Mifepristone versus Placebo in the Treatment of
`Psychosis in Patients with Psychotic Major Depression
`Charles DeBattista, Joseph Belanoff, Steven Glass, Arif Khan, Robert L. Horne, Christine Blasey,
`Linda L. Carpenter, and Gustavo Alva
`
`Background: Abnormalities in the hypothalamic pituitary adrenal axis have been implicated in the pathophysiology of psychotic
`major depression (PMD). Recent studies have suggested that the antiglucocorticoid, mifepristone might have a role in the treatment of
`PMD. The current study tested the efficacy of mifepristone treatment of the psychotic symptoms of PMD.
`Methods: 221 patients, aged 19 to 75 years, who met DSM-IV and SCID criteria for PMD and were not receiving antidepressants or
`antipsychotics, participated in a double blind, randomized, placebo controlled study. Patients were randomly assigned to either 7 days
`of mifepristone (n ⫽ 105) or placebo (n ⫽ 116) followed by 21 days of usual treatment.
`Results: Patients treated with mifepristone were significantly more likely to achieve response, defined as a 30% reduction in the Brief
`Psychiatric Rating Scale (BPRS). In addition, mifepristone treated patients were significantly more likely to achieve a 50% reduction
`in the BPRS Positive Symptom Scale (PSS). No significant differences were observed on measures of depression.
`Conclusion: A seven day course of mifepristone followed by usual treatment appears to be effective and well tolerated in the treatment
`of psychosis in PMD. This study suggests that the antiglucocorticoid, mifepristone, might represent an alternative to traditional
`treatments of psychosis in psychotic depression.
`
`Key Words: Psychotic major depression, mifepristone, cortisol, GR
`
`A lthough psychotic major depression (PMD) is simply classi-
`
`fied as a severe form of depression in DSM IV(TR), PMD may
`represent a unique subtype of depression with its own
`phenomenology, treatment response, and biology (Schatzberg and
`Rothschild 1996). Psychotic features occur in about 14 –25% of
`patients with major depression (Coryell 1996; Johnson et al 1991).
`PMD may be associated with a more chronic course, more frequent
`hospitalizations, higher risk of suicide, and greater disability than
`other forms of depression (Angst 1986; Coryell et al 1986).
`The most commonly employed treatments for PMD are the
`combination of an antidepressant with an antipsychotic (Amore et al
`1996; Rothschild et al 1993; Schatzberg 1992; Simpson et al 1999;
`Wheeler Vega et al 2000) or electroconvulsive
`therapy (ECT)
`(Buchan et al 1992; Minter and Mandel 1979). ECT appears to be
`effective in the treatment of PMD even when pharmacotherapy is
`unsuccessful (Avery and Lubrano 1979; Petrides 2001). Unfortu-
`nately, both ECT and combination pharmacotherapy have draw-
`backs, including substantial side effects, social stigma, and a delayed
`onset of therapeutic benefit (Challiner and Griffiths 2000; Datto
`2000; Fogg-Waberski and Waberski 2000).
`Abnormalities in the hypothalamic pituitary adrenal axis (HPA)
`have long been implicated in the pathophysiology of PMD. Patients
`with psychotic depression consistently show a high rate of non-
`suppression on the dexamethasone suppression test (DST) and/or
`high post dexamethasone cortisol levles (Arana et al 1983; Ayuso-
`Gutierrez et al 1985; Bond et al 1986; Caroff et al 1983; Mendlewicz
`et al 1982; Rothschild et al 1982). Other HPA abnormalities found in
`
`From Corcept Therapeutics (CD, JB, CB), Menlo Park, California; Department
`of Psychiatry (CD, CB), Stanford University, Stanford, California; North-
`west Clinical Research Center (AK), Bellevue, Washington; Department
`of Psychiatry (GA), University of California Irvine Medical Center, Orange,
`California; Brown University School of Medicine (LLC), Butler Hospital,
`Providence, Rhode Island; Montevista Hospital (RLH), University of Ne-
`vada School of Medicine, Las Vegas, Nevada.
`Address reprint requests to Charles DeBattista, M.D., Mood Disorders Clinic,
`Stanford University, 401 Quarry Road, Stanford, CA 94305; E-mail:
`debattista@stanford.edu.
`Received December 9, 2005; revised May 3, 2006; accepted May 25, 2006.
`
`0006-3223/06/$32.00
`doi:10.1016/j.biopsych.2006.05.034
`
`PMD include higher 24 hour urinary free cortisol relative to non-
`psychotic major depression (NPMD) patients (Anton 1987) and
`higher serum ACTH and nocturnal cortisol levels in PMD patients
`compared to NPMD patients (Keller et al, in press). It has been
`suggested that psychosis in PMD is driven in part by the effects of
`glucocorticoids on dopamine synthesis and activity (Schatzberg et al
`1985). Exogenous glucocorticoid administration as well as Cush-
`ing’s disease may be associated with changes in mood, cognition,
`and perception that parallel symptoms seen in PMD (de Quervain et
`al 2000; Forget et al 2000; Gifford and Gunderson 1970; Jeffcoate et
`al 1979; Mauri et al 1993; Starkman 1993).
`Cortisol synthesis inhibitors, such as ketoconazole and metyrap-
`one, may have therapeutic benefits in some depressed patients. For
`example, ketoconazole was shown to have antidepressant effects in
`a subset of depressed patients with hypercortisolemia (Wolkowitz
`et al 1999; Wolkowitz et al 1993). More recently, Jahn and col-
`leagues (Jahn et al 2004) found that the addition of metyrapone to
`a standard serotonergic antidepressant was more effective than the
`addition of a placebo in augmenting antidepressant response. In
`addition, metyrapone treated patients exhibited a more rapid anti-
`depressant response than did placebo treated patients. However,
`cortisol synthesis inhibitors are limited by a variety of potentially
`serious side effects at doses necessary to suppress cortisol synthesis
`(Sonino 1987).
`Mifepristone is a potent and specific antagonist of the type II
`glucocorticoid receptor (GR-II) and the progesterone receptor
`(Gaillard et al 1984; Herrmann 1982; Lamberts et al 1984;
`Proulx-Ferland et al 1982). Although mifepristone is a potent
`GR-II antagonist,
`it has little effect on the mineralocorticoid
`receptors MR (previously named GR-1). In addition, mifepristone
`has no known affinity to monoamine, histamine, or cholinergic
`receptors. The GR-II receptor has a low affinity for cortisol and
`appears to play a part in the termination of the stress response.
`Mifepristone does not appear to be associated with suppression
`of glucocorticosteroid actions peripherally (Bertagna et al 1994;
`Gaillard et al 1984).
`Several studies have indicated that mifepristone might be
`effective in the treatment of PMD and that
`the effects on
`psychosis might be more consistent and robust than the effects
`on depression (Belanoff et al 2001; Belanoff et al 2002; Simpson
`et al 2005; Flores et al 2006). These studies also suggested that
`
`BIOL PSYCHIATRY 2006;60:1343–1349
`© 2006 Society of Biological Psychiatry
`
`

`

`1344 BIOL PSYCHIATRY 2006;60:1343–1349
`
`C. DeBattista et al
`
`the benefits of mifepristone might be seen with only 6 or 7 days
`of treatment and the effects might be sustained for at least 3
`weeks after treatment with mifepristone was discontinued. Sim-
`ilarly, mifepristone might have value in the treatment of bipolar
`depression (Young 2004). These preliminary studies have been
`small, and only limited conclusions can be drawn from them. A
`larger double blind study of 208 PMD patients examined the
`effect of adding 7 days of mifepristone or placebo to usual
`treatment in patients hospitalized for the purposes of the study
`(DeBattista 2003). Both treatment groups improved significantly
`from baseline but did not differ from each other on the primary
`end point (a 30% reduction in the BPRS at 7 and 28 days).
`However, in post hoc analyses patients who received mifepris-
`tone were more likely to achieve a rigorous response (i.e., HamD
`ⱕ 7, BPRS ⱕ 25) require less antipsychotics, and were more
`likely to be discharged earlier from the hospital
`than were
`placebo-treated patients. The concomitant use of treatments
`known to be effective (concurrent antidepressant/antipsychotic
`use and hospitalization) may have reduced the ability to dem-
`onstrate a difference between groups on the primary endpoint.
`While most of the work involving anti-glucocortioid strategies
`has focused on antidepressant effects, there is reason to believe
`that the antipsychotic effects of these drugs in the treatment of
`psychotic depression might be greater than their effects on
`depression. In 1985, Schatzberg and colleagues proposed a
`corticosteroid hypothesis/dopamine hypothesis for psychotic
`depression which postulated that steroid mechanisms were
`driving the psychotic symptoms of PMD (Schatzberg et al 1985).
`Specifically,
`they stated “This hypothesis is not
`intended to
`primarily account for why patients become depressed but rather
`why some depressed patients become psychotic.” While prelim-
`inary studies with mifepristone in the treatment of PMD have
`suggested an antidepressant effect, the most robust and consis-
`tent effects have been seen on scales that measure psychosis,
`such as the BPRS. In particular, mifepristone appeared to impact
`the positive symptom subscale, which measures core psychotic
`symptoms including delusions, paranoia and hallucinations.
`In this study, the antipsychotic efficacy of mifepristone is
`compared with placebo in PMD patients who are not taking
`antidepressants or antipsychotics during and prior to study drug
`administration. We hypothesized that patients taking mifepris-
`tone would have a rapid reduction of psychotic symptoms
`evident after 7 days of treatment, and that this response would be
`sustained for three weeks post
`treatment. Furthermore, we
`hypothesized that differences between response rates to treat-
`ment and placebo would be more pronounced among patients
`with higher baseline psychotic symptom severity. It was ex-
`pected that improvements in psychotic symptoms would be
`greater than improvements in depressive symptoms.
`
`Methods and Materials
`
`Twenty-nine sites in the continental United States participated
`in this study after obtaining institutional review board approval.
`All patients provided written informed consent before participa-
`tion.
`Patients were included if they met DSM-IV criteria for PMD by
`clinical interview and by SCID. In addition, hospital admission
`notes were reviewed by the sponsor’s medical monitor to further
`confirm the diagnosis of PMD. Enrolled patients were required to
`achieve a score of 38 or greater on the BPRS and 20 or greater on
`the HamD-24. Patients were required to have a negative serum
`pregnancy test and to use two acceptable methods of contracep-
`tion throughout the study.
`
`www.sobp.org/journal
`
`Exclusion criteria included an unstable medical condition, the
`use of systemic or inhaled corticosteroids, ECT in the 3 months
`prior to randomization, antidepressant and/or antipsychotic use
`during the 7 days before randomization, a history of illicit drug
`use in the previous month or alcohol or drug dependence in the
`previous 6 months.
`Patients who met the study criteria were randomized 1:1 to 7
`days of inpatient treatment in a double blind, placebo controlled,
`parallel group design. Patients received either mifepristone
`600mg/day or placebo for seven consecutive days. Patients were
`evaluated prior to dosing at day 0 and then daily during dosing
`(days 1–7). Psychiatric assessments included the BPRS and
`HAMD scales performed at days 0, 3, 7, 14, and 28. Antipsychot-
`ics and antidepressants were not allowed for at least 7 days prior
`to randomization and for the 7 days of study drug administration.
`From day 8 onward, the investigator could prescribe any medi-
`cation regimen or treatment that was clinically indicated. At the
`request of the FDA, a subset of patients (chronologically, the
`patients enrolled in the latter part of the study) had efficacy
`measures at day 56 to further assess the durability of response.
`All patients were hospitalized for at least the first three days of
`the trial. Thereafter, any patient deemed stable for discharge
`could continue to receive study medication as an outpatient.
`Patients discharged from the inpatient setting before day 7 were
`seen daily by research staff for clinical assessments and wit-
`nessed administration of study medication.
`Safety was assessed by spontaneous report of adverse events,
`physical examination, and laboratory assessments including a
`serum chemistry panel, a complete blood count with differential,
`and an electrocardiogram at days 0 and 7.
`
`Statistical Analysis
`The primary endpoint measure for efficacy, defined a priori,
`was a “responder analysis” based on BPRS scale level of re-
`sponse. This endpoint compared the percentage of patients who
`had Rapid Response (at least a 30% reduction in the BPRS Total
`at days 7 and 28), Response (at least a 30% reduction in BPRS
`Total at day 28 but not at day 7), and No Response. A categorical
`30% reduction in the BPRS had been used in a previous trial of
`mifepristone
`in PMD (DeBattista 2003) and was
`thought
`to
`represent a clinically meaningful reduction in psychotic symp-
`toms. There were two prominent secondary measures of
`efficacy: 1) a 50% reduction in the BPRS positive symptom
`subscale (PSS), the 4 core psychotic items including suspi-
`ciousness, hallucinatory behavior, disorganized thinking, and
`unusual thought content; and 2) a 50% reduction on the HamD
`at day 7 and sustained to day 28.
`Efficacy analyses were performed on the Intent to Treat (ITT)
`sample (n ⫽ 221), which consisted of all randomized subjects
`who received at least one dose of study medication. Data were
`observed at day 28 for 170 of the 221 patients (77%). For the 51
`patients with missing data at either day 7 or day 28, BPRS and
`HAMD data were imputed using a mixed effects model for
`repeated measurements (MMRM). The response variable was the
`natural
`logarithm of the BPRS total, BPRS PSS (rescaled by
`subtracting 4 from each value), and HAMD scores. Values of zero
`were replaced with .25. The model included fixed effects cate-
`gorical terms for treatment group, visit, and their interaction. An
`unstructured covariance matrix was used to model the interde-
`pendence of the within-subject repeated measures. The calcula-
`tions were carried out using SAS PROC MIXED. Scores on the
`three outcome measures were imputed using the MMRM model,
`
`

`

`C. DeBattista et al
`
`BIOL PSYCHIATRY 2006;60:1343–1349 1345
`
`Table 1. Demographic Characteristics of the ITT Sample (n ⫽ 221)
`
`Mifepristone
`(n ⫽ 105)
`
`Placebo
`(n ⫽ 116)
`
`p-Value
`
`Age (Mean, SD)
`Age group (n, %)
`18 to 34
`35 to 64
`65⫹
`Gender (n, %)
`Male
`Female
`Race, (n, %)
`White
`Black
`Asian
`Hispanic or Latino
`Baseline Measures
`BPRS Total
`BPRS PSS Scale
`HAMD
`
`40.9 ⫾ 10.8
`
`41.6 ⫾ 11.0
`
`31 (29.5%)
`72 (68.6%)
`2 (1.9%)
`
`56 (53.3)
`49 (46.7)
`
`57 (54.3)
`38 (36.2)
`1 (1.0)
`9 (8.6)
`
`55.8 ⫾ 11.6
`13.7 ⫾ 3.6
`37.3 ⫾ 8.4
`
`33 (28.4%)
`79 (68.1%)
`4 (3.4%)
`
`56 (48.3)
`60 (51.7)
`
`59 (50.9)
`42 (36.2)
`4 (3.4)
`10 (8.6)
`
`55.7 ⫾ 9.2
`13.4 ⫾ 3.2
`37.3 ⫾ 7.5
`
`.62a
`
`.86b
`
`.45c
`
`.74b
`
`.56a
`.95a
`.96a
`
`aSignificance level from a one-way ANOVA with treatment as a
`factor.
`bFrom a Fisher’s Exact test.
`cFrom a Pearson chi-square test.
`
`and then responder status was determined based on the defini-
`tions described above.
`The proportion of rapid and sustained responders, responders,
`and non-responders were compared across treatments using Coch-
`ran-Mantel-Haenszel tests adjusting for site. The test statistic is the
`Cochran-Armitage linear trend test with p-values computed using a
`z-score approximation (equivalent to the Cochran-Mantel-Haenszel
`test for nonzero correlation). The test was carried out using a
`permutation approach,
`in which 10,000,000 resamples without
`replacement will be drawn independently within centers. The
`calculations were carried out using SAS PROC MULTTEST.
`Results are also presented for the completer sample or
`observed cases (OC) data.
`
`After completing the efficacy analyses on the ITT sample,
`statistical analyses targeted a focal population of interest: patients
`with more substantial psychotic symptoms (n ⫽ 159). This group
`was defined a priori as patients having a BPRS PSS ⱖ 12 at
`baseline. The cutpoint of 12 was derived from a moderator
`analysis of efficacy data from a previous double blind PMD
`mifepristone trial which found that patients with a BPRS PSS ⱖ 12
`at baseline were more likely to have marked response. For this
`subset of the ITT sample (n ⫽ 159), proportions of rapid and
`sustained responders were again compared using Cochran-
`Mantel-Haenszel tests, adjusted for site. For exploratory pur-
`poses, analyses were conducted on a subset of patients who
`were asked, based on a FDA request, to complete a follow-up
`efficacy assessment at day 56. Chronologically, these patients
`were enrolled at the latter end of the study. All statistical tests
`were conducted using an alpha of .05 (two-tailed).
`
`Results
`A total of 221 patients (n ⫽ 221) were randomized and
`received at least one dose of study medication. The demograph-
`ics of the ITT sample are described in Table 1. A randomization
`check indicated no significant differences between groups at
`baseline on demographic, BRPS, and HAMD measures. There
`were no statistical differences between groups on the rate of
`antidepressant use after day 7 (58% mifepristone, 62% placebo)
`or in the rate of antipsychotic use after day 7 (36% mifepristone,
`42 % placebo), In addition, there was no significant difference
`between groups in the rate of ECT use after day 7 (2% on
`mifepristone, 3% on placebo) or the rate of combination treat-
`ment with both an antidepressant and antipsychotic after day 7
`(29% mifepristone, 37% placebo).
`
`Efficacy Analyses: Primary and Secondary Endpoints
`As shown in Table 2, patients in the mifepristone treated
`group were more likely to achieve the response criterion on the
`primary measure, a 30% improvement in the total BPRS (Rapid
`and Sustained Response and Response). This difference was
`statistically significant in the ITT sample (p ⫽ .041) and compl-
`
`Table 2. Primary and Secondary Endpoints: Response Status by Treatment Group
`
`BPRS Total
`Rapid and Sustained
`Response
`Non-Response
`
`BPRS PSS
`Rapid and Sustained
`Response
`Non-Response
`
`HamD
`Rapid and Sustained
`Response
`Non-Response
`
`Intent-to-Treat
`
`Observed Cases
`
`Mifepristone
`(n ⫽ 105)
`
`Placebo
`(n ⫽ 116)
`
`Mifepristone
`(n ⫽ 78)
`
`Placebo
`(n ⫽ 92)
`
`51 (48.6%)
`31 (29.1%)
`23 (21.9%)
`p valueb ⫽ .041
`
`50 (47.6%)
`23 (21.9%)
`32 (30.5%)
`p valueb ⫽ .006
`
`50 (47.6%)
`21 (20.0%)
`34 (32.4%)
`p valueb ⫽ .668
`
`49 (42.2%)
`23 (19.8%)
`44 (37.9%)
`
`40 (34.5%)
`17 (14.7%)
`59 (50.9%)
`
`56 (48.3%)
`15 (12.9%)
`45 (38.8%)
`
`40 (51.3%)
`18 (23.1%)
`20 (25.6%)
`p valueb ⫽ .020
`
`39 (50.0%)
`13 (16.7%)
`26 (33.3%)
`p valueb ⫽ .019
`
`37 (47.4%)
`22 (28.2%)
`19 (24.4%)
`p valueb ⫽ .546
`
`36 (39.1%)
`17 (18.5%)
`39 (42.4%)
`
`30 (32.6%)
`14 (15.2%)
`48 (52.2%)
`
`42 (45.7%)
`30 (32.6%)
`20 (21.7%)
`
`aRapid and Sustained Response, Response achieved at day 7 and sustained at day 28; Response, Response
`achieved by day 28; Non-Response, Response achieved at day 7 but not sustained at day 28, or not achieved at days 7
`and 28.
`bFrom a Cochran-Mantel-Haenszel test adjusted by pooled site.
`
`www.sobp.org/journal
`
`

`

`1346 BIOL PSYCHIATRY 2006;60:1343–1349
`
`C. DeBattista et al
`
`Table 3. Patients with Baseline PSS Scores ⬎ ⫽ 12 (n ⫽ 159): Response Status by Treatment Group
`
`Intent-to-Treat
`
`Observed Cases
`
`Mifepristone
`(n ⫽ 74)
`
`Placebo
`(n ⫽ 85)
`
`pa
`
`Mifepristone
`(n ⫽ 53)
`
`Placebo
`(n ⫽ 65)
`
`pa
`
`BPRS PSS
`Rapid and Sustained
`Response
`Non-response
`
`43 (58.1%)
`11 (14.9%)
`20 (27.0%)
`
`28 (32.9%)
`11 (12.9%)
`46 (54.1%)
`
`32 (60.4%)
`6 (11.3%)
`15 (28.3%)
`
`20 (30.8%)
`9 (13.8%)
`36 (55.5%)
`
`.001
`
`.003
`
`aRapid and Sustained Response, 50% reduction in BPRS PSS score achieved at day 7 and sustained at day 28;
`Response, 50% reduction in BPRS PSS score achieved by day 28; Non-Response, 50% reduction in BPRS PSS score
`achieved at day 7 but not sustained at day 28, or not achieved at days 7 and 28.
`bFrom a Cochran-Mantel-Haenszel test adjusted by pooled site.
`
`eter samples (p ⫽ .020). Mifepristone treated patients were more
`likely to achieve response on one of two secondary measures, a
`50% improvement in the BPRS PSS. This difference was statisti-
`in both the ITT (p ⫽ .006) and completer
`cally significant
`populations (p ⫽ .019). Both mifepristone and placebo treated
`patients had improvement on their HAMD scores, but
`the
`difference between groups on responder status was not statisti-
`cally significant (ITT: p ⫽ .668; OC: p ⫽ .546). Table 2 shows the
`proportion of rapid responders by treatment group for primary
`and secondary endpoints.
`
`Other Analyses
`After completing the efficacy analyses on the ITT sample,
`statistical analyses targeted a focal population of interest, defined
`a priori as patients having a BPRS PSS ⱖ 12 at baseline. As stated
`earlier, the cutpoint of 12 was derived from a moderator analysis
`of efficacy data from a previous double blind PMD mifepristone
`trial which found that patients with a BPRS PSS ⱖ 12 at baseline
`were more likely to have marked response. Of the 221 patients in
`the ITT sample, 159 patients had a BPRS PSS score ⱖ 12 at
`baseline (n ⫽ 159), indicating the presence of at least minimal
`psychotic symptoms. These patients did not differ from other
`patients in the ITT sample on baseline demographic variables
`(age: t ⫽ .17, df ⫽ 219, p ⫽ .87; gender: X2 ⫽ .02, df ⫽ 1, p ⫽
`.90). Among this a priori designated group of interest, differences
`between the treatment and placebo groups were larger in terms
`of effect size. As shown in Table 3, a greater percentage of
`patients receiving mifepristone had a rapid reduction in psy-
`chotic symptoms, measured by the BPRS PSS, by day 7 and
`sustained their response at day 28. The difference in responder
`rate was statistically significant for the ITT (p ⫽ .003) and
`observed cases samples (p ⫽ .001). The group receiving mife-
`pristone showed a significantly greater mean reduction in PSS
`scores at day 28 using both ITT (treatment: ⫺7.2 ⫾ 3.4; placebo:
`⫺5.2 ⫾ 4.4; t ⫽ ⫺3.2, df ⫽ 157, p ⫽ .001) and observed cases
`data (treatment: ⫺7.0 ⫾ 3.6; placebo: ⫺4.7 ⫾ 4.5; t ⫽ ⫺3.04,
`df ⫽ 121, p ⫽ .003).
`Among this target group of interest, forty-two participants
`(mifepristone: n ⫽ 19; placebo: n ⫽ 23) were observed at the
`FDA’s request; these patients were assessed 7 weeks after the
`end of study drug administration. Responder analysis of the day
`56 data indicated that a greater percentage of patients in the
`mifepristone group showed a 50% reduction in the PSS at days 7
`and 56. Fifty-three (53%) of patients receiving treatment and
`twenty-two percent (22%) of patients randomized to placebo
`responded by day 7 and sustained their response at day 56 (p ⫽
`.038). Mean change from baseline PSS scores at both day 7 and
`day 56 were significantly different across groups (Day 7 treat-
`
`www.sobp.org/journal
`
`ment ⫺6.1 ⫾ 3.4; placebo 4.7 ⫾ 4.2 df ⫽ 157, t ⫽ 2.21, (p ⫽
`.028); Day 56 treatment: ⫺8.1 ⫾ 3.5; placebo: ⫺4.9 ⫾ 3.6; t ⫽
`⫺2.86, df ⫽ 40, p ⫽ .007). There was also a trend favoring
`mifepristone on improvement in the HamD absolute change
`from baseline at day 56 (treatment ⫺20.2 ⫾ 11.3; placebo
`⫺13.0 ⫾ 12.0, df ⫽ 40, t ⫽ 1.96, (p ⫽ .056).
`
`Adverse Events
`tolerated with no AEs
`Mifepristone appeared to be well
`occurring at significantly higher rates than with placebo (see
`Table 4). While the rates of nausea, vomiting, rash and toothache
`trended higher in the mifepristone group, these differences did
`not achieve statistical significance. Likewise, the higher trended
`rates of constipation, dyspepsia, abdominal pain, and somno-
`lence in the placebo groups were not statistically different. The
`rate of serious adverse events in the placebo group was higher
`than in the mifepristone group but these differences were not
`statistically significant. The rate of withdrawal due to adverse
`events in both the placebo and drug treated was very low, with
`1 patient withdrawing from both the placebo and C-1073 treated
`groups secondary to AEs.
`
`Table 4. Adverse Events in Most Frequently Affected Body Systems (ⱖ 5%
`for Any Group)
`
`Mifepristone
`n (%)a
`
`Placebo
`n (%)
`
`Total
`n (%)
`
`pb
`
`Patients Studied
`Total Patients Studied
`Total Patients with TEAEs
`Body System Affected
`Headache NOS
`Nausea
`Vomiting NOS
`Constipation
`Dizziness
`Insomnia
`Sedation
`Abdominal pain NOS
`Rash NOS
`Abdominal pain upper
`Toothache
`
`105
`71 (67.6)
`
`116
`85 (73.3)
`
`221
`156 (70.6)
`
`17 (16.2)
`12 (11.4)
`10 (9.5)
`5 (4.8)
`6 (5.7)
`5 (4.8)
`7 (6.7)
`6 (5.7)
`6 (5.7)
`1 (1.0)
`6 (5.7)
`
`21 (18.1)
`7 (6.0)
`5 (4.3)
`12 (10.3)
`10 (8.6)
`6 (5.2)
`8 (6.9)
`4 (3.4)
`2 (1.7)
`6 (5.2)
`1 (0.9)
`
`38 (17.2)
`19 (8.6)
`15 (6.8)
`17 (7.7)
`16 (7.2)
`11 (5.0)
`15 (6.8)
`10 (4.5)
`8 (3.6)
`7 (3.2)
`7 (3.2)
`
`.73
`.23
`.18
`.14
`.45
`1.00
`1.00
`.52
`.16
`.12
`.06
`
`TEAE, treatment-emergent adverse event.
`a The denominator for the percentages is the total number of patients in
`each treatment.
`bFisher’s Exact.
`
`

`

`C. DeBattista et al
`
`Discussion
`
`Mifepristone appears to significantly reduce psychotic symp-
`toms in patients with PMD as measured by the BPRS. The effects
`of mifepristone were most evident on the positive symptom
`subscale, which assesses core psychotic symptoms including delu-
`sions, hallucinations, suspiciousness, and disorganized thinking. As
`previously seen in patients with PMD, the effects of mifepristone
`were seen in psychosis but no significant effects were seen in
`depression. There might be several reasons for a differential effect
`on psychotic
`symptoms
`relative to depressive symptoms.
`Schatzberg and colleagues (Schatzberg et al 1985) have postulated
`a corticosteroid-mediated abnormality in dopamine function in
`PMD. Thus, an antiglucocorticoid might relieve psychotic symptoms
`in PMD more than depressive symptoms. Another possibility is that
`depressive symptoms might be more sensitive to non-specific
`treatment effects, such as being in the hospital, or to concurrent
`therapy than psychotic symptoms (Stolk et al 2001). Thus, even
`though mifepristone might have had effects on depressive symp-
`toms, a high placebo response rate on depressive symptoms might
`have been difficult to overcome. Finally, it is possible that antipsy-
`chotic effect might occur early and antidepressant effects later.
`While there were no differences in the HamD between groups at
`days 7 or 28 in the secondary analysis, a post hoc analysis at day 56
`suggests a strong trend favoring mifepristone (p ⫽ .056). However,
`the post hoc nature of this analysis in a subset of the total sample
`that were evaluated at day 56 renders this finding inconclusive.
`Another notable finding of the study is the placebo response
`rate. Historically the placebo response rate in PMD was thought to
`be quite low with rates have ranging from 0 to 28 (Spiker and
`Kupfer 1988; Kocsis et al 1990; Schatzberg and Rothschild 1992).
`More recently, higher placebo responses have been observed in
`two PMD inpatient trials (DeBattista et al 2003; Rothschild et al
`2004). There are a number of possibilities for the relatively high
`placebo response rate in this trial. One is that hospitalization and
`concurrent medications contributed to a higher placebo response
`rate than has been historically reported. Despite the precautions
`taken, it is also conceivable that patients may have entered the trial
`with a diagnosis that has a higher placebo response rate than does
`PMD. However, the high baseline HamD and BPRS scores suggest
`that the population was quite ill. The most likely alternative diag-
`noses were depression without psychosis, schizoaffective disorder
`or schizophrenia. Placebo response in depression without psychotic
`features is inversely related to HamD scores and the high HamD
`scores in this population would predict a comparably lower placebo
`response rate (Khan et al 2002). Neither schizophrenia nor schizo-
`affective disorder is associated with high placebo response rates.
`Another possible contributor to the relatively high placebo response
`rate is that an open label extension study may have confounded
`results. Patient who had an adequate response to study drug in the
`current study were eligible to participate in an open label extension
`study if they relapsed. Thus, it is possible that the motivation to be
`assured of open label treatment in the extension study might have
`contributed in some way, to the higher response rate in the acute
`study.
`The effects of mifepristone on psychosis seemed to be evident
`both 3 weeks (day 28) and 7 weeks (day 56) after the drug was
`stopped. Studies employing antiglucocorticoid agents such as keto-
`conazole and metyrapone have reported that the clinical effects of
`these agents might persist for up to 8 months after the drug is
`stopped (Ghadirian 1995; Murphy 1998). The saturation of the GR II
`receptor for 7 days might also have effects on the HPA axis for a
`prolonged period. It is speculative but conceivable that an overac-
`
`BIOL PSYCHIATRY 2006;60:1343–1349 1347
`
`tive HPA axis might be reset by acute blockade of the GRII receptor
`(Belanoff et al 2002). The persistent effects of mifepristone on the
`treatment of psychosis are not adequately explained by antipsy-
`chotic use after day 7. An antipsychotic effect was evident by day 7
`when patients were not on antipsychotics. In addition, a minority of
`patients in both arms of the study (36% of mifepristone and 41% of
`placebo treated patients) were put on antipsychotics after day 7,
`predominately atypical antipsychotics such as olanzapine and ris-
`peridone. Since the majority of patients were not treated with
`antipsychotics, it seems unlikely that the difference between groups
`could be explained by concurrent medications.
`There are a number of limitations to this study. Among them is
`that antipsychotics were allowed after day 7. Perhaps the antipsy-
`chotic effects seen were primarily related to the use of concurrent
`antipsychotics. However, this is unlikely for at least 2 reasons. One,
`only a minority of patients were administered antipsychotics at any
`point in the study. In addition, placebo treated patients were
`numerically more likely to have received follow-on antipsychotics
`than were patients who received mifepristone. Another possibility is
`that the concurrent antidepressant use after day 7 contributed to
`antipsychotic effects in these PMD patients. While a few studies
`have suggested antidepressants might treat the entire syndrome of
`PMD (Gatti et al 1996; Zanardi et al 1997; Zanardi et al 2000), most
`studies have not found this to be true. In any case, the rate of
`antidepressant use was similar in both groups and thus would not
`adequately explain a difference in the BPRS score. Missing data
`were imputed by MMRM, and inherent to all imputation methods is
`the risk of misestimation. However, MMRM is considered advanta-
`geous to other available methods (e.g., Last Observation Carried
`Forward).
`Despite the limitations, this study suggests that there may be
`therapeutic benefits of mifepristone in the treatment of psychosis in
`psychotic major depression. If proven in subsequent trials, the
`targeting of a purported pathophysiology with a specific pharma-
`cotherapy would be largely unprecedented in psychiatry. The
`discovery of psychotropics to date has rested on serendipity and
`repetition of drugs with similar pharmacological profiles.
`Currently available therapies for PMD have significant limita-
`tions. While antipsychotics are effective, they are sometimes asso-
`ciated with extrapyramidal symptoms and potentially serious met-
`abolic effects. In addition, antipsychotics tend to work in a slow
`measured manner in the treatment of any psychosis. Likewise ECT,
`while effective, requires repeated treatments under general anes-
`thesia,
`is associated with cognitive side effects, and carries a
`significant stigma. An effective alternative to these treatments would
`be welcome.
`This study suggests that mifepristone may have clinically signif-
`icant antipsychotic effects in the treatment of PMD. Additional
`controlled trials are needed to replicate this effect. However, if
`replicated in additional studies, it may indicate that antiglucocorti-
`coid drugs might have important applications in the treatment of
`psychotic depression.
`
`This work was sponsored by Corcept Therapeutics, Menlo
`Park, California.
`Presented at the 30th annual meeting of the American College
`of Neuropsychopharmacology, San Juan, Puerto Rico, 2004.
`We acknowledge the following disclosures; CD: Speakers
`Bureau; Wyeth, Cephalon, Pfizer, GSK, Lilly, BMS, Cyberonics.
`Grant Support; Wyeth, GSK, Cephalon, Pritzker Foundation,
`NARSAD, NIMH, Neuronetics, Cyberonics. Consultant; Corcept
`Therapeutics, Wyeth, Lilly, Roche, BMS. Stock-holder; Corcept
`Therapeutics. JB: CEO and equity-holder; Corcept Therapeutics,
`
`www.sobp.org/journal
`
`

`

`1348 BIOL PSYCHIATRY 2006;60:1343–1349
`
`C. DeBattista et al
`
`Menlo Park, CA. CB: Statistical Consultant; Corcept Therapeu-
`tics, Menlo Park, CA. LL