ORIGINAL CONTRIBUTION
`
`Two 6-Week, Randomized, Double-Blind, Placebo-Controlled
`Studies of Ziprasidone in Outpatients With
`Bipolar | Depression
`Did Baseline Characteristics Impact Trial Outcome?
`
`llise Lombardo, MD,* Gary Sachs, MD,7% Sheela Kolluri, PhD,* Charlotte Kremer, MD,*
`and Ruoyong Yang, PhD*
`
`Abstract: Two randomized, double-blind, placebo-contralled, 6-week
`studies comparing ziprasidone versus placebo for treatment of bipo-
`lar depression (BPD) failed to meet their primary study objectives,
`indicating that either ziprasidone is ineffective in the treatment of BPD
`or the study failed.
`Adult outpatients with bipolar | depression with 17-1tem Hamilton
`Rating Scale for Depression total score more than 20 ai screening and
`baseline received either ziprasidone 40 to 80 mg/d, 120 to 160 mg/d, or
`placebo(study 1). or ziprasidone 40 to 160 mg/d or placebo (study 2).
`Primary efficacy measure in both studies was change from baseline in
`Montgomery-Asberg Depression Rating Scale total scores at week 6
`(end of the study). Mixed-model repeated-measures methodology was
`used to analyzethe primary efficacy measurein both studies. Secondary
`efficacy measures in both studies included Hamilton Rating Scale for
`Depression total score and Clinical Global Impression-Improvement
`score. Post hoc analyses were conducted for both studies to examine
`potential reasons for study failure. In both, ziprasidone treatment groups
`failed to separate statistically from placebo for change from baseline
`Montgomery-Asberg Depression Rating Scale score at week 6. Response
`rates were 49%, 53%, and 46%for placebo, ziprasidone 40 to 80 mg/d,
`and ziprasidone 120 to 160 mg/d, respectively (study 1), and 51%and
`53% for placebo and ziprasidone 40 to 160 mg/d, respectively(study 2).
`Ziprasidone 40 to 160 mg/d did not show superiority over placebo
`at week 6 in the treatment of BPD. Post hoc analyses revealed serious
`inconsistencies in subject rating that mayhavelimited the ability to detect
`a difference betweendrug andplacebo response. Rating reliability warrants
`further investigation to improve clinical trial methodology in psychiatry.
`Key Words:bipolar depression, atypical antipsychotic, placebo
`response
`
`(J Clin Psychopharmacol 2012:32: 470-478)
`
`monotherapy oras an adjunct to mood stabilizers.! > Other classes
`of medication have, however, demonstrated efficacy for BPD.'4*
`In 2003. Tohen et al? reported that olanzapine and the
`combination of olanzapine andfluoxetine (OFC) were superior
`to placebofor treatment of BPD, and the US Food and Drug
`Administration granted approval to OFC in December 2003.
`Interest in atypical antipsychotic medication as treatment for
`BPDfollowed this success in the hopesthat, as a class, atypical
`antipsychotics mightbe effective for the treatment of BPD,* 7
`but results from clinical
`trials have been mixed. Whereas
`quetiapine®” and OFC® have demonstrated efficacy for the
`treatment of BPD, bifeprunoex and aripiprazole failed to dem-
`onstrate superiority to placebo in 2 recent clinical trials based
`on the change in Montgomery-Asberg Depression Rating Scale
`(MADRS)score from baseline to the end of the study.*-? Studies
`of lamotrigine for BPD have also produced inconsistentresults.'°
`Placeboresponse is a common problem inclinicaltrials for
`psychiatric disorders.'! [In randomized trials for bipolar disorder,
`there has been a pronounced increase in placeboresponse during
`the last several years.'* Someinvestigators have suggested that a
`component of the apparent placebo response may be attributable
`to a phenomenonreferred to as baseline inflation, in which the
`baseline scores of subjects entering trial may be exaggerated so
`as to be above the threshold requiredfor study entry.'*
`Ziprasidone is an atypical antipsychotic that,
`like most
`commonly prescribed antidepressants, inhibits the reuptake of
`serotonin and norepinephrine. Giventhat several small studies
`supported the use of ziprasidone for BPD,'* '® the primary
`objective of the present studies was to comparetheefficacy of
`ziprasidone with placebo during a 6-week course of treatment
`in outpatients with bipolar I disorder. [n an effort to mitigate
`baseline inflation of the primary efficacy measure,the | 7-item
`Hamilton Rating Scale for Depression (HAM-D-17) was used
`to determine eligibility, and the MADRS was the primary
`Fe\eeced bipolar depression (BPD) is defined by a major de-
`measure of efficacy. Here, we describe the findings of the
`pressive episode inapatient with bipolar disorder. Episodes
`2 studies; in both, ziprasidone failed to separate statistically
`of BPDshare diagnostic criteria such as sadness, anxiety, guilt,
`from placebo for the change from baseline MADRSscore at
`anger, andsleep disturbances, with episodes of major depressive
`week 6. To better understand the outcome of the 2 present
`studies, we further examined the relationship between the
`disorder (Diagnostic and Statistical Manual ofMental Disorders,
`Fourth Edition [DSM-IV]). Despite the cross-sectional clinical
`HAM-D-17 and MADRSscores at screening and at baseline.
`similarities, BPD responds poorly to standard antidepressants as
`The concurrent use of 2 rating scales allowed for evaluation
`
`of the reliability of illness severity ratings and may provide
`insights applicable to broader clinical trial methodology.
`
`From “Pfizer Inc, New York, NY; (Massachusetts General Hospital, Boston;
`and {Harvard University, Cambridge, MA.
`Received February 22, 2010; accepted after revision January 9, 2012.
`Repnints;Ilise Lombardo, MD,PfizerInc, 235 East 42nd St, New York, NY10017
`(e-mail: ilisc.lombardo@pfizer.com).
`This stadywas fundedby Pfizer Inc. Editorial support was provided by Annic
`Neild, PhD, of PAREXEL and was funded byPfizer Inc.
`Copyright © 2012 by Lippincott Williams & Wilkins
`ISSN: 0271-0749
`DO! 10.1097/ICP.ObO 13e3 182S5cedeS
`
`MATERIALS AND METHODS
`
`Study Population
`Inclusion Criteria
`
`Subjects who met the following criteria were included in
`both studies: (1) men and womenaged 18 years orolder at the
`time of consent, with a primary diagnosis of bipolar I disorder,
`
`470 | www.psychopharmacology.com
`
`Journal of Clinical Psychopharmacology * Volume 32, Number 4, August 2012
`
`Copyright © 2012 Lippincott Williams & Wilkins, Unauthorized reproductionofthis article is prohibited.
`
`1
`
`Exhibit 2053
`Slayback v. Sumitomo
`IPR2020-01053
`
`Exhibit 2053
`Slayback v. Sumitomo
`IPR2020-01053
`
`

`

`fourna!ofClinical Psychopharmacology * Volume 32, Number 4, August 2012
`
`Ziprasidone in Bipolar | Depression
`
`most recent episode depressed, with or without rapid cycling,
`and without psychotic features, as defined in the DSM-/}-Text
`Revision (296.SX) and confirmed by the Mini International
`Neuropsychiatric Interview version 5.0.0.'7; (2)lifetime history
`of at least 1 bipolar manic or mixed-manic episode (the initial
`protocol required atleast | lifetime hospitalization for a bipolar
`manic or mixed-mani¢ episode; this requirement was dropped
`in May 2007);
`(3) HAM-D-17 total score more than 20 at
`screening and at baseline (HAM-D-17 score was derived from
`the first 17 items of the HAM-D-25'%), obtained at least 3 days
`apart, and screening-to-baseline improvement in HAM-D-17
`total score less than 25%; (4) Young Mania Rating Scale
`(YMRS'") score less than 12 at screening and at baseline,
`Study 2 (A1281139, February 2006-March 2008)
`obtained at least 3 days apart; duration of the current bipolar
`Subjects were randomly assigned toa ziprasidone flexible-
`I disorder depressive episode of more than 2 weeks and less
`than 6 months.
`dose treatment group or placebo ina|:1 ratio as follows:
`* Ziprasidone flexible-dose treatment group: subjects were
`started at 20 mg bid fixed dose on days 1 and 2, 40 mg bid on
`days 3 to 6, and flexible dosing starting on day7 (ie, 20-80 mg
`bid. adjustable by 20 mg bid at each visit) for the remainder
`ofthe 6-week study.
`* Placebo: subjects were given placebo with the same flexible
`dosing schedule as ziprasidone for the entire 6-week study.
`
`Ziprasidone low-dose (40-80 mg/d): subjects started dosing
`at 20 mg bid on days | to 6, thenflexible dosing started on day
`7 (20-40 mg bid [20 mg bid or 40 mg bid at the discretion of
`the investigator]) for the remainder of the 6-week study.
`Ziprasidone high-dose (120-160 mg/d): subjects started at
`20 mgbid on days | to 2, then 40 mg bid on days3 to 4, then
`60 mg bid on days5 to6, then flexible dosing started on day 7
`(60-80 mg bid [60 mg bid or 80 mg bid at the discretion of
`the investigator]) for the remainderof the 6-week study.
`* Placebo: subjects were given placebo with the sameflexible
`dosing schedule as ziprasidone for the entire 6-week study.
`
`Concomitant Medication
`Foragitation or intolerable anxiety. lorazepam up to 2 mg/d
`was allowed during the screening period and the first 2 weeks
`of double-blind treatment up to 4 days per week. For insomnia,
`nonbenzodiazepine sleep agents (all approved agents, eg, zol-
`pidemup to 10 mg/d, eszopiclone up to 3 mg/d, zaleplon up to
`20 mg/d, or ramelteon up to 8 mg/d) were allowed during the
`screening period and the first 2 weeks of double-blind treat-
`ment up to 4 days per week and for the remainderof the study
`up to 2 days per week. Benztropine (up to 6 mg/d) for extra-
`pyramidal symptoms and propranolol (up to 120 mg/d) for
`akathisia were allowed only on an as-needed basis and not on
`a continuous daily basis prophylactically to treat extrapyramidal
`symptoms/akathisia. These medications were not allowed within
`the 12 hours before cognitive testing. All other psychoactive
`medications were prohibited during the subject’s participation
`in the study.
`
`Efficacy, Safety, and Post Hoc Analyses
`Efficacy
`The primary efficacy measure in both studies was the
`change in MADRStotal score”? from baseline to week 6 (end
`of the study). Response on the MADRSscale was defined as
`a 50% or greater reduction from baseline in the MADRS to-
`tal score. Secondary efficacy measures included baseline and
`postbaseline measurement ofthe range ofdepressive symptoms
`(using HAM-Dscore), anxiety (HAM-A score), mania (using
`YMRS), global clinical severity, and global improvement of
`symptoms (via Clinical Global Impression [CGI] of Severity
`and CGI of Improvementscores, respectively); global assess-
`ment of functioning; change in quality of life, enjoyment, or
`satisfaction; occupational/psychosocial impact of symptoms;
`and cognition.
`
`Safety
`Safety and tolerability assessments included AEs, vital
`signs, laboratory tests, serum prolactin, and weight. Movement
`disorder symptoms were measured using the Simpson-Angus
`Scale (SAS).?! the Barnes Akathisia Scale (BAS),*? and the
`Abnormal Involuntary Movement Scale (AIMS).
`
`Exclusion Criteria
`
`The following subjects were excluded from bothstudies: (1)
`subjects diagnosed with schizophrenia or schizoaffective disor-
`der, schizophreniform disorder, delusional disorder, or psychotic
`disorders not otherwise specified; (2) subjects who failed 3 or
`more adequate studies (more than 4 weeks at an adequate dose)
`of an antidepressant either as monotherapy or in combination
`therapy (with lithium or an anticonvulsant) in a previous de-
`pressive episode or within the current episode; (3) subjects with
`psychotic features associated with bipolar I depression within the
`index (ie, current) episode; (4) subjects with ultrafast rapid cy-
`cling (defined as 8 or more moodepisodes during the 12-month
`period preceding the screening visit); (5) subjects with YMRS
`score more than 16 at screening or at baseline were discontinued
`from the study and provided with appropriate treatment or re-
`ferral by the investigator; (6) subjects with a YMRSscore greater
`than or equal to 16 at any postbaseline visit; (7) subjects with
`DSM-IV-Text Revision—defined alcohol or psychoactive sub-
`stance abuse in the 3-month period preceding the screening visit
`or significantrisk of self-injurious/suicidal or violent/homicidal
`behavior; (8) subjects with a history of inadequate response to
`ziprasidone(at least 6 weeks’ duration) for the treatment of BPD
`or a history ofintolerance to ziprasidone; (9) subjects who had
`ever been discontinued from ziprasidone treatment because of
`lack ofefficacy or significant adverse events (AEs).
`In addition, subjects were required to have discontinued
`use of previous psychotropic agents (including anticonvulsants)
`for a minimumof | week; lithium for a minimum of 2 weeks;
`monoamine oxidase inhibitors, fluoxetine, or the OFC for a
`minimum of 4 weeks; and any depot neuroleptic agent for a
`minimum of 6 months before being randomized into the study.
`Women of childbearing age agreed to use birth control. All
`subjects provided written informed consent.
`Study Design
`Studies | and 2 were 6-week, randomized, double-blind,
`multicenter, flexible-dose, placebo-controlled studies conducted
`in the United States evaluating the efficacy and safety of oral
`ziprasidone in outpatient subjects aged 18 years and older with
`bipolar I disorder. The first study recruited participants from
`56 of 70 investigational sites in 25 states, whereas the second
`recruited at 45 sites from a total of 48 sites in 22 states. Fifteen
`states contributed to both trials.
`
`Study 1 (A1281136, July 2005-February 2008)
`Subjects were randomly assigned to a ziprasidone fixed-
`flexible dosing group (20-40 mg twice daily [bid] or 60-80 mg
`bid) or placebo in a 1:1:1 ratio as follows:
`
`© 2012 Lippincott Williams & Wilkins
`
`www.psychopharmacology.com | 471
`
`Copyright © 2012 Lippincott Williams & Wilkins, Unauthorized reproductionofthis article is prohibited.
`
`2
`
`

`

`Lombardo et a!
`
`Journal of Clinical Psychopharmacology * Volume 32, Number 4, August 2012
`
`Post Hoc Analyses
`Post hoc analyses were performed for both studies to ob-
`tain a better understanding ofpotential reasonsfor study failure
`such as high placebo response and rating inconsistencies. Spe-
`cifically, comparisons were made between the MADRSscores
`and HAM-D-17 scores at baseline and between MADRSactual
`scores and predicted MADRSscores (ie, MADRS scores derived
`from HAM-Dscoresusing the formula developed by Zimmerman
`et al). Subgroup analyses to study the influence of baseline
`illness severity (as measured by the MADRSscore) were also
`performed.
`Statistical Analysis
`The safety population for both studies included all ran-
`domized subjects who were administered at least
`1 dose of
`double-blind study medication. The intent-to-treat (ITT) pop-
`ulation for both studies included all subjects included in the
`safety population and for whomatleast | postbaseline primary
`efficacy evaluation was obtained. The primary efficacyanalysis
`in both studies used the ITT population.
`
`Efficacy
`Similar mixed-mode] repeated-measures (MMRM) analy-
`ses were used for the primary efficacy evaluation in both studies.
`The primary comparisons of interest in both studies were the
`mean changes from baseline to week 6 in MADRSscore between
`ziprasidone and placebo. In study 1, the specific treatment com-
`parisons of interest were ziprasidone 120 to 160 mg/d versus
`placebo and ziprasidone 40 and 80 mg/d versus placebo. The
`primary analysis was based on the ITT population using ob-
`served cases (OCs) data. The Hochberg procedure for adjusting
`for multiple treatment comparisons was used only for the change
`frombaseline MADRSscore (at each time point) in the MMRM
`analysis only. The MMRM model in both studies included fixed
`categorical effects of treatment, rapid cycling, center, visit, pre-
`vious hospitalization status (with or without previous bipolar
`manic or mixed-manic episode hospitalization), and treatment-by-
`visit interaction, as well as fixed continuous effect of baseline
`MMRS total score in the model. The subject effect was included
`as a randomeffect. The restricted maximum likelihood estima-
`tion method was used for the MMRManalysis with a sandwich
`estimator of the variance-covariance matrix of the fixed-effects
`parameters. An unstructured variance-covariance matrix was used.
`The assumptions of the MMRM analyses were evaluated. In ad-
`dition, changes from baseline in MADRStotal score at each visit
`week(last observation carried forward [LOCF] at week 6 and OC
`data at each visit week) were analyzed with an analysis of co-
`variance (ANCOVA) modelthat included the following model
`terms: treatment, rapid cycling, center, previous hospitalization
`status, and baseline score as a covariate.
`For the secondary efficacy evaluations in both studies, the
`MMRM model described above for the MADRStotal score was
`applied to the change from baseline in CGIof severity and CGI
`of improvement scores. Change from baseline in HAM-D-17
`at each visit week was analyzed using the ANCOVA model
`described above for the MADRStotal score.
`The Cochran-Mantel-Haenszeltest stratified by study center
`and rapid cycling strata was used in both studies to compare re-
`sponse rates between ziprasidone and placebo, where response
`based on the MADRS scales was defined as a more than 50%
`reduction from the baseline MADRStotal score.
`
`Safety
`Standard safety summaries of AEs, vital signs, laboratory
`tests. serumprolactin, and weight were generated, Analysis of the
`
`change from baseline to the end of the study in SAS total score,
`BASglobalclinical assessment of akathisia, AIMS total score,
`AIMSglobal severity score, and AIMSincapacitation score was
`performed using the same ANCOVA model described above in
`the analysis of the MADRStotal! score.
`
`Post Hoc Analyses
`Additionalpost hoc analyses were conducted for both studies
`to outline possible reasons for study failure. These included:
`(1) Comparison of the distribution of MADRStotal scores and
`HAM-D-17 scores at baseline using graphical displays of
`MADRS total scores and the HAM-D-17 scores at baseline
`with the inclusion cutoff highlighted for both scales.
`(2) Analysis comparing the actual and predicted (ie, derived from
`HAM-D-17) MADRStotal scores at both baseline and the end
`ofthe study (using LOCF data) were performed as a measure of
`rating reliability. Specifically. predicted MADRStotal scores
`were calculated from the HAM-D-17 total scores using the
`formula developed by Zimmermanet aF?(as MADRSpredicted
`total score = 1.43 x HAM-D-17 total score + 0.87). A sum-
`mary of the divergence (calculated for each subject as actual
`MADRStotal score — predicted MADRS total score) was
`reported at baseline and at the end of the study.
`(3) To studythe influence ofbaseline illness severity, the primary
`MMRMefficacy analysis was repeated within subgroups
`based on baseline MADRSscore categories. The protocoleli-
`gibility criterion called for a score ofleast 20 on the HAM-D-17
`at baseline, which correspondsto a predicted MADRSscore
`greater than 29.5. Hence, the subgroups cligible (>29.5)
`versus ineligible (<29.5) were created for this analysis. Of
`note is that, as the conversion calculation gave a cutoff score
`of 29.5, and it is not possible for MADRSscore to be other
`than an integer, the criterion of 29 or less was used.
`(4) To characterize placebo response by site, graphical displays
`showing placebo responserates at the end of the study for
`each site (for sites with at least 10 subjects enrolled) are
`presented.
`
`RESULTS
`Oftreated subjects, a total of 102 (61.8%)of 165, 91 (53.2%)
`of 171, and 111 (66.1%) of 168 subjects in the ziprasidone 40- to
`80-me/d group, ziprasidone | 20- to 160-mg/d group. and placebo
`group,respectively, completed study |. For study 2, of the treated
`subjects a total of 112 (60.5%) of 185 and 134 (68.4%) of 196
`ziprasidone and placebo subjects, respectively, were completers.
`The overal] mean daily dose of ziprasidone for study | was 113.1
`(+27.2) mg/d for the higher dose group and 53.9 (+15.3) mg/d for
`the lower dose group; for study 2, the overall mean daily dose of
`ziprasidone was 83.9 (+29.6) mg/d. Ofthe treated subjects, the
`proportion of study entrants hospitalized for mania did not dif-
`fer significantly between groupsin study| (range, $2.5%-84.8%)
`and in study 2 (range, 80.6%-85.4%). In study |, benzodiazepine
`(lorazepam) usage was reported by 10.9%, 8.2%. and 8.9% in the
`40- to 80-mg/d, 120- to 160-mg/d. and placebo groups, respec-
`tively. In study 2, benzodiazepine usage was reported by 10.3%
`and 6.6% in the ziprasidone and placebo groups, respectively.
`
`Primary Efficacy Analysis
`The primary efficacy analysis (MMRM) indicated that
`both the high- and low-dose ziprasidone groups in study | and
`the ziprasidone group in study 2 failed to demonstrate statistical
`superiority over placebo in change from baseline MADRSscore
`at week 6 (Fig. 1). In both studies, the results from the ANCOVA
`
`472 | www.psychopharmacology.com
`
`©2012 Lippincott Williams & Wilkins
`
`Copyright © 2012 Lippincott Williams & Wilkins, Unauthorized reproductionofthis article is prohibited.
`
`3
`
`

`

`fourna!ofClinical Psychopharmacology * Volume 32, Number 4, August 2012
`
`Ziprasidone in Bipolar | Depression
`
`analyses of week 6 data (both LOCF and OCdata) were generally
`consistent with the primary MMRManalysis results.
`Secondary Efficacy Analysis
`In both studies, MADRS response rates (=50% improve-
`ment from baseline MADRSscores) were similar to placebo,
`ranging from 46% to 53% of subjects (Table 1). In study 1, re-
`sponse rates at the end ofthe study for subjects indicated by the
`MADRSscores were 52.5%, 45.8%, and 49.4% for lower dose
`ziprasidone, higher dose ziprasidone, and placebo, respectively.
`In study 2, response rates at the end of the study were 52.8%
`for ziprasidone subjects and 51.1% for placebo subjects. The
`ziprasidone groups did not demonstrate a statistically significant
`difference over placeboin response rates in either study.
`Results of the ANCOVAanalysis ofthe secondary efficacy
`end point, change from baseline in the HAM-D-17 total score,
`showed nosignificant difference between ziprasidone and pla-
`cebo in both studies (unadjusted P > 0.05 for all comparisons
`between ziprasidone and placebo). In study 1, the least squares
`(LS) mean (SE) for change from baseline to the end ofthe study
`(OC) were —10.5 (0.9) (n = 151), —11.5 (0.95) (n = 150) in the
`
` > @- Zipresidone (120-160 mad)
`
`~t~ Ziprasidone (40-80 mg/d
`
`Weeks
`6
`§
`a
`2
`1
`3
`ppeiccnnnsnnscnamnencanbuccssnsnasascasdasneenioinasescatesoenessocedicnisansiomsusesiensalosaiomstadsseniscnsnAcaccadsemennsssnseea
`
`BS
`
`tudy 2¢
`0
`ed
`
`—@ Ziprasidone ($0-160 mg/d)
`ie Placebo
`
`.
`
`‘
`2 “
`iw
`Q 44
`Se!
`id
`£84
`&
`|
`5 ~10-
`&
`i
`“124
`&
`i
`= -14+
`~16 4
`
`
`
`ziprasidone high- and low-dose groups, respectively, and —10.6
`(0.95) (n = 153) in the placebo group. In study 2, the LS mean
`(SE) for change from baseline to the end of the study (OC) were
`—6.9 (1.4) (n= 168) and —7.1 (1.3) (n= 181) in the ziprasidone
`and placebo groups. respectively. Response rates based on the
`HAM-D-17 total score (response defined as 250% reduction
`from baseline HAM-D-17 total score) also showed no signif-
`icant difference between ziprasidone and placebo groups in
`both studies (nominal P > 0.05 for all comparisons between
`ziprasidone and placebo).
`
`Safety and Tolerability
`In study 1. the most frequently reported treatment-emergent
`AEs(all causalities) in the higher dose ziprasidone group (at
`twice the rate of placebo) were somnolence (17.5%) and sedation
`(11.7%). In the lower dose ziprasidone group, the most frequently
`reported AE was somnolence (15.2%). In study 2, among sub-
`jects randomized to the ziprasidone group, the 3 mostfrequently
`reported AEs were somnolence (13.5%), sedation (11.9%). and
`headache (11.4%) compared with nausea and headache (each
`10.7%) and diarrhea (7.7%) for subjects in the placebo group. In
`both studies, mean changesin vital sign values, body mass index,
`weight, and waist circumference were similar among treatment
`groups. Clinically significant weight gain or loss was not com-
`monly observed. Vital signs among the treatment groups did not
`change appreciably frombaseline to the end of the study.
`For both studies, changes from baseline across treatment
`groups and across movement scales were very small and not
`clinically relevant, although some differences did reach statis-
`tical significance. In study 1, significant changes from baseline
`were observedat the end ofthe study for the comparison between
`the ziprasidone higher dose treatment group and placebo group
`for SAS total score (nominal P = 0.0277). In study 2, the LS
`mean change (SE) from baseline to the end of the study in SAS
`total score was —0.07 (0.08) and —0.23 (0.07)in the ziprasidone
`and placebo groups, respectively; this difference was significant
`(nominal P = 0.0174). The LS mean change (SE) from baseline
`to the end of the study in BAStotal score was 0.08 (0.17) and
`—0.37 (0.16) in the ziprasidone and placebo groups, respec-
`tively; this difference was significant (nominal P = 0.0033). The
`LS mean change (SE) from baseline to the end of the study
`in AIMStotal score was 0.01 (0.09) and —0.00 (0.08) in the
`ziprasidone and placebo groups, respectively;this difference was
`notsignificant (nominal P = 0.8399).
`Post Hoc Analyses
`Results of the post hoc analyses conducted for both studies
`to examine potential reasons for study failure are described below.
`
`Distribution of HAM-D-17 and MADRS
`Scores at Baseline
`In both studies, baseline HAM-D-17 scores determined
`subject inclusion, buta distribution of these scores does notfully
`correspond to baseline MADRS scores—a similar measure of
`depression and the primary efficacy measure in both studies.
`Figures 2 and 3 illustrate the distribution of actual HAM-D-17
`and MADRSscores,respectively, at baseline. Whereas HAM-D-
`17 scores show the inclusion of appropriate study subjects,
`MADRSscores suggest the inclusion of many individuals with
`depression oflesser severity than was required by study inclusion
`criteria.
`The actual baseline MADRS scores of 29 or less show
`that most ofthe subjects in both studies (52.9% of486 in study 1
`and 50.5% of 370 in study 2) had scores below the threshold
`considered to be the minimal severity threshold required for
`
`FIGURE 1. Primary efficacy analysis* comparing ziprasidone
`versus placebo (intent-to-treat [ITT] population, observed cases).
`“The mixed-model repeated-measures (MMRM)with model
`terms: treatment, rapid cycling, center, visit, previous
`hospitalization status, treatmentby visit interaction, and_
`baseline as covariate. °P < 0.05. “Baseline Montgomery-Asberg
`Depression Rating Scale (MADRS) scores were 27.1
`(ziprasidone 120-160 mg/d), 28.7 (ziprasidone 40-80 mg/d),
`and 28.9 (placebo). “Baseline MADRS scores were 28.6
`(ziprasidone 40-160 mg/d) and 28.2 (placebo).
`Week 6 results represent the primary efficacy analysis.
`
`© 2012 Lippincott Williams & Wilkins
`
`www.psychopharmacology.com | 473
`
`Copyright © 2012 Lippincott Williams & Wilkins, Unauthorized reproductionofthis article is prohibited.
`
`4
`
`

`

`Lombardo et a!
`
`Journal of Clinical Psychopharmacology * Volume 32, Number 4, August 2012
`
`TABLE 1. Results From Clinical Studies of Atypical Antipsychotics for the Treatment of Bipolar | Depression
`
`Study
`Study |
`Placebo
`Ziprasidone 40-80 mg/d
`Ziprasidone 120-160 mg/d
`Study 2
`Placebo
`Ziprasidone 40-160 mg/d
`Thase et al, 20067
`Placebo
`Quetiapine 300 mg/d
`Quetiapine 600 mg/d
`Calabrese et al, 2005®
`Placebo
`Quetiapine 300 mg/d
`Quetiapine 600 mg/d
`Tohen et al*
`Placebo
`Olanzapine/fluoxetine
`Olanzapine
`Thase ct al (study 1)”?5
`Placebo
`Aripiprazole
`Thaseet al (study 2)?*>
`176
`44
`56/188 (29.8)
`Placebo
`175
`45
`77/187 (41.2)
`Aripiprazole
`*50% decrease in Montgomery-Asberg Depression Rating Scale (MADRS) score from baseline. SE data missing, when unavailable.
`LSindicates least squares; SE, standard error.
`
`Change in MADRSScore at Last
`Assessment
`
`n
`
`162
`158
`166
`
`190
`180
`
`161
`155
`151
`
`169
`172
`170
`
`355
`82
`351
`
`177
`162
`
`LS Mean (SE)
`
`—13.3 (1.0)
`~14.8 (0.97)
`— 13.8 (1.0)
`
`—13.2 (0.9)
`—14.9 (1.0)
`
`~—11.9 (0.99)
`— 16.9 (0.99)
`—16.0 (1.01)
`
`-103
`~16.4
`—16.7
`
`—11,9 (0.8)
`—18.5 (1.3)
`—15.0 (0.7)
`
`~10.6
`-11.9
`
`11.5
`7Az3
`
`Discontinuation Rate
`(Discontinued/Randomized), %
`
`Response*
`Rate, %
`
`Duration, wk
`6
`
`57/174 (32.8)
`63/176 (35.8)
`80/186 (43.0)
`
`62/200 (31.0)
`73/192 (38.0)
`
`58/168 (34.5)
`71/172 (41.3)
`79/169 (46.7)
`
`74/181 (40.9)
`60/181 (33.1)
`82/180 (45.5)
`
`232/377 (61.5)
`31/86 (36.0)
`191/370 (51.6)
`
`66/188 (35.1)
`87/186 (46.8)
`
`49
`53
`46
`
`51
`53
`
`45
`60
`58
`
`36
`58
`58
`
`30
`56
`39
`
`39
`43
`
`6
`
`8
`
`8g
`
`8
`
`8
`
`g
`
`study enrollment. Furthermore, 12 (3%, study 1) and 19 (5%,
`study 2) subjects at baseline would be considered in remission
`at baseline according to their MADRS scores (MADRSscores
`<12). At the time of last observation, 98 (20%) and 54 (15%)
`subjects had MADRSscoresof4 or less in study | and study 2,
`respectively, including 28 and 12 subjects, respectively, with an
`MADRSscore of 0).
`
`Comparison of Actual and Predicted MADRSScore at
`Both Baseline and at the End of the Study
`The mean (=SD) and median (minimum, maximum) of the
`divergence between actual and predicted MADRS scores ob-
`served at baseline were —7.90 (+ 5.52) and —8.09 (—28.63,
`13.54), and —8.63 (+ 6.18) and —8.32 (—34.34, 7.1), for studies
`| and 2, respectively.
`In a quarter of subjects,
`the predicted
`MADRSscore was more than [1 and 12.3 points greater than
`the actual MADRSscore in studies | and 2. respectively. In 10%
`of subjects, the derived MADRSscore was more than 15 and
`16.6 points greater than the actual MADRS score in studies | and
`2, respectively. At the end of the study, the mean and median
`divergences were markedly less thanat baseline, at 4.67 (+ 4.99)
`and —4.3 (23.33, 9.84) for study |, and —4.29 (+ 5.69) and
`—4.31 (-22.04, 21.11) for study 2. In a quarter of the subjects,
`the predicted MADRSscore was more than 7.8 (study 1) and
`7.9 (study 2) points greater than the actual MADRSscore; in
`10% of subjects, the predicted MADRS score was more than
`11.3 (study 1) and 12.1 (study 2) points greater than the actual
`MADRSscore.
`
`Influence of Baseline IlIness Severity
`Results of the primary efficacy MMRM analysis repeated
`for cach of the 2 subgroups(ineligible vs eligible based on
`baseline MADRStotal scores <29.5 and >29.5) are presented in
`Table 2. Not unexpectedly, subjects with lower baseline MADRS
`scores experienced less change during the courseofthe study than
`subjects with higher MADRSscores. Results in the 2 subgroups
`based on baseline MADRSscores were consistent with the re-
`sults from the primary efficacy MMRM analysis. In study |, the
`placeboresponse rate in the ineligible group was greater than the
`proportion in the eligible group (57.1% vs 41.0%); in study 2,
`however, the placebo response rate in the eligible group was
`greater than that in the ineligible group (54.8% vs 47.4%). For
`both studies, there were no meaningful differences between the
`response rates for ziprasidone subjects in the ineligible group
`versus the eligible group.
`
`Placebo Response
`Among the 21 sites in study | that had at least 10 sub-
`jects, placebo responserates greater than 40% were observed in
`14 sites (66.7%); and among the 15 sites in study 2 that had at
`least 10 subjects, placebo response rates greater than 40% were
`observed in 13 sites (86.7%) (Fig. 4, A and B).
`
`DISCUSSION
`Randomized clinical trials can generate positive or negative
`results or they can fail to provide meaningful results. Positive
`
`474 | www.psychopharmacology.com
`
`©2012 Lippincott Williams & Wilkins
`
`Copyright © 2012 Lippincott Williams & Wilkins, Unauthorized reproductionofthis article is prohibited.
`
`

`

`fourna!ofClinical Psychopharmacology * Volume 32, Number 4, August 2012
`
`Ziprasidone in Bipolar | Depression
`
`Study t
`24
`
`4
`
`
`
`teeepeeeeepc
`ct2saascs 827 6@
`
`& WM 2 HAM-D-17 Baseline Scores:
`cdrenehrenserernsheececcsehsreeessomsBenostei24ii
`=~weevscsederemessrebsemeendAD
`ctS&
`o
`
`Percert
`
`oa
`
`deeseremed,
`
`
` oyeeae se ae ~ ’ ? , Poe,
`
`
`
`
`
`QL FF 4 FH 7 FF OMUMIMH WUT RS UWBRAABWzsZMBRMMUMNURHAKNBDMDIBWBA
`HAMM-D-17 Baseline Scores
`FIGURE 2. Distribution of 17-item Hamilton Rating Scale for Depression (HAM-D-17) scores at baseline.
`
`results require a statistically significant difference between the
`test conditions. Negative results refer to outcomes in which the
`test drug is not significantly different from placebo in a study
`with demonstrated assay sensitivity (eg, results indicating that a
`comparator drug of knownorestablished benefit is significantly
`
`more effective than placebo but the test drug is not). A failed
`study maybe declared in the absence of assay sensitivity (eg.
`when both the test drug and the comparator failed to separate
`from placebo). In the absence ofa known active comparator,it is
`not possibleto distinguish between a negative study and a failed
`
`AS
`
`tudy 4
`12
`
`Percent
`
`
`
`Orz2s4+367
`
`27 B2WIDI VK WH WI WB 41 62 43 te 4546 47 45 4955
`SAADRS Basetine Scores
`
`o e
`
`16 17 16 19-20 28 27 23.24 25.25 27 28 29.20.31
`tAADAS Baseline Scores
`.
`FIGURE 3. Distribution of Montgomery-Asberg Depression Rating Scale (MADRS)scoresat