.5 (7.9)
j14.0
LS means
j5.6
j2.81 (27)
0.005y
j1.85 (23)
0.0383
Period 2, wk 1 (day 8)
n
Arithmetic mean (SD)
LS means
14
12
j14.2 (10.1)
j13.7
j7.1 (6.1)
j7.6
12
10
j8.1 (10.6)
j7.6 (3.8)
j7.7
LS means
j8.0
0.08 (19)
0.468
0.14 (17)
0.447
j0.17 (27)
0.432
j2.79 (27)
0.005
j1.47 (23)
0.077
j0.03 (19)
0.487
0.58 (17)
0.284
12
j10.5 (10.3)
j9.0
j7.6 (10.2)
j9.7
Change from baseline in 17-item HDRS total score by time point, CP-101,606 vs placebo
Period 2, wk 0 (day 2)
n
Arithmetic mean (SD)
15
15
j4.2 (5.4)
j3.7 (5.6)
j4.1
LS means
j3.7
Period 2, wk 0 (day 5)
n
Arithmetic mean (SD)
15
j9.6 (4.8)
j9.5
LS means
15
j3.5 (6.4)
j3.6
Period 2, wk 1 (day 8)
n
Arithmetic mean (SD)
14
12
j8.5 (4.8)
j4.5 (6.7)
j8.3
LS means
j4.8
12
j5.1 (6.0)
j4.7
LS means
10
j4.3 (4.5)
j4.7
12
j7.3 (5.7)
j6.6
8
j7.3 (6.6)
j8.2
*Least-squares mean.
y
This represents the P value for the primary analysis. Note that all other P values are presented only for descriptive purposes.
Copyright @ 2008 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
635
Preskorn et al
DISCUSSION
In this study, the NR2B subunitselective NMDA
antagonist, CP-101,606, produced a significant antidepressant effect in patients with major depression refractory to
treatment with SSRI therapy. On day 5 of period 2, there
was an 8.4-point greater reduction in the MADRS score in
patients receiving CP-101,606 versus those receiving placebo. To provide context for these results, the magnitude of
this effect is 2 to 3 times greater than those seen in the positive registration trials of SSRIs.26 The response and remission rates were 2.5 times higher than those seen in the
Sequenced Treatment Alternative to Relieve Depression
study when nonresponders to citalopram were crossed over
to bupropion, sertraline, or venlafaxine19 and comparable to
those seen in the 2 ketamine studies.16,23
Methodologically, this study differed from the ketamine
studies in several ways: (1) the use of a parallel group versus
a crossover design, (2) the use of a prospective adequate trial
of an SSRI (paroxetine) to confirm treatment refractoriness,
and (3) the addition of the NMDA antagonist to SSRI
therapy. Whereas the ketamine studies looked at earlier time
points to assess antidepressant activity, the statistical plan
for this study prespecified assessing antidepressant response
on day 5 after the double-blind infusion because the HDRS
and MADRS, as originally designed, assess change over
days rather than over hours. That is particularly true for the
HDRS, which assesses items such as insomnia, appetite,
work and activities, and sex drive. Nevertheless, the improvement in the mental status examination including mood was
frequently apparent on day 2 after infusion of CP-101,606 in
those patients who met response and remission criteria on
day 5.
636
Dissociative responses occurred in 6 participants on CP101,606 and 2 on placebo, raising the possibility that the
improvement after CP-101,606 could be related to the unblinding of the rater. There are 3 factors against this possibility. First,
the staff doing the efficacy ratings did not observe the patients
from the ratings done immediately before the infusion to those
done 24 hours after the infusion, at which time any cognitive
adverse effects had resolved. Second, the staff who did the
efficacy ratings was instructed to not ask the patients how they
responded to the infusion on day 1, and patients and infusion
staff were similarly instructed to not comment to the efficacy
rating staff how the patient had done on the infusion day. Third,
5 of the 9 CP-101,606treated patients who met criteria for
antidepressant response did not experience a dissociative
response as a result of the infusion. Fourth, the post hoc analysis
of the 23 patients after the reduction in the infusion found a
statistically significant better response in the CP-101,606
treated compared with placebo-treated patients, and the difference in MADRS scores between CP-101,606 and placebo in
these 23 patients was numerically the same as that seen in the
entire group of 30 patients (ie, both the high- and low-infusion
patients). With this lower infusion, there was no difference in the
rate of dissociative reactions reported by CP-101,606 and
placebo-treated patients (2 of 11 vs 2 of 12, respectively), and all
4 of these dissociative reactions were mild in severity.
In contrast to the ketamine studies, the lower infusion of
CP-101,606 produced a rapid and robust antidepressant response without producing a dissociative reaction. There are
several possible reasons for this difference. First, lower doses
of ketamine that do not cause dissociative states may nonetheless be antidepressive. Second, the administration of CP101,606 in combination with paroxetine may have reduced
the severity of the neuropsychiatric side effects. Third, the
NMDA receptor subunit selectivity and allosteric mechanism
of action may yield a bona fide greater therapeutic index for
CP-101,606 compared with ketamine.
This study demonstrated statistical superiority of the
active versus placebo treatment despite the small number of
patients studied (n = 30). To accomplish this goal, the study
had to have a low placebo response rate, which was made
more daunting because the study drug had to be given
intravenously and thus required admission to a clinical
research unit. For these reasons, the study design included
the following elements: (a) an inclusion criteria requiring a
history of nonresponse to an adequate trial of at least 1 SSRI,
(b) prospective evidence of nonresponse to an adequate trial
of paroxetine plus a single-blind placebo infusion which was
identical to the subsequent randomized infusion, (c) the use
of HDRS score to qualify the patient for entry into the openlabel trial (period 1) and to progress to the double-blind trial
(period 2) but the use of the MADRS as the primary outcome
measurement, and (d) the use of different and blinded staff to
do the efficacy ratings from those who cared for the patient
during their time on the clinical research unit. The exclusion
of patients who had not benefited from adequate trials of
more than 2 mechanistically different forms of antidepressant
treatment was done to avoid enrolling patients whose
depressive illness was so treatment resistant that they could
produce a false-negative result.
* 2008 Lippincott Williams & Wilkins
Copyright @ 2008 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Copyright @ 2008 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
637