Death following high-dose AAV9 gene therapy in a patient with advanced SMA-PME

Abstract

Spinal muscular atrophy with progressive myoclonic epilepsy (SMA-PME) is an ultra-rare, fatal autosomal-recessive disorder caused by ASAH1 mutations, with no curative treatment. We report the first-in-human intravenous administration of an AAV9 vector carrying the human ASAH1 coding sequence in a 15-year-old female with advanced SMA-PME (heterozygous ASAH1 c.456A>C and c.918-2A>G mutations). The patient presented with severe neuromuscular impairment, deafness, and recurrent myoclonic status epilepticus, without detectable anti-AAV9 neutralizing antibodies. A marked clinical deterioration warranted compassionate use (GNT-017-ASAH-CU). A dose of 2.2 × 10¹⁴ vector genomes per kilogram of body weight was administered under prophylactic prednisolone and sirolimus. The patient experienced rapid onset of complement activation leading to cytokine-mediated capillary leak syndrome, culminating in refractory shock, multiorgan failure, and death on day 8. A postmortem examination revealed acute circulatory failure as the cause of death without myocarditis or thrombotic microangiopathy. Some endothelial injury was suggested by a rise in von Willebrand factor from days 4 to 8, paralleled by increased hyaluronic acid on days 7 and 8. These findings underscore the potential for life-threatening innate immune activation in patients with advanced SMA-PME receiving high-dose systemic AAV9 gene therapy, highlighting the need to identify high-risk patients and proactively monitor biomarkers of endothelial injury.

Graphical abstract

First-in-human intravenous AAV9-ASAH1 gene therapy in advanced SMA-PME led to fatal cytokine-mediated capillary leak syndrome with multiorgan failure. Biomarker changes suggested endothelial injury. These findings highlight the risk of severe innate immune activation with high-dose systemic AAV9 and the need for patient risk stratification and biomarker monitoring.

Introduction

Spinal muscular atrophy with progressive myoclonic epilepsy (SMA-PME) is an ultra-rare autosomal-recessive disorder caused by mutations in the ASAH1 gene, which encodes acid ceramidase (ACDase).¹

SMA-PME is an often-misdiagnosed lysosomal storage disorder, typically presenting between ages 2 and 7 with lower motor neuron signs, though adolescent onset is possible.^1,2,3,4 Progression involves weakness, tremors, myoclonus, seizures, cognitive decline, and hearing loss.^1,5,6 Death usually occurs in early adulthood due to respiratory failure or refractory seizures.^1,6 No curative medication exists, and current treatment is supportive.

We report the case of a 15-year-old girl with SMA-PME who received a single intravenous infusion of an AAV9-based ASAH1 gene therapy, administered under a compassionate use authorization from the French Agence Nationale de Sécurité du Médicament (ANSM) and in accordance with CARE guidelines.

The investigational product was a recombinant AAV9 vector containing the human ASAH1 open reading frame, which encodes ACDase, under the control of the cytomegalovirus early enhancer/chicken beta-actin (CAG) promoter (AAV9-CAG-hASAH1 or GNT0009) (Denard, Marinello et al., data not shown).

This case represents the first reported use of gene therapy in a patient with SMA-PME and offers important insights into both the clinical presentation of this ultra-rare disorder and the potential mechanisms driving toxicity following high-dose intravenous AAV9 administration. The findings provide timely and relevant evidence that informs broader challenges of AAV-mediated gene delivery and underscore the need for continued refinement of vector design, dosing strategies, and immune management.

Patient details and clinical findings

A 15-year-old female (born in 2008), presenting with early-onset dyspraxia, fine motor impairment, and sensorineural hearing loss, was diagnosed with SMA PME with bi-allelic pathogenic ASAH1 variants (c.456A>C, p.Lys152Asn [exon 6 skipping]; c.918-2A>G, p.Glu306fsStop17). Absence epilepsy began at age 7–8 years. Progressive neurologic deterioration followed, with worsening myoclonus, atonia, and emerging cerebellar signs including ataxia. The disease course was marked by progressive motor and cognitive decline, with loss of ambulation, denervation on electromyography, and mild cerebellar atrophy on MRI, progressing by age 14 to severe myoclonic epilepsy with recurrent status epilepticus and wheelchair dependence (2022–2023) (Figure 1A).⁶ Neurological function was relatively stabilized with lamotrigine (400 mg daily), clobazam (40 mg daily), and perampanel (6 mg daily), despite resting myoclonia affecting all four limbs and impaired communication.

Figure 1.

Clinical course and biological data

(A) Timeline of administered treatments from 7 days before vector injection (gene therapy), starting with an immunosuppressive treatment (sirolimus, PO [per os]), until the death of the patient, and the major events that occurred during the clinical course. GNT0009 intravenous (i.v.) administration was performed on day 1 at 5:00 p.m. (B) Chest X-ray on day 4. New ill-defined alveolar opacities involving the lower lung zones and perihilar regions can be observed, with obscuration of the cardiac borders (suggestive of ARDS), as well as subtle linear atelectasis at the left lung base. No pleural effusion was observed. (C–F) Days indicated on the x axis are based on the compassionate use protocol, in which the day of gene therapy administration is day 1. (C) Platelets and D-dimer kinetics during the clinical course. Dotted lines represent the range of normal platelet values (between 150 and 450 × 109/L). The dashed line represents the limit of a normal D-dimer value (<500 ng/mL). (D) Hepatic enzyme (AST, ALT, and GGT) kinetics during the clinical course. The dashed line represents the limit of a normal hepatic enzyme value (<32 UI/L). (E) Cardiac markers (troponin I and BNP) kinetics during the clinical course. The dashed line represents the limit of a normal troponin I value (<38 UI/L). The dotted line represents the limit of a normal BNP value (<100 UI/L). (F) Hyaluronic acid (HA) and von Willebrand factor (vWF) kinetics during the clinical course. The HA limit of detection (LOD) was 5 ng/mL; all values from days −47 to 6 were below the LOD. The dashed line represents the healthy donor vWF threshold (10.7 μg/mL, based on 6 volunteers).

At screening, spirometry suggested a restrictive ventilatory defect, with a reduced functional residual capacity (FRC; 0.54 L, 30% of predicted). A chest X-ray performed 2 months prior to treatment was unremarkable. Cardiac ultrasonographic parameters were within normal limits, with a left ventricular ejection fraction (LVEF) of 67.3% and fractional shortening (FS) of 36.2%. Biochemistry and hematology tests were normal, except for gamma-glutamyl transferase (GGT) at 62 UI/L (N < 32 UI/L). The baseline serum calprotectin level was 32,185.9 ng/mL and decreased to 1,167.85 ng/mL (N = 1–∼10 μg/mL)⁷ following initiation of immunosuppressive prophylaxis. Immunologic screening prior to GNT0009 injection demonstrated the absence of anti-AAV9 neutralizing antibodies (NAbs). Retrospective analyses showed that total IgM antibodies against AAV9 capsid were below cut-point levels, whereas total low IgG antibodies were detected prior to vector injection, with a 1:157 titer at day −1, as assessed by ECLA. The levels of complement system components (C4, C3, CH50, and sC5b9) were normal at inclusion. Before treatment, the patient received recommended immunizations for Neisseria meningitis and Streptococcus pneumoniae a few weeks prior to treatment. She received preventive treatment with amoxicillin-clavulanic acid (3 g daily; May 6–14, 2023) for mild bronchial congestion. The clinical course was rapidly favorable with supportive respiratory physiotherapy, and no infection was documented. A radiologic control 1 day before infusion was normal. This patient was treated under a compassionate use program authorized by the ANSM. Approval was granted given the rapid and severe clinical deterioration and reduced life expectancy. The patient and her parents were informed of the risks and provided written consent.

On May 16, 2023 (day 1), the patient (54.3 kg) received 2.2 × 10¹⁴ viral genomes per kilogram of body weight (vg/kg) of GNT0009 (total dose: 1.2 × 10¹⁶ vg). Prophylactic immunosuppressive therapy included prednisolone and sirolimus (Figure 1A). During the first 3 days, the patient was clinically stable with normal laboratory values. Sirolimus levels in blood were within the target range of 4–8 ng/mL. Between days 3 and 4, she developed fever, asthenia, and worsening myoclonia. The patient had inflammatory syndrome with elevated D-dimers and liver enzymes (Figures 1C and 1D). There was no anemia or acute kidney injury. This episode was followed by mild proteinuria and a mild decrease in platelet count, prompting a dose of eculizumab (900 mg daily) on day 5 to prevent complement-driven platelet consumption. On day 5, the patient experienced worsening of both respiratory and cardiovascular function, characterized by a rapid decline in cardiac output and progression to cardiogenic shock (Figures 1A, 1B, and 1E), necessitating high-dose catecholamine and extracorporeal membrane oxygenation support. On day 8, life-sustaining care was withdrawn because of multiorgan failure and ischemic neurologic injury, which rapidly led to her death.

Laboratory and postmortem studies

A retrospective multiplex serum cytokine analysis revealed elevated levels of CD40 ligand (CD40L), interleukin-18 (IL-18), and VEGF-A prior to treatment (Figure 2A); these elevations normalized following immunosuppressive therapy. On day 1, a retrospective cytokine assessment showed elevated circulating levels of IL-1α (60-fold higher than 6 healthy donors), eotaxin (3-fold), and IL-17F (2-fold). Following AAV administration, IL-1α levels decreased, whereas eotaxin remained stable through day 3 before gradually normalizing over time (Figure 2B). These findings indicate a pre-existing inflammatory state prior to dosing. From days 2 to 5, longitudinal monitoring showed that the levels of different pro-inflammatory cytokines increased (Figure 2C), preceding a cytokine storm (Figure 2D).

Figure 2.

Immune response assessments

For each image, the days on the x axis are based on the compassionate use protocol, in which the day of vector administration (gene therapy) is day 1. (A) Cytokines with elevated levels before gene therapy that normalized with immunosuppressive treatments (CD40-L, IL-18, and VEGF-A). (B) Cytokines exhibiting high levels at the time of inclusion (day −1) (eotaxin, IL-17f, and IL-1a). (C) Cytokine kinetics exhibiting early elevation following gene therapy (IP10, MCP1, MCSF, and IFNγ). (D) Cytokine kinetics representative of the cytokine storm occurring before death of the patient (IL-1Ra, IL-6, GCSF, FGF2, IL-8, IL-10, and IL-15). (E) Complement element kinetics (CH50 [normal value (norm. val.) between 70% and 130%], C4 [norm. val. between 93 and 380 mg/L], C3 [norm. val. between 660 and 1,250 mg/L], sC5b9 [norm. val. <300 ng/mL], and C4d). (F) Anti-AAV9 Ig levels during clinical course. The dashed line represents the IgM cut point (1.2 μg/mL). IgG antibodies are expressed as a titer 1:X.

A mixed picture of complement components was observed from day 4, with an important elevation of sC5b9 to above 1,650 ng per mL (normal range: <300 ng per mL) on day 5. Total hemolytic complement (CH50) dropped on day 7 due to the administration of eculizumab on day 5. Complement factors C3 and C4 dropped on day 7 (Figure 2E), and von Willebrand factor (vWF; ELISA, Thermo Fisher Scientific) showed a slight but consistent increase from days 4 to 8, paralleling high values of hyaluronic acid (HA) on days 7 and 8 (Figure 1F). Anti-AAV9 IgM antibodies were below the assay cutoff until day 4, then rose to peak values of 8.8 and 6.7 μg/mL on days 7 and 8, respectively. The anti-AAV9 IgG titers increased from day 2, reaching peak levels on days 4 and 7 (1:10,240) (Figure 2F).

Parental consent for autopsy was obtained. Postmortem analysis identified acute circulatory failure as the cause of death, with no overt signs of myocarditis or thrombotic microangiopathy (TMA). The lungs were edematous with intra-alveolar hyaline membranes and perivascular inflammatory infiltrates, suggestive of acute respiratory stress syndrome (ARDS). Skeletal muscle histopathology showed neurogenic myofiber atrophy, chronic reinnervation, and lipid overload in type 1 fibers. Central nervous system (CNS) analysis revealed ballooned neurons with PAS-positive inclusions and microglial activation (Figure 3A).

Figure 3.

Postmortem tissue analysis and biodistribution

(A) Histology images of heart, lung, and C4 hippocampal area. The heart shows the left and right atrial fibrosis (top left, hematoxylin-eosin staining). The lungs were edematous with intra-alveolar hyaline membranes (top right, hematoxylin-eosin staining, hyaline membrane indicated with black arrows) and light bilateral perivascular inflammatory infiltrates, indicating acute respiratory distress syndrome. In the CNS, microglial activation was observed in the C4 hippocampal area (bottom left, Iba1 antibody immunohistochemistry [IHC]), as well as in the posterior horn and temporal cortex. Ballooned neurons were also observed in the striatum, medulla oblongata, spinal cord, and C4 hippocampal area with PAS-positive cytoplasmic inclusions (bottom right, PAS staining, ballooned neurons indicated with the black arrow). (B) Vector biodistribution assessed by droplet digital PCR (ddPCR) and expressed as the number of vector genomes (VCN) per diploid genome in each tissue. (C) Protein levels of vector-derived human acid ceramidase (hACDase; ASAH1) in fibroblasts and postmortem tissues (SMA-PME) relative to healthy donor controls. The blot shows the α (13 kDa) and β (40 kDa) subunits of hACDase, with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) serving as a loading control. In addition to the mature subunits, a higher-molecular-weight band (∼55 kDa), consistent with the ACDase precursor form (indicated by asterisks), was detected in the samples. The amount of total protein loaded per tissue sample is indicated (μg/well). CTRL, control; CERV., cervical; SP., spinal cord; LUMB., lumbar.

Vector distribution and transgene expression

Vector biodistribution was evaluated using droplet digital PCR, revealing abundant vector genome copies in peripheral tissues, with highest levels in liver and spleen. In contrast, vector copy numbers (VCNs) in the CNS were substantially lower, showing a 16- to 550-fold decrease compared to the liver (Figure 3B). RT-qPCR analysis confirmed ASAH1 transgene expression across all examined tissues, with the highest ASAH1 mRNA levels detected in the liver and adrenal glands. In the CNS, the highest transcript levels were found in the cerebellum, followed by the cortex (data not shown). ACDase protein expression was assessed in various tissues by immunoblotting and compared with endogenous levels from protein extracts of healthy adult donors. In the patient, the highest protein levels were detected in the liver and spleen, followed by the heart, adrenal glands, and the lungs (Figure 3C). In peripheral tissues, ACDase levels were generally higher than in controls, whereas they remained below endogenous levels in various regions of the CNS 8 days after vector injection.

Discussion

A 15-year-old female with advanced SMA-PME developed severe cardiovascular toxicity within 5 days of receiving an intravenous dose of 2.2 × 10¹⁴ vg/kg of AAV9-CAG-hASAH1. Despite mitigating therapies, she died from refractory cardiovascular and multiorgan failure on day 8.

Severe adverse events (SAEs), including TMA and atypical hemolytic uremic syndrome (aHUS), have been observed following high-dose (>1.0 × 10¹³ vg/kg) intravenous AAV gene therapies,^8,9 particularly with AAV9 across multiple conditions, including Danon disease, Duchenne muscular dystrophy (DMD), and SMA.^10,11 However, in the present case, despite evidence of mild thrombocytopenia and complement activation, no features of TMA or aHUS were observed, suggesting a distinct pathophysiological profile.

We propose a mechanism driven by the early innate immune response, in which complement-mediated endothelial injury leads to a cytokine-mediated capillary leak syndrome, culminating in refractory shock and multiorgan failure. Retrospective analyses showed that the levels of several cytokines were elevated prior to or at inclusion, suggesting a pre-existing inflammatory state, either systemic or organ specific (e.g., pulmonary), that may have contributed to the dysregulated hyperinflammatory response. Direct endothelial cell transduction may also contribute to this vascular toxicity, as suggested by findings from Hordeaux et al. in a patient with DMD who developed fatal acute respiratory distress syndrome.¹² In our study, ACDase expression was clearly detected in several organs 8 days after vector administration, and levels generally correlated with VCNs. The highest level of ACDase protein was found in the liver, which was associated with an increase in ALT, AST, and HA, a specific indicator of liver endothelial cell injury. In lungs, the amount of ACDase protein was close to the endogenous level, although cell-type-specific vector transduction in this tissue was not assessed. In the CNS, ACDase protein levels were lower than in peripheral tissues and below endogenous levels 8 days post-injection, reflecting reduced vector transduction due to restricted passage across the blood-brain barrier and possibly rate-limiting intracellular vector processing in this tissue.

Multiple factors likely contributed to this fatal outcome, including pre-existing inflammation, high vector dose, AAV serotype, intravenous delivery, and a low level of anti-AAV9 IgG prior to injection (despite undetectable NAbs). Notably, this presentation aligns with emerging reports of severe cardiopulmonary toxicity without TMA, as described by Lek et al., in which cytokine-mediated capillary leak and pulmonary involvement were predominant.^13,14

While complement inhibition or intensified immunosuppression may be considered, evidence from sepsis and related hyperinflammatory states suggests that delayed initiation—particularly after cytokine storm onset—may markedly reduce the efficacy of agents such as eculizumab or cytokine-targeted therapies.¹⁵ Importantly, sirolimus did not appear to mitigate early immune response in this patient, consistent with its predominant effects on adaptive immunity rather than the innate and complement-mediated pathways implicated in endothelial injury and capillary leak syndrome after systemic AAV gene transfer. Although the respective contributions of SMA-PME-related inflammation and recent bronchial congestion without documented infection remain difficult to disentangle, these observations underscore the importance of identifying high-risk patients, conducting careful pre-treatment inflammatory assessment, proactively monitoring endothelial injury, and implementing timely mechanism-based interventions.

Acknowledgments

The authors gratefully acknowledge the patient’s family and members of the ASAP for Children Association for their support. We also thank the members of the safety advisory board (SAB), Prof. Olivier Benveniste, and Prof. Enrico Bertini for data review and advice and Genosafe, a clinical research organization (CRO), for contribution to the analyses. The study was supported by Genethon, a non-profit organization. The graphical abstract was created in https://BioRender.com.

Author contributions

Study design, O.B.-T., A.V., A.B.-B., and G.P.; data analysis and interpretation, O.B.-T., A.V., A.B.-B., and G.P.; writing – original draft, O.B.-T., A.V., A.B.-B., and G.P.; writing – review & editing, O.B.-T., A.V., A.B.-B., G.P., L.P., T.C., N.K.-L., B.B., M.L., M.S., K.A., A.R., J.R., P.-L.L., I.P., D.S., V.F.-B., and T.L., ; project management, L.P., T.C., N.K.-L., B.B., M.L., M.S., K.A., and A.R.; pharmacovigilance and regulatory support, L.P., T.C., N.K.-L., B.B., M.L., M.S., K.A., and A.R.; clinical data collection and analysis, J.R. and P.-L.L.; neuropathological analysis, I.P. and D.S.; postmortem data interpretation, I.P. and D.S.; complement system analysis, V.F.-B.; genetic analysis, T.L.; vector biodistribution, protein expression, and immunological analyses, M.M., J.D., M.D., P.-R.l.B., N.G., and P.V. All authors reviewed the manuscript draft and approved the final version for publication. O.B.-T. and A.V. had full access to all data and took responsibility for the integrity of the data and the accuracy of data analysis.

Declaration of interests

M.M., J.D., and A.B.-B. are inventors of a patent on gene therapy for ACDase deficiency.