Main

Globally, heart failure is a growing health and economic burden, in part due to the aging population1,2. An estimated nearly 60 million people globally have the clinical syndrome of heart failure, with approximately 50% having left ventricular ejection fraction (LVEF) <40% (refs. 1,3). The current standard-of-care treatments for systolic heart failure—including angiotensin-converting enzyme inhibitors, angiotensin receptor blockers (ARBs), angiotensin receptor–neprilysin inhibitors, β-adrenergic receptor blocking agents (β-blockers), mineralocorticoid receptor antagonists and sodium–glucose cotransporter 2 inhibitors—reduce symptoms, delay disease progression and increase survival4. However, systolic heart failure remains a lethal condition, particularly for patients who do not respond to conventional medical therapy. Thus, there is an important need for innovative therapies to reverse ventricular dysfunction and to modify the course of the disease4.

The contractile state of the heart is regulated by the catecholamine-dependent activation of myocardial β-adrenergic receptors—resulting in increased levels of cyclic adenosine monophosphate, activation of the cyclic-adenosine-monophosphate-dependent protein kinase A (PKA) and PKA-mediated phosphorylation of enzymes and key regulatory proteins involved in excitation–contraction coupling5,6. Phospholamban (PLN) affects contractility by regulating the activity of the sarcoplasmic–endoplasmic reticulum calcium ions (Ca2+) ATPase (SERCA2a), which transports Ca2+ from the cytosol into the sarcoplasmic reticulum during cardiac relaxation. Dephosphorylated PLN inhibits the activity of SERCA2a7. When PLN is phosphorylated by PKA, inhibition is lost, leading to increased SERCA2a activity along with cardiac relaxation7. Protein phosphatase 1 (PP1) acts as a key regulator of SERCA2a via dephosphorylation of PLN8,9. The activity of PP1 is in turn regulated by inhibitor of PP1 (I-1), an endogenous protein that is activated when it is phosphorylated by PKA9,10. The inhibition of PP1 by I-1 results in the phosphorylation of PLN, which in turn increases SERCA2a activity, Ca2+ uptake and contractility10,11,12.

As demonstrated in humans13 and animal/experimental models with failing hearts14, Ca2+ uptake by the sarcoplasmic reticulum is deficient in cardiomyocytes during relaxation. In patients with end-stage heart failure, PP1 activity is increased15, and β-receptors are downregulated, leading to decreased PLN phosphorylation and reduced activity of SERCA2a10. Increases in PP1 activity in failing hearts are accompanied by increases in the dephosphorylated (inactive) state of I-1. Because I-1 is less active in failing hearts, interventions aimed to increase its activity may be beneficial. Preclinical studies have shown that the constitutively active I-1 of PP1 (amino acids 1–65; T35D) not only restores contractility but also reverses adverse remodeling by directly decreasing fibrosis and cardiac hypertrophy10,16,17. Notably, mice expressing I-1c maintained higher phosphorylation of PLN at both Ser16 and Thr17 and had normal survival and similar cardiac function to wild-type mice over 20 months, with 20-month-old mice having no increases in arrhythmias under stress conditions17. In addition, the inhibition of PP1 has other effects that may be important in heart failure, including the inhibition of apoptosis and the inhibition of the activity of mitogen-activated protein kinases10,11,18.

The use of adeno-associated virus (AAV) vectors is a promising option for gene therapy due to its potential for durable expression in nondividing cells, low immunogenicity and selective transduction based on the capsid protein19,20. Furthermore, AAVs can be delivered locally to the heart via intracoronary infusion19,20,21,22, which is a minimally invasive method that has been used previously in clinical trials of cardiac gene therapy23,24 to target the administration to the heart as opposed to an intravenous, systemic administration. AB-1002 is an investigational gene therapy product composed of a chimeric cardiotropic AAV2/AAV8 vector capsid (AAV2i8) that delivers the gene sequence for the expression of the constitutively active PP1 inhibitor 1 (I-1c). AAVs have shown favorable safety in cardiovascular applications21,25 and can transduce slowly dividing or nondividing cells such as cardiomyocytes21,22, suggesting that they may be strong candidates for delivering gene therapy in heart failure. In porcine models of ischemic26 and nonischemic16 heart failure, the intracoronary delivery of AB-1002 significantly improved hemodynamic parameters and cardiac function.

The primary objective of this first-in-human phase 1 study (NCT04179643) was to evaluate the safety and feasibility of a single antegrade coronary artery infusion of AB-1002 in patients with nonischemic cardiomyopathy (LVEF 15–35%) and New York Heart Association (NYHA) class III symptoms of heart failure over 12 months. A secondary objective was to explore the biologic activity and efficacy of AB-1002 to identify appropriate doses for future studies.

Results

Patients

Between 30 January 2020 and 17 August 2023, a total of 11 patients with NYHA class III heart failure and LVEF 15–35% were treated (cohort 1, n = 6; cohort 2, n = 5) at three study centers out of four active sites in the USA (Fig. 1). Overall, six patients (cohort 1 (3.25 × 1013 viral genomes (vg) per patient), n = 3; cohort 2 (1.08 × 1014 vg per patient), n = 3) completed the full 12-month primary observation period and had 24 months of follow-up; two patients in cohort 2 died, one during the 12-month primary observation period (~6 months after infusion) and one during long-term follow-up. The results are reported for the 12-month primary study period.

Fig. 1: Patient disposition.
figure 1

In cohort 1, patient 2 could be treated 30 days after patient 1, and patient 3 could be treated 14 days after patient 2. In cohort 2, patients were treated sequentially after 14 days. vg, viral genomes.

Full size image

Most patients (82%) were male. The median patient age in cohort 1 (73.5 years) was greater than in cohort 2 (57 years) (Table 1). The mean (s.d.) baseline LVEF was also higher in cohort 1 (30.3 (3.9)) than in cohort 2 (23.3 (5.9)). The baseline heart rate and systolic blood pressure did not differ between the cohorts. Overall, baseline mean (s.d.) resting heart rate was 70 (9.8) bpm, and systolic blood pressure was 104 (14) mm Hg. In total, four patients in cohort 1 and one patient in cohort 2 had atrial fibrillation at baseline. All patients received a total dose volume of 50 ml across the left (all patients) and right (six in cohort 1 and four in cohort 2) coronary arteries. The distribution between the left and right coronary arteries was variable, and one patient in cohort 2 had a portion of their dose administered via circumflex.

Table 1 Patient demographics and baseline clinical characteristics
Full size table

Safety

During the 12-month study period, 83 treatment-emergent adverse events (TEAEs) (grade 1, 35 events; grade 2, 32 events; grade 3, 11 events; grade 4, 4 events; and grade 5, 1 event) were reported in nine patients; four patients (67%) in cohort 1 and five (100%) in cohort 2 experienced ≥1 TEAE (Table 2). No TEAEs were considered by the investigator to be related to the study treatment or the administration procedure, and no predefined dose-limiting toxicities (DLTs; myocarditis, myositis, neutropenia, cardiac tamponade) were observed. The most commonly reported (occurring in more than two patients) TEAEs overall were blood creatinine increased and hypotension (each n = 3 patients) (Table 2 and Extended Data Table 1).

Table 2 Summary of AEs
Full size table

Overall, 14 serious AEs (SAEs) were reported in four patients (cohort 1, n = 1; cohort 2, n = 3) (Table 2). An SAE (cardiomyopathy) unrelated to study treatment resulted in a fatal outcome for one patient in cohort 2, occurring 204 days after infusion; the cause of death was deemed as probably related to substance abuse and noncompliance with concurrent medications. This patient experienced four cardiac-related hospitalizations occurring at 98 (heart failure), 157 (acute on chronic heart failure), 170 (acute on chronic heart failure, severe mitral regurgitation and ventricular tachycardia) and 198 (coronavirus disease 2019 (COVID-19) and cardiomyopathy) days after infusion.

One patient in cohort 1 experienced one cardiac-related hospitalization (due to atrial flutter with rapid ventricular response), which occurred 61 days after infusion. A total of two patients in cohort 2 experienced cardiac-related hospitalizations (one of which is noted above). The other patient experiencing cardiac-related hospitalizations was hospitalized twice, with the first occurring 223 (acute on chronic systolic heart failure) and the second occurring 305 (cardiogenic shock) days after infusion.

One patient in cohort 1 experienced an increase in capsid-specific interferon-γ (IFNγ) enzyme-linked immunospot (ELISpot) followed by a transient but marked increase in ALT and AST without associated elevations in bilirubin at month 2, which coincided with the above-described hospitalization and SAE of atrial flutter with rapid ventricular response (assessed unrelated to study treatment by the investigator) and normalized after 10 days. In cohort 2, mild, self-limiting increases in ALT and AST that did not exceed two times the upper limit of normal were observed in four of five patients between weeks 4 and 12 (Extended Data Fig. 1). There were no grade ≥3 (that is, severe or worse) hematology, chemistry or liver function test abnormalities during the 12-month primary observation period, and changes in clinical laboratory parameters were small and were not considered clinically significant. Changes from baseline in vital signs (Extended Data Fig. 2) and electrocardiogram parameters (Extended Data Table 2) were small and were not considered clinically significant.

There was no predefined DLT window between patients dosed. However, in cohort 1, the time interval between each patient dosed was >1 month. The waiting period between the first patient in cohort 1 and the first patient in cohort 2 was approximately 10.5 months, and the time from the third patient dosed in cohort 1 and the first patient dosed in cohort 2 was approximately 6 months. Furthermore, the duration between the death of the patient in cohort 2 and the subsequent patient receiving a dose was 23 months.

Efficacy and quality-of-life outcomes

Efficacy outcomes for cohorts 1 and 2, including NYHA class, LVEF, maximal oxygen consumption (pVO2) and 6-min walk test (6MWT), were improved or remained stable over the 12-month study period for most patients. In cohort 1, two of the three patients with available data at 3 months were classified as NYHA class II, and five of the six patients were classified as NYHA class I or II at 6 months, which was sustained at 12 months (Fig. 2a). Among the four patients in cohort 2 who survived through the 12-month study period, two were classified as NYHA class II at 3 months, which was sustained at 12 months, and two did not experience any change.

Fig. 2: Efficacy and quality-of-life parameters.
figure 2

a–f, The timecourses showing the data in cohorts 1 and 2 for NYHA class (a), LVEF (assessed by the central reader) (b), pVO2 (c), 6MWT (d), MLHFQ (e) and KCCQ (f). Each dot and associated line indicates one patient. The circles indicate males, and the squares indicate females. 6MWT, 6-minute walk test; FC, functional classification; KCCQ, Kansas City cardiomyopathy questionnaire; LVEF, left ventricular ejection fraction; MLHFQ, Minnesota living with heart failure questionnaire; NYHA, New York Heart Association; pVO2, peak oxygen consumption.

Full size image

At baseline, the mean (s.d.) LVEF (assessed by the central reader) was 30.5% (6.2%) and 24.5% (7.6%) for cohorts 1 (n = 5) and 2 (n = 5), respectively; at 12 months, the mean (s.d.) LVEF was 42.2% (6.8%) and 36.7% (7.6%), respectively (cohort 1, n = 6; cohort 2, n = 3) (Fig. 2b and Extended Data Table 3). For those with both baseline and 12-month assessments (cohort 1, n = 5; cohort 2, n = 3), all had improvement in LVEF of ≥5%.

At baseline, the mean (s.d.) pVO2 was 11.7 (3.5) and 19.0 (3.8) ml kg−1 min−1 for cohorts 1 (n = 5 patients) and 2 (n = 5), respectively. At 12 months, the mean (s.d.) pVO2 was 13.2 (3.4) ml kg−1 min−1 in cohort 1 (n = 4) and 17.0 (3.6) ml kg−1 min−1 in cohort 2 (n = 4), representing an 18.8% and −8.7% change, respectively, among those with both baseline and 12-month assessments (cohort 1, n = 4; cohort 2, n = 4) (Fig. 2c).

Patients in cohort 1 showed a 36.2% increase from baseline to 12 months in the 6MWT (mean (s.d.): baseline, 297.7 (113.4) m, n = 6; 12 months, 376.1 (143.1) m, n = 5), and patients in cohort 2 showed a −1.6% change from baseline to 12 months (mean (s.d.): baseline, 392.9 (53.5) m, n = 5; 12 months, 374.8 (88.9) m, n = 4) (Fig. 2d).

In both cohorts, mean (s.d.) Minnesota Living with Heart Failure Questionnaire (MLHFQ) total scores decreased from baseline (cohort 1, 42.3 (26.0); cohort 2, 58.0 (29.0)) to 12 months (cohort 1, 33.8 (19.2); cohort 2, 45.0 (35.6)), indicating an improved quality of life for patients in both cohorts (Fig. 2e).

In cohort 1, the mean (s.d.) Kansas City Cardiomyopathy Questionnaire (KCCQ) overall summary scale scores were stable from baseline (73.3 (2.6)) to 12 months (71.2 (8.5)); in cohort 2, scores increased from baseline (39.8 (18.3)) to 12 months (59.6 (28.4)), indicating an improved quality of life in patients assigned to cohort 2 (Fig. 2f). Similar results were seen in other KCCQ scale scores at various timepoints following treatment, with generally stability in cohort 1 and improvements in cohort 2 (Extended Data Table 4).

Cardiac biomarkers and complement proteins

N-terminal pro-B-type natriuretic peptide (NT-proBNP)/pro-B-type natriuretic peptide (BNP) levels generally remained stable in cohort 1 and increased in one patient in cohort 2 (Fig. 3a). Troponin I remained low throughout the study in both cohorts (Fig. 3b). Complement proteins C3 and C4 remained at stable levels throughout the study in both cohorts (Fig. 3c,d).

Fig. 3: Cardiac and immunological biomarkers.
figure 3

a–d, The timecourses showing data in cohorts 1 and 2 for the cardiac biomarkers NT-proBNP/BNP (a) and troponin I (b) and complement C3 (c) and complement C4 (d). Each dot and associated line indicates one patient. For a, the data points with black circles indicate BNP values. For a and b, the y axis is on a logarithmic scale. The circles indicate males, and the squares indicate females. BNP, pro-B-type natriuretic peptide; NT-proBNP, N-terminal pro-B-type natriuretic peptide.

Full size image

Immune responses

Increases from baseline in capsid-specific neutralizing antibodies were observed at week 1 and persisted through month 12 in both cohorts (Extended Data Fig. 3).

Because three patients in cohort 1 received study treatment during the early phase of the COVID-19 pandemic, adequate blood sampling could not be obtained to complete all laboratory tests and assays, and therefore, not all scheduled IFNγ ELISpot assays were performed in these patients. An additional three patients were enrolled in cohort 1 following the enrollment of cohort 2; blood sampling was performed in each of these patients and in all patients in cohort 2. All patients in cohort 2 and all three patients in cohort 1 who provided all samples showed T cell reactivity to the capsid and/or to I-1c between weeks 4 and 12; in addition, reactivity to I-1c was observed in one of the first three patients in cohort 1 (Extended Data Fig. 4a,b). The IFNγ ELISpot responses were temporally associated with transient increases in ALT and/or AST in some individuals (Extended Data Fig. 4c,d). The T cell responses waned to below the limit of detection by 12 months in both cohorts.

Tissue analysis

One patient in cohort 2 underwent left ventricle assist device (LVAD) implantation 13 months after AB-1002 infusion. The myocardium from the left ventricle (LV) apical core, removed at the time of device implant, was used to assess the AAV2i8 I-1c transduction efficiency and PLN Ser16 phosphorylation. High myocardial transduction efficiency was observed with AB-1002 at a dose of 1.19 vg dg−1 (Fig. 4a). In addition, the phosphorylation of PLN was comparable with levels in LV samples from healthy controls without heart failure; control sample 3 gave high reads across two runs and was believed to be contaminated (Fig. 4a and Extended Data Fig. 5).

Fig. 4: Transduction efficiency and PLN phosphorylation.
figure 4

a, The myocardial transduction efficiency with AB-1002 represented as vg per μg DNA (left) and per diploid genome (right). Samples 1–5 represent biopsy samples from a single patient; samples 6–7 represent healthy control samples (n = 2). Data are not shown for one control sample that gave high reads across two runs and was believed to be contaminated. b, The phosphorylation of PLN normalized to calsequestrin as a loading control in biopsy sample (performed in duplicate) and healthy control samples (n = 3). PLN, phospholamban; vg, vector genomes.

Full size image

Discussion

AB-1002 is an AAV-based gene therapy comprising a cardiotropic capsid designed to transduce a gene promoting the increased expression of I-1c to block the action of PP1, which has increased activity in systolic heart failure and is speculated to have a pathologic role in the disease27. The genetic material being delivered within AB-1002 is nonintegrating, meaning it remains separate from a patient’s genome. The combination of the cardiotropic AAV vector that selectively transduces cardiac muscle tissues and a ubiquitous promoter facilitates efficient transgene expression in cardiac muscle cells21,28. Unlike other naturally occurring AAVs, AAV2i8 displays markedly reduced hepatic tropism28, and the minimally invasive percutaneous intracoronary infusion approach allows access to the heart for localized delivery21,22,29.

In this first-in-human phase 1 trial in patients with severe NYHA class III systolic heart failure and a high baseline risk of progression, safety data suggest that a single antegrade coronary artery infusion of AB-1002 was generally well tolerated. No TEAEs or SAEs were attributed to the investigational product or administration procedures. Although 82% of patients experienced TEAEs, most AEs (~81%) were mild to moderate in severity and were self-limited, without sequelae. One patient in cohort 2 experienced SAEs of transient ischemic attack, acute on chronic cardiac failure, COVID-19 infection and, ultimately, cardiovascular death. This fatal event was probably related to substance abuse and noncompliance with concurrent medications as opposed to investigational treatment. No other deaths were reported during the 12-month follow-up period. Mild elevations in ALT and AST observed in four of five patients in cohort 2 are probably suggestive of a dose-dependent effect. In some patients, ALT and/or AST elevations were temporally associated with IFNγ ELISpot responses directed to the capsid or I-1c. Because patients with heart failure cannot safely use prophylactic immunosuppression with steroids to limit immune responses to gene therapy, owing to the potential risk of myopathy, it was considered particularly important to identify a potentially effective dose that was not associated with immune responses to capsid or I-1c. Thus, a decision was made not to enroll patients in a planned, higher-dose (3.0 × 1014 vg per patient) cohort because of similar efficacy findings between cohorts 1 and 2, coupled with concerns regarding elevations in liver enzymes and immune responses directed to the capsid or I-1c observed in patients from cohort 2.

The study patients in both cohorts exhibited trends toward stabilization or improvement in NYHA functional class, low-level exercise capacity (6MWT) and left ventricular systolic performance (LVEF) at 6 and 12 months. Reduced LVEF is associated with increased mortality30, and improvements in LVEF have been associated with improved cardiovascular outcomes31. In addition, most patients in both cohorts reported stabilization or improvements in quality of life, as measured by the MLHFQ total score and KCCQ overall summary score, at 12 months posttreatment, although the interpretation of changes in quality-of-life parameters is limited by the small sample size. The changes in pVO2 following gene therapy were variable, with stabilization or improvement in pVO2 at 6 and 12 months in most patients in cohort 1 (3.25 × 1013 vg) but worsening in cohort 2 (1.08 × 1014 vg) by 12 months. Notably, baseline pVO2 was higher in cohort 2 compared with cohort 1, suggesting that the reduction observed in this cohort may represent a regression to the mean. Furthermore, although 6MWT performance has shown correlation with pVO2 in several studies in patients with heart failure with reduced ejection fraction32, corresponding reductions in 6MWT were not consistently observed in patients with reduced pVO2 in our study. The patients were enrolled sequentially, and the baseline clinical characteristics including age and LVEF varied between cohorts. Patients in cohort 1 were older with higher baseline LVEF, whereas patients in cohort 2 were younger with lower LVEF but higher exercise capacity. In addition, the patients could receive dose adjustments in other heart failure treatments, although the patients were generally on individually optimized guideline-directed medical therapy at study entry. The apparent differences between cohorts in efficacy parameters should be interpreted with caution, as they may reflect the small sample size, as well as differences in baseline characteristics and treatment patterns, rather than the effects of the study treatment.

Gene therapy with AAV-based vectors holds promise for the management of heart failure. Not surprisingly, SERCA2a has emerged as an important gene therapy target based on its pivotal role in altered Ca2+ cycling in failing hearts33. However, despite the demonstrated safety and efficacy of antegrade coronary delivery of AAV1.SERCA2a in the CUPID phase 1 and early phase 2 trials in patients with advanced heart failure25,34, the subsequent CUPID-2 trial showed that treatment did not change the clinical course of patients with heart failure, and surprisingly, this intervention was associated with reduced ejection fraction35. One possible reason cited for the failure of AAV1/SERCA2a in CUPID-2 was inadequate delivery and uptake of the vector in the hearts of enrolled patients, underscoring the challenges of delivering heart-targeted gene therapy36. Notably, the current study used a more cardiotropic vector with better uptake in human myocardium at a greater than three- to tenfold higher dose compared with CUPID-2; this may result in improved myocardial gene delivery and ultimately efficacy. Another possible explanation could be that the activity of virally delivered SERCA2a in patients with heart failure was inhibited by PP1, similar to endogenous SERCA2a, limiting its efficacy. In this scenario, we might expect the expression of I-1c to reduce the PP1-mediated inhibition of SERCA2a, thereby increasing SERCA2a activity more efficiently than directly increasing SERCA2a expression. Consistent with a model where the posttranslational modification of SERCA2a activity plays a larger role in heart failure than the SERCA2a expression, a recent study found no difference in SERCA2a and PLN protein abundance between cardiac tissues from patients with heart failure (including patients with dilated cardiomyopathy and ischemic heart disease) and controls37. Animal models have suggested that targeting both SERCA expression and posttranslational modification may have synergistic effects in enhancing myocyte contractility38,39. PP1 is involved in many fundamental processes in cardiac myocytes and other cell types and has been implicated in numerous diseases including heart failure progression and atrial fibrillation40. The selective modulation of the PP1 signaling cascade may be an effective therapeutic option for patients with these disorders.

AB-1002 gene delivery led to clear biological effects, including a transduction efficiency ranging from 0.57 to 2.6 vg dg−1 (30-fold higher compared with transduction measured in a porcine model of heart failure26), transient elevation of liver enzymes and a transient T cell response. The expression of I-1c in cardiomyocytes is intended to inhibit PP1, restoring normal PLN phosphorylation status, SERCA2a activity, Ca2+ uptake and contractility. A tissue sample provided by a patient in cohort 2 had PLN phosphorylation levels similar to samples from healthy controls without heart failure, suggesting that I-1c expression has the expected biologic activity.

The therapeutic window observed for AB-1002 in this phase 1 study formed the basis for the 2 doses (3.25 × 1013 and 6.5 × 1013 vg) and clinical endpoints chosen for the phase 2 study (GenePHIT; NCT05598333). The doses of AB-1002 in the phase 2 study will use the same low dose and a second dose that represents a midpoint between the two doses used in the phase 1 study; this may help to reduce the mild elevations in liver enzymes and immune responses to viral capsid or I-1c seen among patients in cohort 2 in this study while maintaining, and potentially improving on, the preliminary signals of efficacy observed in cohort 1 at a dose of 3.25 × 1013 vg. Importantly, patients with neutralizing antibodies to AAV at baseline were excluded in this study but will be eligible to participate in GenePHIT. Up to 150 patients with nonischemic NYHA class III heart failure are planned to be enrolled in GenePHIT.

The small sample size and single-arm, open-label, non-placebo-controlled design of this phase 1 study limits the interpretation of data and does not control for confounding by any natural trajectories of the collected outcomes. Owing to the small sample size, we cannot exclude the possibility that the efficacy results may be related to imbalances between the cohorts in demographics and the baseline clinical characteristics of the patients or natural variability in disease progression. Although most patients received the optimal standard-of-care treatment at baseline, changes in the dose of standard concomitant medications were allowed during the study, which may have influenced the outcomes including LV systolic function, NYHA class, exercise performance and quality of life. The potential effects of dose changes on outcomes cannot be assessed in this study. Nevertheless, the results suggest that both doses of AB-1002 may be associated with beneficial clinical effects and acceptable safety and tolerability that support the ongoing development of a phase 2 clinical trial.

For patients with NYHA class III systolic heart failure, survival remains poor despite the currently available therapies. AB-1002 is an investigational gene therapy consisting of a cardiotropic AAV2i8 vector delivering the gene sequence expressing the constitutively active PP1 inhibitor I-1c, a protein that enhances SERCA2a function and calcium cycling in cardiomyocytes to increase cardiac contractility while improving relaxation. In this single-arm phase 1 clinical trial, a single intracoronary infusion of AB-1002 in patients with heart failure resulted in no treatment-related AEs, and the preliminary efficacy results showed a relative stability of outcomes and potential for clinically meaningful improvement, particularly with regard to LV systolic performance. Based on these data, the further investigation of AB-1002 is warranted for the possible treatment of patients with heart failure.

Methods

Study design

This is a phase 1, prospective, multicenter, open-label, sequential dose-escalation study (ClinicalTrials.gov registration number NCT04179643) designed to explore the safety, feasibility and efficacy of a single intracoronary infusion of AB-1002 in patients with NYHA class III systolic heart failure. The study was conducted at four sites located within the USA, including Minnesota, Ohio and Wisconsin. All patients were evaluated over a primary observation period of 12 months after treatment intervention, the details of which are reported here. Following the 12-month primary observation period, patients entered a long-term (24-month) follow-up period consisting of semistructured telephone questionnaires every 6 months, for a total of 3 years of follow-up; long-term follow-up is ongoing for three patients.

Originally, three dose groups were planned. The dose escalation was determined on the basis of the occurrence of DLTs—including myocarditis, myositis, neutropenia and cardiac tamponade—attributed to AB-1002. There was no independent oversight of safety in this trial. Safety was reviewed by a clinical safety committee consisting of the study principal investigators and a cardiologist employed by the study sponsor. The DLT window was not prespecified. A dose escalation was only to occur if no DLT event was observed between 6 h postdosing and following the review of safety data in that cohort by the clinical safety committee. The first cohort (dose of 3.25 × 1013 vg per patient) was planned to include three patients; if one of the three patients experienced a DLT, an additional patient was to be enrolled in that cohort. The second cohort (1.08 × 1014 vg per patient) and the third cohort (dose of 3.0 × 1014 vg per patient) were planned to include up to six patients each. If one of the six patients experienced a DLT, an additional patient was to be enrolled in that cohort. If there was only one DLT in the expanded cohort, the dosing would proceed to the next dosing level. However, if a second patient experienced a DLT, the enrollment would be paused to determine the next steps for dosing. Based on the preliminary efficacy findings in cohort 1, elevations in liver enzymes observed in patients from cohort 2 and antibodies against the viral vector in both cohorts, the protocol was amended to enroll three additional patients in cohort 1 (3.25 × 1013 vg) rather than enroll patients into a higher dose cohort as originally planned. For more information, the study protocol and statistical analysis plan are included in the Supplementary Appendix.

Study population

Key inclusion criteria

Eligible patients were men and women >18 years of age with chronic nonischemic cardiomyopathy, LVEF of 15–35% (inclusive) as determined by transthoracic echocardiography within 6 months before enrollment and NYHA class III heart failure for ≥1 month despite appropriate stable pharmacologic therapy (including but not limited to β-blocker therapy, angiotensin-converting enzyme inhibitor, ARB, sacubitril–valsartan or cardiac resynchronization therapy). Patients of childbearing potential or capable of fathering a child must have used acceptable forms of birth control throughout the study and for 6 months following AB-1002 administration.

Key exclusion criteria

Patients with ischemic, hypertrophic, infiltrative, inflammatory and restrictive forms of cardiomyopathy or uncorrected third-degree heart block were excluded. Patients who were supported with intravenous inotropic therapy, an intraaortic balloon pump or another form of percutaneous mechanical cardiac assist device within 30 days before enrollment were ineligible, as were patients who had cardiac surgery or percutaneous coronary intervention, were dependent on dialysis or had serum creatinine >2.5 mg dl−1 during the same time frame. Patients were also excluded if they had clinically relevant myocardial infarction within 6 months before enrollment; had prior heart transplant, LV reduction surgery, cardiomyoplasty, passive restraint device or surgically implanted LVAD or cardiac shunt; or were anticipated to undergo cardiac resynchronization therapy, LV reduction surgery, heart transplant, percutaneous or surgical coronary revascularization or valvular repair or replacement within 3 months of AB-1002 dosing. Patients with neutralizing antibodies against AAV at a titer of >1:5 within 6 months before the administration of investigational product were not eligible to participate. Patients were similarly excluded if they had known hypersensitivity to contrast dyes for angiography or a history of or likely need for high-dose steroid treatment before the contrast angiography; an expected survival <1 year in the judgment of the investigator; active or suspected infection within 48 h before enrollment; cancer within 5 years of AB-1002 administration (with the exception of those with a negligible risk of metastasis or death); known intrinsic liver disease or liver function tests >2× the upper limit of normal; bleeding diathesis or thrombocytopenia, anemia, neutropenia, known acquired immunodeficiency syndrome or human-immunodeficiency-virus-positive status or other condition that in the opinion of the investigator would preclude study participation; or documented history of noncompliance with medications or illicit drug use during screening. Pregnancy, breastfeeding, previous participation in a gene therapy trial or receiving other investigational interventions were also exclusionary.

Outcomes

Prespecified safety endpoints were assessed over the 12-month follow-up period. These included adverse events (AEs) and observed changes from baseline in clinical laboratory tests, vital signs and electrocardiograms. AEs were graded according to the National Cancer Institute Common Terminology Criteria of Adverse Events (v5.0)41. TEAEs were defined as new AEs or those that worsened in intensity or frequency after the administration of the investigational product. All-cause mortality, cardiovascular mortality, cardiac-related hospitalization, cardiac transplantation and LVAD implantation were recorded over the 12-month follow-up period and will continue to be collected over the course of the additional 24-month long-term follow-up period (until month 36 post intervention).

This phase 1 study did not prespecify primary and secondary efficacy endpoints but did include prespecified efficacy assessments of functional status, physiology and quality of life. The prespecified endpoints assessing functional status included pVO2 as determined by cardiopulmonary exercise testing, a 6MWT and NYHA classification and were performed at screening and at months 3, 6 and 12. Echocardiography was prespecified as a physiologic assessment and was used to assess the impact of treatment on LV systolic function; studies were interpreted by local readers at each site and by a central reader to ensure consistency and to avoid interobserver bias. These assessments were performed at screening and at 3, 6 and 12 months after administration of AB-1002. The prespecified quality-of-life assessments were conducted at screening; week 8; and months 3, 6 and 12 using the MLHFQ. The protocol was amended after the first three patients were enrolled to include KCCQ, which was administered to patients in cohort 2 and the cohort 1 expansion at screening; week 8; and months 3, 6 and 12. For the MLHFQ, lower scores indicate better quality of life; for the KCCQ, higher scores indicate better quality of life.

Additional prespecified laboratory testing was performed for relevant biomarkers and immunologic markers including NT-proBNP or BNP, as well as troponin I, complement profile and anti-AAV neutralizing antibodies. Cellular immunity to AAV was assessed using IFNγ ELISpot assays to quantitate T cells specific to the capsid and to I-1c. The assays were conducted 1 day before screening, at screening, immediately following infusion, and at 18–24 h; 2, 4, 6, 8, 10 and 12 weeks; and 6 and 12 months following infusion.

Neutralizing antibody assay

The neutralizing antibody assay was a transduction–inhibition assay. The assay used an AAV2i8-secNLuc vector containing a NanoLuc Luciferase reporter gene modified with a secretion sequence. Successful transfection of HEK293 cells with the AAV2i8-secNLuc vector yielded an increase in luminescence detected in the culture supernatant, which could be inhibited by the presence of neutralizing antibodies.

On the day of assay initiation, a vial of assay-ready HEK293 cells (passage 5) was removed from liquid nitrogen storage. The cells were thawed and resuspended in room temperature CTS AIM-V assay medium. The cells were centrifuged at 125–200g for 5 min and then resuspended in the same medium. The cell density and viability were determined using the ViCell XR automated cell counter, and the amount of cell suspension needed to achieve a density of 6 × 105 cells ml−1 was calculated. A total of 50 μl of cell suspension containing 30,000 cells was dispensed into each well of a 96-well cell culture plate. The plates were incubated at 37 °C with 5% CO2 for ≥1 h.

The anti-AAV2-positive control antibody study samples were diluted at 1:2 in the assay medium, and the AAV2i8-secNLuc vector dilutions were prepared in assay medium in a codilution plate. A total of 60 μl of AAV2i8-secNLuc vector dilution was mixed with an equal volume of human serum samples or controls and incubated at 2–8 °C for 60 (±10) min. After the 30-min coincubation period, the cell plates were removed from the incubator and chilled at 2–8 °C for 30 (±5) min. After chilling, 50 μl of the vector–serum or vector–control mixture was added to the cell plates, bringing the final volume to 100 μl and achieving 10,000 multiplicity of infection of vector per cell. The cell plates were incubated at 2–8 °C for 90 (±15) min, then transferred to a humidified incubator set to 37 °C with 5% CO2.

After 48 (±4) h, the plates were inspected under a microscope at room temperature for growth and to ensure no contamination. A total of 50 μl of supernatant from each well was transferred to a 96-well black polystyrene microplate containing 50 μl of cell-culture-grade water, achieving a final volume of 100 μl per well. The Nano-Glo reagent was prepared by adding 200 μl of the substrate to a thawed 10-ml bottle of assay buffer. A total of 100 μl of Nano-Glo reagent was added per well to the assay read plates, which were incubated for 3 min in the dark with gentle mixing. The luminescence was read immediately, ensuring <15 min elapsed from the addition of the substrate to the luminescence reading. The results were normalized, and the changes from baseline for neutralizing antibody titer are reported.

IFNγ ELISpot assay

Cryopreserved peripheral blood mononuclear cells isolated from patient whole-blood (Precision For Medicine) were thawed and plated at a concentration of 300,000 cells per well on a 96-well filter plate precoated with anti-human IFNγ antibody (Cellular Technology Limited). Cells were stimulated overnight (20–24 h) at 37.0 °C with a pool of peptides consisting of 15-mers offset by four amino acids spanning the sequence of AAV2i8 capsid (181 peptides; JPT) or I-1c protein (14 peptides; JPT). The final concentration of each peptide was 1 μg ml−1. Roswell Park Memorial Institute medium containing dimethyl sulfoxide peptide diluent was used as a negative control, and anti-CD3 and CEF-X peptide pools were used as positive controls. All antigens were tested in triplicate.

Following incubation, cells were removed, and captured cytokine was detected using a biotinylated secondary antibody, followed by colorimetric development using streptavidin and an enzymatic substrate (Cellular Technology Limited). The plate was dried overnight, and spots were counted using a CTL ImmunoSpot S6 Analyzer. The mean value of each triplicate was calculated, and the mean assay background (negative control) was subtracted from the antigen-stimulated conditions. The results were normalized and reported as spot-forming units per million peripheral blood mononuclear cells.

Transduction efficiency and molecular assessments

Patients were given the option to provide a sample of cardiac tissue, obtained either by elective endomyocardial biopsy or at the time of a clinically indicated open chest procedure, for the molecular and cellular assessment of AAV2i8 I-1c transduction efficiency and PLN Ser16 phosphorylation (prespecified as exploratory endpoints). The tissue samples were flash frozen upon collection. In this study, cardiac tissue was collected from a single patient who underwent LVAD implantation at 13 months postinfusion of AB-1002. Postmortem LV tissue from three individuals without cardiac pathology served as controls; these tissues were obtained commercially from BioIVT or through the National Disease Research Interchange.

To assess the transduction efficiency, DNA was extracted from the myocardium of a treated patient and healthy human control samples using the QIAamp 96 DNA kit and QIAcube high-throughput workstation via the manufacturer’s protocols. The DNA was measured using a microvolume spectrophotometer (NanoDrop), and vector genomes were quantified using droplet digital PCR (Bio-Rad) and probes targeting the polyA sequence in the transgene (Supplementary Table 1). To assess the PLN Ser16 phosphorylation, the total protein was extracted from the treated patient’s myocardium and three healthy human control tissue samples. Western blotting was performed on a 12% (w/v) sodium dodecyl sulfate–polyacrylamide gel electrophoresis gel using a PLN-pSer16 primary antibody (Badrilla; 1:500 dilution) and an IRDye 800CW goat anti-rabbit secondary antibody (1:5,000 dilution) with 30 μg sample per lane normalized to a calsequestrin, which served as a loading control. The levels of myocardial PLN phosphorylation in the treated patient (in duplicate) were similar to that in the control samples.

Ethics

The study was conducted according to the principles of the Declaration of Helsinki, the International Conference on Harmonisation Guidance on Good Clinical Practice and the requirements of the US Food and Drug Administration regarding the conduct of human clinical studies. The protocols, including for biopsy and control tissue samples, were reviewed and approved by an institutional review board or independent ethics committee at each site: The Christ Hospital (Cincinnati, OH, USA), IRB00001448; University of Wisconsin (WIRB, Puyallup, WA, USA) and Minneapolis Heart Institute (WIRB, Puyallup, WA, USA), IRB00000533; and Ohio State University (Columbus, OH, USA), IRB0000181. All the patients and healthy controls provided written informed consent. The control tissue samples used for the biopsy analysis were deidentified and provided by a commercial source (BioIVT) or through the National Disease Research Interchange.

Statistical analysis

Data were summarized descriptively (mean, s.d. and percentages). Safety data were analyzed using the safety analysis set, which included all enrolled patients who received any amount of study therapy, and efficacy analyses were assessed on the basis of the full analysis set, which included all enrolled patients who received the complete infusion of investigational product. No formal hypothesis testing was planned, and no sample size calculations were performed. The sample size was based on feasibility; up to 18 evaluable participants were planned to be enrolled, with up to 6 within each cohort. If a DLT occurred in a cohort, an additional participant could be enrolled in that cohort for a total population of up to 18 participants. All data were analyzed and reported using SAS v.9.4.

Reporting summary

Further information on the research design is available in the Nature Portfolio Reporting Summary linked to this Article.