AAV gene therapy for Tay-Sachs disease

Tay-Sachs disease (TSD) is an inherited neurological disorder caused by deficiency of hexosaminidase A (HexA). Here, we describe an adeno-associated virus (AAV) gene therapy expanded-access trial in two patients with infantile TSD (IND 18225) with safety as the primary endpoint and no secondary endpoints. Patient TSD-001 was treated at 30 months with an equimolar mix of AAVrh8-HEXA and AAVrh8-HEXB administered intrathecally (i.t.), with 75% of the total dose (1 × 1014 vector genomes (vg)) in the cisterna magna and 25% at the thoracolumbar junction. Patient TSD-002 was treated at 7 months by combined bilateral thalamic (1.5 × 1012 vg per thalamus) and i.t. infusion (3.9 × 1013 vg). Both patients were immunosuppressed. Injection procedures were well tolerated, with no vector-related adverse events (AEs) to date. Cerebrospinal fluid (CSF) HexA activity increased from baseline and remained stable in both patients. TSD-002 showed disease stabilization by 3 months after injection with ongoing myelination, a temporary deviation from the natural history of infantile TSD, but disease progression was evident at 6 months after treatment. TSD-001 remains seizure-free at 5 years of age on the same anticonvulsant therapy as before therapy. TSD-002 developed anticonvulsant-responsive seizures at 2 years of age. This study provides early safety and proof-of-concept data in humans for treatment of patients with TSD by AAV gene therapy. First-in-human combined intrathalamic and intrathecal gene therapy in two patients with Tay-Sachs disease provides early evidence on the safety and feasibility of the approach.

T SD and Sandhoff disease (SD) are recessively inherited and caused by mutations in HEXA and HEXB genes, respectively, encoding heterodimeric β-N-acetylhexosaminidase A enzyme (HexA). Severity correlates with residual activity; infantile patients typically have <0.1% normal enzyme activity and are diagnosed in the first 12-14 months of life 1 . Disease symptoms include inability to sit, difficulty swallowing, seizures and progressive loss of acquired developmental milestones. Presently, there is no effective treatment.
Here, we describe the initial results of an expanded-access clinical trial with two children with infantile TSD treated by a similar approach using a combination of AAVrh8-HEXA and AAVrh8-HEXB vectors through a combination of thalamic and CSF delivery. This study reports the first human use of AAV gene therapy for TSD and provides the basis for future studies.
Dosing. The study agent consists of an equimolar mix of AAVrh8-HEXA and AAVrh8-HEXB vectors. Dose scaling for patients was based on the nonhuman primate (NHP) to human brain weight ratio (Supplementary Table 1).
TSD-001 was treated at 30 months by i.t. administration only because of severe thalamic degeneration. The dose (1 × 10 14 vg) was split between cisterna magna (75%; 9 ml) and thoracolumbar junction (25%; 3 ml) using a SL-10 microcatheter in the intrathecal space 23 . TSD-002 was treated at 7 months by bilateral thalamic injections followed by i.t. delivery as above, for a total combined dose of 4.2 × 10 13 vg. The dose was 3.08 × 10 12 vg divided between thalami with a total volume of 180 µl per thalamus and 3.89 × 10 13 vg in 4.5 ml i.t. divided 75% in the cisterna magna and 25% in the lumbar intrathecal space.
General safety. Study procedures were well tolerated, resulting in no neurologic deficits or other notable acute procedural complications, with exception of a low-grade, transient postoperative fever occurring in the first 48 h (TSD-002), without any identifiable source. All other AEs were judged to be unrelated to the study (Table 1). Transaminases were higher than the non-TSD reference ranges but remained within the range for TSD 24 throughout the study (Fig. 1a,b). A very brief (<2-week) deviation from baseline (but remaining within the TSD range) was observed after the operative procedures and was deemed to be due to a combination of anesthesia medications and underlying liver dysfunction in TSD (Supplementary Table 2).
There were two serious AEs (SAEs), both of which were unrelated to the vector or infusion procedures. TSD-001 was hospitalized 9 months after treatment (3 months after immune suppression) due to viral pneumonia, which was subsequently complicated by ventilator-associated bacterial pneumonia and gastroesophageal reflux-associated aspiration. TSD-001 recovered and was discharged. TSD-002 was hospitalized 4 months after treatment with a febrile urinary tract infection from Klebsiella pneumoniae that responded to antibiotics.

Immune suppression and immune responses.
Neither patient showed sustained increases in total anti-capsid immunoglobulin G (IgG) from baseline to 6 months ( Fig. 1c), which is in stark contrast to the rapid increase in a patient with amyotrophic lateral sclerosis who was treated with 60 mg prednisolone per day after intrathecal AAV gene delivery 25 . The transient increase in total IgG documented at 2 months after treatment is likely related to i.v. immunoglobulin administration at 1 month (denoted by a gray arrow in Fig. 1c). TSD-001 showed a rise in anti-capsid neutralizing antibody titers, peaking at week 2 and again at 6 months after treatment (Fig. 1d). Increases in neutralizing antibody titers coincided with i.v. immunoglobulin treatment (not tested for anti-AAV antibodies). Mild T cell responses to AAVrh8 capsid were detected at months 3.5 and 6 in TSD-001 and up until month 1 in TSD-002 (Fig. 1e,f). Positive ELISpot assays in TSD-002 occurred when trough sirolimus levels dropped below 4 ng ml −1 . Neither patient showed T cell responses to HEXA or HEXB (Fig. 1g,h).

Bioefficacy: protein and enzyme activity in CSF and serum.
In both patients, baseline CSF HexA activity was ~0.3 nmol ml −1 per hour and increased to 0.5-0.6 nmol ml −1 per hour after treatment (Fig. 2a). Baseline serum HexA activity was ~0.3 nmol ml −1 per hour in both patients (Fig. 2a, b) and increased slightly at all time points (except at 6 months in TSD-002). Western blot of CSF showed HEXA protein levels increasing in both patients from 0 to 6 months after treatment (Fig. 2d). Various GM2 ganglioside species in CSF (Fig. 1c) were increased in TSD-001 and to a lesser degree in TSD-002, suggesting that GM2 levels may be directly proportional to disease state. There was a disparity between GM2 species, where GM2(22:0) and GM2(24:0) plateaued after treatment and remaining species continued to increase.

Radiologic assessments.
Postinjection magnetic resonance imaging (MRI) showed successful thalamic targeting in TSD-002, without any detectable deleterious effects (Fig. 3a). Fluid signal averaged 474 mm 3 per thalamus, ~2.6 times the infused volume. At the time of treatment, TSD-001 had advanced cortical atrophy, mild ventricular enlargement and diffusely affected white matter (Fig. 3b). The significance of this finding is unclear due to the advanced stage of disease in this patient and lack of longitudinal natural history data. MRIs from healthy infants at 7 and 12 months of age 26 and from an 11-month-old child with infantile TSD are provided for comparison (Extended Data Fig. 1). At baseline, TSD-002 MRI was suggestive of mild dysmyelination that appeared unchanged at 3 months after treatment (Fig. 3b). At 6 months, areas of increased myelination were noted in several sites, including the anterior and posterior corpus callosum. Volumetric analyses from TSD-001 suggested stabilization of corpus callosum volumes after treatment (Fig. 3d). Total brain volume of TSD-002 increased over the 6-month period, a finding also noted in the caudate, corpus callosum and cerebellum ( Fig. 3d and Extended Data Fig. 2). On diffusion tensor imaging (DTI), TSD-002 showed improved myelination of the external capsule and optic radiations (Fig. 4b). Interestingly, the improved myelination as visualized by tractography is comparable to that in a healthy ~9-month-old infant 27 (Fig. 4a and Extended Data Fig. 3). Quantification of fractional anisotropy of various brain structures showed maintained anisotropy that indicates preserved myelination, whereas myelination loss is typically expected in patients with TSD over this time period ( Fig. 4b and Extended Data Fig. 4).
Neurologic and neurodevelopmental outcomes. CHOP-INTEND scores reflected the advanced disease stage in TSD-001 and the early stage in TSD-002. TSD-001 remained stable, with a score of ~20 (Fig. 2e). Presently, TSD-001 is 5 years old and has been seizure-free since treatment with the same anticonvulsant therapy (Keppra) and nonadjusted dose that was in place before therapy, when seizures were active. Electroencephalograms, auditory brainstem responses and retinal examination findings did not change significantly in the first 6 months after therapy, after which time data were not available due to health-related travel restrictions.
The score for TSD-002 remained ~60 until 6 months, when it decreased to 52. At 13 months of age, TSD-002 was still sitting (an important milestone for infantile TSD 1 ), but with some truncal stability loss and exaggerated startle response. At 16 months of age, TSD-002 could sit for 5 seconds and retained visual tracking and response to auditory stimuli, but this ability was lost by 24 months of age. Seizures developed by 24 months of age (17 months after therapy). As with the first patient, electroencephalograms, auditory brainstem responses, and retinal examination findings did not change significantly in the first 6 months after therapy, after which time data were not available.
Pandemic-related interruption and telehealth follow-up. TSD-001 was no longer able to travel after her illness 9 months after vector, but the clinical assessment, neurologic examination and CHOP-INTEND score (scored at 26) were completed at 13 months after vector (3.5 years old) with a home visit by the investigator. Additional telehealth assessment was completed at 24 months after vector (4.5 years old) and at 30 months after vector (5 years old). At this time point, the patient remained seizure-free and retained subjective improvements in body tone. TSD-002 was unable to travel due to pandemic restrictions after 6 months post-treatment. A telehealth assessment was performed at 12 months after vector (19 months old) and showed a loss of stability of sitting to less than 5 seconds and a new onset of a seizure disorder, which was managed   Fig. 1 | Longitudinal safety and immunological studies. a,b, Serum levels of liver enzymes alanine aminotransferase (a; ALT) and aspartate aminotransferase (b; AST) enzymes in TSD-001 and TSD-002 patients before treatment (Pre-TX); weeks 1, 2 and 3 (W1, W2 and W3) and months 1, 2, 3 and 6 (M1, M2, M3 and M6). Shaded bars (a,b) represent the normal interval for both assays. Dashed line (b) indicates the maximum value reported in patients with TSD 24 . c, Total anti-AAVrh8 IgG in serum quantified by enzyme-linked immunosorbent assay. Vertical gray arrow indicates that both patients received i.v. immunoglobulin 1 month after treatment. d, Neutralizing antibody titers to AAVrh8 capsid quantified by a transduction assay. e-h, Peripheral blood mononuclear cells were isolated from patients TSD-001 and TSD-002 before treatment and at various time points thereafter to assess T cell responses by interferon-γ eLISpot assays using stimulation with pools of overlapping peptides spanning AAVrh8 VP1 (e,f) or human HeXA and HeXB proteins (g,h). Unstimulated or CD3/CD28-stimulated peripheral blood mononuclear cells were included in all assays as negative and positive controls. Data are presented as mean ± s.d. and were calculated from triplicate technical replicates for each sample. Statistical analysis was performed using two-way analysis of variance or mixed effects analysis (due to missing samples), followed by a Dunnett post hoc test to determine statistical significance of peptide pool stimulation versus the unstimulated sample (*P < 0.05). TSD-001 data in a-d are represented by blue circles and connecting lines, whereas TSD-002 data are represented by red triangles and connecting lines. D, day; NAb, neutralizing antibody; SFU, spot forming units.   with Keppra monotherapy. At the time of this report, TSD-001 is 5 years old and remains stable in every respect, including retention of improved body tone, and remains seizure-free clinically and by EEG. TSD-002 is 2 years and 5 months old, and outside physician reports indicate some additional deterioration of neurodevelopment but no recurrence of seizures on Keppra.

Discussion
This is the first-in-human test of AAV gene therapy in TSD. This trial achieved a mild increment in CSF HexA activity, showed good general safety in both patients and demonstrated feasibility of bilateral thalamic injections in a 7-month-old infant. There were no vector-related SAEs, and mild AEs after treatment, including a brief, mild postoperative fever, were judged to be related to procedural aspects of the protocol. Transaminase levels remained within the historical range reported for TSD 25 . No antitransgene immune responses were observed, and anti-capsid immune responses were short-lived and mild. Posttreatment CSF HexA activity in the range observed could alter the trajectory of disease progression in TSD, as patients with 0.5% of normal HexA activity exhibit a late infantile/juvenile phenotype and survival into the second decade of life, and those with 2% HexA activity can have a near-normal life span 1,28,29 . HexA levels in CSF may not accurately represent brain tissue activity, and it must be noted that the intrathalamic route could result in variable levels of enzyme activity in different brain structures. Another important aspect to consider in drawing conclusions from enzymatic assays alone is the fact that the artificial substrate used in the assays, MUGS, can be hydrolyzed by both HexA (αβ heterodimer) and HexS (αα homodimers) isozymes 30 . Therefore, the increase in hexosaminidase activity can be due to either isozyme or a combination of both at unknown ratios. The molecular basis for the increase in hexosaminidase activity is relevant, because in humans, HexA is the only enzyme that hydrolyzes GM2 ganglioside 31 , whereas HexS hydrolyzes sulfated glycosphingolipids and anionic glycans 32 . Presently, we are unable to address this issue in patient samples, as isozyme analysis by DEAE cellulose anion-exchange chromatography requires large amounts of biological material not available from either patient. However, it is important to consider that intraparenchymal infusion of two monocistronic AAV vectors encoding HEXA and HEXB results in a high degree of cotransduction in target structures 20,33 . In addition, our therapeutic experiments in GM2 cats treated with monocistronic AAVrh8 vectors (encoding feline HEXA and HEXB proteins) infused bilaterally in the thalamus and intracerebroventricularly showed clear evidence of functional enzyme in the spinal cord, which is only possible when the vectors are administered by the combined routes 22,34,35 . These observations are supported by our studies in GM1 gangliosidosis mice, where we showed that bilateral thalamic AAV gene delivery alone is insufficient to provide therapeutic levels of lysosomal β-galactosidase to the spinal cord or cerebellum 15 . Preclinical studies in Tay  deliver AAV to the cerebellum and spinal cord 36 . To improve the distribution to these structures, we developed an intrathecal catheterization method to safely deliver AAV to the cisterna magna and intrathecal space 37 . The combined evidence from AAV gene therapy experiments in various animal models of GM2 suggests that delivery of monocistronic AAVrh8 vectors into CSF is likely to result in cotransduction of cells 19,22,36 . Unfortunately, longitudinal data on GM2 levels in CSF of TSD children are not available, primarily because there is no clinical justification to do repeated lumbar punctures is the absence of a treatment. Storage of GM2 ganglioside in the brain of TSD patients has been extensively reported [38][39][40] . Infantile TSD and SD fetal brains have a GM2 level approximately fivefold higher than normal fetuses, and this level can reach ~200-fold higher than normal age-matched tissues at the time of death (3-5 years of age). Currently, there is no information in the literature suggesting that GM2 ganglioside levels in the brain or CSF of patients with infantile TSD plateau or decrease but suggests that its increase is exponential. The rapid disease progression could be construed as an indication of a relentless increase of GM2 levels, as could be reasonably expected of cells maintaining a regular catabolic rate leading to GM2 storage in a stable progressive manner. Therefore, we infer the modest increases or stable levels of GM2 ganglioside in the CSF of AAV-treated patients to be indicative of a therapeutic effect, albeit modest.
Regardless of the exact isozyme composition that underlies the increase in hexosaminidase activity in CSF, MRI of TSD-002 showed some degree of disease burden stabilization and increased myelination in several structures, in agreement with the attenuation of clinical disease progression. If sustained, this would represent a meaningful benefit when compared to siblings and natural course of infantile TSD. This impact occurred in TSD-002, even though this patient received half the dose of TSD-001, pointing to the value of early treatment. A retrospective study of infantile TSD showed that 98% of patients developed seizures with an onset at 16.8 ± 5.5 months of age that are often refractory to antiepileptic drugs 1 . TSD-001 has remained free of seizures at 4.5 years of age (24 months after therapy) while remaining on the same anticonvulsant therapy (Keppra) and dose since gene therapy started. By this age, TSD patients are either deceased or at end-stage disease, where they are refractory to antiepileptics 1 , suggesting a positive benefit of the gene therapy intervention. Seizures are uncontrollable in infantile patients at equivalent age, and infantile TSD is often fatal before 5 years of age. The outcome in TSD-001 is viewed very positively by the patient's parents, in spite of the fact that her status has stabilized at a very low functional level. In TSD-002, seizures were noted by 24 months of age (17 months after therapy) but are also controlled with antiepileptic therapy (Keppra). This is the first report of combined intrathalamic and intrathecal gene therapy in patients. Thalamic injection holds potential for treatment of other diseases, because axonal transport of AAV vectors and lysosomal enzymes enables global CNS distribution 21,22 . The thalamus is a high-risk target because of its critical functions as an integration and relay center of the cerebral cortex. Therefore, to minimize risk to the patient, the thalamic injection volume in TSD-002 (180 µl) was roughly equivalent to that safely injected in NHPs (150 µl; data not shown). The dose/volume used corresponds to ~1/10 of the dose projected to be most efficacious based on preclinical experiments. Because the thalamic volume is estimated at ~4,500 μl 41 and infusate spread was ~474 μl, a substantial increase in dose/volume will be needed to achieve maximal therapeutic benefit.
Animal studies suggest that a 10-fold higher thalamic injection volume is safe 42 ; however, risk of thalamic injury in patients drove the decision to inject a lower volume. Therapeutic effect could also be augmented by an increased intrathecal AAV dose, and both options are under consideration for future human clinical trials.
Because TSD is fatal, taking the risk associated with injecting larger volumes/doses is warranted and necessary to achieve transformative therapeutic outcomes.
In summary, expanded-access use of AAVrh8-HEXA/HEXB vectors was safe after intrathecal and bilateral thalamic delivery and accompanied by a modest increase in CSF HexA activity. Clinical findings in TSD-002, who was treated at a younger age, suggest a possible deviation from the natural history of TSD despite receiving a lower dose than indicated by preclinical experiments. This expanded-access trial provides the basis for future trials of AAVrh8-HEXA/HEXB vectors in patients with TSD and SD.

Online content
Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/ s41591-021-01664-4.