ABBREVIATED STATISTICAL ANALYSIS PLAN A Phase 2b, Double-Blind, Placebo-Controlled, Multinational, Multicenter, Randomized Study Evaluating the Safety and Efficacy of Intracoronary Administration of MYDICAR® (AAV1/SERCA2a) in Subjects with Heart Failure PROTOCOL No: CELL-004, Version 6, Amendment 5 (CELL-004-V6-A5) EudraCT number 2012-001700-37 NIH OBA Protocol No: 1205-1167 STUDY NAME: Calcium Up-Regulation by Percutaneous Administration of Gene Therapy In Cardiac Disease Phase 2b (CUPID Phase 2b Trial) Name of Investigational Medicinal Product: MYDICAR® (AAV1/SERCA2a) Development Phase: Sponsor: 2b Celladon Corporation 12760 High Bluff Drive, Suite 240 San Diego, CA 92130-2019 Version 3 26 November 2013 25 May 2012, Version 1 31 July 2013, Version 2 Current Version Number: Release Date: Version History: Confidentiality Statement This document contains confidential information that must not be disclosed to anyone other than the Sponsor and Sponsor representatives, the Investigator Team, regulatory authorities, and members of institutional review boards, ethics committees or safety review boards. 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 1 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 Table of Contents TABLE OF CONTENTS .................................................................................................. 2 ABBREVIATIONS, ACRONYMS, AND DEFINITIONS............................................ 4 1. PREFACE .............................................................................................................. 6 2. PURPOSE .............................................................................................................. 6 3. STUDY OBJECTIVES, TREATMENTS AND ENDPOINTS ......................... 6 3.1 3.2 3.3 OBJECTIVES............................................................................................................................ 6 TREATMENT GROUP COMPARISONS ....................................................................................... 7 STUDY ENDPOINTS ................................................................................................................. 7 3.3.1 3.3.2 3.3.3 4. STUDY DESIGN ................................................................................................... 8 4.1 4.2 5. RANDOMIZATION AND BLINDING........................................................................................... 9 SAMPLE SIZE .......................................................................................................................... 9 ANALYSIS POPULATIONS............................................................................. 10 5.1 5.2 5.3 5.4 6. INTENT-TO-TREAT POPULATION .......................................................................................... 10 MODIFIED ITT POPULATION ................................................................................................ 10 PER PROTOCOL POPULATION ............................................................................................... 10 SAFETY POPULATION ........................................................................................................... 10 CONSIDERATIONS FOR DATA HANDLING, PRESENTATION, AND ANALYSIS .......................................................................................................... 10 6.1 6.2 PROGRAMMING ENVIRONMENT ........................................................................................... 10 TABLES, FIGURES, AND LISTINGS ........................................................................................ 10 6.2.1 6.2.2 6.3 6.4 6.5 6.6 TABLES AND FIGURES ....................................................................................................... 10 LISTINGS ........................................................................................................................... 10 DATA ANALYSIS .................................................................................................................. 11 MISSING DATA ..................................................................................................................... 11 STUDY PERIOD AND TIME POINT DEFINITIONS .................................................................... 11 6.5.1 6.5.2 6.5.3 6.5.4 6.5.5 6.5.6 6.5.7 6.5.8 7. EFFICACY ENDPOINTS ......................................................................................................... 7 EXPLORATORY EFFICACY ENDPOINTS ................................................................................ 8 SAFETY ENDPOINTS ............................................................................................................ 8 SCREENING PERIOD ........................................................................................................... 11 BASELINE .......................................................................................................................... 11 DAY 0 ............................................................................................................................... 11 12-MONTH ACTIVE OBSERVATION PERIOD....................................................................... 11 LTFU PERIOD ................................................................................................................... 11 PRIMARY ANALYSIS DATA CUTOFF .................................................................................. 12 EOS, SUBJECT-LEVEL ....................................................................................................... 12 EOS, STUDY-LEVEL.......................................................................................................... 12 VISIT WINDOWS ................................................................................................................... 12 STUDY POPULATION ..................................................................................... 13 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 2 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 8. EFFICACY ANALYSES .................................................................................... 13 8.1 8.2 8.3 8.4 PRIMARY EFFICACY ANALYSIS ............................................................................................ 13 SECONDARY EFFICACY ANALYSES ...................................................................................... 14 EXPLORATORY EFFICACY ANALYSIS ................................................................................... 14 STATISTICAL METHODOLOGY FOR EFFICACY ANALYSES .................................................... 14 8.4.1 8.4.2 8.4.3 9. SURVIVAL ANALYSES ....................................................................................................... 14 CHANGE FROM BASELINE ANALYSES ............................................................................... 19 MULTIPLICITY ................................................................................................................... 19 SAFETY ANALYSES ......................................................................................... 19 9.1 9.2 9.3 PRIMARY ANALYSIS DATA CUTOFF SAFETY PARAMETERS ................................................. 19 STUDY-LEVEL EOS SAFETY PARAMETERS .......................................................................... 20 STATISTICAL METHODOLOGY FOR SAFETY ANALYSES ....................................................... 20 10. DATA MONITORING COMMITTEE ............................................................ 20 11. CLINICAL ENDPOINTS COMMITTEE........................................................ 21 12. SEQUENCE OF PLANNED ANALYSES ....................................................... 21 12.1 12.2 12.3 13. INTERIM ANALYSES ............................................................................................................. 21 PRIMARY ANALYSIS ............................................................................................................. 21 STUDY-LEVEL EOS ANALYSES ........................................................................................... 21 REFERENCES .................................................................................................... 21 FIGURES Figure 1. Schematic Diagram of Study ............................................................................... 9 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 3 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 Abbreviations, Acronyms, and Definitions 12-Month Active Observation Period Subject-level period from the day of investigational medicinal product administration (Day 0) to completion of the 12-month visit or death, transplant, LVAD, or permanent loss to follow-up prior to Month 12. 6MWT 6-Minute Walk Test AAV adeno-associated virus AAV1 adeno-associated virus serotype 1 AE adverse event, adverse experience ANCOVA analysis of covariance βR logarithm of the hazard ratio for recurrent events βT logarithm of the hazard ratio for terminal events BNP brain natriuretic peptide CEC Clinical Endpoints Committee COMPANION Comparison of Medical Therapy, Pacing, and Defibrillation in Chronic Heart Failure trial CSR Clinical Study Report CUPID Calcium Up-regulation by Percutaneous Administration of Gene Therapy in Cardiac Disease trial CV Cardiovascular DMC Data Monitoring Committee DNA DRP deoxyribonucleic acid DNAase resistant particle ECG Electrocardiogram ELISpot enzyme-linked immunospot EOS End of Study EOS, Study-Level The point in time after the Primary Analysis Data Cutoff has been reached and all subjects have completed a cumulative total of at least 24 months of follow-up since IMP administration EOS, SubjectLevel The point in time after the Primary Analysis Data Cutoff has been reached and the subject has completed a cumulative total of at least 24 months of follow-up since Investigational Medicinal Product Administration exp exponent of HF heart failure HR hazard ratio 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 4 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 Abbreviations, Acronyms, and Definitions ICD implantable cardioverter defibrillator ICH International Conference on Harmonization IMP investigational medicinal product ITT intent to treat JFM joint frailty model KCCQ Kansas City Cardiomyopathy Questionnaire LTFU Period Defined by subject and starts immediately after the subject completes the 12-Month Active Observation Period and ends at the subject-level End of Study. LVAD left ventricular assist device MedDRA Medical Dictionary for Regulatory Activities MITT modified intent to treat NAb neutralizing antibody NT-proBNP N-terminal prohormone brain natriuretic peptide NYHA New York Heart Association PP per protocol Primary Analysis Data Cutoff The study-level point in time at which (1) all patients except those who experienced a terminal event or were lost to follow-up have completed the full 12-Month Active Observation Period and (2) at least 186 adjudicated HF-related hospitalizations and/or episodes of ambulatory worsening HF have occurred. The Primary Analysis Data Cutoff is the trigger for locking the clinical database, unblinding treatment assignments, and performing all planned analyses. PT Preferred Term SAE serious adverse event SAP Statistical Analysis Plan SERCA sarcoendoplasmic reticulum Ca2+ ATPase SERCA2a 2a isoform of SERCA SOC System Organ Class TEAE treatment-emergent adverse event WHF worsening heart failure WHOdd World Health Organization drug dictionary 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 5 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 1. PREFACE This Statistical Analysis Plan (SAP) describes the planned statistical analyses and reporting for Celladon Corporation (Celladon) Protocol Number CELL-004, entitled A Phase 2b, Double-Blind, Placebo-Controlled, Multinational, Multicenter, Randomized Study Evaluating the Safety and Efficacy of Intracoronary Administration of MYDICAR® (AAV1/SERCA2a) in Subjects with Heart Failure, Version 6, Amendment 5 (CELL-004-V6-A5), dated 21 October 2013. It is a supplement to the clinical protocol, which should be referred to for details regarding the study objectives and design. This purpose of this phase 2b study is to assess the safety and efficacy of MYDICAR® (AAV1/SERCA2a) in the treatment of patients with moderate to advanced heart failure due to systolic dysfunction. The structure and content of this SAP provides sufficient detail to meet the requirements identified by the Food and Drug Administration (FDA) and International Conference on Harmonisation (ICH) E9: Guidance on Statistical Principles in Clinical Trials.1,2 All work planned and reported for this SAP will follow internationally accepted guidelines published by the American Statistical Association for statistical practice. The following documents were reviewed in preparation of this SAP: Clinical Research Protocol CELL-004 ICH E9 Guidance on Statistical Principles for Clinical Trials ICH E6 Guideline on Good Clinical Practice ICH E8 General Considerations for Clinical Trials ICH E3 Structure and Content of Clinical Study Reports 2. PURPOSE The purpose of this SAP is to outline the analyses planned to support the completion of the Clinical Study Report (CSR) for Protocol CELL-004. The planned analyses identified in this SAP will be included in regulatory submissions and/or future manuscripts. Exploratory analyses not identified in this SAP may be performed to support the clinical development program. Any unplanned analyses performed but not identified in this SAP will be clearly identified in the CSR. 3. STUDY OBJECTIVES, TREATMENTS AND ENDPOINTS 3.1 Objectives The primary objective of this study is to determine the efficacy of a single intracoronary infusion of 1 x 1013 DNAase resistant particles (DRP) MYDICAR® added to an optimal HF regimen in patients with ischemic or non-ischemic cardiomyopathy and moderate to advanced symptoms of HF by reducing the frequency of and/or delaying HF-related hospitalizations 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 6 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 and episodes of ambulatory worsening HF (WHF) (recurrent events) compared to placebotreated subjects. The target population for this study is moderate to advanced HF patients, especially those with high risk of recurrent events. Based on published studies3-5, this population is characterized by having either elevated N-terminal prohormone brain natriuretic peptide (NT-proBNP)/BNP within 30 days of screening and/or a recent (within 6 months of screening) hospitalization for HF. Secondary objectives of the study include assessment of MYDICAR® safety. 3.2 Treatment Group Comparisons Two treatment groups will be compared to each other: MYDICAR® and placebo. 3.3 Study Endpoints 3.3.1 Efficacy Endpoints 3.3.1.1 Primary Efficacy Endpoint The primary efficacy endpoint is time-to-recurrent events, defined as adjudicated HF-related hospitalizations and/or episodes of ambulatory WHF. Note: all clinical events defined as efficacy endpoints from this point forward in the SAP refer to adjudicated events. 3.3.1.2 Secondary Efficacy Endpoint The secondary efficacy endpoint is time to first terminal event, defined as all-cause death, heart transplant, or left ventricular assist device (LVAD) implantation. 3.3.1.3 Efficacy Sensitivity Analyses Endpoints 3.3.1.4 Alternate Definitions of Recurrent and Terminal Events The following alternate definitions of recurrent and terminal events will be analyzed to assess the sensitivity of the primary and secondary efficacy analyses to the definitions of recurrent and terminal events: Recurrent events defined as HF-related hospitalizations, ambulatory WHF, LVAD implantation, and/or heart transplant; Recurrent events defined as HF-related hospitalizations only; Recurrent events defined as all-cause hospitalizations and/or episodes of ambulatory WHF; Recurrent events defined as CV-related hospitalizations only; Terminal events defined as all-cause death only; Terminal events defined as all-cause death, urgent transplant, or LVAD. 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 7 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 3.3.1.5 Alternate Endpoints Rather than analyzing time-to-recurrent events and time-to-first terminal event, the following alternate endpoints will be analyzed: Time-to-first clinical event, defined as HF-related hospitalization, episode of ambulatory WHF, LVAD, transplant, or death; Number of hospitalizations per patient-year; Days of hospitalization per patient-year. 3.3.2 Exploratory Efficacy Endpoints Exploratory efficacy endpoints will be changes from baseline in: New York Heart Association (NYHA) class 6-Minute Walk Test (6MWT) KCCQ NT-proBNP 3.3.3 Safety Endpoints Safety endpoints will include: Subject disposition; Time-to-cardiovascular (CV) related death; AEs, including serious AEs (SAEs) and adjudicated clinical events; Changes from baseline in safety laboratory evaluations, HF-related medications, vital signs, physical examination findings, 12-lead electrocardiogram (ECG) parameters, and implanted cardioverter defibrillator (ICD) interrogation parameters. 4. STUDY DESIGN This is a phase 2b, multinational, multicenter, double-blind, placebo-controlled, randomized study of a single intracoronary administration of 1 x 1013 DRP MYDICAR® versus placebo added to an optimal HF regimen in the treatment of HF. Up to 250 adult patients with chronic systolic HF due to ischemic or non-ischemic cardiomyopathy will be randomized in parallel in a ratio of 1:1. A schematic of the study design is shown in Figure 1 below. 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 8 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 The duration of long-term follow-up is variable for each subject. Each subject should be followed quarterly in the Long-Term Follow-Up Period until End of Study. For some subjects enrolled early in the study, Long-Term FollowUp will continue past Month 24. This applies to any subject who completed or was discontinued from the 12-Month Active Observation Period. Clinical events occurring during the Long-Term Follow-Up Period will be adjudicated by the Clinical Endpoint Committee until End of Study. Figure 1. Schematic Diagram of Study Study visits will occur at Screening, Day 0 for IMP administration, and at Months 1, 3, 6, 9 and 12 for safety and efficacy assessments. After the 12-Month Active Observation Period, subjects will be followed quarterly (e.g., Months 15, 18, 21, 24, etc.) in the Long-Term Follow-Up (LTFU) Period for collection of information on clinical outcomes until death or their (subject-level) EOS. 4.1 Randomization and Blinding Subjects will be assigned in a 1:1 ratio to MYDICAR® or placebo. Randomization will be stratified by country and the ability to walk between 150 to 425 meters in the 6MWT. Subjects, all sponsor personnel, all site personnel performing direct care and assessments of subjects, and the Clinical Endpoints Committee (CEC) (Section 11) will be blinded to randomized treatment assignments until the Primary Analysis Data Cutoff is reached, all data queries have been resolved, and the database has been locked for the primary analysis. The DMC (Section 10) will be unblinded to treatment assignment throughout the study. 4.2 Sample Size The sample size for this study was based on Monte-Carlo simulations. Up to 250 subjects randomized with an estimated total of 186 recurrent events at a median follow-up time of 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 9 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 18 months will provide approximately 83% power at the 0.05 two-sided significance level to detect at least a 45% risk reduction (hazard ratio [HR] of 0.55) based on the primary efficacy analysis of time-to-recurrent events in the presence of terminal events. Simulation parameters were based on 12-month and 3-year data from the previous phase 2 trial and a previously published large HF clinical trial, COMPANION.6 5. ANALYSIS POPULATIONS 5.1 Intent-to-Treat Population The Intent-to-Treat (ITT) Population will be all subjects who were randomized. Subjects will be summarized and analyzed in the treatment group to which they were randomized, regardless of the treatment they actually received. 5.2 Modified ITT Population The Modified ITT (MITT) population will be all subjects in the ITT Population who received the IMP. 5.3 Per Protocol Population The Per Protocol (PP) population will be subjects in the ITT Population with no major protocol violations (including early unblinding). The list of subjects in the PP Population will be compiled in a blinded fashion prior to database lock. 5.4 Safety Population The Safety Population will be the same as the MITT Population. 6. CONSIDERATIONS FOR DATA HANDLING, PRESENTATION, AND ANALYSIS 6.1 Programming Environment SAS® version 9.3 or higher (SAS Institute, Cary, NC) will be used for statistical analyses and the production of tables, figures, and listings. 6.2 Tables, Figures, and Listings 6.2.1 Tables and Figures Tables and figures will present data summaries and/or analyses for the appropriate study population; e.g., summaries of safety parameters will be shown for the Safety Population only. 6.2.2 Listings Listings will present all data collected for subjects in the ITT Population. 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 10 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 6.3 Data Analysis Categorical variables will be summarized as frequencies and percentages in each category. Percentages will be reported to one decimal place. Continuous variables will be summarized by numbers of subjects, means, standard deviations, medians, and ranges. All statistical tests will be two-sided. All statistical hypotheses will be tested at the 0.05 significance level. 6.4 Missing Data Listings will present data as reported. Missing or partially missing dates that are required for date-dependent definitions (e.g., treatment-emergent AEs, concomitant medications) will be assumed to be the most conservative date possible. 6.5 Study Period and Time Point Definitions 6.5.1 Screening Period The Screening Period is defined by subject as the 30 days prior to IMP administration. 6.5.2 Baseline Baseline is defined by subject as immediately prior to IMP administration. A baseline observation is the last non-missing observation prior to IMP administration. 6.5.3 Day 0 Day 0 is defined by subject as the day of IMP administration. 6.5.4 12-Month Active Observation Period The 12-Month Active Observation Period is defined by subject as the period from IMP administration (Day 0) to completion of the Month 12 study visit, terminal event during the first 12 months, or permanent loss to follow-up during the first 12 months after IMP administration. Subjects who are still alive and not lost to follow-up at the end of the 12Month Active Observation Period will be rolled over into the LTFU Period (see Section 6.5.5). 6.5.5 LTFU Period The LTFU Period is defined by subject and starts immediately after the subject completes the 12-Month Active Observation Period and ends at their subject-level EOS (see Section 6.5.7). The purpose of the LTFU Period is to monitor for delayed toxicity that might be associated with gene transfer. Subjects must complete a minimum cumulative total of 24 months of observation/follow-up (time in the 12-Month Active Observation Period plus time in the LTFU Period). 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 11 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 6.5.6 Primary Analysis Data Cutoff The Primary Analysis Data Cutoff is declared at the study level when: 1. All subjects have completed the full 12-Month Active Observation Period; and 2. A total of at least 186 adjudicated HF-related hospitalizations and/or episodes of ambulatory WHF have occurred. The Primary Analysis Data Cutoff is the trigger for locking the clinical database, unblinding the treatment assignments, and performing all planned analyses. 6.5.7 EOS, Subject-Level EOS is declared at the subject-level after the Primary Analysis Data Cutoff has been reached and the subject has completed a cumulative total of at least 24 months of observation/followup since IMP administration. Subjects can reach their EOS before the overall study is completed (see Section 6.5.8). 6.5.8 EOS, Study-Level EOS at the study level is declared after the Primary Analysis Data Cutoff has been reached and all subjects have reached the end of their LTFU period; i.e., have each completed a cumulative total of at least 24 months of follow-up since IMP administration. 6.6 Visit Windows Study visits will include Screening, Baseline, Day 0 and nominal months indicated in Table 1 below. Visit windows will be used in the analysis of this study and will be based on number of relative days, defined as: Number of relative days = date of observation – Day 0 date + 1 Visit windows per protocol and to be used in analyses are shown in Table 1. 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 12 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 Table 1. Study Time Point Relative Day Ranges and Minimum/Maximum Days for Analyses Nominal Visit (months) 1 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 Nominal Day 31 92 183 274 366 457 548 640 731 822 914 1005 1096 1188 1279 1370 1461 Min Day Per Protocol 24 85 176 267 359 443 534 626 717 808 900 991 1082 1174 1265 1356 1447 Max Day Per Protocol 38 99 190 281 373 471 562 654 745 836 928 1019 1110 1202 1293 1384 1475 Min Day for Analyses 1 63 139 230 321 409 504 595 687 778 869 961 1052 1143 1235 1326 1417 Max Day for Analyses 62 138 229 320 408 503 594 686 777 868 960 1051 1142 1234 1325 1416 1507 7. STUDY POPULATION The ITT Population will be used for listings and summaries of study population parameters. Some listings and summaries may also be done for the MITT and/or PP populations. The study population will be described in terms of subject disposition; analysis populations; dates of screening, informed consent, randomization, and IMP administration; other details of IMP administration; satisfaction of inclusion/exclusion criteria and inclusion/exclusion waivers granted; demographics, HF etiology, medical history (CV and non-CV), results of chest x-ray and angiography at Screening; dates of completion of study visits and study periods; and protocol violations and deviations. 8. EFFICACY ANALYSES The ITT Population will be used for listings of efficacy parameters. Efficacy summaries and analyses will be done for the ITT, MITT and PP populations. Efficacy will be described in terms of clinical events, hospitalization rates and duration, NYHA class, 6MWT, KCCQ, and NT-proBNP. 8.1 Primary Efficacy Analysis The primary efficacy analysis will be the estimation of treatment effect on the primary efficacy endpoint, time-to-recurrent events defined as HF-related hospitalizations and/or 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 13 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 episodes of ambulatory WHF, in the ITT Population at the Primary Data Analysis Cutoff. The null and alternative hypotheses for the primary efficacy analysis are: Ho: βR = 0 HA: βR < 1 where exp(βR) is the recurrent event HR for MYDICAR® compared to placebo. Occurrence, timing, and frequency of HF-related hospitalizations, episodes of ambulatory WHF, and all other clinical events will be presented as listings, summary tables, and/or graphical displays. 8.2 Secondary Efficacy Analyses Secondary efficacy analyses will comprise the secondary efficacy endpoint analysis as well as analyses of both the primary and secondary endpoints in alternate analysis populations and at different study time points. The secondary efficacy endpoint is time-to-first terminal event defined as all-cause death, transplant, or LVAD implantation. The null and alternative hypotheses for analysis of the secondary efficacy endpoint are: Ho: βT = 0 HA: βT < 1 where exp(βT) is the terminal event HR for MYDICAR® compared to placebo. 8.3 Exploratory Efficacy Analysis Exploratory efficacy endpoints will be change from baseline in NYHA class, 6MWT, KCCQ, and NT-proBNP. 8.4 Statistical Methodology for Efficacy Analyses 8.4.1 Survival Analyses The semi-parametric JFM will be used to estimate the recurrent event HR in the presence of terminal events, i.e., informative censoring, and to simultaneously estimate the terminal event HR. Observation ends for subjects who experience a terminal event. Subjects who do not experience a terminal event by the analysis cutoff time point are censored at that time point. Subjects lost to follow-up before the analysis cutoff time point are censored at the time of last contact. A recurrent event and a terminal event cannot happen at the same time; e.g., if a subject gets a heart transplant during a hospitalization, the heart transplant will be counted as a terminal event and the hospitalization will not be counted. The JFM will be implemented using the SAS® procedure NLMIXED.7 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 14 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 8.4.1.1 Statistical Model Background and Rationale The study objective is to demonstrate the safety and efficacy of a single intracoronary infusion of 1 x 1013 DRP MYDICAR® in patients with ischemic or non-ischemic cardiomyopathy and moderate to advanced symptoms of HF by reducing the frequency and/or delaying recurrent events (HF-related hospitalizations and/or ambulatory WHF) compared to placebo-treated subjects. Several aspects of the moderate to advanced HF patient population may complicate the estimation of treatment effect on disease-related events. This patient population is characterized by frequent HF-related hospitalizations, high mortality and continued worsening of the disease, often leading to surgical interventions such as LVAD implantation and/or heart transplant. Patients who are hospitalized or experience ambulatory WHF may be at an increased risk of additional occurrences of these events. Therefore, recurrent events within a patient may not be independent. Patients with high recurrent event rates may also have increased risk of terminal events; i.e., risks of recurrent and terminal events are likely to be correlated and may need to be jointly estimated to avoid biased estimates of treatment effect. Further, terminal events in this context represent informative censoring, a violation of one of the assumptions of traditional survival analysis. Other challenges include the possibility that the impact of a treatment on the risk of recurrent events may be different than on the risk of a terminal event, and random between-patient variations (unobserved covariates) may impact treatment outcomes. The semi-parametric JFM addresses these challenges. The JFM accommodates: Multiple outcomes within subjects by accounting for the resulting within-subject correlation through the estimation of a subject-specific frailty parameter, thereby avoiding underestimation of the treatment effect variance. Within-subject correlation between recurrent and terminal events by modeling the subject-specific frailty parameter that is shared between recurrent and terminal event processes and is further described by an estimated frailty coefficient that allows frailties to have differing effects on recurrent and terminal events. The competing risk of terminal events in the analysis of recurrent events (subjects who are “censored” due to a terminal event can theoretically no longer experience a recurrent event). Simultaneous estimation of treatment effect on terminal events. Differential treatment effect for recurrent and terminal events. Following the model of Liu, et al., 20047, and as specified in Rondeau, et al., 20108, the joint frailty model for the hazard functions for recurrent and terminal events, respectively, is: 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 15 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 𝑟𝑖 (𝑡|𝜔𝑖 ) = 𝜔𝑖 𝑟0 (𝑡) 𝑒𝑥𝑝(𝛽1′ 𝑍𝑖 (𝑡)) = 𝜔𝑖 𝑟𝑖 (𝑡) { 𝜆𝑖 (𝑡|𝜔𝑖 ) = 𝜔𝑖𝛼 𝜆0 (𝑡)𝑒𝑥𝑝(𝛽2′ 𝑍𝑖 (𝑡)) = 𝜔𝑖𝛼 𝜆𝑖 (𝑡) where: 𝜔𝑖 = patient-specific frailty allowing for random between-subject differences in baseline hazards due to unobserved covariates, assumed to be independently gamma-distributed with mean 1 and variance θ; 𝑟𝑖 = recurrent events hazard function for the i-th patient conditioned on 𝜔𝑖 (frailty is shared between recurrent and terminal events); 𝜆𝑖 = terminal events hazard function for the i-th patient conditioned on 𝜔𝑖 ; 𝛽1′ , 𝛽2′ = covariate coefficients for recurrent and terminal events, respectively; 𝑍𝑖 = covariate value (e.g., treatment group) for the i-th patient; α = frailty coefficient allowing for differential effects of 𝜔𝑖 on recurrent and terminal events. 8.4.1.2 Sensitivity Analyses A series of sensitivity analyses will be conducted using the ITT population. Statistical significance of treatment effects using alternative approaches will not be required. 8.4.1.2.1 Alternative Definitions of Recurrent and Terminal Events The following JFM analyses will be done to evaluate the effect of alternative definitions of recurrent and terminal events: 1. Sensitivity Analysis #1 a. Recurrent events defined as HF-related hospitalizations, ambulatory worsening HF, LVAD implantation, and heart transplant. b. Terminal events defined as death only. c. Two different ending time points (subjects still alive at the ending time point will be censored): i. Primary Analysis Data Cutoff ii. Study-level EOS (subjects still alive will be censored) 2. Sensitivity Analysis #2 a. Recurrent events defined as HF-related hospitalizations only. b. All other specifications are the same as primary efficacy analysis. 3. Sensitivity Analysis #3 a. Recurrent events defined as all-cause hospitalizations and ambulatory WHF. b. All other specifications are the same as primary efficacy analysis. 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 16 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 4. Sensitivity Analysis #4 a. Recurrent events defined as CV-related hospitalizations only. b. All other specifications are the same as primary efficacy analysis. 5. Sensitivity Analysis #5 a. Terminal events defined as death, LVAD implantation and adjudicated urgent heart transplant. If a subject has status other than 1A or 1B on the transplant list or an equivalent priority system, the subject will be censored at the time of transplant. b. All other specifications are the same as primary efficacy analysis. 6. Sensitivity Analysis #6 a. Terminal events defined as death only. Subjects receiving heart transplants or LVAD implantations will be censored at the time of surgery. Note: this differs from the primary efficacy analysis in that transplant and LVAD subjects are censored rather than terminated, meaning future clinical events may occur but will not be observed. b. All other specifications are the same as primary efficacy analysis. 8.4.1.2.2 Alternative Analytical Methods All of the following analyses will be done at the Primary Analysis Data Cutoff only. 7. Sensitivity Analysis #7 a. Treatment effect on recurrent events will be analyzed using both a nonparametric approach9 and a semi-parametric marginal model10 developed by Ghosh and Lin. 8. Sensitivity Analysis #8 a. Three “time-to-first” analyses will be done: i. Time-to-first terminal event (all-cause death, heart transplant, or LVAD); ii. Time-to-first clinical event, defined as all-cause death, heart transplant, LVAD implantation, HF-related hospitalization, or ambulatory WHF; iii. Time-to-all-cause death. b. Product-limit estimates of median time-to-first event, including 95% confidence intervals, will be calculated. The log-rank test will be used to compare treatment groups. 9. Sensitivity Analysis #9 a. Number of hospitalizations per subject-year and days of hospitalization per subject-year from randomization to the Primary Analysis Data Cutoff will be analyzed using an appropriate analytical method depending on the variables’ distributions (e.g., Poisson regression). 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 17 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 b. Two categories of hospitalization will be considered: i. HF-related ii. All-cause 10. Sensitivity Analysis #10 a. In the analysis of the primary efficacy endpoint, a log-normal rather than a gamma distribution for the frailty random variable in the JFM will be assumed. 11. Sensitivity Analysis #11 a. An “as randomized” analysis of the primary endpoint will be done in which the randomization strata (country and ability to walk between 150 to 425 meters in the 6MWT) will be included in the model as fixed-effect covariates. Countries with very few subjects (< 10) may be combined with other countries with similar HF management. 12. Sensitivity Analysis #12 a. Univariable analyses of the following covariates will be done to assess confounding and modifying effects in the analyses of the primary and secondary efficacy endpoints: i. Geography (U.S., non-U.S.); ii. HF etiology (ischemic/non-ischemic); iii. NYHA Class (II vs. III/IV); iv. Age at the time of signing informed consent (continuous and categorized); v. Intervention (e.g., percutaneous coronary intervention, coronary bypass artery graft) performed during the first infusion attempt; vi. The number (nevent) of recurrent events (from the time of randomization) preceding any given recurrent event, initially set to 0 for each subject and increased by 1 for each recurrent event (nevent = 0 for the first event, nevent = 1 for the second event, etc.). b. Confounding effects will be assessed both by evaluating changes in HRs with and without the potential confounder as well as by subgroup analyses. c. Treatment effect modification will be assessed by subgroup analyses. As this study is not powered to statistically test for significant interactions or equivalence, we will evaluate effect modification descriptively and consider HRs in the same direction (i.e., both > 1 or both < 1) in each subgroup as evidence that treatment is similar across subgroups. d. Covariate effects on the secondary efficacy endpoint will also be assessed using a Cox proportional hazards model. 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 18 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 13. Sensitivity Analysis #13 a. As efficacy may differ depending on AAV NAb status, the primary efficacy endpoint at the Primary Analysis Data Cutoff will be analyzed both with and without patients with baseline AAV NAb titers > 1:2 (positive) or with titers < 1:2 that are classified as equivocal (although titer < 1:2 is a study inclusion criterion, some patients with positive titer at Baseline may still be infused as results from NAb samples taken immediately before infusion are not yet known at the time of infusion). In addition, separate summaries of primary and secondary efficacy endpoints will be done for patients with positive or equivocal AAV NAb titers. b. The primary efficacy endpoint will be analyzed at the Primary Analysis Data Cutoff both with and without patients with positive AAV NAb titers at Baseline, and separate summaries of primary and secondary efficacy endpoints will be done for patients with positive AAV NAb titers. 8.4.2 Change From Baseline Analyses Change from Baseline analyses of the exploratory efficacy endpoints 6MWT, KCCQ and NTproBNP will be done using analysis of covariance (ANCOVA) with the baseline observation as the covariate. Change in NYHA class will be analyzed by comparing distributions of change scores between treatment groups using a chi square test of association (or Fisher’s exact test if any change category within treatment group contains less than 5 subjects). 8.4.3 Multiplicity Although the JFM estimates treatment effect on both recurrent and terminal events simultaneously, this study has a single primary efficacy endpoint: time-to-recurrent events. Therefore, no multiplicity adjustment is required. 9. SAFETY ANALYSES The Safety Population will be used for listings, summaries and analyses of safety parameters. Safety will be assessed at the Primary Analysis Data Cutoff and at study-level EOS. 9.1 Primary Analysis Data Cutoff Safety Parameters Safety parameters at the Primary Analysis Data Cutoff will include subject disposition; timeto-CV-related death; AE incidence, severity, and relationship to IMP and IMP administration; time-to-first serious AE (SAE); changes from Baseline in safety laboratory evaluations, vital signs, physical examination findings, and 12-lead ECG parameters; ELISpot assay results (negative/positive); concomitant medications, including changes from Baseline in HF-related medications; concomitant medical/surgical procedures; ELISpot assay; and ICD interrogations. 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 19 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 AEs will be coded according to MedDRA® (Medical Dictionary for Regulatory Activities). Each reported AE will be mapped to a Preferred Term (PT) and a System Organ Class (SOC). A treatment-emergent AE (TEAE) will be defined as an AE that began or worsened during or after IMP administration. AEs recorded prior to IMP administration will be considered nontreatment-emergent. AEs with insufficient date or time information to determine whether or not they were treatment-emergent will be considered treatment-emergent. Prescription, over-the-counter, and alternative medication use will be coded to generic terms using the World Health Organization drug dictionary (WHOdd). Medications that were stopped before IMP administration will be considered “pre-infusion.” All other medications will be considered “concomitant.” Medications recorded with insufficient exposure dates to determine whether or not they were concomitant will be considered concomitant. 9.2 Study-Level EOS Safety Parameters Safety parameters to be assessed at study-level EOS will include subject disposition, adjudicated clinical events and responses to an in-person questionnaire designed to ascertain new conditions and exacerbations of existing conditions. 9.3 Statistical Methodology for Safety Analyses For all statistical analyses of safety parameters, it is understood that non-significance is not synonymous with equality. Time-to-CV-related death and time-to-first SAE will be compared between treatment groups using product-limit estimates of median survival times with 95% confidence intervals and will be graphically presented using Kaplan-Meier step function plots; p-values from log-rank tests will be reported. In CV-related death analyses, subjects still alive at the data cutoff will be censored at that time point and subjects who died from non-CV reasons prior to the data cutoff will be censored at their death dates. In the SAE analysis, subjects with no reported SAEs at the data cutoff will be censored. In both analyses, subjects lost to follow up will be censored at the date of last contact. 10. DATA MONITORING COMMITTEE Unblinded DMC reviews will be conducted at periodic, pre-specified milestones of the study. The DMC will be advisory to the sponsor. The sponsor will be responsible for reviewing the DMC recommendations promptly, to decide whether to continue or terminate the trial, and to determine whether amendments to the protocol or changes in study conduct are required. Study termination based on findings from the DMC can only be due to safety issues. The study will not be terminated due to efficacy, even if recommended by the DMC, because subject follow-up data must be acquired to fulfill the safety requirements of the study. The DMC will be an independent multidisciplinary group consisting of at least three individuals, 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 20 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 one of whom one must be an interventional cardiologist, one a cardiologist with expertise in HF, and one a statistician. 11. CLINICAL ENDPOINTS COMMITTEE A blinded CEC will be established to review clinical events for adjudication in a consistent and unbiased manner throughout the course of the study; i.e., the CEC will ensure that all clinical events that are reported by sites are judged uniformly, using the same criteria across all sites and throughout the study. No member from the sponsor or the sponsor’s agents, consultants or designees will be allowed to be present during the CEC deliberations. All policies and procedures will be specified in the written CEC Manual of Operations (or equivalent). 12. SEQUENCE OF PLANNED ANALYSES 12.1 Interim Analyses There are no planned interim analyses. 12.2 Primary Analysis At the time of the Primary Analysis Data Cutoff and after the clinical database has been quality-checked and cleaned, the database will be locked. 12.3 Study-Level EOS Analyses Analyses that require all subjects to reach EOS (i.e., select safety and sensitivity analyses) will be done at study-level EOS using the locked, unblinded database. 13. REFERENCES 1. E9 Guidance on Statistical Principles in Clinical Trials. International Conference on Harmonisation (ICH); 1998. 2. Guidance for Industry: E9 Statistical Principles for Clinical Trials. Food and Drug Administration: Center for Drug Evaluation and Research (CDER) and Center for Biologics Evaluation and Research (CBER); 1998. 3. Packer M. Effect of Carvedilol on the Morbidity of Patients With Severe Chronic Heart Failure: Results of the Carvedilol Prospective Randomized Cumulative Survival (COPERNICUS) Study. Circulation 2002;106:2194-9. 4. Fonarow G, Committee. ASA. The Acute Decompensated Heart Failure National Registry (ADHERETM): Opportunities to Improve Care of Patients Hospitalized With Acute Decompensated Heart Failure. Rev Cardiovasc Med 2003;4:S21-30. 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 21 of 22 Celladon CUPID Phase 2b CELL-004 Abbreviated Statistical Analysis Plan Version 3, 26 November 2013 5. Rogers J, McMurray J, Pocock SJ, et al. Eplerenone in Patients With Systolic Heart Failure and Mild Symptoms: Analysis of Repeat Hospitalizations. Circulation 2012;126:2317-23. 6. Anand IS, Carson P, Galle E, et al. Cardiac resynchronization therapy reduces the risk of hospitalizations in patients with advanced heart failure: results from the Comparison of Medical Therapy, Pacing and Defibrillation in Heart Failure (COMPANION) trial. Circulation 2009;119:969-77. 7. Liu L, Wolfe R, Huang X. Shared Frailty Models for Recurrent Events and a Terminal Event. Biometrics 2004;60:747-56. 8. Rondeau V. Statistical models for recurrent events and death: Application to cancer events. Mathematical and Computer Modelling 2010;52:949-55. 9. Ghosh D, Lin DY. Nonparametric analysis of recurrent events and death. Biometrics 2000;56:554-62. 10. Ghosh D, Lin D. Marginal regression models for recurrent and terminal events. Statistica Sinica 2002;12:663-88. 2012-2013 Celladon Corporation. Printed in USA. All rights reserved. CELLADON CORP. Page 22 of 22