Gene Therapy for Heart Failure: Time to Go Back to the Drawing Board

This article refers to ‘Effect of intracoronary administration of AAV1/SERCA2a on ventricular remodelling in patients with advanced systolic heart failure: results from the AGENT-HF randomized phase 2 trial’, by J.-S. Hulot et al., published in this issue on pages 1534–1541. Over the past four decades there have been substantial advances in the treatment of heart failure. The addition of new drugs and devices as well as improved understanding of how to use them more effectively has increased both the quality and quantity of life of millions of people around the world. Despite this progress, however, heart failure patients experience a high likelihood of hospitalization, a marked reduction in survival, and quality of life that is amongst the lowest for any chronic disease. Although the need for new approaches that advance heart failure care beyond existing paradigms is both obvious and pressing, therapeutic breakthroughs have been intermittent at best. In particular, strategies for treating heart failure based on cell transplantation or gene transfer have thus far failed to achieve the goals that were expected when they were first introduced. In this issue of the Journal, Hulot and colleagues present the results of AGENT-HF, a trial that was designed to assess the effects of AAV1/SERCA2a gene therapy on cardiac remodelling in patients with symptomatic heart failure and a reduced ejection fraction.1 The study was meant to be a companion piece for the larger phase 2b CUPID 2 trial, which assessed the effects of AAV1/SERCA2a on the clinical course of 250 chronic heart failure patients who were selected based on similar entry criteria to those used in AGENT-HF.1, 2 The exact same dose and mode of delivery of AAV1/SERCA2a were employed in the two trials. Despite considerable promise and enthusiasm based on the results of the pilot phase 2a CUPID 1 study,3 however, CUPID 2 failed to demonstrate evidence of an improvement either in its primary endpoint of recurrent heart failure hospitalizations, its main secondary endpoint of risk of terminal events, or in any of the other efficacy outcomes that were assessed.4 Further analysis of the CUPID 2 results failed to identify any subgroups of patients that benefited from AAV1/SERCA2a therapy. A consequence of these disappointing findings was the premature termination of AGENT-HF by the sponsor. At the time of study closure, only 9 of the planned 40 patients had been treated, with 5 of them receiving active therapy and the others receiving placebo. No evidence of improvement was noted for the primary endpoint of AGENT-HF, which was the change in left ventricular end-systolic volume (LVESV), or in any other efficacy parameter when the AAV1/SERCA2a-treated patients were compared with those treated with placebo. Cardiac computed tomography (CCT) was used in AGENT-HF and the reduced variability in LVESV measurement using this technique allowed the study to aim for enrolling a smaller number of patients than would be needed had echocardiography been used to assess cardiac size and function. Even using CCT measurements, however, the very small number of patients that were actually treated makes it impossible to draw any meaningful conclusion regarding the results of AGENT-HF. It is worth mentioning that enthusiasm for AAV1/SERCA2a was based largely on improvements in clinical parameters, changes in cardiac structure (particularly LVESV), and a reduction in clinical events in the cohort of nine patients who received the highest dose (i.e. 1013 DNase-resistant particles) of drug in CUPID 1.4 The message for future studies regarding our ability to reliably detect efficacy signals from changes occurring in small groups of patients could not be more clear. The disappointing results seen in CUPID 2 and AGENT-HF raise a number of questions, the answers to which should help in the design of future gene transfer studies in heart failure patients. The failure of any promising treatment modality, and particularly one with as high a profile as gene therapy, raises the possibility that the design of the pivotal studies might have inadvertently undermined their ability to detect favourable effects in heart failure patients. For AAV1/SERCA2a therapy, however, the decision to include a mixture of patients with moderately severe symptoms due to heart failure of either ischaemic or non-ischaemic aetiologies, selection of recurrent heart failure hospitalizations as the primary endpoint, duration of patient follow-up, and other features of trial design appeared to have had virtually no impact on the outcome of CUPID 2 (and AGENT-HF) as the lack of efficacy of AAV1/SERCA2a extended across all of the endpoints that were evaluated and was seen in all of the subgroups that were analysed.3 Another consideration is whether targeting SERCA2a alone would be sufficient to improve the clinical course of heart failure patients. Perhaps changes in regulatory proteins (e.g. phospholamban) or other pathways controlling intracellular calcium levels in cardiomyocytes5 might have emerged to counteract a rise in SERCA2a activity with the net effect being little or no improvement in calcium homeostasis in the AAV1/SERCA2a-treated patients. Unfortunately, the results of CUPID 2 and AGENT-HF cannot provide a definitive answer to this question as tissue recovered from the hearts of patients from these studies failed to demonstrate evidence of expression of the SERCA2a gene at levels that would be expected to correct deficiencies seen in the failing heart.3 However, the importance of calcium homeostasis in regulating both systolic and diastolic functions of the heart and the central role of SERCA2a in maintaining calcium levels within a normal range in cardiomyocytes are well established.5-8 Evidence that the activity of SERCA2a is reduced in the failing human heart9 and that gene therapy designed to increase SERCA2a is effective in correcting abnormalities in calcium flux and cardiomyocyte function suggests that this enzyme is an appropriate target for therapeutic interventions.10-12 Reports that AAV1/SERCA2a gene therapy improves cardiac function and the natural history of experimental animal models of heart failure strongly support the selection of SERCA2a as a therapeutic target.13, 14 The failure to demonstrate evidence of adequate SERCA2a gene expression in tissue samples of myocardium from patients in the CUPID 2 and AGENT-HF trials focuses attention on factors that determine the efficacy of cardiomyocyte transfection and persistence of gene expression within these cells. These include the vector that was selected for transporting the SERCA2a gene, its mode of delivery, the dose that was administered, and the role of other factors that could affect the cardiomyocyte transfection efficiency or the persistence of gene expression within the cells. The use of viral vectors to deliver genes to treat human diseases has emerged as the most promising approach for gene therapies. Although a number of different viral vectors have been used, the adeno-associated viruses (AAVs) appear to be the most promising based on their avidity for cardiac tissue, failure to elicit an inflammatory response, relative lack of immunogenicity, and evidence of sustained gene expression in animal models.13-17 In both the CUPID studies and AGENT-HF, the SERCA2a gene driven by a strong promoter was inserted into an AAV1 vector from which the intrinsic viral genome had been removed. Other AAV subtypes, however, have been reported to have strong affinity for cardiac tissue17 and it is possible that use of one of these (e.g. AAV9) might have resulted in more efficient cardiomyocyte transfection. The choice of a dose of 1013 DNase-resistant particles of AAV1/SERCA2a was based on preliminary studies in heart failure patients in which there was an efficacy signal in the absence of any safety concerns.2, 18 Higher doses, however, were not tested in these earlier studies and it is possible that they would have proved to be safe and more effective in raising SERCA2a levels. Another possibility is that gene delivery by either direct intra-myocardial injection or by retrograde coronary perfusion through the coronary sinus might have augmented the efficiency of cardiomyocyte transfection and increased transgene expression in the heart.19, 20 An important limitation of the use of AAVs for gene therapy is the presence of neutralizing antibodies (NAbs) that develop as a result of prior infection. The presence of NAbs at inhibitory concentrations of ≥1:2 in plasma were detected in approximately 60% of the population being considered for the CUPID 2 study21 and resulted in their exclusion from the trial as titres <1:2 were required for entry.2 It is possible, however, that even the low levels of NAbs present in patients enrolled into the study decreased transfection efficiency. The inhibitory effects of low levels of NAbs or other substances in blood that could interfere with uptake of viral particles into cells can be overcome by increasing the dose of particles containing the SERCA2a gene, emphasizing once again the importance of determining optimal dose in designing clinical trials. These effects can also be reduced by increasing the number of empty viral capsids (i.e. ones that do not contain genetic material) in the preparation that is administered. There is evidence that a strategy of increasing the percentage of empty viral capsids can improve transfection efficiency by serving as decoys for NAbs and other substances in blood that inhibit cell transfection.22 Unfortunately, a change in preparation of AAV1/SERCA2a between the earlier CUPID studies and CUPID 2 and AGENT-HF resulted in greater packaging efficiency and a marked reduction in empty capsids at equal doses of active gene-containing viral particles used in the later studies. It is possible that the reduction in number of empty capsids resulted in the low level of gene expression in the CUPID 2 and AGENT-HF patients from whom tissue was available for analysis.4 Finally, although sustained transgene expression of SERCA2a has been demonstrated in animal models, there is evidence that T-cell-mediated immune responses may limit expression over time in human patients.23 It is possible that in both CUPID 2 and AGENT-HF beneficial effects of AAV1/SERCA2a occurring early after drug administration might have been mitigated by the development of a late immune response. However, neither the primary endpoint (i.e. recurrent heart failure hospitalizations over time), the other efficacy endpoints that were measured in CUPID 2, nor the remodelling endpoints that were measured in AGENT-HF at 12 months after drug administration provided any insight into whether this biphasic response might have occurred. Thus, whether or not late immune responses may have played a role in the negative results is uncertain. Do the results of CUPID 2 and AGENT-HF bring to an end the promise of gene therapy for heart failure? Based on the discussion above I believe this is not the case. The bright spot of the otherwise dismal results of CUPID 2 and AGENT-HF is that gene therapy was not associated with any safety concerns. While vigilance will still be required in future trials using strategies that will hopefully result in therapeutic levels of gene expression, the safety profile of the CUPID studies and AGENT-HF are reassuring. We should, however, pay careful attention to the questions raised by the disappointing results using AAV1/SERCA2a in these studies. The very low level of gene expression tells us that the studies that have been conducted up until now have not adequately addressed the question of whether increasing SERCA2a levels in the failing heart of human patients can improve the clinical course. The reasons for the low levels seen should direct the focus of future research towards better understanding of why this occurred and more importantly, towards developing strategies to enhance gene expression to a level where the hypothesis can be adequately tested. Whether overcoming low gene expression levels can be accomplished by improving the vector or method of delivery, increasing the dose used, strategies to inhibit factors that interfere with uptake of viruses into cells, or altering an immune response that diminishes gene expression over time needs to be determined. By addressing these issues and improving our understanding of the multiple factors involved in achieving sustained gene expression, it is hoped that we will be able to develop successful strategies that will enable gene therapy to fulfil its considerable promise of improving outcomes in heart failure patients. Conflict of interest: B.G. chaired the Steering Committee for the CUPID 2 study and received honoraria for this activity.