`RESEARCH
`
`
`APPLICATION NUMBER:
`
`022063Orig1s000
`
`STATISTICAL REVIEW(S)
`
`
`
`
`
`
`
`
`
`
`
`
`
`U.S. Department of Health and Human Services
`Food and Drug Administration
`Center for Drug Evaluation and Research
`Office of Translational Sciences
`Office of Biostatistics
`
`
`S T A T I S T I C A L R E V I E W A N D E VA L U A T I O N
`CLINICAL STUDIES
`
`NDA/BLA #:
`Supplement #:
`Drug Name:
`
`Indication(s):
`Applicant:
`Date(s):
`
`NDA 022-063
`O-1 (SN0022 SN0023)
`
`Mydayis (mixed salts of a single-entity amphetamine product) Sustained-
`Release Capsules
`Attention-deficit/Hyperactivity Disorder (ADHD)
`Shire
`Letter date: Dec 20, 2016
`PDUFA due date: Jun 20, 2017
`Priority (resubmission)
`
`Division of Biometrics I
`
`Review Priority:
`
`Biometrics Division:
`Statistical Reviewer:
`Yang Wang, Ph.D.
`Concurring Reviewers: Peiling Yang, Ph. D., Team Leader
`H.M. James Hung, Ph.D., Division Director
`
`Division of Psychiatry Products
`Medical Division:
`Clinical Team:
`Nancy Clark Dickinson, M.D., Clinical Reviewer
`Javier Muniz, M.D., Clinical Team Leader
`Latrice Wilson, Pharm.D.
`
`
`Project Manager:
`
`Keywords:
`Link to keywords:
`http://intranetapps.fda.gov/scripts/ob apps/ob/eWork/uploads/eWork/2009/Keywords-in-
`DFS.htm
`
`
`
`Reference ID: 4100074
`
`
`
`Table of Contents
`1. EXECUTIVE SUMMARY ................................................................................................................................. 5
`
`INTRODUCTION ............................................................................................................................................... 6
`2.
`2.1
`OVERVIEW ...................................................................................................................................................... 6
`2.2
`DATA SOURCES .............................................................................................................................................. 7
`3. STATISTICAL EVALUATION ........................................................................................................................ 9
`3.1
`DATA AND ANALYSIS QUALITY ..................................................................................................................... 9
`3.2
`EVALUATION OF EFFICACY ............................................................................................................................ 9
`3.2.1
`Study Design and Endpoints .................................................................................................................. 9
`3.2.2
`Statistical Methodologies ..................................................................................................................... 12
`3.2.3
`Patient Disposition, Demographic and Baseline Characteristics........................................................ 17
`3.2.4
`Results and Conclusions ...................................................................................................................... 22
`3.3
`EVALUATION OF SAFETY .............................................................................................................................. 37
`4. FINDINGS IN SPECIAL/SUBGROUP POPULATIONS ............................................................................. 38
`4.1
`GENDER, RACE, AGE, AND GEOGRAPHIC REGION ........................................................................................ 38
`4.1.1.1 Change from Baseline by Region ......................................................................................................... 38
`4.1.1.2 Change from Baseline by Gender ........................................................................................................ 38
`4.1.1.3 Change from Baseline by Race ............................................................................................................ 38
`4.1.1.4 Change from Baseline by Age Group .................................................................................................. 39
`4.1.2.1 Change from Baseline by Region ......................................................................................................... 40
`4.1.2.2 Change from Baseline by Gender ........................................................................................................ 40
`4.1.2.3 Change from Baseline by Race ............................................................................................................ 41
`5. SUMMARY AND CONCLUSIONS ................................................................................................................ 42
`5.1
`STATISTICAL ISSUES ..................................................................................................................................... 42
`5.2
`COLLECTIVE EVIDENCE ................................................................................................................................ 42
`5.3
`CONCLUSIONS AND RECOMMENDATIONS ..................................................................................................... 42
`
`
`
`
`
`Reference ID: 4100074
`
`2
`
`
`
`
`LIST OF TABLES
`
`
`Table 1: List of all studies included in analysis ............................................................................................................. 7
`Table 2: Schedule of Assessments – SHP465-305 ...................................................................................................... 10
`Table 3: Dosing Schedule ............................................................................................................................................ 11
`Table 4: Schedule of Assessments – SHP465-306 ...................................................................................................... 12
`Table 5: Disposition by Withdrawal Reason - SHP465-305 ....................................................................................... 19
`Table 6: Disposition by Withdrawal Reason - SHP465-306 ....................................................................................... 21
`Table 7: Demographic and Baseline Characteristics by Treatment Group - SHP465-305 .......................................... 21
`Table 8: Demographic and Baseline Characteristics by Treatment Group - FAS - SHP465-306................................ 22
`Table 9 : Testing Hierarchy - SHP465-305 ................................................................................................................. 23
`Table 10: Primary Analysis of ADHD-RS-IV Total Score at Visit 6 (Week 4) - FAS - SHP465-305 ........................ 23
`Table 11: Primary Analysis of CGI-I Scores at Visit 6 (Week 4) - FAS - SHP465-305 ............................................. 24
`Table 12: Sesitivity Analysis Results of ADHD-RS-IV Total Score - FAS - SHP465-305 ........................................ 27
`Table 13: Sesitivity Analysis Results of CGI-I - FAS - SHP465-305 ........................................................................ 28
`Table 14: Testing Hierarchy - SHP465-306 ................................................................................................................ 30
`Table 15: Primary Analysis of ADHD-RS with Prompts Total Score at Visit 6 (Week 4) - FAS - SHP465-306 ....... 30
`Table 16: Primary Analysis of CGI-I Scores at Visit 6 (Week 4) - FAS - SHP465-306 ............................................. 31
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`3
`
`Reference ID: 4100074
`
`
`
`
`
`
`
`
`LIST OF FIGURES
`
`Figure 1: Study Design Schematic of SHP465-305 ..................................................................................................... 10
`Figure 2: Study Design Flow Chart - SHP465-306 ..................................................................................................... 11
`Figure 3: Subject Disposition - SHP465-305............................................................................................................... 18
`Figure 4: Subject Disposition - SHP465-306............................................................................................................... 20
`Figure 5: Least Squares Mean (±SE) of Change from Baseline in ADHD-RS-IV Total Score by Visit and Treatment
`Group - FAS - SHP465-305 ........................................................................................................................................ 25
`Figure 6: Dichotomized CGI-I by Visit and Treatment Group - FAS - SHP465-305 ................................................. 26
`Figure 7: Histogrm of Change from Baseline in ADHD-RS-IV Total Score to Week 4 - SHP465-305 ..................... 29
`Figure 8: Least Squares Mean (±SE) of Change from Baseline in ADHD-RS with Prompts Total Score by Visit and
`Treatment Group - FAS - SHP465-306 ....................................................................................................................... 32
`
`
`
`Reference ID: 4100074
`
`4
`
`
`
`
`1. EXECUTIVE SUMMARY
`
`
`The efficacy results from two Phase 3 studies (SHP465-305 and SHP465-306) supported
`Sponsor’s claim that SHP465 is efficacious as a long-acting stimulant at dose levels of 12.5 mg
`and 37.5 mg in adults and at a dose range between 12.5 and 25 mg in pediatrics (based on body
`weight) for the treatment of ADHD.
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Reference ID: 4100074
`
`5
`
`
`
`2. INTRODUCTION
`
`2.1 Overview
`
` on
`The original NDA for SHP465 (NDA 22-063) was submitted under the trade name
`July 21, 2006. An Approvable Letter was issued to Shire on May 18, 2007, tentatively approving
`the dosage strengths of 12.5 and 25 mg for the treatment of ADHD in adults in addition to
`specifying 2 post-marketing commitments. References are made to FDA’s April 25, 2014
`Written Responses Only (WRO) and June 17, 2015 meeting minutes in which FDA classified
`this complete response to the May 18, 2007 Approvable Letter as a 505(b)(1) Class 2
`Resubmission.
`
`The SHP465 clinical development program consists of a total of 16 clinical studies, 13 of which
`were included in the original NDA, and 3 of which are new and included in this NDA
`resubmission.
`
`The initial clinical development program for SHP465 included the following 13 clinical studies:
`• 7 phase 1 PK studies conducted in healthy adult subjects,
`• 4 controlled Phase 2 and 3 efficacy studies conducted in adults with ADHD (SPD465-
`201, SPD465-203, SPD465-301, SPD465-303),
`• 1 placebo- and active-controlled Phase 2 efficacy study conducted in adolescents with
`ADHD (SPD465-202),
`• 1 long-term (12 months), open-label Phase 3 safety study with efficacy as a secondary
`objective in adults with ADHD (SPD465-304).
`
`
`Both pivotal phase 3 studies included in the original NDA (SPD465-301 and SPD465-303)
`demonstrated that adults with ADHD treated with 25, 50, or 75 mg SHP465 experienced
`statistically significantly reduced ADHD symptoms (as assessed by Attention-Deficit
`Hyperactivity Disorder Rating Scale-IV (ADHD-RS-IV)) and a global improvement (as assessed
`by the clinical global impression improvement scale (CGI-I)) compared with adults treated with
`placebo.
`
`The NDA resubmission includes 3 additional clinical studies to support a proposed indication for
`SHP465 for the treatment of ADHD with dosage strengths of 12.5, 25, 37.5, and 50 mg in both
`adult and pediatric patients:
`• SHP465-111, an open-label Phase 1 pharmacokinetics study;
`• SHP465-305, a Phase 3, randomized, double-blind, multicenter, placebo-controlled, dose-
`optimization, safety and efficacy study of SHP465 12.5 mg to 25 mg in children and
`adolescents aged 6 to 17 years with ADHD, conducted as a premarketing study in
`pediatric patients with ADHD, as a component in the complete response to the
`Approvable Letter; and
`• SHP465-306, Study SHP465-306 is a Phase 3, randomized, double-blind, multicenter,
`placebo-controlled, forced-dose titration, safety and efficacy study of SHP465 in adults
`aged 18 to 55 years with ADHD, conducted as a post-marketing commitment for
`exploration of dose response for effectiveness as requested in the Approvable Letter.
`
`
`
`Reference ID: 4100074
`
`6
`
`(b) (4)
`
`
`
`The two Phase 3 studies SHP465-305 and SHP465-306 are the main focus of this review and
`listed below.
`
`Table 1: List of all studies included in analysis
`
`Phase and
`Treatment
`Design
`Period
`Phase 3
`4 weeks
`
`SHP465-305
`
`Follow-up
`Period
`7 days
`
`SHP465-306
`
`Phase 3
`
`4 weeks
`
`7 days
`
`Study
`Population
`Children and
`adolescents
`(aged 6-17
`years) with
`ADHD
`Adults (aged
`18-55 years)
`with ADHD
`
` # of Subjects
`per Arm
`257 in FAS:
`129 in
`Placebo and
`128 in
`SHP465
`263 in FAS:
`86 in placebo,
`89 in SHP465
`12.5 mg, and
`88 in SHP465
`37.5 mg.
`
`
`Objectives of SHP465-305:
`Primary: to evaluate the efficacy of SHP465 administered as a daily morning dose compared to
`placebo in the treatment of children and adolescents (aged 6-17 years, inclusive) diagnosed with
`ADHD.
`Key secondary: to assess the efficacy of SHP465 compared to placebo using a global clinical
`measure of improvement, the Clinical Global Impression - Global Improvement scale (CGI-I).
`
`Objectives of SHP465-306:
`Primary: to evaluate the efficacy of each SHP465 dose (12.5 and 37.5 mg) administered daily in
`the morning compared to placebo in the treatment of adults (18-55 years of age, inclusive)
`diagnosed with ADHD.
`Key secondary: to evaluate the efficacy of each SHP465 dose (12.5 and 37.5 mg) compared
`with placebo using a global clinical measure of improvement, the Clinical Global Impression –
`Improvement (CGI-I).
`
`2.2 Data Sources
`
`The following data sources were considered in this review:
`a) Applicant’s study report
`(\\CDSESUB1\evsprod\NDA022063\0022\m5\53-clin-stud-rep\535-rep-effic-safety-
`stud\adhd\5351-stud-rep-contr\shp465-305)
`(\\CDSESUB1\evsprod\NDA022063\0022\m5\53-clin-stud-rep\535-rep-effic-safety-
`stud\adhd\5351-stud-rep-contr\shp465-306)
`b) Data sets
`(\\CDSESUB1\evsprod\NDA022063\0022\m5\datasets\shp465-305\analysis\adam\datasets)
`(\\CDSESUB1\evsprod\NDA022063\0022\m5\datasets\shp465-306\analysis\adam\datasets)
`c) Software code
`(\\CDSESUB1\evsprod\NDA022063\0023\m5\datasets\shp465-305\analysis\programs)
`
`
`7
`
`Reference ID: 4100074
`
`
`
`(\\CDSESUB1\evsprod\NDA022063\0023\m5\datasets\shp465-306\analysis\programs)
`d) Response to FDA information request
`(\\CDSESUB1\evsprod\NDA022063\0022\m1\us)
`(\\CDSESUB1\evsprod\NDA022063\0023\m1\us)
`
`
`
`
`Reference ID: 4100074
`
`8
`
`
`
`3. STATISTICAL EVALUATION
`
`3.1 Data and Analysis Quality
`
`The sponsor has complied with our requests for providing necessary datasets, definition files,
`and statistical programs for their analyses. This reviewer found the quality of their submissions
`acceptable and was able to replicate the primary results from the sponsor’s Clinical Study Report
`(CSR).
`
`3.2 Evaluation of Efficacy
`
`3.2.1 Study Design and Endpoints
`
`3.2.1.1 Study SHP465-305
`
`SHP465-305 was a randomized, multicenter (36 sites in the United States (US)), double-blind,
`placebo-controlled, dose-optimization study in children and adolescent subjects (6-17 years of
`age inclusive) with ADHD.
`
`After the screening visit, a washout period was included to prevent any carry-over effects of
`residual prior medications before randomization. Subjects were stratified within each age group
`(6-12 years vs 13-17 years), and randomly assigned in a 1:1 ratio to SHP465 or placebo at
`baseline (Visit 2) and then would have 4 weeks of double-blind evaluation (2 weeks of dose-
`optimization and 2 weeks of dose-maintenance periods) of safety and efficacy. All enrolled
`subjects who completed the study or discontinued early were to complete Visit 6/early
`termination (ET). The follow-up period for this protocol was 7 days (+2 days) from the last dose
`of the investigational product.
`
`
`
`
`Reference ID: 4100074
`
`9
`
`
`
`Figure 1: Study Design Schematic of SHP465-305
`
`
`Source: figure 1 on page 29 of Sponsor’s CSR.
`
`Table 2: Schedule of Assessments – SHP465-305
`
`
`
`
`
`Source: table 3 on page 41-42 of Sponsor’s CSR.
`
`The primary measure of efficacy was the ADHD-RS-IV, consisting of 18 items designed to
`reflect current symptomatology of ADHD based on DSM-IV-TR criteria. Each item is scored on
`a 4-point scale ranging from 0 (reflecting no symptoms) to 3 (reflecting severe symptoms) with
`total scores ranging from 0-54. The 18 items may be grouped into 2 subscales:
`hyperactivity/impulsivity (even-numbered items 2-18) and inattentiveness (odd-numbered items
`1-17).
`
`The primary efficacy endpoint was the change from baseline in the ADHD-RS-IV Total Score at
`Visit 6 (Week 4). The baseline ADHD-RS-IV Total Score was defined as the last valid ADHD-
`RS-IV Total Score assessment prior to taking the first dose of double-blind investigational
`product, usually at Visit 2 (Week 0).
`
`The key secondary measure was CGI-I to assess the 3 target areas of improvement recorded at
`the baseline visit (Visit 2) by a 7-point scale ranging from 1 (very much improved) to 7 (very
`much worse).
`
`
`
`Reference ID: 4100074
`
`10
`
`
`
`The full analysis set (FAS) consisted of all subjects who signed informed consent, had been
`assigned a randomization number, had taken at least 1 dose of investigational product, and had at
`least 1 post-dose ADHD-RS-IV Total Score assessment.
`
`3.2.1.2 Study SHP465-306
`
`SHP465-306 was a Phase 3, randomized, multicenter (43 sites in the United States), double-
`blind, parallel-group, placebo-controlled, forced-dose titration study.
`
`The study had 4 periods: screening and washout, forced-dose titration (Weeks 1 and 2), dose
`maintenance (Weeks 3 and 4), and safety follow-up. The duration of the double-blind evaluation
`period (forced-dose titration and dose maintenance periods) was 4 weeks. Subjects were
`randomly assigned at baseline (Visit 2) in a 1:1:1 ratio to 1 of 3 treatment groups: SHP465 12.5
`mg, SHP465 37.5 mg, or placebo. Subjects received an oral dose of investigational product each
`morning for 4 weeks as detailed in the following table.
`
`Table 3: Dosing Schedule
`
`Source: table 2 on page 27 of Sponsor’s CSR.
`
`All randomly assigned subjects who completed the study or discontinued early were to complete
`Visit 6/early termination (ET). The follow-up period was 7 (+2) days from the last dose of the
`investigational product.
`
`Figure 2: Study Design Flow Chart - SHP465-306
`
`
`
`Source: figure 1 on page 28 of Sponsor’s CSR.
`
`
`
`
`11
`
`Reference ID: 4100074
`
`
`
`Table 4: Schedule of Assessments – SHP465-306
`
`
`
`
`
`Source: table 4 on page 39-40 of Sponsor’s CSR.
`
`The primary measure of efficacy was the clinician-administered adult ADHD-RS with prompts
`consisting of 18 items designated to reflect current symptomatology of ADHD based on the
`DSM-5 criteria. Each item is scored on a 4-point scale ranging from 0 (no symptoms) to 3
`(severe symptoms), with the total score for the rating scale ranging from 0 to 54. The scale is
`subdivided into 2 subscales of 9 symptoms each: hyperactivity/impulsivity and inattentiveness.
`
`The primary efficacy endpoint was defined as the change from baseline of the adult ADHD-RS
`with prompts total score at Visit 6 (Week 4). Baseline adult ADHD-RS with prompts total score
`was defined as the last valid adult ADHD-RS with prompts total score assessment prior to taking
`the first dose of double-blind investigational product, usually at Visit 2 (Week 0).
`
`The key secondary efficacy endpoint was the CGI-I score.
`
`The full analysis set (FAS) consisted of all subjects who signed informed consent, had been
`assigned a randomization number, took at least 1 dose of investigational product, and had at least
`1 post-dose baseline primary efficacy assessment (ADHD-RS with prompt total score) on
`treatment.
`
`
`3.2.2 Statistical Methodologies
`
`3.2.2.1 Study SHP465-305
`
`3.2.2.1.1 Primary Analyses
`
`The primary efficacy endpoint was analyzed by using the linear mixed-effects model for
`repeated measures (MMRM) with treatment group, visit, age group (6-12 years vs 13-17 years),
`and the interaction of treatment group with visit as factors, baseline ADHD-RS-IV Total Score
`as a covariate, and the interaction of baseline ADHD-RS-IV Total Score with visit adjusted in
`the model.
`
`The key secondary efficacy measurement, CGI-I, was analyzed using the same analysis method
`(MMRM) as for the primary efficacy endpoint, including treatment group, nominal visit, age
`group, interaction of the treatment group with the visit as factors, baseline CGI-S as a covariate,
`and an adjustment for the interaction of the baseline CGI-S with the visit. The model is based on
`a restricted maximum likelihood (REML) method of estimation and utilizes an unstructured
`covariance type. The primary contrast of interest was at Visit 6 (Week 4) for SHP465 compared
`with placebo.
`
`
`12
`
`Reference ID: 4100074
`
`
`
`The fixed-sequence test procedure was applied to protect the study-wide Type I error at the 2-
`sided 0.05 for testing across the primary and the key secondary hypotheses.
`
`3.2.2.1.2 Interim Analysis
`
` A
`
` blinded interim analysis was planned, when approximately 75% of all randomly assigned
`subjects had either completed or discontinued from the study, to reassess the sample size in case
`of an underestimated variability postulated at the design stage.
`
`If the re-estimated pooled standard deviation (SD) by a blinded analysis of the cumulative real
`data was larger than the 10.0 postulated at the design stage, the final total number of subjects to
`be enrolled was to be calculated using the re-estimated pooled SD together with the assumed
`treatment difference of 6.0. If the re-estimated pooled SD was smaller than 10.0, the sample size
`was not to be adjusted.
`
`No data monitoring or review committee was planned for this study.
`
`3.2.2.1.3 Additional Analyses
`
`The CGI-I categories were dichotomized into 2 categories: “improved” (which included the
`categories of “very much improved” and “much improved”) and “not improved” (which included
`all other assessed categories grouped together). The key secondary efficacy measurement was
`analyzed using the proportion of subjects with an “improved” CGI-I measurement at Visit 6
`(Week 4) using a Cochran-Mantel-Haenszel test that was stratified by age group and CGI-S
`value at baseline. If missing data exist at Visit 6 (Week 4), the visit was imputed by carrying
`forward the last post-baseline observation value.
`
`3.2.2.1.4 Sensitivity Analyses
`
`MMRM relies on the assumption that the missing data mechanism follows the missing at random
`(MAR) scenario, assuming that the probability of missing data is unrelated to the unobserved
`value itself, after controlling for observed data. For both the primary efficacy endpoint and the
`key secondary efficacy endpoint, two sensitivity analysis models that assume different missing
`not at random (MNAR) mechanisms were carried out to examine the robustness of the MMRM
`analysis results using pattern-mixture models.
`
`Model 1 - Placebo multiple imputation based on the distribution of placebo group responses
`over time, assuming a subject on the active treatment with missing data follows the distribution
`of the placebo responses, i.e., the means and the intra-subject correlations based on the placebo
`responses will apply. The MNAR assumption is implemented by applying penalties to missing
`items in a multiple imputations process based on treatment-specific multivariate normal
`distribution for response. The penalty applied is a fraction of the estimated standard deviation for
`the primary endpoint.
`
`
`
`
`Reference ID: 4100074
`
`13
`
`
`
`Step 1: Imputations
`A total of 200 sets of posterior mean and co-variance estimates are extracted from the SAS MI
`procedure using the available non-missing placebo data, 100 of which applied to the active
`treatment group and the other 100 to the placebo group. One set of imputations for all missing
`values will be generated based on each variation of posterior estimates. All 100 sets for
`imputations within a treatment group will be ordered from 1 to 100, and combined between
`active treatment and placebo, for a total of 100 completely imputed data sets.
`
`Step 2: Analysis of complete data sets
`The primary endpoint will be analyzed for each of the 100 complete data sets with imputed data
`using an ANCOVA with treatment group and age group as factors, and the baseline value as a
`covariate.
`
`Step 3: Inference
`The LS mean difference estimates will be averaged and the associated SEs will be summarized
`based on within-imputation and between-imputation variance using the SAS MIANALYZE
`procedure to yield a final estimate with associated 95% CI and p-value.
`
`Model 2 – Multiple imputations with penalties applied to dropouts, assumed subjects who
`discontinue have worse changes than that predicted using MAR after discontinuation by a
`penalty. The MNAR assumption is implemented by extracting posterior mean and covariance in
`a multiple imputations process based on placebo patients, and applied to all SHP465 dropouts.
`Penalties were fractions of the SD for the primary and key secondary endpoints, and 5 different
`penalties were applied. SD is the estimated standard deviation for the primary endpoint (the
`square root of the estimated element for Visit 6 of the covariance matrix R from the primary
`MMRM).
`
`Step 1a: Imputations
`Based on the MAR assumption, missing data will be multiply imputed for 100 times on a
`treatment specific, multivariate normal distribution of the response over time using the SAS MI
`procedure with treatment in the BY statement.
`
`Step 1b: Application of penalty
`A fraction of the estimated standard deviation (SD) for the primary endpoint: (0*SD), (0.25*SD),
`(0.5*SD), etc. will be applied as a penalty to the multiply imputed values at Visit 6 (Week 4). SD
`is the square root of the estimated element for Visit 6 of the co-variance matrix R from the
`primary MMRM model.
`
`Step 2 (analysis of complete data sets) and Step 3 (inference) are the same as Step 2 and Step
`3, respectively, for Model 1.
`
`
`
`Reference ID: 4100074
`
`14
`
`
`
`3.2.2.1.5 Sample Size Determination
`
`To detect an assumed difference of 6.0 for the change from baseline in the ADHD-RS-IV Total
`Score between the SHP465 treatment group and the placebo group with the assumed common
`SD of 10.0, 60 subjects per group were needed to provide 90% power for a 2-sided t-test with an
`α level of 0.05. This yielded a total of 120 subjects (60 subjects on active treatment and 60
`subjects on placebo). Taking into account an expected post-randomization dropout rate of 20%,
`the randomization target was set at 150 subjects in total. It was estimated that approximately
`25% of the enrolled subjects would be 6-12 years old. The final total number of subjects
`randomly assigned between the 2 groups was to be calculated in the blinded interim analysis,
`based on the estimate of the pooled variance.
`
` A
`
` blinded interim analysis for sample size re-estimation was performed based on all subjects
`randomly assigned as of Aug 12, 2015 (the interim cohort), among which 118 subjects were in
`the FAS and 107 subjects completed the study. Based on the derived pooled SD of 13.66, the
`recalculated sample size was 110 subjects in each treatment group, or 220 subjects in total,
`without changing other original assumptions. Taking into account an expected post-
`randomization dropout rate of 20% for subjects not in the interim cohort, the overall
`randomization target was set at 264 subjects.
`
`
`3.2.2.2 Study SHP465-306
`
`3.2.2.2.1 Primary Analyses
`
`The primary efficacy endpoint was analyzed by using the linear mixed-effects model for
`repeated measures (MMRM) with treatment group, visit, and the interaction of treatment group
`with visit as factors, baseline adult ADHD-RS with prompts total score as a covariate, and the
`interaction of baseline adult ADHD-RS with prompts total score with visit adjusted in the model.
`
`The key secondary efficacy endpoint was analyzed using the same analysis method (MMRM) as
`for the primary efficacy endpoint. The baseline CGI-S score was used as the covariate. The
`primary contrast of interest was at Visit 6 (Week 4) for the specific SHP465 treatment group
`compared with placebo.
`
`In order to protect the study-wide Type I error at the 2-sided 0.05 for testing across the primary
`and key secondary hypotheses, the fixed-sequence test procedure was applied in the following
`order based on the MMRM:
`
`
`• SHP465 37.5 mg vs. placebo on change from baseline adult ADHD-RS with prompts
`total score at Visit 6 (Week 4)
`• SHP465 12.5 mg vs. placebo on change from baseline adult ADHD-RS with prompts
`total score at Visit 6 (Week 4)
`• SHP465 37.5 mg vs. placebo on CGI-I at Visit 6 (Week 4)
`
`
`
`Reference ID: 4100074
`
`15
`
`
`
`• SHP465 12.5 mg vs. placebo on CGI-I at Visit 6 (Week 4)
`
`3.2.2.2.2 Additional Analyses
`
`The CGI-I categories were dichotomized into 2 categories: “improved” (which included the
`categories of “very much improved” and “much improved”) and “not improved” (which included
`all other assessed categories grouped together). The key secondary efficacy measurement was
`analyzed using the proportion of subjects with an “improved” CGI-I measurement at Visit 6
`(Week 4) using a Cochran-Mantel-Haenszel test that was stratified by age group and CGI-S
`value at baseline. If missing data exist at Visit 6 (Week 4), the visit was imputed by carrying
`forward the last post-baseline observation value.
`
`3.2.2.2.3 Sensitivity Analyses
`
`MMRM relies on the assumption that the missing data mechanism follows the missing at random
`(MAR) scenario, assuming that the probability of missing data is unrelated to the unobserved
`value itself, after controlling for observed data. For both the primary efficacy endpoint and the
`key secondary efficacy endpoint, two sensitivity analysis models that assume different missing
`not at random (MNAR) mechanisms were carried out to examine the robustness of the MMRM
`analysis results using pattern-mixture models.
`
`Model 1 - Placebo multiple imputation based on the distribution of placebo group responses
`over time, assumed dropouts with missing values on the active treatment follow placebo pattern,
`i.e., the means and the intra-subject correlations based on the placebo responses will apply. The
`MNAR assumption is implemented by extracting posterior mean and covariance in a multiple
`imputations process based on placebo patients, and applied to all SHP465 dropouts.
`
`Step 1: Imputations
`A total of 300 sets of posterior mean and co-variance estimates are extracted from the SAS MI
`procedure using the available non-missing placebo data. One hundred of the posterior sets will
`be applied to each SHP465 treatment group respectively, the other 100 applied to the placebo
`group. One set of imputations for all missing values will be generated based on each variation of
`posterior estimates. All 100 sets for imputations within a treatment group will be ordered from 1
`to 100, and combined between SHP465 treatment groups and placebo, for a total of 100
`completely imputed data sets.
`
`Step 2: Analysis of complete data sets
`The primary endpoint will be analyzed for each of the 100 complete data sets with imputed data
`using an ANCOVA with treatment as the factor and the baseline value as a covariate.
`
`Step 3: Inference
`The LS mean difference estimates will be averaged and the associated standard errors will be
`summarized based on within-imputation and between-imputation variance using the SAS
`MIANALYZE procedure to yield a final estimate with associated 95% CI and p-value.
`
`
`
`
`Reference ID: 4100074
`
`16
`
`
`
`Model 2 - Multiple imputations with penalties applied to dropouts, assumed subjects who
`discontinue have worse changes than that predicted using MAR after discontinuation by a
`penalty. The MNAR assumption is implemented by applying penalties to missing items in a
`multiple imputations process based on treatment-specific multivariate normal distribution for
`response. The penalty applied is a fraction of the estimated standard deviation for the primary
`and key secondary endpoints, and 5 different penalties were applied.
`
`Step 1a: Imputations
`Missing data will be multiply imputed for 100 times based on a treatment specific, multivariate
`normal distribution of the response over time using the SAS MI procedure with treatment in the
`BY statement. This step is based on the MAR assumption.
`
`Step 1b: Application of penalty
`A penalty will then be applied to the multiply imputed values at Visit 6 (Week 4). The penalty
`wil