`for Acute Treatment of Depression in Patients With Bipolar I Disorder:
`A Randomized, Double-Blind, Placebo-Controlled Trial
`
`Gary S. Sachs, MD; Kathleen S. Ice, PhD; Phillip B. Chappell, MBA, MD;Jeffrey H. Schwartz, PhD;
`Oksana Gurtovaya, PhD; Douglas G. Vanderburg, MPH, MD; and Bryce Kasuba, MA, MBA
`
`ABSTRACT
`
`Objective: To assess efficacy and safety of adjunctive
`ziprasidone in subjects with bipolar depression treated
`with lithium, lamotrigine, or valproate.
`Method: 298 adult outpatients with bipolar|
`disorder (OSM-1Y criteria) were randomized to
`receive ziprasidone, 20-80 mg twice a day, or placebo
`twice a day for 6 weeks plus their preexisting mood
`stabilizer. The primary efficacy variable was change
`in Montgomery-Asberg Depression Rating Scale
`(MADRS) total scores from baseline to 6 weeks. The
`key secondaryefficacy endpoint was change from
`baseline to week 6 in Clinical Global Impressions
`Severity (CGES) scores. Computer-administered
`assessments for diagnostic confidence were included
`for quality control and to evaluate study performance.
`The study was conducted between October 2007 and
`December 2008.
`
`Results: The mean +SD daily dose of ziprasidone was
`89.84 29.1 mg. Least squares mean + standard error
`changes from baseline to week 6 on MADRStotal
`score for ziprasidone and placebo treatment groups
`were ~ 13.24 1.2 and 12.9 1.1, respectively, with
`a 2-sided ? value of .792. There was no significant
`difference on the key secondaryvariable (CGI-S}.
`Adjunctive ziprasidone was well tolerated. Poor
`quality ratings et baseline were associated with
`a trend for better improvement on placebo than
`Ziprasidone. Among 43 placebo-tieated subjects
`with poor baseline quality ratings, 29 (67.4%) had
`baseline MADRSscores > 10 points higher on the
`computer-administered assessment than the MADRS
`administered by the site-based rater. The response
`favoring placebo over ziprasidone observed in this
`subgroup suggesis that poor signal detection in
`some clinical trials can be a consequence of “subject
`inflation’ as well as “raterinflation”
`
`Conclusions: Adjunctive ziprasidone treatrnent failed
`to separate from moodstabilizer alone on primary and
`secondary endpoints. Possible contributionsto this
`result include enrollment of a substantial numberof
`subjects with low diagnostic confidence, low quality
`ratings on the MADRS, and overzealous reporting of
`symptoms bysubjects.
`Trial Registration:clinical trials.qoyIdentifier:
`NCT004 83548
`
`J Clin Psychiatry 2011;72(10):1413~1422
`© Copyright 2011 Physicians Postgraduate Press, Ine.
`
`
`Submitted: December 22, 2009; accepted August | 9, 2010.
`
`Online ahead ofprint: May 3, 2011 (dot: 10.4088/KP09M05934).
`Corresponding author: Gary S. Sachs, MD. Bipolar
`Clinic ond
`Research Program, Massachusetts GenerelHospital, 50 Staniford
`St, 5th Floor, Boston MA 02174 SachsG@acl.con}).
`
`B ipolar I disorder is a common complex, chronic illness that is
`
`associated with considerable functional impairment.' This
`dynamic, pleomorphic disorder challenges researchers as well as cli-
`nicians and, as a consequence,relatively little high quality data are
`available to guide clinical practice. The managementof depressionin
`patients with bipolar | disorder remains an area of significant unmet
`need.’ In the Systematic Treatment Enhancement Program for Bipo-
`lar Disorder (STEP-BD), subjects with bipolar disorder experienced
`high rates of depressive relapse despite maintenance treatment with
`lithium, valproate, or other US Food and Drug Administration (FDA)-
`approved antimanic agents.’ In view of the unmet need for adjunctive
`treatments for patients suffering from bipolar depression despite pre-
`scribed maintenance treatment at dosages considered adequate, we
`undertook a study of adjunctive ziprasidone.
`Like other agents classified as atypical antipsychotics, ziprasidone
`is a dopamine D, and 5-HT>2, antagonist and interacts with numerous
`other receptors. Ziprasidone shows agonist activity at 5-HT |, receptors
`and antagonistactivity at 5-HT), and 5-HTp receptors. The affinity
`of ziprasidone for 5-HT |p receptors, and its serotonin-norepinephrine
`reuptake inhibition, is comparable to that of the tricyclic antidepres-
`sant imipramine, and provides a rationale for studying ziprasidone
`as an antidepressant.4 Data from prior small, open studies suggest
`that ziprasidone may reduce depressive symptomsassociated with
`bipolar I disorder.>-7
`Only 2 treatments have FDA approval for treatment of bipolar
`depression:the atypical antipsychotics quetiapine® and olanzapine-
`fluoxetine combination’ have demonstrated more efficacy than
`placeboin reducing depressive symptoms in patients with bipolar| dis-
`order. However, both drugs are associated with undesirable metabolic
`effects such as weight gain and disturbances of glucose homeosta-
`sis.!°"! Ziprasidone has a lower propensity for weight gain and other
`metabolic disturbances than olanzapine or quetiapine.'*
`Adjunctive treatment with standard antidepressant medicationsis
`the most commonly prescribed intervention for patients with bipolar
`depression.'* The STEP-BD showed no benefit, however,for adjunctive
`treatment with antidepressants (bupropion or paroxetine) compared
`to moodstabilizer plus placebo. To date, only one placebo-controlled
`study has succeeded in demonstrating the efficacy of any agent as an
`adjunct to lithiam or valproate.'* Although successful in an adjunct
`study'* and commonly used for maintenance treatment for bipolar
`disorder, lamotrigine failed to separate from placebo in 5 of 5 bipolar
`depression monotherapy studies on primary outcome measure and
`4 of 5 studies on key secondary outcome measures.'> Anotheratypi-
`cal antipsychotic, aripiprazole, studied for bipolar depression, has also
`produced negative or failed results.'®
`There are no double-blind data available to guide the care of
`depressed bipolar patients who have not respondedto lithium, lamo-
`trigine, or valproate. As preliminary clinical studies have suggested that
`
`f41QOPYRIGHT 207 7 PHYSICIANS POSTGRADBATE PRiggcRArRisTBON MIGHT 2071
`PHYSGHAPHANGBHOORRBM201.
`
`
`1
`
`Exhibit 2054
`Slayback v. Sumitomo
`IPR2020-01053
`
`Exhibit 2054
`Slayback v. Sumitomo
`IPR2020-01053
`
`
`
`VW
`
`a)
`
`=i
`
`he
`
`A B
`
`e)
`cs
`beeor
`
`ziprasidone may have an antidepressant effect in subjects
`with bipolar disorder or with other psychiatric diagnoses,
`the present study was designed to investigate the efficacy
`and safety of ziprasidone as add-on therapy in patients with
`bipolar I disorder who were treated with lithium, valproate,
`or lamotrigine. In view of the frequency at which bipolar
`depression studies have failed or produced negative results,
`weincorporated an innovative computer-based rating man-
`agementsystem into the study design.
`
`METHOD
`
`The study(clinicaltrials.gov registry: NCT00483548) was
`a randomized, double-blind, placebo-controlled,trial con-
`ducted at 78 centers located in Australia (4), India (6), and
`the UnitedStates (68). The protocol was approved by insti-
`tutional review boardsor independent ethics committees at
`each center, andthe trial was conducted in accordance with
`the Declaration of Helsinki, the International Conference
`on Harmonisation Good Clinical Practice guidelines, and
`all appropriate local regulatory requirements.
`The primaryaim of the study was to investigate the effi-
`cacy andsafety of ziprasidone as add-on adjunctive therapy
`in the treatmentof depression associated with bipolar | disor-
`der. Secondary objectives included examination ofthe effects
`of ziprasidone on global functioning and quality oflife.
`
`Subjects
`Adult (2 18 years old) outpatients of either sex were eli-
`gible for the study if they had a primary diagnosis of bipolar
`I disorder, with the most recent episode depressed (296.5x),
`with or without rapid cycling, and without psychotic fea-
`tures, as defined in the Diagnostic and Statistical Manualof
`Mental Disorders, Fourth Edition (DSM-IV)."” The diagnosis
`wasestablished by consensus betweena certified site-based
`rater using the Mini-International Neuropsychiatric Inter-
`view'S and an independent expert employed by Concordant
`Raters Systems in Boston, Massachusetts; Philadelphia,
`Pennsylvania; or San Francisco, California. The expert was
`a psychiatrist or psychologist with clinical experience and
`research experience who reviewed details of prior manic
`or mixed episodes collected directly from the subject by
`the computer and who validated the subjects’eligibility for
`randomization, if at least | episode met full DSM-IVcri-
`teria for mania or a mixed episode. The Bipolarity Index, a
`measure of diagnostic confidence,'? was also used for cases
`in whichit was not possible to confirmthe diagnosis based
`on the computer assessment. In these cases, subjects were
`included onlyif sufficient additional diagnostic information
`was obtained from the investigator (or designee) to estab-
`lish acceptable diagnostic confidence.*” The onset of the
`depressive episode was required to be between 2 weeks and
`6 monthsofscreening.In addition,subjects were required to
`have a scoreofat least 20 on the 17-item Hamilton Depres-
`sion Rating Scale (HDRS-17)?! and a score of <12 on the
`Young Mania Rating Scale (YMRS)”at both screening and
`randomization.
`
`Adjunctive Ziprasidone in Bipolar Depression
`
`# The frequentfailure of randomized controlled studies
`to detect differences between study medication and
`placeboisa significant obstacle to drug development.
`# Although somestudiesinclude active comparators,this
`componentalone doeslittle to inform thefield as to why
`randomizedclinical trials often lack assay sensitivity.
`= Using data from tandem assessments madebysite-
`based raters and computer-administered assessments,
`this report examined the impact of protocol-specific
`eligibility criteria, diagnostic confidence, and rating
`quality on signal detection. The results suggest that
`variability in study quality can lead to study failure and
`that future clinical trials could benefit from procedures
`that do notrely exclusively on assessments madeby a
`singlerater.
`
`Subjects were excluded from the study if they had any
`DSM-IV-TR Axis I or Axis I] disorder that was clinically
`unstable or required treatment or if they showed ultrafast
`rapid cycling (defined as >8 mood episodes during the 12
`months before screening). Other psychiatric exclusion cri-
`teria included a suicide attempt within the 3 months before
`screening or a score ofat least 4 on the suicide itemof the
`Montgomery-Asberg Depression Rating Scale (MADRS),*
`DSM-IV-TR-defined alcohol or psychoactive substance
`dependency within 6 months prior to screening or docu-
`mented abuse of such substances within 3 months before
`screening, electroconvulsive therapy (ECT) within 3 months
`before screening, a history of nonresponse to ECT, treatment
`with any psychotropic medication other than lithium, val-
`proate or lamotrigine within | week prior to screening, or
`depot neuroleptic treatment within the previous 6 months.
`In addition, subjects were excluded if they had clinically
`significant electrocardiogram (ECG) abnormalities, a his-
`tory of QT interval prolongation or any medical condition
`or treatment that could produce such prolongation, or sig-
`nificant medical conditions, including a history of seizures,
`cardiovascular disease, neuroleptic malignant syndrome, or
`tardive dyskinesia that did not respond to treatment. Women
`of childbearing potential were required to use appropriate
`contraceptive precautions during the study.
`Written informed consent was to be obtained before
`inclusion in the study. In the caseofilliterate subjects, the
`subject provided an alternative indication, such as a thumb-
`print, and an impartial witness was required to provide
`signed confirmation that the informed consent procedure
`had been appropriate.
`
`Study Design and Treatment
`Study subjects comprised (1) subjects already on a
`moodstabilizer at screening and (2) subjects initiated on a
`moodstabilizer at screening. In both cases, moodstabilizer
`treatment had to remain stable, as defined by the protocol
`
`w
`OFAPNthiath)73.16/OabEPSOHPOTSRADBAYE PRES poyety)
`
`*
`
`Te tae
`
`oe x
`
` SSICANS POSTGRADUATE PRESS, Hiidy,
`
`
`
`Adjunctive Ziprasidonein Bipolar Depression
`
`requirements for lamotrigine dose (100-200 mg/d) or blood
`concentrations of lithium or valproate (0.6-1.2 mEq/L for
`lithium or 50-125 pg/mLforvalproate), and was to be main-
`tained for at least 4 weeks before randomization. Subjects
`whose moodstabilizer therapy had remainedstable as per-
`protocol requirementsfor at least 4 weeks were randomized
`in a 1:1 ratio to receive adjunctive ziprasidone or placebo
`for 6 weeks.
`Randomization was performed using a unique identifica-
`tion numberfor each subject and wasstratified according
`to the type of moodstabilizer therapy (lithium, valproate,
`or lamotrigine). An internet-/telephone-based random-
`ization and drug management system was used to provide
`the identification number and to assign cither ziprasidone
`or matching placebo capsules to each subject throughout
`the trial. Blinding was to be broken only in the event of an
`emergency that required knowledge of the treatment for
`subject safety. One formal interim analysis was to be per-
`formed when approximately 60% of the planned subjects
`had either completed the study or discontinued prematurely,
`The Data Safety Monitoring Committee had the option to
`recommend stopping the study early for efficacy (nominal
`P value <.0076, 2-sided) or for futility (nominal P value
`>.5099, 2-sided).
`Subjects were instructed to take all study medication
`with food. The starting dose of ziprasidone was 40 mg in
`the evening onthe day of randomization, followed by 40 mg
`twice daily on the second day(ie, 80 mgtotal daily dose).
`Thereafter, subjects were titrated twice daily with total daily
`doses in the range of 40-160 mg, depending on symptoms
`andtolerability. Compliance was assessed bypill counts, and
`blood levels of lithium and valproate were monitored via
`samples taken at screening, baseline, and week 6,or at the
`early terminationvisit to ensure the subject met the required
`therapeutic blood level specified in the protocol.
`All other psychotropic medications were withdrawn at
`least 7 days or 4 half-lives (whichever was longer) before
`randomization. Lorazepam, or analternative short-acting
`benzodiazepine, could be given at doses of up to 2 mg/d for
`up to 4 days per week during screening and thefirst 2 weeks
`of the double-blind treatment period to treat agitation or
`anxiety. Regulatory agency-approved nonbenzodiazepine
`medications could be used to treat sleep disturbances for
`up to 4 days per week until the end of the second week
`of double-blind treatment and for up to 2 days per week
`thereafter. The benzodiazepines and sleep agents were not
`to be given on the same day and were notto be used within
`24 hoursofefficacy assessments. Benztropine (< 6 mg/d)
`or an equivalent agent could be used to treat extrapyrami-
`dal symptoms. Propranolol (s 120 mg/d) could be used to
`treat akathisia.
`
`Assessments
`Efficacy assessments were made at baseline (randomiza-
`tion) and at weekly intervals thereafter. The primaryefficacy
`endpoint was the change from baseline to week 6 in the
`MADRStotal score. The key secondary efficacy endpoint
`
`wasthe change from baseline to week6 in the Clinical Global
`Impressions-Severity scale (CGI-S)*4 score. Additional sec-
`ondary efficacy endpoints included change from baseline in
`Hamilton Anxiety Rating Scale (HARS)** total score; change
`from baseline in YMRStotal score; change from baseline
`in Global Assessment of Functioning (GAF) scale’? score;
`change from baseline in Sheehan Disability Scaletotal
`score; and change from baseline in Quality of Life Enjoy-
`mentandSatisfaction Scale (Q-LES-Q)*’ total score.
`Only qualified raters who meteducational and experience
`requirements participated in the trial. Prior to the start of the
`trial, rater training was conducted on-line and at an inves-
`tigators’ meeting for all participating centers. The MADRS
`data at cach study visit were monitored using a remotesite
`management system developed by Concordant Rater Sys-
`tems, the vendor responsible for rater training and remote
`site management. Raters completed the training program
`and then received “provisionalcertification’; “full certifica-
`tion” was granted on raters demonstrating proficiency with
`concordance betweensite-based ratings and computerrat-
`ings within the acceptable concordance range overthefirst
`3-6 actual subjectratings. Raters not meetingproficiency re-
`quirements were not allowed to enroll additional subjects.
`Each site was provided with a laptop computer with
`the remote site management software (Concordant Rater
`Systems). The MADRSitem scores as determined by the
`site-based ratings were entered on the laptop. In addition,
`(withoutassistance or input from thesite rater), the subject
`completed an interactive interview on the computer, which
`selected a sequence of questions as necessary to map the
`subject's responses to the MADRSanchor points for cach
`scale item. A computer-generated score was assigned based
`on the subject’s input. Prior studies have demonstrated that
`site-based ratings and computer-administered MADRSare
`highly correlated.?*
`[tem ratings scores on which the site-based ratings and
`computer scores differed by no more than | point were
`considered to be concordant. Concordant Rater Systems
`contacted raters by telephone to discuss the potential causes
`for discordantratings, if the total score differential was 26
`points or more than2 items with a differential of = 3 points.
`No further action was taken with raters who provided
`supporting information for their ratings; however, raters
`with unresolved discordance received remediation on use
`of appropriate probes and/or scoring conventions for the
`MADRS.In allcases, site-based raters were instructed not
`to changetheir original scores.
`The same procedure was applied to the YMRS data
`at screening and baseline and to the HDRS-17 data at
`screening. Rater quality scores were categorized as better
`quality, lower quality, and poor quality ifthe absolute value of
`the difference between the computer- and site-based ratings
`was $5,>5< 10, or > 10, respectively. The poorquality ratings
`were designatedrater inflation if the site-based ratings score
`was > 10 points higher than the computer score and subject
`inflation when the computer score was >10 points higher
`than the site-based ratings score. Confidence in the lifetime
`
` CHKPEGPODS201K
`faSOPYRIGRT 201 7 PHYSICIANS POSTGRADUATE PRiggoRAArRidtaA SIGHT 2077
`
`
`
`diagnosis of bipolar I disorder was assessed with the Bipo-
`larity Index.”® This scale quantifies the process suggested
`by Robins and Guze*? for validating psychiatric diagnosis
`by scoring 5 illness domains (episode characteristics, age at
`onset, response to treatment, course of illness, and family
`history) on a 0-20 scale, on which higherscores are given
`to characteristics most associated with the Kraepelinian
`conception of bipolar disorder. Prior psychometric studies
`indicate that acceptable confidence for bipolar | disorder
`lifetime diagnosis corresponds to scores above 60 or having
`at least 3 domainsscored 15 or higher.”*
`Safety and tolerability were assessed by recording of
`adverse events, physical examination, and measurementof
`vital signs, 12-lead ECG, andclinical laboratory evaluation,
`Extrapyramidal symptoms, akathisia, and dyskinesia were
`assessed by meansofthe Simpson-AngusScale,” the Barnes
`Akathisia Scale,** and the Abnormal Involuntary Movement
`Scale (AIMS).??
`
`Statistical Analysis
`Statistical analyses were performed on the intent-to-
`treat (ITT) population, which consisted ofall subjects who
`were randomized, received at least | dose of double-blind
`medication, and had at least 1 postbaseline primary effi-
`cacy assessment. In addition, the primary and key secondary
`efficacy endpoints were analyzed in the per-protoco] popu-
`lation, which included all subjects in the ITT population
`with no major protocolviolations.
`The primary efficacy variable, the mean change in
`MADRSscores from baseline to week 6, was analyzed
`using a mixed model repeated measures (MMRM) analysis
`with fixed categorical effects of treatment, country, type of
`moodstabilizer, visit and treatment-by-visit interaction, and
`a fixed, continuouseffect of baseline MADRSscore; subject
`effect was included as a random effect. The mixed model
`repeated measures analysis used the restricted maximum
`likelihood estimation method, with a sandwich estimatorof
`variance-covariance matrix of the fixed effects parameters.
`The analysis was performed using the SAS PROC MIXED
`procedure (SAS Institute Inc, Cary, North Carolina). An
`unstructured variance-covariance matrix was used in the
`REPEATED statement. Supplemental analyses of the pri-
`mary endpointincluded analysis of covariance (ANCOVA)
`of the change in MADRS scores from baseline to week 6,
`with missing data imputed using last observation carried
`forward (LOCF) principle; ANCOVA of change from base-
`line in MADRSscores at week 6 on observed cases only,
`the primary analysis using log-transformed total MADRS
`score; and a pattern mixture, mixed model repeated mea-
`sures analysis of change from baseline in MADRSscores.
`The change in CGI-S score from baseline to week 6 was
`analyzed by mixed model repeated measures as described
`above, and supplementary analyses were performed by
`ANCOVAon both LOCF and observedcases data. Adjusted
`for the interim analysis, the P value threshold for the pri-
`mary analysis was .0476, For change from baseline in total
`score for HARS and YMRS, ANCOVAsimilarto that for the
`
`Adjunctive Ziprasidone in Bipolar Depression
`
`primary endpoint was conducted at each postbaseline col-
`lection time point on the basis of both LOCFand observed
`cases, For change from baseline in scores for GAF, Sheehan
`Disability Scale, Q-LES-Q, Simpson-Angus Scale, Barnes
`Akathisia Scale, and AIMS, ANCOVAsimilar to that for
`the primary endpoint was conducted on the basis of the
`observed cases.
`
`The sample size calculation was performed with EAST 4
`software (Cytel Inc, Cambridge, Massachusetts) to account
`fora preplanned interim analysis, when approximately 60%
`of the planned numberof subjects had either completed the
`study or discontinued prematurely. It was calculated that a
`sample size of 141 subjects per group (282 in total) would
`provide 85% powerto detect a treatmentdifference in the
`mean change in MADRS scores from baseline to week 6 of
`4.0 points, with a standard deviation of 11.0, using a 2-sided
`test at a significance level of .05.
`
`Rating Quality Data Analysis
`After completionof the efficacy analysis, the study spon-
`sor sent unblinded treatment assignments to Concordant
`Rater Systems and matchedwith the rater quality datafiles.
`The files were reviewed for accuracy, and analyses were car-
`ried out using Stata version 11.0 statistical software.
`The analysis plan compared keyresults from theefficacy
`analysis to those derived from the Rater Quality data set
`and evaluateda list ofa priori competing hypotheses. These
`involved comparing results from prespecified subgroups
`defined by variables derived from computer-administered
`scales.
`Quality ratings were defined based on the absolute value
`of the difference between the computer- andsite-based rat-
`ings scores on the MADRS:better quality (difference <5),
`low quality (absolute value of difference from 5-10), or
`poor quality (difference > 10). Baseline MADRSratings in
`which the site-based ratings score was > 10 higher than the
`computer scored wereclassified as indicating likely rater
`inflation. Baseline MADRSratings in which the site-based
`ratings score was > 10 lower than the computerscored were
`classified as indicating likely subject inflation.
`
`RESULTS
`
`Between October2007 and December 2008, 792 subjects
`were screened, of whom 298 were randomized and 294 (147
`in each group) received treatment (Figure 1). Of the 294
`whoreceived treatment, 102 subjects discontinued treat-
`ment, mainly due to adverse events and protocolviolations.
`Thus, 192 subjects (88 in the ziprasidone group, 104 in the
`placebo group) completed the study(Figure 1). The sample
`characteristics are summarized in Table 1.
`
`Interim Analysis
`The interim analysis was performed on 168 subjects
`(84 subjects in each treatment group; 59.6% of the planned
`final sample size). At the interim analysis, the least squares
`mean + standard error (SE) changes from baseline to week
`
`© fOMyPHhing73.10POSENSOTPSCRADBA PRES PoyATRISTOMONT 2071 PHYSICIANS POSTGRADUATE PRESS, Tygyg
`
`
`
`Adjunctive Ziprasidonein Bipolar Depression
`
`
`Figure 1. Study Profile
`
`
`
`43 (29.3%) Discontinuations
`|
`
`
`14 (9.5%) Adverse events
`|
`
`2 (1.494) Leboretory abnormalities
`
`B (5.4%) Lack ofefficacy
`
`7 4.8%) Protocol violations
`
`12 (8.2%) Other reasons
`
`
`
`
` 104 Completed Study
`hri359Chtenseaig
`
` 4s
`
`|
`i
`|i
`
`Ziprasidone (n = 147)
`
`Placebo (n= 147)
`
`58 (39.5)
`89 (60.5)
`
`56 (38.1)
`91 (61.9)
`
`116 (78.9)
`19 (12.9)
`7 (4.8)
`5.4)
`40.44 114 (18-64)
`84.3 £214 (45.0-156.1)
`168.2 + 10.2 (138.0-188.0)
`16.2 (0.07-50.7)
`
`111 (75.5)
`20 {13.5}
`117.5)
`5 (3.4)
`40.44 11.9 (18-66)
`89.94 23.2 (45.4-174.8)
`168.04 10.0 (139.7-195.0)
`16.6 (0.1-45.2)
`
`
`Table 1. Baseline Characteristics
`Characteristic
`Sex, n (%)
`Male
`Female
`Race, n (%)
`White
`Black
`Asian
`Other
`Age, meant SD(range), ¥
`Weight, mean + SD (range), kg
`Height, mean + SD(range), cm
`Time since first diagnosis ofbipolar [ disorder,
`mean (range), y
`Duration of current episode, mean (range), d
`No. of episodesin previous 12 mo, mean (range)
`Suicidal ideation in previous 12 mo, n (%)
`History of suicide attempt in previous |2 months, n (%)
`Moodstabilizer, n®
`Lithium
`Valproate
`Lamotrigine
`*One subjectin the ziprasidone group did not receive a moodstabilizer during the double-blind period and was
`therefore excluded from the analysis.
`
`76.2 (15-254)
`2.7 (0-20)
`60 (41.4)
`6 (4.1)
`
`53
`52
`4t
`
`$2.9 (16-207}
`2.3 (0-10)
`45 (31.5)
`4 (2.8)
`
`S4
`52
`4h
`
`6o0n the MADRS for the ziprasidone and placebo treatment
`groups were —11.3 (2.18) and -13.3 (2.06), respectively, with
`a 2-sided P value of .2690 favoring placebo.
`Enrollmentin the study was faster than expected, and the
`results of the interim analysis were not available until enroll-
`ment was almost completed. On the basis of the results of
`the interim analysis, the Data Safety Monitoring Committee
`recommended that, due to study futility, already random-
`ized subjects could complete the study but that no further
`subjects should enter thetrial. Enrollment was completed
`before this recommendation was implemented.
`
`Efficacy
`Table 2 describes changesin efficacy rating scores. At
`baseline, there were no significant differences between the
`
`groups onany efficacy measure. The mean+SDdaily dose
`of ziprasidone was 89.8+29.1 mg.
`There was no significant difference on the primary out-
`comevariable, the key secondaryvariable (CGI-S), or most
`of the other secondary measures, including YMRS, HARS,
`and Q-LES-Q. There was, however, a significant difference
`favoring ziprasidone over placebo on the GAF scale and the
`Sheehan Disability Scale.
`Theleast squares mean + SE change from baseline at week
`6 in MADRStotalscore for ziprasidone- and placebo-treated
`subjects was —13.24 1.2 and -12.9 + L.1, respectively (Table
`2), corresponding to a least squares mean + SE treatment
`difference of ~0.36 + 1.37 (95% CI, -3.07 to 2.34) that was
`not statistically significant (P=.7921). The results of the
`per-protocolanalysis (P = .3989) and the sensitivity analysis
`
`nroceree ecw “SSR
`rexen wer Boawe ge Baye 5 2
`4
` be201.
`{ChiaPayehineey7346
`1017 PHYSKIANS Paste
`Sed
`SADE © TBGYCRAATRISTIOON O°? <
`
`
`
`
`
`Adjunctive Ziprasidone in Bipolar Depression
`
`28.8461
`-12.94 11
`
`~0A414(-3.1t0 2.3)
`
` .7921
`
`Table 2. Changes in Efficacy Rating Scores
`
`Ziprasidone——Placebo Treatment Difference, P
`
`Scale
`(n= 145)
`(n= 145)
`LS Meant SE (95% C1) Value
`MADRS*
`3.0455
`Baseline score, mean 4 $D
`Change frombaseline to week 6, LS mean ESE ~13.241.2
`CGI-S*
`AALOT
`AALOS
`Baseline score, mean 4 $D
`
`
`
`Changefrombaseline to week 6, LS mean #SE 0.1.0.2 (-0.3 to0.4)—.7223-1.4402 -1440.2
`YMRS?
`Baseline score, mean i SD
`Change trom baseline to week 6, LS meantSE
`HARS®
`Baseline score, mean + SD
`Change from baseline to week 6,LS mean LSE
`GAF
`Baseline score, mean + SD
`Change frombaseline to week 6,1S meantSE
`Sheehan Disability Scale‘
`Raseline score, mean + $D
`Changefrombaseline to week 6, LS mean £SE
`Q-LES-Q*
`Baseline score, mean 4 $D
`Change frombaseline to week 6,US mean £SE
`“Mixed model repeated measures.
`Last observation carried forward.
`“Observedcases.
`Abbreviations: CGI-§ « Clinical Global Impressions-Severity scale; GAP « Global Assessment of Functioning;
`HARS= Hamilton Anxiety Rating Scale; LS=least squares; MADRS = Montgomery -Asberg Depression Rating
`Scale: Q-LES-Q = Quality of Life, Enjoyment, and Satisfaction Scale; SE = standard error; YMRS= Young
`Mania Rating Scale,
`
`6.84 3.1
`0.8411
`
`20.1462
`-8.6414
`
`§2.4475
`16.8424
`
`18.1471
`~7.442.1
`
`FALB0
`~1O4 1.0
`
`19.9+7.0
`9,241.3
`
`52.1469
`12.5423
`
`16.9473
`-2.842.1
`
`O.240.7 (-L.1to 1.5)
`
`.7647
`
`0.640.9 (-~L.1t02.3)
`
`4770
`
`4.22 1.7 (1.0 to 7.5)
`
`.O108
`
`-46414(-7.2to-19)
`
`.001
`
`AVALIS2
`12.14.46
`
`43.24 166
`10.6445
`
`1.542.5(-3.5 to 6,5)
`
`.5519
`
`
`
`Figure 2. (A) Least Squares Mean Change in Montgomery-Asberg Depression Rating Scale (MADRS) Scores (last
`observation carried forward) and (8) Correlation Between Site-Based Rater and Computer-Administered MADRS
`Scores Over the Course of the Study
`
`
`
`RAADRSScore,LeastSquaresMeanChange
`
`5
`
`Baseline Wk} Wk2>Wk3 Last
`Baseline Wk1 Wk2 Wk3 Wk4 WkS Wk6
`Wk4 WkS Wk6
`Observation
`
`9
`
`@ Ziprssidone
`O Placebo
`
`Correlation, 4
`
`B. Correlation Between Site-Based Rater and
`Computer-Administered MADRS Scores
`10
`
`09
`
`08 0.578
`
`A. Change in MADRSScore
`
`
`
`
`
`Timepoint
`
`Timepoint
`
`were consistent with this result. Figure 2A shows change in
`MADRS score with time.
`
`Rater Quality
`Results from 1 or more computer-administered scales
`were available for 282 subjects at 66 of the 78sites partici-
`pating in thetrial. Results from at least | postrandomization
`visit were available for 265 subjects (placebo, n= 137; zipra-
`sidone, n=1]28). Overall, the rater quality data produced
`results (Table 3) similar to the overall results of the effi-
`cacy analysis. The mean+$D Bipolarity Index score was
`
`71.7+10.1. The correlation between site-based rater and
`computer-administered MADRSscores improved consis-
`tently over the course ofstudy visits (Figure 2B).
`Noneof the subgroup analyses based on the computer
`assessments foundstatistically significant differences be-
`tween adjunctive placebo versus adjunctive ziprasidone.
`The computer assessments, however, suggest that many
`subjects were enrolled who did not meet protocol-specified
`eligibility criteria. On the basis ofthe computer assessments,
`about 30% ofall participants failed to meet severity criteria
`{computer-administered HDRS score <20 criteria n=22
`
`Pn
`OFGWPYGhiatry72.10OSSberS0TPSTOADSAE PRESpeyerATRIstOMe? 2077
`
`YSICQANS POSTGRADUATE
`
`
`
`PRESS, 14idyg
`
`
`
`Adjunctive Ziprasidonein Bipolar Depression
`
`
`
`Table 3, impact ofEligibility Criteria Based on Computer Assessments:
`Comparison of Last Observation Carried Forward Site-Based Rater and
`Computer-Administered Montgomery-Asberg Depression Rating Scale (MADRS}
`Change Scores
`
`Ziprasidone
`Change From
`Baseline MADRS,
`Mean (SD)
`
`n Difference
`
`P Value,
`2-Sided
`{Test
`
`Variable
`Full sample
`Site-based rater
`Computer
`
`Baseline HDRS,oayp2 20
`Eligible
`Sile-based rater
`Computer
`Noteligible
`Site-based rater
`Computer
`
`Placebo
`Change From
`Bascline MADRS,
`Mean(SD)
`
`12.1 (11.0)
`—10.3 (12.7)
`
`10.7 ALL)
`10.3 (12.5)
`
`~—13.0 (10.5)
`~10.7 (13.6)
`
`Baseline YMRSconp <3 symptomsof mania
`Eligible
`Site-based rater
`Computer
`Noteligible
`Site-based rater
`Computer
`
`11.5 (10.9)
`—11.0 (12.8)
`
`~9,.0 (11.7)
`~7.2 (12.0)
`
`n
`
`137
`137
`
`M5
`HLS
`
`22
`22,
`
`Lid
`Li4
`
`23
`23
`
`~1b4 (11.2)
`—11.4 (12.3)
`
`10.9 (11.1)
`11.2 (12.3)
`
`—14.3 (11.6)
`-12.$ (12.2
`
`12.3 (11.2)
`—12.6 (12.7)
`
`~7.8 (10.7)
`~6.2 (8.6)
`
`—14.7 (9.9)
`~ 13.0 (10.2)
`
`~ 18.2 (12.1)
`—10.2 (13.6)
`
`128
`128
`
`108
`108
`
`20
`20
`
`104
`104
`
`24
`24
`
`56
`56
`
`72
`72
`
`10.3
`+h
`
`+0.20
`40,94
`
`+4h3
`«+18
`
`40.8
`+16
`
`—1.2
`1.0
`
`+1.2
`#2.2
`
`0
`+02
`
`> 80
`> AS
`
`> 89
`> 57
`
`> 72
`> .66
`
`> .60
`> 34
`
`>.7
`>.74
`
`> .67
`> .34
`
`> 98
`>.92
`
`Diagnostic confidence (Bipolarity Index mean score = 71.7)
`Above mean
`Sile-based rater
`Computer
`Below mean
`Site-based rater
`Computer
`
`—10.9 (12.3)
`10.8 3.7)
`
`~11.2 GL)
`—10.0 (12,0)
`
`58
`58
`
`79
`7
`
`and/or met criteria for a current mixed
`episode(3 clinically significant symptoms
`on computer-administered YMRS, n=23),
`Only 89 (33.6%) of the 265 subjects met
`the eligibility requirements based on the
`computer ratings. The computer ratings
`de