* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download 2015 08 26 CUPID 2 manuscript CLEAN FINAL
Survey
Document related concepts
Transcript
1 1 2 Calcium Up-Regulation by Percutaneous Administration of Gene Therapy in Cardiac Disease Phase 2b (CUPID 2): a Randomised, Multinational, Double-Blind, Placebo-controlled Trial 3 4 5 Barry Greenberg, Javed Butler, G. Michael Felker, Piotr Ponikowski, Adriaan A. Voors, Akshay 6 S. Desai, Denise Barnard, MD; Alain Bouchard, Brian Jaski, Alexander R. Lyon, MD, PhD; Janice 7 M. Pogoda, Jeffrey J. Rudy, Krisztina M. Zsebo 8 9 Affiliations: UCSD Sulpizio Cardiovascular Center, La Jolla, CA, USA (Prof B Greenberg MD, 10 Prof D Barnard MD); Stony Brook University, Stony Brook, NY, USA (Prof J Butler MD); Duke 11 University School of Medicine, Durham, NC, USA (Prof G M Felker MD); Wroclaw Medical 12 University and Military Hospital, Wroclaw, Poland (Prof P Ponikowski MD); University of 13 Groningen, Groningen, Netherlands (Prof A A Voors MD); Cardiovascular Division, Brigham and 14 Women’s Hospital, Boston, MA, USA (A S Desai MD); Cardiology, PC, Birmingham, AL, USA (A 15 Bouchard MD); San Diego Cardiac Center, Sharp Memorial Hospital, San Diego, CA, USA (B Jaski 16 MD); Royal Brompton Hospital and Imperial College London, London, UK (A R Lyon MD); 17 Celladon Corporation, San Diego, CA, USA (J M Pogoda PhD, J J Rudy BS); Santa Barbara, CA, 18 USA (K M Zsebo PhD) 19 20 Corresponding Author: 21 Barry Greenberg, MD 22 Distinguished Professor of Medicine 23 Director Advanced Heart Failure Treatment Program 24 UCSD Sulpizio Cardiovascular Center 25 9444 Medical Center Dr., #7411 26 La Jolla, CA 92037-7411 27 Phone: 858-657-5267 28 Email: [email protected] 29 30 31 32 Key Words: Gene transfer therapy, heart failure, SERCA2a 33 34 Funding statement: The clinical study, data analyses, and manuscript support were 35 funded by Celladon Corporation. 2 36 Summary 37 Background Sarco/endoplasmic reticulum Ca2+ ATPase (SERCA2a) activity is deficient 38 in the failing heart. Correction of this abnormality by gene transfer may improve 39 cardiac function. CUPID 2 investigated the clinical benefits and safety of gene therapy 40 through infusion of adeno-associated virus 1 (AAV1)/SERCA2a in heart failure patients 41 with reduced ejection fraction. 42 Methods CUPID 2 was a phase 2b, multinational, double-blind, placebo-controlled 43 study of high-risk ambulatory patients with New York Heart Association class II-IV 44 symptoms, ischemic or non-ischemic aetiology, and left ventricular ejection fraction 45 ≤0·35. The study was conducted at 67 clinical centres and hospitals in the United 46 States, Europe, and Israel. Patients were randomised 1:1 via an interactive voice and 47 web response system to receive a single intracoronary infusion of 1x1013 DNase- 48 resistant particles of AAV1/SERCA2a or placebo. Randomisation was stratified by 49 country and by 6 minute walk test distance. Patients were followed for ≥12 months. 50 The primary efficacy endpoint was time to recurrent events (hospitalization, 51 ambulatory worsening heart failure treatment) analysed using a joint frailty model to 52 account for multiple, correlated events within subjects. Primary efficacy endpoint 53 analyses and safety analyses were performed on all treated patients. The trial was 54 registered with clinicaltrials.gov, number NCT01643330, and is now closed. 55 Findings Between July 9, 2012 and February 5, 2014, 1558 patients were screened and 56 250 were enrolled; 121 were infused with AAV1/SERCA2a and 122 with placebo. 57 Compared with placebo, AAV1/SERCA2a did not improve the primary endpoint (128 58 recurrent events versus 104 recurrent events; hazard ratio 0·93; 95% CI 0·53—1·65; 59 p=0·81). No safety issues were noted. 3 60 Interpretation CUPID 2 was the largest gene transfer study performed in heart failure 61 patients to date. Despite promising results from earlier studies, a single intracoronary 62 infusion of AAV1/SERCA2a at the dose tested did not improve the clinical course of 63 heart failure patients with reduced ejection fraction. 64 65 66 67 Funding Celladon Corporation. 4 68 Introduction 69 Despite advances in treatment, morbidity and mortality remain unacceptably high for 70 patients with heart failure (HF)1 and new approaches for improving outcomes are 71 needed. Identification of derangements in key pathways that regulate cardiac function 72 has provided potential novel targets for gene therapy, and evidence that vectors such 73 as adeno-associated viruses (AAVs) can deliver genes of interest to cardiomyocytes, 74 resulting in sustained transgene expression in the heart, has stimulated interest in 75 gene transfer as a strategy for treating HF. The sarco/endoplasmic reticulum Ca2+ 76 ATPase (SERCA2a) regulates cardiomyocyte contraction and relaxation by transporting 77 Ca2+ from the cytosol into the sarcoplasmic reticulum during diastole.2,3 A deficiency 78 of SERCA2a is related to HF progression.4,5 Correction of this deficiency has been 79 shown to favourably affect calcium flux and improve the function of cardiomyocytes 80 derived from failing hearts. Gene transfer of SERCA2a has also been shown to improve 81 cardiac performance and survival in experimental models of HF.4,5 Recently, we 82 reported that a single intracoronary infusion of recombinant AAV serotype 1 (AAV1) 83 delivering the SERCA2a gene to the heart had favourable effects in patients with 84 advanced HF in a pilot study.6,7 On the basis of these promising results, the Calcium 85 Up-Regulation by Percutaneous Administration of Gene Therapy in Cardiac Disease 86 Phase 2b (CUPID 2) study was designed to further assess the effects of AAV1/SERCA2a 87 therapy on clinical outcomes in a larger group of patients with moderate to severe HF 88 and reduced ejection fraction.8 89 90 METHODS 91 Study design 5 92 The CUPID Phase 2b trial (CUPID 2; NCT01643330) was a multinational, double-blind, 93 placebo-controlled, randomised study designed to investigate whether gene transfer 94 therapy with SERCA2a improved outcomes in patients with HF and reduced ejection 95 fraction. The study design has been published.8 The study was conducted at 67 centres 96 and hospitals in the United States (US), Europe, and Israel according to the principles 97 of the International Conference on Harmonisation Guideline on Good Clinical Practice 98 and the principles of the World Medical Association Declaration of Helsinki. All 99 relevant Institutional Review Board and Institutional Bio-Safety Committee approvals 100 were obtained at each site. Manufacturing information is provided in the Appendix (p 101 4). 102 103 Participants 104 Eligible patients were between 18 and 80 years of age with a diagnosis of stable New 105 York Heart Association (NYHA) class II-IV chronic HF due to ischemic or non-ischemic 106 cardiomyopathy and left ventricular ejection fraction ≤0·35 on optimal tolerated 107 stable medical therapy for at least 30 days prior to randomisation. In response to a 108 lower than anticipated pooled event rate during the early period of the trial, a 109 protocol amendment designed to increase risk for future HF events was initiated after 110 enrolment of 101 patients. This amendment required eligible patients to have 111 elevated N-terminal pro–B-type natriuretic peptide (NT-proBNP) (>1,200 pg/mL, or 112 >1,600 pg/mL if atrial fibrillation was present) or HF-related hospitalization within 6 113 months of enrolment into the study. Patients were required to have <1:2 or equivocal 114 anti-AAV1 neutralizing antibody (NAb) titres at screening. 115 Exclusion criteria included cardiac surgery, percutaneous coronary intervention, 116 valvuloplasty, or intravenous (IV) therapy for HF within 30 days prior to screening. A 6 117 comprehensive list of exclusion criteria has been published. 8 All patients provided 118 written informed consent. 119 120 Randomisation and masking 121 Following screening, patients were randomised in parallel in a 1:1 ratio to receive 122 either 1x1013 DNase resistant particles (DRP) AAV1/SERCA2a or placebo. 123 Randomisation was conducted through a fully validated and controlled interactive 124 voice and web response system provided by Almac Clinical Technologies. 125 Randomisation was stratified by country and the ability to walk between 150 and 425 126 meters or outside of these distances on the 6 minute walk test (6MWT). A blinded kit 127 was shipped to the investigative site following randomisation. All patients and 128 physicians were blinded to treatment assignment, and the company that conducted 129 randomisation was not involved with other facets of the trial. 130 131 Procedures 132 Drug was administered a single time to each patient. On day 0, before infusion of the 133 investigational product, coronary angiography was performed to determine the 134 strategy for administering AAV1/SERCA2a and to confirm that at least one coronary 135 artery had Thrombolysis in Myocardial Infarction (TIMI) flow grade 3. Infusion of the 136 investigational product was tailored to the patient and multiple infusion scenarios 137 were possible depending on the extent and distribution of coronary artery stenosis, 138 collateralization patterns, and anatomic variations. During the single administration of 139 drug, operators were instructed to provide delivery using at most three infusions 140 according to the distribution of left ventricular blood flow.8 The overall goal was to 141 achieve homogeneous delivery to the myocardium with two-thirds of the dose to the 7 142 anterolateral and one-third to the posterolateral myocardium. It was recognized that 143 multiple coronary infusion scenarios were possible based on occlusive disease and 144 collateralization patterns and investigators received instruction regarding perfusion 145 options at the time their sites were activated. An IV nitroglycerin infusion was started 146 10 to 25 minutes prior to infusion of the investigational product to enhance uptake of 147 AAV1/SERCA2a in cardiomyocytes by increasing vasodilation of the capillary bed.9 148 During the 12-month active observation period, assessments of efficacy, safety, and 149 quality of life were undertaken at months 1, 3, 6, 9, and 12. Data collection on 150 clinical endpoints continued until the primary analysis data cutoff was reached, which 151 was when all patients completed the 12-month active observation period and at least 152 186 adjudicated HF-related recurrent events had occurred. 153 154 Outcomes 155 The primary efficacy endpoint was time to recurrent events, defined as 156 hospitalizations due to HF or ambulatory treatment for worsening HF. The secondary 157 efficacy endpoint was time to first terminal event, defined as all-cause death, heart 158 transplant, or durable mechanical circulatory support device (MCSD) implantation. All 159 primary and secondary endpoints were reviewed by a blinded clinical endpoints 160 committee (Appendix p 3) and adjudicated according to standardized definitions. 161 Adjudication criteria for these events, as well as detailed statistical methods, have 162 been previously described.8 Exploratory analyses included the effect of the 163 investigational product on change from baseline in NYHA class, exercise ability as 164 assessed by the 6MWT, quality of life as assessed by the Kansas City Cardiomyopathy 165 Questionnaire (KCCQ), and NT-proBNP. 8 166 Safety was assessed in all patients who received treatment with AAV1/SERCA2a or 167 placebo. Safety parameters included incidence and severity of adverse events and 168 time to cardiovascular-related death. 169 170 Post-treatment tissue and serum processing 171 During follow-up of patients enrolled in the study, participating centres were 172 instructed to try to obtain tissue samples from treated patients at the time of cardiac 173 transplantation, implantation of a MCSD, or at autopsy. The levels of AAV1/SERCA2a 174 were determined using methods previously described.7 In addition, AAV1 NAb testing 175 was performed in study patients using serum collected at the 6 month follow-up visit. 176 177 Statistical analysis 178 Monte Carlo simulation using background rates and correlations similar to those 179 observed in CUPID 1 estimated that 186 recurrent events in 250 patients with a 180 median follow-up time of 18 months would provide 80% power at the 0·05 two-sided 181 significance level to detect a recurrent event hazard ratio (HR) of 0·55 using a joint 182 frailty model. 183 The intention-to-treat (ITT) analysis population was defined as all randomised 184 subjects.10 A modified ITT (mITT) analysis population was also pre-specified, 185 comprising only randomised patients who received study medication.10,11 The primary 186 analysis of the primary and secondary endpoints was done at the primary analysis data 187 cutoff using the mITT population ; secondary analyses were done using the ITT 188 population (all randomised patients) and additional pre-specified populations 189 (Appendix p 5). Treatment effects on the primary and secondary endpoints were 190 estimated simultaneously by a semi-parametric joint frailty model12 implemented 9 191 using the NLMIXED procedure13 in SAS (SAS Institute, Inc., Cary, NC). This model 192 accounts for correlated recurrent events within patients and the correlation between 193 recurrent and terminal events (i.e., informative censoring). The reference time point 194 was randomisation date for the ITT population and treatment date for the mITT 195 population and for the additional pre-specified populations (e.g. excluding patients 196 who had major protocol deviations and excluding patients who were positive or 197 equivocal for neutralizing antibodies). Primary and secondary endpoints were 198 graphically depicted using the mean cumulative function14 and the survival function 199 (estimated by the PHREG procedure in SAS), respectively. Sensitivity analyses using 200 alternative models for both endpoints were also performed. 201 The trial was registered with clinicaltrials.gov, number NCT01643330. 202 203 Role of the funding source 204 This trial, including patient management, data collection, and data analysis, was 205 funded by Celladon Corporation. Celladon also provided funding for manuscript and 206 graphics support. The corresponding author had full access to all data in the study 207 and, with the support of the full author group, had final responsibility for the decision 208 to submit for publication. 209 210 Results 211 From July 9, 2012 through February 5, 2014, 1558 patients at 67 centres in the US, 212 Europe, and Israel underwent NAb prescreening for CUPID 2 (Figure 1). Of these 213 patients, 921 (59·1%) were NAb positive and 284 (18·2%) were considered ineligible for 214 other reasons, leaving 353 (22·7%) with a qualifying NAb titre (<1:2 or equivocal) who 215 were eligible for further screening. Of these patients, 103 (29·2%) were excluded for 10 216 reasons summarized in Figure 1, and 250 patients were enrolled into the study and 217 randomised. Two of 123 patients allocated to receive AAV1/SERCA2a and five of 127 218 patients allocated to placebo did not receive study drug infusion (Figure 1). The 219 remaining 121 patients who received AAV1/SERCA2a and 122 patients who received 220 placebo constituted the mITT population that was the pre-specified population for the 221 primary efficacy analysis. Over the course of the study, 5 patients (3 in mITT) 222 withdrew consent and 1 (in mITT) was lost to follow-up. 223 The participants were predominantly white and male with two-thirds from the US 224 (Table 1). A total of 135/250 (55·6%) patients had coronary artery disease and HF was 225 ascribed to an ischemic aetiology in 125/250 (51·4%) patients. Patients had moderate 226 to severe HF as evidenced by NYHA Functional Class, ejection fraction, 6MWT 227 distance, KCCQ score, and NT-proBNP level. Baseline characteristics were balanced 228 between groups. 229 Median follow-up was 17·5 months since the study extended over 30 months in order 230 to allow all randomised patients to be followed for at least 12 months. At the time the 231 last patient had been followed for 12 months, a total of 232 recurrent and 65 terminal 232 events had occurred in the mITT population. Of the 232 recurrent events that 233 qualified as primary endpoints, 128 were in the placebo group and 104 were in the 234 AAV1/SERCA2a group; most were HF hospitalizations. Treatment with AAV1/SERCA2a 235 failed to improve the rate of recurrent events (HR, 0·93; 95% confidence interval [CI] 236 0·53 to 1·65; p=0·81; Figure 2A and Table 2). Of the 65 terminal events that qualified 237 as secondary endpoints, 29 were in the placebo group and 36 were in the 238 AAV1/SERCA2a group; most were deaths (Table 2). AAV1/SERCA2a administration 239 failed to improve time to first terminal event (HR, 1·27; 95% CI 0·72 to 2·24; p=0·40; 11 240 Figure 2B). AAV1/SERCA2a treatment also did not improve time to all-cause death 241 (Figure 2C). 242 No differences between treatment groups were detected in subgroup analyses of the 243 primary endpoint (Figure 3). In a pre-specified subgroup analysis of the secondary 244 endpoint, there was a significant interaction between treatment and geography 245 (Figure 3), with a higher HR in non-US patients compared with US patients. However, 246 the number of events in the analysis of non-US patients was small (22 events in 85 247 patients), and baseline disease characteristics suggest that non-US AAV1/SERCA2a 248 patients may have had more severe illness than non-US placebo patients (Appendix 249 Web Table 1). There was no such interaction for the primary endpoint. No other 250 significant interactions were detected for pre-specified subgroup analyses, although a 251 significant interaction was observed for the non-pre-specified subgroup of patients 252 with diabetes. Post-hoc analyses of the primary and secondary endpoints stratified by 253 randomisation in the study “pre” or “post” initiation of the protocol amendment 254 designed to increase the risk for future HF events showed that there was no 255 meaningful difference in treatment effect between these subgroups. For the primary 256 endpoint, the HRs were 0·86 (95% CI 0·32 to 2·27) and 1·05 (95% CI 0·53 to 2·08) for 257 “pre” and “post” amendment patients, respectively, while for the secondary endpoint 258 the HRs were 1·14 (95% CI 0·53 to 2·44) and 1.38 (95% CI 0·59 to 3·25), respectively. 259 There were no significant differences between treatment groups for any of the 260 exploratory efficacy analyses (change from baseline in NYHA class, exercise ability as 261 assessed by the 6MWT, quality of life as assessed by the KCCQ, or levels of NT-proBNP) 262 over 12 months of follow-up. No significant treatment group differences were 263 observed in the ITT analyses or in analyses conducted in other pre-specified 264 populations (Appendix p 5). 12 265 In safety evaluations, there were 262 clinical events in placebo and 190 in 266 AAV1/SERCA2a patients (Table 3); most were hospitalizations. There were 20 deaths in 267 placebo and 25 deaths in AAV1/SERCA2a patients, 18 and 22 of which were 268 adjudicated as being due to cardiovascular causes. Comparisons of treatment- 269 emergent serious adverse events occurring in ≥2% of either treatment group identified 270 only one significant difference between groups: placebo patients had a higher rate of 271 implantable defibrillator insertion than AAV1/SERCA2a patients (6/122 [4·9%] versus 272 0/121 [0%]; p=0·03) (Appendix Web Table 2). 273 Since the results of CUPID 2 were divergent from those of CUPID 1, which showed a 274 beneficial effect of AAV1/SERCA2a on HF outcomes, post-hoc analyses were performed 275 to provide potential insights into the differences in the efficacy of AAV1/SERCA2a 276 therapy in CUPID 2 compared with CUPID 1. There were no obvious important 277 differences in study population characteristics between these trials except for a 278 higher use of cardiac resynchronization therapy in CUPID 1 (Appendix Web Table 3), 279 which reflected the higher usage of this treatment modality in the exclusively US 280 population in CUPID 1 as compared with the international population enrolled in CUPID 281 2. A review of manufacturing processes identified a difference in the proportion of 282 empty viral capsids (containing only the protein capsid and not the single stranded 283 DNA) between CUPID 1 (85%) and CUPID 2 (25%) (Appendix Web Table 4), which may 284 have affected transduction efficiency (Appendix p 4 and Web Figure 1). 285 We assessed the presence of AAV1/SERCA2a in cardiac tissues from patients whose 286 condition deteriorated requiring either transplant, or MCSD implantation and patients 287 from whom cardiac tissue was obtained at autopsy. A total of 23 heart tissue samples 288 were obtained from 7 patients (Appendix Web Table 5). The levels of vector DNA in 289 these tissues (approximate median of 43 copies/μg DNA; range <10 to 192) were at the 13 290 lower end of the threshold for dose response curve in pharmacology studies (<500 291 copies/μg DNA). Although it is difficult to determine the number of cells that were 292 transduced due to the variable ploidy of cardiomyocytes in advanced HF patients,15 293 these levels are most consistent with the likelihood that only a very small percentage 294 of cardiomyocytes were expressing AAV1-delivered SERCA2a in the myocardium of 295 these patients. 296 Testing for the presence of AAV1 NAbs showed the expected high rate of 297 seroconversion in patients treated with AAV1/SERCA2a, but not in those who were 298 treated with placebo (Appendix p 5 and Web Table 6). These NAbs are not expected to 299 have influenced the level of SERCA2a expression, as an antibody response occurs days 300 to weeks after the cells take up the AAV vector. Testing for the presence of an anti- 301 AAV1 specific CD8 T cell response was conducted and found to be mostly negative, so 302 a cellular immune response cannot explain the low level of transduced cells and lack 303 of efficacy. 304 305 Discussion 306 CUPID 2 was the largest study of gene transfer performed in a HF population to date 307 and the first to look at clinical events as the primary endpoint. On the basis of strong 308 evidence demonstrating that a deficiency in SERCA2a adversely affects cardiac 309 function and favourable results with AAV1/SERCA2a gene transfer in both 310 experimental models and patients treated in pilot studies,6,7,16-20 CUPID 2 was designed 311 to assess whether AAV1/SERCA2a administration improves the clinical course of 312 moderate to severe HF patients with reduced ejection fraction who were receiving 313 contemporary guideline-recommended therapy. The results showed that 314 AAV1/SERCA2a at the dose used did not reduce either recurrent HF events (primary 14 315 efficacy endpoint) or terminal events (secondary efficacy endpoint) in the overall 316 study population or in pre-specified subgroups. However, no evidence of worsening of 317 the clinical course of study patients emerged during the study.The negative results of 318 CUPID 2 raise important questions. Although gene transfer is a promising approach for 319 treating human disease, there has been limited success to date with this approach in 320 treating patients with cardiovascular disease. Previous experimental studies showed 321 that reduced SERCA2a activity was associated with abnormalities in calcium 322 homeostasis and cardiomyocyte function and that correction of these abnormalities by 323 gene transfer improved cardiac function and survival.2-5,16-19 In a pilot dose-finding 324 study of AAV1/SERCA2a (CUPID 1) in patients with HF, administration of 1x1013 DRP 325 was associated with stabilization or improvement in several independent measures of 326 patient wellbeing and cardiac function. There was also a reduction in the recurrent 327 event rate compared with patients who were treated with placebo.6,7, These results 328 provided the rationale for and informed the design (including dose) for CUPID 2. The 329 reasons for the failure of AAV1/SERCA2a to improve the clinical course of HF patients 330 and the differences between the results of CUPID 1 and CUPID 2 are unclear. The 331 entry criteria and treatment algorithms were similar between the studies, and 332 although CUPID 2 added the requirement for elevated natriuretic peptide levels or a 333 recent hospitalization during the course of the study to enrich for recurrent HF events, 334 comparison of the profile of the patients included in the studies reveals no striking 335 differences. Moreover, post-hoc analyses indicated that the amendment did not 336 meaningfully affect the response to treatment for either the primary or secondary 337 efficacy endpoints. However, the crude recurrent event rate in placebo patients was 338 higher in CUPID 1 compared with CUPID 2 (1·27 per patient/yr vs. 0·7 per patient/yr, 339 respectively), and CUPID 1 was a small study, with only 14 patients receiving placebo 15 340 and nine receiving 1x1013 DRP AAV1/SERCA2a. These factors raise the possibility that 341 the positive results observed in CUPID 1 were due to chance and/or to greater severity 342 of illness in patients randomised to placebo. The negative results of CUPID 2, 343 however, do not appear to be related to the high percentage of coronary artery 344 disease (56%) in the population enrolled in the study, as there were no differences in 345 outcomes by HF aetiology. 346 Another possibility is inadequate delivery and uptake of the vector in the hearts of 347 patients enrolled in CUPID 2. It is possible that other approaches for introducing 348 AAV1/SERCA2a to the heart might have enhanced uptake into cardiomyocytes.21,22 349 Intracoronary delivery of AAV1/SERCA2a, however, is simpler and more practical than 350 other modes of delivery, and this technique was associated with significant increases 351 in SERCA2a gene expression both in a large animal model using the same vector as in 352 CUPID 218 and in pilot studies in which HF patients were treated with intracoronary 353 delivery of AAV1/SERCA2a.6,7,20 However, in the intervening period between CUPID 1 354 and CUPID 2, the work of Mingozzi et al. showed that not only is the quantity of full 355 AAV viral capsid particles (containing the single stranded DNA and used for dose 356 determination) important for in vivo activity, but also the total viral particle dose, 357 including the proportion of empty capsid particles contained in the preparation.23 358 Though perhaps counterintuitive, the possibility that a higher proportion of empty 359 capsids improves gene transfer is supported by results presented in Appendix p 4 and 360 Web Figure 1. These findings differ from results of earlier studies showing improved 361 gene delivery with fewer empty capsids,24 likely related to the fact that previous work 362 did not address the neutralization that might occur with vascular delivery of AAVs in 363 vivo. Thus, empty capsids may serve as “decoy” proteins that block the inhibitory 364 activity of antibodies and possibly of other serum-based interfering substances.23 The 16 365 presence of even low titres of NAb (<1:2) or other interfering substances in vivo can 366 shift the dose response curve; with a lower percentage of empty capsids in the 367 preparation, higher doses are required in order to achieve the same level of gene 368 transfer. The difference in the proportion of empty capsids in preparations used in 369 CUPID 1 and CUPID 2, with a lower total particle dose infused in this study, may have 370 contributed to a reduction in gene transfer efficiency in CUPID 2. Although the low 371 level of vector DNA in the limited number of CUPID 2 patients from whom tissue was 372 available is consistent with this possibility, these patients may not be representative 373 of the overall study population since their condition had deteriorated to the point 374 where they required advanced therapies. A fundamental question is whether SERCA2a 375 was an appropriate target for therapy. Although deficiencies in SERCA2a activity in the 376 failing heart and their correction by gene transfer have been demonstrated in 377 experimental models,16-19 it is possible that these findings are not applicable in human 378 HF and that, regardless of the level to which SERCA2a activity is raised, the impact 379 would be insufficient to alter the trajectory of the disease. It is also possible that 380 post-transcriptional or post-translational regulatory factors in patients may have 381 negated enhanced transgene expression or enzyme activity in treated patients and 382 that the earlier findings in animal studies showing significant improvement using this 383 same vector do not translate to humans with HF. 384 During the course of the study no signals regarding safety emerged. While it is 385 reassuring that the intracoronary delivery of the drug can be safely performed in 386 patients with moderate to severe HF due to reduced ejection fraction, concerns about 387 the efficiency of AAV1/SERCA2a delivery raise the point that conclusive data on the 388 safety of AAV1/SERCA2a will require demonstration of greater uptake and expression 389 of the transgene in cardiomyocytes. 17 390 Although we did note a significant interaction between treatment group and 391 geography, suggesting that the risk of terminal events might be greater in the 392 AAV1/SERCA2a group than in placebo in non-US patients, the number of patients and 393 terminal events in this sub-group was small and the non-US AAV1/SERCA2a patients 394 appear to have been somewhat sicker at baseline than non-US placebo patients. Thus, 395 this finding was likely due to chance, sicker patients being randomized to 396 AAV1/SERCA2a, or both. The lack of an increased HR for recurrent events, which 397 should have been influenced in the same direction, suggests chance as the most likely 398 explanation. 399 While the results of CUPID 2 show that antegrade coronary delivery of 1x1013 DRP of 400 AAV1/SERCA2a does not alter the clinical course of HF patients with reduced ejection 401 fraction, they raise a number of questions that will need to be addressed if future 402 studies in this area are to be successful. For the development of AAV1/SERCA2a, 403 evidence that efficiency might have been compromised by the lower number of empty 404 capsids raises the possibility that the latter was responsible for the negative results of 405 CUPID 2 and it provides the rational for further studies using drug with higher numbers 406 of total capsid proteins, which is best achieved by increasing the dose of 407 AAV1/SERCA2a. In addition, the issues raised by the negative results of CUPID 2 need 408 to be considered in designing trials with other constructs meant to enhance gene 409 expression in the failing heart in the future,25 and they suggest that it will be 410 important to characterize serum effects from the target patient population for their 411 potential impact on the biological potency of drugs used for gene transfer. 412 413 414 18 415 Contributors 416 All authors contributed to the interpretation of the results and writing of the manuscript and 417 all authors approved the decision to submit the manuscript for publication. BG, JB, GMF, PP, 418 AAV, ASD, DB, AB, BJ, and ARL were investigators in this study. JMP was the study statistician. 419 JJR, KMZ, and JMP were involved in study design. BG wrote and prepared the first draft of the 420 manuscript, with input from the other authors. 421 422 Declaration of interests 423 BG, JB, GMF, PP, AV, AD, DB, AB, BEJ, and ARL received financial support from Celladon 424 Corporation, the sponsor of this trial, in the form of grants, personal fees, and other financial 425 support. JMP, JJR, and KMZ were employees of Celladon during the CUPID 2 trial. 426 427 Acknowledgments 428 This trial was funded by Celladon Corporation. Celladon also provided funding for manuscript 429 and graphics support. We would like to thank all of the investigators (Appendix p 3) and 430 patients involved in this study. We wish to thank Roger Hajjar, M.D. (Mt. Sinai, New York, NY), 431 for his guidance on AAV1/SERCA2a development. ARL wishes to acknowledge support from the 432 National Institute for Health Research Cardiovascular Biomedical Research Unit, Royal 433 Brompton Hospital, and the British Heart Foundation. We wish to thank Sharon L. Cross, Ph.D. 434 for providing manuscript support and Julia Andres for providing graphics support on behalf of 435 Celladon. 436 437 19 438 439 REFERENCES 440 441 442 1 National Heart, Lung, and Blood Institute. Morbidity and Mortality: 2012 Chart Book on Cardiovascular, Lung, and Blood Diseases. Bethesda, MD: US Department of Health and Human Services, National Institutes of Health, 2012. 443 444 445 2 Hasenfuss G, Reinecke H, Studer R, et al. Relation between myocardial function and expression of sarcoplasmic reticulum Ca2+-ATPase in failing and nonfailing human myocardium. Circ Res 1994; 75: 434–42. 446 447 3 Hasenfuss G, Pieske B. Calcium cycling in congestive heart failure. J Mol Cell Cardiol 2002; 34: 951–69. 448 449 4 Kho C, Lee A, Hajjar RJ. Altered sarcoplasmic reticulum calcium cycling—targets for heart failure therapy. Nat Rev Cardiol 2012; 9: 717–33. 450 451 5 Eisner D, Caldwell J, Trafford A. Sarcoplasmic reticulum Ca-ATPase and heart failure 20 years later. Circ Res 2013; 113: 958–61. 452 453 454 455 6 Jessup M, Greenberg B, Mancini D, et al. Calcium upregulation by percutaneous administration of gene therapy in cardiac diseases (CUPID): a phase 2 trial of intracoronary gene therapy of sarcoplasmic reticulum CA2+-ATPase in patients with advanced heart failure. Circulation 2011; 124: 304–13. 456 457 458 7 Zsebo K, Yaroshinsky A, Rudy JJ, et al. Long-term effects of AAV1/SERCA2a gene transfer in patients with severe heart failure. Analysis of recurrent cardiovascular events and mortality. Circ Res 2014; 114: 101–8. 459 460 461 462 8 Greenberg B, Yaroshinsky A, Zsebo KM, et al. Design of a phase 2b trial of intracoronary administration of AAV1/SERCA2a in patients with advance heart failure: the CUPID 2 trial (calcium up-regulation by percutaneous administration of gene therapy in cardiac disease phase 2b). J Am Coll Cardiol HF 2014; 2: 84–92. 463 464 465 9 Karakikes I, Hadri L, Rapti K, et al. Concomitant intravenous nitroglycerin with intracoronary delivery of AAV1.SERCA2a enhances gene transfer in porcine hearts. Mol Ther 2012; 20: 565–71. 466 467 10 Moher D, Hopewell S, Schulz KF, et al. CONSORT 2010 explanation and elaboration: updated guidelines for reporting parallel group randomised trials. BMJ 2010; 340: c869. 468 469 470 471 11 Center for Drug Evaluation and Research (CDER) and Center for Biologics Evaluation and Research (CBER). Guidance for Industry: E9 Statistical Principles for Clinical Trials. Rockville, MD: US Department of Health and Human Services, Food and Drug Administration, 1998. 472 473 12 Liu L, Wolfe RA, Huang X. Shared frailty models for recurrent events and a terminal event. Biometrics 2004; 60: 747–56. 20 474 475 13 Liu L, Huang X. The use of Gaussian quadrature for estimation in frailty proportional hazards models. Stat Med 2008; 27: 2665–83. 476 477 478 14 Nelson WB. Recurrent events data analysis for product repairs, disease recurrences, and other applications. Schenectady, NY: Society for Industrial and Applied Mathematics, 2003. 479 480 15 Beltrami CA, Di Loreto C, Finato N, Yan SM. DNA content in end-stage heart failure. Adv Clin Path 1997; 1: 597–3 481 482 483 16 del Monte F, Williams E, Lebeche D, et al. Improvement in survival and cardiac metabolism after gene transfer of sarcoplasmic reticulum CA2+-ATPase in a rat model of heart failure. Circulation 2001; 104: 1424–9. 484 485 486 17 Sakata S, Lebeche D, Sakata N, et al. Restoration of mechanical and energetic function in failing aortic-banded rat hears by gene transfer of calcium cycling proteins. J Mol Cell Cardiol 2007; 42: 852–61. 487 488 489 18 Kawase Y, Ly H, Prunier F, et al. Reversal of cardiac dysfunction after long-term expression of SERCA2a by gene transfer in a pre-clinical model of heart failure. J Am Coll Cardiol 2008; 51: 1112–9. 490 491 492 19 Byrne M, Power J, Preovolos A, Mariani J, Hajjar R, Kaye D. Recirculating cardiac delivery of AAV2/1SERCA2a improves myocardial function in an experimental model of heart failure in large animals. Gene Ther 2008; 15: 1550–7. 493 494 495 20 Jaski BE, Jessup ML, Mancini DM, et al. Calcium upregulation by percutaneous administration of gene therapy in cardiac disease (CUPID trial), a first-in-human phase 1/2 clinical trial. J Card Fail 2009; 15: 171–81. 496 497 498 21 Mariani JA, Smolic A, Preovolos A, Byrne MJ, Power JM, Kaye DM. Augmentation of left ventricular mechanics by recirculation-mediated AAV2/1-SERCA2a gene delivery in experimental heart failure. Eur J Heart Fail 2011; 13: 247–53. 499 500 22 Wolfram JA, Donahue JK. Gene therapy to treat cardiovascular disease. J Am Heart Assoc 2013; 2: e000119. 501 502 23 Mingozzi F, Anguela XM, Pavani G, et al. Overcoming preexisting humoral immunity to AAV using capsid decoys. Sci Transl Med 2013; 5: 194ra92. 503 504 505 24 Urabe M, Xin KQ, Obara Y, et al. Removal of empty capsids from type 1 adenoassociated virus vector stocks by anion-exchange chromatography potentiates transgene expression. Mol Ther 2006; 13: 823–8. 506 507 25 Zouein FA, Booz GW. AAV-mediated gene therapy for heart failure: enhancing contractility and calcium handling. F1000Prime Rep 2013; 5: 27. 508 21 509 Table 1: Characteristics at baseline in the modified intention-to-treat population. Placebo AAV1/SERCA2a All patients (n = 122) (n = 121) (N = 243) Age (years) 58·4±12·26 60·3±9·77 59·3±11·11 Female sex 24 (19·7%) 21 (17·4%) 45 (18·5%) White 98 (81·1%) 99 (81·8%) 198 (81·5%) Black/African American 22 (18·0%) 18 (14·9%) 40 (16·5%) American Indian/Alaskan Native 0 (0·0%) 1 (0·8%) 1 (0·4%) Native Hawaiian/Pacific Islander 0 (0·0%) 1 (0·8%) 1(0·4%) Other 1 (0·8%) 2 (1·7%) 3 (1·2%) 118 (96·7%) 114 (94·2%) 232 (95·5%) 4 (3·3%) 7 (5·8%) 11 (4·5%) United States 79 (64·8%) 79 (65·3%) 158 (65·0%) Non-United States* 43 (35·2%) 42 (34·7%) 85 (35·0%) Coronary artery disease 67 (54·9%) 68 (56·2%) 135 (55·6%) 336·6 (71·29) 319·9 (91·47) 328·2 (82·23) 24·0 (6·26) 23·0 (6·48) 23·5 (6·37) II 21 (17·2%) 22 (18·2%) 43 (17·7%) III 100 (82·0%) 96 (79·3%) 196 (80·7%) IV 1 (0·8%) 3 (2·5%) 4 (1·6%) KCCQ (overall score) 59·2 (22·7) 58·4 (19·76) 58·8 (21·02) NT-proBNP (pg/mL) 1504 1754 1679 (849-3031) (843-3785) (843-3561) Characteristic Race Ethnicity Not Hispanic Hispanic Country Six-minute walk test (meters) Left ventricular ejection fraction (%) NYHA functional class 22 Placebo AAV1/SERCA2a All patients (n = 122) (n = 121) (N = 243) Ischemic 63 (51·6%) 62 (51·2%) 125 (51·4%) Idiopathic 50 (41·0%) 48 (39·7%) 98 (40·3%) Hypertensive 5 (4·1%) 5 (4·1%) 10 (4·1%) Familial 1 (0·8%) 2 (1·7%) 3 (1·2%) Peripartum 2 (1·6%) 0 (0·0%) 2 (0·8%) Other 1 (0·8%) 7 (5·8%) 8 (3·3%) ACE inhibitor/ARB 110 (90·2%) 111 (91·7%) 221 (90·9%) Aldosterone antagonist 74 (60·7%) 83 (68·6%) 157 (64·6%) Beta blocker 117 (95·9%) 117 (96·7%) 234 (96·3%) Diuretic 109 (89·3%) 111 (91·7%) 220 (90·5%) Digoxin 48 (39·3%) 45 (37·2%) 93 (38·3%) OAC/NOAC 81 (66·4%) 76 (62·8%) 157 (64·6%) Cardiac resynchronization therapy 39 (32·0%) 53 (43·8%) 92 (37·9%) Implantable cardioverter-defibrillator 89 (73·0%) 98 (81·0%) 187 (77·0%) Chronic renal insufficiency 37 (30·3%) 36 (30·0%) 73 (30·2%) Type 2 diabetes 49 (40·2%) 59 (48·8%) 108 (44·4%) Atrial fibrillation 49 (40·2%) 44 (36·4%) 93 (38·3%) COPD 18 (14·8%) 15 (12·5%) 33 (13·6%) Characteristic Heart failure aetiology Heart failure regimen Other medical history 510 511 Data are mean (standard deviation) or n (%) except for NT-proBNP, which is median (IQR). 512 There were no significant differences between the two groups in baseline demographic or 513 disease characteristics. 23 Sweden (8/7/15), Great Britain (6/8/14), Denmark (5/6/11), Poland (6/5/11), Germany 514 * 515 (5/5/10), Hungary (5/3/82), Israel (5/3/8), Belgium (3/4/7), and the Netherlands (0/1/1) for 516 Placebo, AAV1/SERCA2, and All Patients, respectively . 517 ACE inhibitor=angiotensin-converting enzyme inhibitor. ARB=angiotensin-receptor blocker. 518 COPD=chronic obstructive pulmonary disease. IQR=interquartile range. KCCQ=Kansas City 519 Cardiomyopathy Questionnaire. NT-proBNP=N-terminal pro-B-type natriuretic peptide. 520 NYHA=New York Heart Association. OAC/NOAC=oral anti-coagulant/novel oral anti-coagulant. 521 522 24 523 Table 2: Primary and secondary endpoints at the primary analysis data cutoff in the 524 modified intention-to-treat population Outcome Placebo (N=122) AAV1/SERCA Hazard ratio 2a (CI) p value (N=121) Primary endpoint Recurrent events 128 (73·9) 104 (62·8) 0·93 0·81 (0·53-1·65) HF- related 121 (69·8) 96 (57·9) 7 (4·0) 8 (4·8) 29 (16·7) 36 (21·7) hospitalizations Ambulatory treatment for worsening HF Secondary endpoint First terminal event 1·27 (0·72–2·24) Death 19 (11·0) 24 (14·5) Heart transplant 2 (1·2) 5 (3·0) Durable MCSD 8 (4·6) 7 (4·2) implant 525 526 Data are n (rate per 100 patient-years). CI=confidence interval. HF=heart failure. 527 MCSD=mechanical circulatory support device. 528 0·41 25 529 Table 3. Rates of adjudicated clinical events in the safety population* Placebo AAV1/SERCA2a (N=122) (N=121) All clinical events† 262 (147) 190 (111) All-cause hospitalizations 240 (135) 172 (100) 121 (67·9) 99 (57·7) Ambulatory treatment for worsening HF 7 (4·0) 8 (4·8) Non-fatal myocardial infarctions 5 (2·8) 3 (1·7) Non-fatal strokes 3 (1·7) 5 (2·9) Heart transplant 4 (2·2) 7 (4·1) Durable MCSD implant 8 (4·5) 7 (4·1) 20 (11·2) 25 (14·6) 2 (1·1) 3 (1·7) 18 (10·1) 22 (12·8) Pump failure 11 (6·2) 14 (8·2) Sudden death 3 (1.7) 7 (4·1) Presumed sudden death 1 (0·6) 0 (0) Arrhythmia 2 (1·1) 0 (0) Fatal stroke 0 (0) 1 (0·6) 1 (0·6) 0 (0) Outcome HF- related hospitalizations Deaths Non-cardiovascular Cardiovascular Non-traumatic subdural hematoma 530 Data are n (rate per 100 patient-years) 531 * The numbers of recurrent and terminal events differ slightly from the numbers in the 532 efficacy analysis shown in Table 2 due to specific definitions used for primary and secondary 533 endpoints. For example, in the primary efficacy analysis, only first terminal events were 534 counted, and recurrent events that occurred after terminal events were not counted. 535 † 536 as a hospitalization. 537 MCSD=mechanical circulatory support device. 538 539 540 541 542 Excluding all heart failure (HF) hospitalizations and any other clinical event already counted 26 543 Figure legends 544 Figure 1. Trial profile 545 546 Fx=failure. HF=heart failure. I/E=inclusion/exclusion. ITT=intention-to-treat. mITT=modified intention-to-treat. NAb=neutralizing antibodies. URI=upper respiratory infection. 547 548 549 Figure 2. Kaplan-Meier curves for cumulative number of recurrent events per patient at 550 the primary analysis data cutoff (A), the probability of being terminal-event free at the 551 primary analysis data cutoff (B), and the probability of death from any cause (C) in 552 patients assigned to AAV1/SERCA2a (blue) or placebo (yellow) in the modified intention-to- 553 treat population 554 CI=confidence interval. 555 556 557 Figure 3. Subgroup analyses 558 Hazard ratios (HR) for recurrent events (primary endpoint) and first terminal event (secondary 559 endpoint) in the listed subgroups. The size of the square corresponds to the number of 560 patients in each subgroup. Pre-specified analyses consisted of overall patient population, 561 geography, heart failure (HF) aetiology, New York Heart Association (NYHA) class, and years of 562 age. CI=confidence interval. ICD=implantable cardioverter-defibrillator. ITT=intent-to-treat. 563 LVEF=left ventricular ejection fraction. mITT=modified ITT. NT-proBNP=N-terminal pro-B-type 564 natriuretic peptide. US=United States