Intracerebral Gene Therapy in Children with Mucopolysaccharidosis Type IIIB Syndrome: an Uncontrolled Phase 1/2 Clinical Trial

Articles

Intracerebral gene therapy in children with mucopolysaccharidosis type IIIB syndrome: an uncontrolled phase 1/2 clinical trial

Marc Tardieu, Michel Zérah, Marie-Lise Gougeon, Jérome Ausseil, Stéphanie de Bournonville, Béatrice Husson, Dimitrios Zafeiriou, Giancarlo Parenti, Philippe Bourget, Béatrice Poirier, Valérie Furlan, Cécile Artaud, Thomas Baugnon, Thomas Roujeau, Ronald G Crystal, Christian Meyer, Kumaran Deiva, Jean-Michel Heard

Summary Background Mucopolysaccharidosis type IIIB syndrome (also known as Sanfilippo type B syndrome) is a lysosomal Lancet Neurol 2017 storage disease resulting in progressive deterioration of cognitive acquisition after age 2–4 years. No treatment is Published Online available for the neurological manifestations of the disease. We sought to assess the safety and efficacy of a novel July 13, 2017 intracerebral gene therapy. http://dx.doi.org/10.1016/ S1474-4422(17)30169-2 See Online/Comment Methods Local regulatory authorities in France allowed inclusion of up to four children in this phase 1/2 study. http://dx.doi.org/10.1016/ Treatment was 16 intraparenchymal deposits (four in the cerebellum) of a recombinant adenoassociated S1474-4422(17)30200-4 viral vector serotype 2/5 (rAAV2/5) encoding human α-N-acetylglucosaminidase (NAGLU) plus immunosuppressive Paediatric Neurology therapy. We assessed tolerance, neurocognitive progression, brain growth, NAGLU enzymatic activity in CSF, and Department specific anti-NAGLU immune response for 30 months after surgery. This trial is registered with EudraCT, number (Prof M Tardieu MD, S de Bournonville MSc, 2012-000856-33, and the International Standard Clinical Trial Registry, number ISRCTN19853672. K Deiva MD), Paediatric Radiology Department Findings Of seven eligible children, the four youngest, from France (n=2), Italy (n=1), and Greece (n=1), (B Husson MD), and aged 20, 26, 30, and 53 months, were included between February, 2012, and February, 2014. 125 adverse events were Pharmacology Toxicology Department (V Furlan PharmD), recorded, of which 117 were treatment emergent and included six classified as severe, but no suspected unexpected Université Paris Sud and serious adverse drug reactions were seen. Vector genomes were detected in blood for 2 days after surgery. Compared Assistance Publique-Hôpitaux with the natural history of mucopolysaccharidosis type III syndromes, neurocognitive progression was improved in all de Paris, Hôpitaux patients, with the youngest patient having function close to that in healthy children. Decrease in developmental quotient Universitaires Paris Sud, Le Kremlin-Bicêtre, France; was –11·0 points in patient one, –23·0 in patient two, –29·0 in patient three, and –17·0 in patient four, compared with Paediatric Neurosurgery –37·7 in the natural history of the disease. NAGLU activity was detected in lumbar CSF and was 15–20% of that in Department (Prof M Zérah MD), unaffected children. Circulating T lymphocytes that proliferated and produced tumour necrosis factor α upon ex-vivo Clinical Pharmacy Department exposure to NAGLU antigens were detectable at 1–12 months and 3–12 months, respectively, but not at 30 months in (Prof P Bourget PharmD), and Anaesthesiology Department three of four patients. (T Baugnon MD), Université Paris Descartes and Assistance Interpretation Intracerebral rAVV2/5 was well tolerated and induced sustained enzyme production in the brain. The Publique-Hôpitaux de Paris, initial specific anti-NAGLU immune response that later subsided suggested acquired immunological tolerance. The Hôpital Universitaire Necker, Paris, France; Antiviral best results being obtained in the youngest patient implies a potential window of opportunity. Longer follow-up is Immunity, Biotherapy and needed to further assess safety outcomes and persistence of improved cognitive development. Vaccine unit, Infection and Epidemiology Department Funding Association Française Contre les Myopathies, Vaincre les Maladies Lysosomales, Institut Pasteur, and UniQure. (M-L Gougeon PhD, B Poirier PhD), Centre for Translational Science, Clinical Introduction recombinant enzyme in cerebral ventricles has not been Core (C Artaud MSc), and Mucopolysaccharidosis type IIIB syndrome (also known efficacious.7,8 Intracerebral gene therapy in rodents and Department of Neuroscience, as Sanfilippo syndrome type B) is a rare lysosomal storage dogs led to the release of therapeutic enzyme in resident Biotherapy and Neurodegenerative Diseases 9 disease that causes progressive deterioration of cognitive cells of the brain that had been transduced. Biochemical Unit, INSERM U1115 abilities after 2–4 years of apparently normal development, defects, neuropathology, and animal behaviour improved (J-M Heard MD), Institut and eventually leads to death.1–3 The disorder is caused by when treatment was given before patent tissue lesions Pasteur, Paris, France; 9–11 Laboratoire de Biochimie and the accumulation of partly degraded heparan sulphate developed, but not after. Lesions outside the CNS, INSERM U1088, Université de oligosaccharides in the absence of degradation by α-N- including in the meninges, choroid plexus, and brain Picardie-Jules Verne, Hôpitaux acetylglucosaminidase (NAGLU) due to its catalytic capillaries, persisted. In dogs, intracerebral gene therapy Universitaires d’Amiens, activity being impaired by mutations in NAGLU.4–6 elicited a specific immune response against NAGLU that Amiens, France (Prof J Ausseil PhD); Department Although recombinant enzyme peripheral infusion is was not present during education of the immune system of Paediatrics, Aristotle beneficial for patients with mucopolysaccharidosis type before treatment. Rejection of NAGLU-expressing cells University, Thessaloniki, Greece IIIB syndrome, the blood–brain barrier impairs delivery suggested the need for concomitant immunosuppression.11 (Prof D Zafeiriou MD); to the CNS and, thus, also treatment of the neurological Previous intracerebral gene therapy trials have been Department of Translational Medical Sciences, Frederico II manifestations of the disease. Direct infusion of done in human beings with aromatic L-aminoacid www.thelancet.com/neurology Published online July 13, 2017 http://dx.doi.org/10.1016/S1474-4422(17)30169-2 1 Articles

University, Naples, Italy (G Parenti MD); Telethon Research in context Institute of Genetics and Medicine, Pozzuoli, Italy Evidence before this study Added value of this study (G Parenti); Neuroscience Mucopolysaccharidosis type IIIB syndrome is an orphan disease This study is the first to report human intracerebral gene Department, Hôpitaux de for which no treatment strategy has been investigated in therapy for a lysosomal disease with extensive follow-up and Montpellier, Montpellier, human beings. In studies in mice and dogs, we saw beneficial investigations. We treated four young children and followed France (Prof T Roujeau MD); Department of Genetic effects with intracerebral gene therapy administered via a them up for 30 months. Brain cells showed sustained enzyme Medicine, Weill Cornell Medical recombinant adenoassociated virus 2/5 vector with production that led to improved cognitive outcomes in all College, New York, NY, USA concomitant immunosuppression. Toxicological studies in rats patients compared with the natural history of (R G Crystal MD); and UniQure, Amsterdam, Netherlands validated the findings and suggested that this vector could be mucopolysaccharidosis type III syndromes, with the outcomes (C Meyer MD) used in human beings . We did a retrospective epidemiological being best in the youngest patient. We saw an initial specific Correspondence to: study of patients with mucopolysaccharidosis type III immune response to the therapeutic enzyme that later Prof Marc Tardieu, Paediatric syndromes in France, the UK, and Greece to specify the natural subsided, which indicated acquired immunological tolerance Neurology Department, disease course. We identified reports by searching PubMed for for three patients. The treatment had a good safety profile Hôpitaux Universitaires Paris papers published up to February, 2017, using the terms “MPS within the follow-up period in all patients. Sud, Le Kremlin Bicêtre 94275, III”, “Sanfilippo syndromes”, “therapy for MPS”, “natural history France Implications of all the available evidence [email protected] of MPS”, and “cognitive evolution in MPS III”, with no language This phase 1/2 study in four patients suggests that larger and restrictions. We identified three prospective studies of the longer trials are feasible. Several factors should be considered for cognitive evolution of patients with mucopolysaccharidosis further assessment: treatment seemed to be most efficacious at type IIIA or type IIIB syndrome from which we derived an early age, before irreversible brain damage had occurred; quantitative indicators of cognitive evolution as reference efficacy of enzyme substitution might be improved if provided values against which to assess cognitive progress in our not only to parenchymal brain cells but also to meninges, brain patients. We found no trials of intracerebral gene therapy in capillaries, and choroid plexus, in which the disease was patients with mucopolysaccharidosis type IIIB syndrome. presumably not prevented by the current treatment protocol; Preliminary reports of enzyme infusion into CSF or intravenous and immunosuppression was beneficial, at least in the first adenoassociated virus 9 administration had been presented at few months after injection, to prevent overt immune reaction. scientific meetings.

decarboxylase deficiency, Canavan’s disease, Parkinson’s atrophy on MRI (maximum hemispheric corticodural disease, mucopolysaccharidosis type IIIA syndrome, or distance >0·6 cm) due to the increased risk of bleeding late infantile neuronal ceroid lipofuscinosis.12–17 Our during neurosurgery, those with loss of ability to walk findings for mucopolysaccharidosis type I and type IIIB independently, and those receiving drugs in the 6 months syndrome in mice and dogs indicated that the enzyme before treatment that might modify the natural course of release after intracerebral gene therapy administration the disease (sleep and mood regulators were accepted). was associated with improvements in biochemical defects, To recruit children, we informed parents’ associations, neuropathology, and abnormal behaviour,11,18 and paediatric neurologists, and metabolic disease specialists toxicological studies in rats suggested that this vector in France, the UK, Germany, Italy, Greece, and Spain could be used in human trials (unpublished). We, about the study. An international independent data therefore, postulated that intracerebral gene therapy monitoring committee assessed progress and safety data combined with immunosuppression would improve before each new child was enrolled. cognitive outcomes in children with mucopolysaccharidosis The study protocol was approved for inclusion of up to type IIIB syndrome. We sought to assess the safety of and four patients by Agence Nationale de Sécurité du tolerance to 16 deposits of a recombinant adenoassociated Médicament et des Produits de Santé and the ethics viral vector serotype 2/5 (rAAV2/5) encoding human committee of Comité de Protection des Personnes NAGLU in the brain and cerebellum combined Île-de-France II, and was endorsed by two parents’ with immunosuppressive therapy in four children with associations, Association Française Contre les Myopathies, mucopolysaccharidosis type IIIB syndrome. and Vaincre les Maladies Lysosomales. Families provided written informed consent. Methods Study design and patients Procedures This was an uncontrolled, phase 1/2 clinical study. Immunosuppression was started 14 days before surgery Inclusion criteria were age 18 months to 4 years (up to with 0·2 mg/kg oral tacrolimus per day and 1200 mg/m² the child’s fifth birthday), clinical manifestations related oral mycophenolate mofetil per day. A 4 h pharmacokinetic to mucopolysaccharidosis type IIIB syndrome, and profile was determined when steady state was achieved NAGLU activity in the blood less than 10% of that in and 7 days after both drugs were started. The dose unaffected children. We excluded children with brain schedule for mycophenolate mofetil was adapted to

2 www.thelancet.com/neurology Published online July 13, 2017 http://dx.doi.org/10.1016/S1474-4422(17)30169-2 Articles

achieve an area under the plasma concentration curve for evolutions,1,22–24 based on data from studies in France, mycophenolic acid (a product derived from myco the UK, Greece, and the USA. From these, develop phenolate mofetil) from 0 h to 12 h greater than mental age evolution and decline in developmental 30 h × mg/L and to maintain it for 6 weeks after surgery. quotient (the ratio of developmental age to chronological We did area under the plasma concentration curve age × 100) were quantified, which provided a reference analysis for tacrolimus to assess its metabolism in the value for decline of –37·7 points over 31 months (95% treated children, with the trough concentrations used as CI –30·5 to –45·2)24 that we used in this study. the target. Tacrolimus doses were adjusted to achieve Haematological and biochemical blood tests and viral trough concentrations in blood of 10–15 ng/mL up to serology tests for varicella zoster, Epstein-Barr virus, 3 months, 7–10 ng/mL from 3 months to 18 months, and and cytomegalovirus were done 1 month before and 5–8 ng/mL from 18 months to 30 months after surgery. 7 days, 1 month, and 3 months after surgery, and every Prednisolone was given to patients from 1 day before to 10 days after surgery at a dose of 1 mg/kg per day. The gene transfer vector, rAAV2/5, encoding human Patient one Patient two Patient three Patient four NAGLU, was produced by UniQure (Amsterdam, Sex Female Male Male Male Netherlands) in the Sf9 insect cell line, grown in serum- Characteristics at time of diagnosis free conditions in accordance with Good Manufacturing Age (months) 15 18 5 40 11 Practice. Surgery was performed as previously Percentage serum enzymatic 0 0 0 0 described,16,19–21 but modified for additional vector activity depositions to the cerebellum. Briefly, two coaxial silica Total urinary 84 69 83 213 glass injection capillaries, manufactured for the study glycosaminoglycan concentration* by PhotonLines (Saint-Germain-en-Laye, France) were Mutations introduced (one deeply and one superficially) through 1 214_237dup Tyr140Cys 214_237dup His414Arg needles inserted into eight burr holes drilled laterally 2 Trp540X Tyr140Cys Not determined Trp218Gly† from the midline, three into the hemispheric white matter and one into the cerebellum on each side. The Initial cognitive development Age at start of independent 17 15 18 13 16 targets were equidistant bilaterally (anterior, median, walking (months) and posterior) in the white matter of both cerebral and Age at start of 22 24 24 48 cerebellar hemispheres to offer the widest distribution communicating in two-word of the vector, to avoid side-effects by crossing a sulcus or combinations (months)‡ a cortical vein or artery, and to allow superficial and deep Ability lost No No No No targets with the same trajectory. Each deposit was 60 µL Age at onset of cognitive No 18 24 18 and contained 2·4 × 10¹¹ vector genomes (vg), and all delay (months)§ were delivered simultaneously over a period of 2 h at a Abnormal behaviour No No No No rate of 0·5 µL/min, giving a total dose of 4·0 × 10¹² vg. Epilepsy No No No No Characteristics at time of inclusion Assessments Age (months) 20 26 30 53 We assessed viral shedding in blood and urine after the Head circumference (cm) 50·5 55·0 53·5 53·0 procedure with real-time quantitative PCR amplification (75th percentile) (90th percentile) (75th percentile) (70th percentile) of DNA (Genosafe, Evry, France; limit of quantification Coarse features Yes Yes Yes Yes 50 vg/µg DNA; amplification of the vg region corre Hepatomegaly Yes Yes Yes Yes sponding to the bovine growth hormone polyadenation Splenomegaly No No No No sequence). Blood samples were collected at 3 min, 1 h, Autistic spectrum disorder No No No No 12 h, 36 h, 60 h, and 6 days, and urine samples were Brain MRI collected at 12 h, 36 h, and every 24 h thereafter until Corpus callosum area 386 344 387 218 values were less than 2·2 × 10³ vg/mL in two successive (mm²) assessments. Bi-insular diameter (mm) 107 119 116 120 Neurological, behavioural, and developmental features Ventricle morphology Thin Slightly enlarged Slightly enlarged Slightly enlarged were assessed 1 month before and 12 and 30 months Myelin maturation¶ Not available|| Delayed Delayed Delayed after surgery by one trained neuropsychologist (SdB) Heparan sulphate 2490 3731 3656 1928 concentration in CSF with the Brunet-Lezine revised test (an equivalent of the (ng/mL)** American Bayley test), the Vineland Adaptive Behaviour Scale, and the Psychoeducational Profile, third edition.16 *Expressed as mg/mmol of creatinine (concentrations are <28 mg/mmol in unaffected children). †Previously unreported. ‡Between ages 13 and 27 months in unaffected children. §Exhaustive data for cognitive assessments are We compared the disease course for mucopoly provided in the appendix. ¶Myelin maturation assessed visually on fluid attenuation inversion recovery or T2-weighted saccharidosis type IIIB syndrome in treated patients MRI scans. ||Patient too young for assessment. **Concentrations <200 ng/mL in unaffected children. with the natural history of mucopolysaccharidosis Table 1: Characteristics of patients at diagnosis and inclusion in the study type IIIA and IIIB syndromes, which have similar www.thelancet.com/neurology Published online July 13, 2017 http://dx.doi.org/10.1016/S1474-4422(17)30169-2 3 Articles

3 min 1 h 12 h 24 h 36 h 42 or 48 h 60 h 6 days 8 days Concentration in blood (vg/µg)* Patient 1 2·0 × 10³ 3·3 × 10³ 8·6 × 10² 3·3 × 10² NM 2·2 × 10² NM NM <15 Patient 2 <15 <15 41 50 NM <15 NM NM NM Patient 3 8·4 × 10² 1·7 × 10³ 1·8 × 10³ 6·3 × 10² NM 7·1 × 10² NM <15 <15 Patient 4 58 4·5 × 10² 1·3 × 10³ 4·5 × 10² NM 1·4 × 10² NM NM <15 Concentration in urine (vg/mL)† Patient 1 NM NM 9·0 × 10³ 8·1 × 10² NM NM 2·2 × 10³ NM NM Patient 2 NM NM 3·8 × 10⁴ 2·2 × 10³ NM NM 7·4 × 10³ <6·4 × 10² <6·4 × 10² Patient 3 NM NM 2·9 × 10³ 3·6 × 10³ NM NM 39 <6·4 × 10² NM Patient 4 NM NM 1·3 × 10³ 2·2 × 10² NM NM 3·1 × 10³ NM <6·4 × 10²

NM=not measured. vg=vector genomes.*Limit of quantification 50 vg/µg DNA; limit of detection 15 vg/µg DNA. †Limit of quantification 2·2 × 10³ vg/mL; limit of detection 6·4 × 10² vg/mL.

Table 2: Viral shedding after intracranial injection

patients with mucopolysaccharidosis type IIIB syndrome Inclusion 3 weeks Follow-up to to surgery* after surgery* 30 months† compared with the normal value in unaffected children of 4·3 nmol h–¹ mL–¹ [0·1]). Heparan sulphate concentrations Upper respiratory tract 5 0 65 26 infection‡ in CSF were measured as previously described. Minor anaesthesia-related 0 8 0 Measurements were made in CSF samples from enrolled Diarrhoea or 1 2 20 patients and compared with mean values in CSF or plasma gastroenteritis samples obtained in the same hospitals from unaffected Minor domestic accident 1 0 0 children with suspected acute meningitis but no CSF Pain 1 0 0 abnormalities. Scar 0 1 0 We assessed T-cell responses against NAGLU by High alanine 0 2 2 measuring the proliferation of freshly isolated CD4 aminotransferase and CD8 T lymphocytes labelled with 2 µg/mL carboxy concentrations fluorescein succinimidyl ester (R&D Systems, Lille, Behaviour 0 0 4 France). Measurements were made 4 days after ex-vivo Other biological 0 0 5 exposure to recombinant human NAGLU in the presence abnormality§ of antibodies against CD28 and CD49d. Intracellular Miscellaneous¶ 0 0 8 production of tumour necrosis factor α (TNFα) after *All events were mild. †Events were mild to moderate. ‡Upper respiratory tract NAGLU exposure was detected by intracellular staining infection, fever, or acute viral infection. §Anaemia, transient increase in alkaline with BD Horizon TNFα-V450 monoclonal antibody phosphate concentration, or transient hydroelectrolytic disorder (appendix). (BD Biosciences, Rungis, France).27 This trial is registered ¶Tonsillectomy, adenoidectomy, minimum mitral insufficiency, or transient loss of appetite. with EudraCT, number 2012-000856-33, and the International Standard Clinical Trial Registry, number Table 3: Adverse events ISRCTN19853672.

3 months thereafter. Protein and glucose concentrations Role of the funding source and cell counts were measured in CSF samples The funder of the study had no role in the study design, collected 1 month before and 1, 3, 12, and 30 months data collection, data analysis, data interpretation, or the after surgery. Other tests, such as ultrasonography, writing of the report. The funder approved the echocardiography, and abdominal echography, were submission for publication. The corresponding author done at 12 and 30 months. had full access to all the data in the study and had final NAGLU catalytic activity was measured in CSF samples responsibility for the decision to submit for publication. concentrated six times and in non-concentrated plasma samples. 25 µL 4-methylumbelliferyl-2-acetamido- Results 2-deoxy-α-D-glycopyranoside, fluorescent at 4 mm, was Of seven eligible children, the four youngest were added to 50 µL concentrated CSF and incubated for 18 h at enrolled between February, 2012, and February, 2014, 37°C before we measured optic deviation.25 As a control of who originated from and were living in France (n=2), sample quality, we simultaneously measured the activity Italy (n=1), or Greece (n=1). The patients’ characteristics of α-L-iduronidase, which is the deficient enzyme in at diagnosis and study enrolment are provided in table 1 See Online for appendix mucopolysaccharidosis type I syndrome and was found to and the appendix. Patient one was aged 20 months at be normal (mean activity 4·2 nmol h–¹ mL–¹ [SD 0·3] in study enrolment and had normal neurocognitive abilities

4 www.thelancet.com/neurology Published online July 13, 2017 http://dx.doi.org/10.1016/S1474-4422(17)30169-2 Articles

and normal brain MRI. Patient two, aged 26 months, and A B patient three, aged 30 months, had mild neurocognitive disability and no sign of cerebral or cerebellar atrophy, but had slightly enlarged ventricles, enlarged perivascular spaces, and delayed myelin maturation. Patient four was aged 53 months and had the most severe neurocognitive disability. Brain MRI findings were similar to those for patients two and three except that the corpus callosum area was smaller. Biological tests on blood, CSF, and urine were normal at enrolment for all patients, except for the absence of NAGLU activity in serum, high concentrations of glycosaminoglycans in urine, and slightly raised concentrations of alanine amino transferase in serum in patients two and four. Cardiac ultrasonography and electrocardiograms were normal in all patients. Abdominal echography showed enlarged liver, normal spleen, and normal kidneys in all patients. The total dose of 4·0 × 10¹² vg rAAV2/5 was deposited C D in all patients. No adverse events occurred during neurosurgery. Vector was detected in blood at the end of the injection period with the greatest concentrations generally seen in the first 12 h after surgery table( 2). We detected vg in urine in all patients for the first 3 days. We saw no interferon α activity in serum samples collected 1 day after surgery. Clinical examination findings remained stable over the 30 months of follow-up, and cardiac scores and liver, spleen, and kidney sizes measured with ultrasonography did not change. 125 adverse events were recorded, of which 117 were treatment emergent (table 3), including six classified as severe (raised alanine aminotransferase concentration [n=1], respiratory tract infection during follow-up [n=3], and diarrhoea during follow-up [n=2]). 54 events were recorded within 12 months of surgery, and Figure 1: Axial T2-weighted brain MRI scans 63 in the following 18 months. No suspected unexpected Patient one, the youngest patient, had normal ventricular size at inclusion in the study, aged 20 months (A), serious adverse drug reactions were reported. Intraparen and at the end of of 30 months of follow-up when aged 51 months (B). In patient four (the oldest patient), chymal tracks were not visible in any patients, and ventricular size at inclusion in the study at age 53 months (C) was slightly larger than that in patient one at the haemosiderin deposits on MRI were minimal after end of follow-up, but no change was seen after treatment to the end of 30 months of follow-up (D). 3 months, and no signs of inflammation, oedema, or necrosis were detected up to 30 months after treatment General and neurological assessments showed (figure 1). Biological testing of serum and CSF samples improvement in everyday activities and behaviour over showed no substantial abnormalities, except for transiently the follow-up period. Cognitive developmental age was increased concentrations of alanine aminotransferase and consistent across the three neuropsychological tests aspartate aminotransferase in patient two 1 month after within each patient (figure 2, appendix). Patient one, surgery (appendix). who was the youngest at inclusion, acquired skills more 7 days after immunosuppressive therapy was started, rapidly during follow-up than in the natural disease the mycophenolic acid area under the curve was course, and almost as rapidly as unaffected children 36 h × mg/L for patient one, 54 h × mg/L for patient two, (figure 2); decrease in developmental quotient was 52 h × mg/L for patient three, and 65 h × mg/L for patient small (–11·0 vs –37·7 points for natural history of four, and for tacrolimus was 140, 203, 149, and disease); and brain MRI remained normal with normal 272 h × ng/mL, respectively. Mean trough tacrolimus age-dependent increase of corpus callosum surface concentrations in blood were 10·8 ng/mL (SD 3·7) up to (84 mm²; final area 470 mm²) and thin ventricles at 3 months, 7·7 ng/mL (2·2) from 3 months to 18 months, 30 months after surgery (figure 1). Patients two and 6·8 ng/mL (1·5) from 18 months to 30 months after and three also acquired skills more rapidly and had surgery. Patient two showed seroconversion for varicella less decrease in developmental quotient (–23·0 and zoster virus and patient three for Epstein-Barr virus –29·0 points, respectively) than expected in the natural during follow-up. disease evolution. Ventricle sizes and bi-insular www.thelancet.com/neurology Published online July 13, 2017 http://dx.doi.org/10.1016/S1474-4422(17)30169-2 5 Articles

A B 45 Brunet-Lezine test Vineland test 40 PEP-3 test 35 Unaﬀected children 30 MPS3A children 25 20 15 10 5

Cognitive developmental age (months) Cognitive 0

C D 45 40 35 30 25 20 15 10 5

Cognitive developmental age (months) Cognitive 0 0310 20 0540 0860 70 0 90 0310 20 0540 0860 70 0 90 Chronological age (months) Chronological age (months)

Figure 2: Trajectories of neurocognitive development by age during follow-up (A) Patient one. (B) Patient two. (C) Patient three. (D) Patient four. Cognitive developmental age was measured by three neuropsychological tests (Brunet-Lezine revised test, Vineland, and PEP-3). Cognitive growth trajectory of developmental age in the children in our study was compared with that in unaffected children and with the mean (range) expected trajectory for those with rapidly progressing MPS3A.24 Arrows indicate time of treatment. PEP-3=Psychoeducational Profile, third edition. Vineland=Vineland Adaptive Behaviour Scale. MPS3A=mucopolysaccharidosis type IIIA syndrome.

follow-up, but slower than in the natural disease course Unaffected Patient one Patient two Patient three Patient four children* and with less decline in developmental quotient (–17·0 points). Brain MRI results remained stable, Activity in CSF (nmol h–1 mL–1) including corpus callosum area. Baseline 57 (12)

Values are means (SEM) of at least two independent technical replicates. LOQ=limit of quantification in CSF and Neutralising antibodies against adenoassociated viral plasma (5 nmol h–1 mL–1). NAGLU=α-N-acetylglucosaminidase. *Mean values in concentrated CSF (n=4) or vector serotype 5 (AAV5) were not detected in serum plasma (n=5). †Significantly different from value at 1 month (p<0·05). samples collected at inclusion or during follow-up. We Table 4: NAGLU catalytic activity in concentrated CSF and plasma saw no proliferation of CD4 or CD8 T lymphocytes upon ex-vivo exposure to NAGLU at inclusion. They were detectable from 1 to 12 months after surgery, then diameters on brain MRI remained unchanged. Corpus disappeared or persisted at very low concentrations up to callosum surface increased in patient two by 49 mm² 30 months (figure 3). T lymphocytes producing TNFα (final area 393 mm²), and decreased in patient three upon exposure to NAGLU became detectable at 3 months by 14 mm² (final area 373 mm²). In patient four, the and persisted at 12 months, but were not detected at oldest at inclusion, neurocognitive skills declined over 30 months.

6 www.thelancet.com/neurology Published online July 13, 2017 http://dx.doi.org/10.1016/S1474-4422(17)30169-2 Articles

A B 20 Proliferating CD4 T cells 5 Percentage NAGLU-speciﬁc TNFα-producing T cells (%) Proliferating CD8 T cells CD4 T cells producing TNFα CD8 T cells producing TNFα 4 15

5 1 Percentage NAGLU-speciﬁc proliferating T cells (%) proliferating Percentage NAGLU-speciﬁc 0 0

C D

20 5 Percentage NAGLU-speciﬁc TNFα-producing T cells (%)

4 15

5 1 Percentage NAGLU-speciﬁc proliferating T cells (%) proliferating Percentage NAGLU-speciﬁc 0 0 01 336 12 0 01 336 12 0 Time after surgery (months) Time after surgery (months)

Figure 3: Ex-vivo response of circulating T lymphocytes after exposure to NAGLU Proliferation and TNFα production of lymphocytes were measured at baseline (indicated as 0) and 1, 3, 6, 12, and 30 months after treatment. NAGLU=α-N-acetylglucosaminidase. TNFα=tumour necrosis factor α.

Discussion additional sites in the cerebellum, which led to no Clinical and brain imaging data showed that adverse events related to the surgical procedure. rAAV2/5-mediated intracranial gene therapy was well The rationale for including immunosuppression was tolerated in four children with mucopolysaccharidosis also based on our previous findings in dogs,11 as well as type IIIB syndrome over an initial follow-up period of identification of NAGLU mutations associated with 30 months. MRI did not detect oedema, inflammation, inefficient synthesis of tolerance-inducing proteins in the or signs of local necrosis at vector delivery sites, which is four patients (table 1). After treatment, we found emergent consistent with our findings in 25 dogs treated with a subsets of memory and effector T lymphocytes sensitised similar protocol and the same material.11 Good tolerance to NAGLU. Ex vivo, these cells proliferated and produced to the delivery vector in the brain had been previously type 1 cytokines when exposed to NAGLU antigenic reported in human beings.12–17 epitopes. Detection of vg in circulating blood for 2 days Our studies in dogs documented AAV5 vector delivery and transient detection of NAGLU in plasma suggest that in brain tissue in the context of a disorder of heparan particles deposited in brain tissue transited into the sulphate degradation, which is relevant because heparan systemic circulation. Contact with peripheral dendritic sulphate molecules are involved in the penetration of cells or tissue macrophages during this period might AAV5 particles into cells.28 The vector was delivered to account for antigenic presentation of NAGLU epitopes to eight sites in the hemispheric white matter, after which lymphocytes. In the absence of immunosuppressive the vector and enzyme spread efficiently within the therapy, permanent stimulation by epitopes present on hemispheres, but neither was seen in the cerebellum, the therapeutic enzyme would be likely and would trigger which remained severely affected.11,18 In the children in cell proliferation when lymphocytes reactive to NAGLU this study, therefore, we delivered the vector to four were produced. Although there is no clinical evidence that www.thelancet.com/neurology Published online July 13, 2017 http://dx.doi.org/10.1016/S1474-4422(17)30169-2 7 Articles

this reaction would have harmful consequences, we quotient were modest, they were better than those in the felt that prevention of potential risks was justified. Reactive natural history of mucopolysaccharidosis type III lymphocytes were not detected 30 months after surgery, syndromes. MRI scans suggested that brain growth had which suggests acquired immunological tolerance to stabilised without atrophy. Patient four, the oldest at NAGLU. These findings support our view that inclusion, showed a decline after surgery, but disease immunosuppression is important for the safety and progression was less severe than expected from the efficacy of treatment, at least initially. natural history. Detection of NAGLU activity in CSF and sustained These results indicate cognitive benefits with production of the enzyme by brain cells indicates the intracerebral gene therapy despite differences between the efficiency of vector delivery and gene transfer to the brain. children treated. The differences in outcomes seem likely We had found previously that NAGLU activity in the CSF to be related to age and, therefore, the severity of existing of treated dogs with mucopolysaccharidosis type IIIB CNS lesions at the time of surgery. A similar pattern was syndrome was about 20–30% of that in unaffected seen in dogs with mucopolysaccharidosis type IIIB controls and was associated with the efficient clearance of syndrome.18 Differences being related to sex or genotype, lysosomal storage lesions throughout brain parenchyma however, cannot be excluded.11 Additionally, irreversible neurons, astrocytes, and microglia. Nevertheless, severe lesions at the time of treatment might not be the sole cause storage lesions persisted in meninges, choroid plexus, of incomplete recovery, because disease progression in and capillary vessels. Heparan sulphate concentrations tissues as important as meninges, brain capillary walls, were consistently low in brain parenchyma tissue extracts and choroid plexus was presumably not stopped. and high in CSF.11 If a similar situation were to be found rAAV2/5-mediated intracerebral gene therapy was well in treated children as in dogs, despite high heparan tolerated in young children within the studied follow-up sulphate levels in CSF, clearance of storage lesions in period. These results suggest that this approach could large parts of the brain parenchyma might be assumed. prevent or slow cognitive decline in children with Neurocognitive progression is the primary endpoint of mucopolysaccharidosis type IIIB syndrome, and possibly studies assessing clinical efficacy of potential treatments in other related conditions. The small number of enrolled for children with mucopolysaccharidosis type IIIB patients and the limited duration of follow-up emphasise syndrome. Various features that might have been the need for additional studies. Longer follow-up is endpoints decline substantially on brain MRI in children needed to determine long-term safety outcomes and to with mucopolysaccharidosis type IIIB syndrome, but assess whether improved cognitive development persists volumetric analysis and other quantitative assessments or whether the disease reverts to its natural course over cannot be accurately measured in young patients because time. This study, therefore, constitutes an essential first there is little contrast between grey and white matter due step towards the design of future protocols aiming to to early changes in grey-matter water content and lack of obtain full reversal of mucopolysaccharidosis type IIIB myelination. Neurocognitive scores obtained in the four syndrome through the treatment of patients younger patients at inclusion and 12 and 30 months after surgery than 2 years, which seems to be the best window of were compared, according to age, with reference values opportunity, and by the delivery of NAGLU both within for the natural history of the disease derived from a and outside the brain. quantitative assessment of cognitive development Contributors trajectory in young children with early onset mucopoly MT, RGC, KD, and J-MH designed the protocols, followed up the saccharidosis type IIIA syndrome (rapid progressors) and patients, and wrote the paper. MZ, TB, and TR were responsible for neurosurgery, M-LG and BP for immunomonitoring, JA for enzymatic evidence of a similar evolution in children with muco studies, SdB for neuropsychological tests, BH for radiological 22–24 polysaccharidosis type IIIB syndrome. We assume that assessments, PB and VF for administering and determining the doses of for the four patients enrolled in this study, the prognosis immunosuppressants, DZ and GP for clinical care in Greece and Italy, with respect to early disease onset, high urinary heparan and CA and CM for clinical coordination on behalf of the study sponsor. All authors read and edited the drafts of the paper. sulphate excretion, and NAGLU mutations associated with severe phenotype would have been poor.2,3,29,30 Indeed, Declaration of interests MT and SdB have received consulting fees from UniQure. CM is an patient one, the youngest at inclusion, had a sibling with employee of UniQure. The other authors declare no competing interests. the same genotype and a highly progressive form of Acknowledgments mucopolysaccharidosis type IIIB syndrome. Nevertheless, Institut Pasteur sponsored the study during the first 30 months and after surgery patient one showed clear improvement in UniQure sponsored it for the last 9 months. The study was supported cognitive scores, almost reaching those of unaffected by the Association Française Contre les Myopathies, Vaincre les children, and brain growth on sequential MRIs was Maladies Lysosomales, Institut Pasteur, the Conny Maeva Fondation, and a gift from the Akbaraly family. We thank N Matthew Ellinwood normal for age during follow-up. The latter finding is and collaborators at the Department of animal science, Iowa State uncommon in children with mucopolysaccharidosis University, Ames, IA, USA, for help in the animal studies; type IIIB syndrome, even if quantified data on natural Cathy Gomila for technical assistance in the assessment of lysosomal history are unavailable. In patients two and three, although enzyme activity; the Laboratory of Genetic Metabolic Diseases, Academic Medical Centre, Amsterdam, Netherlands, for doing the cognitive progression and decline in developmental

8 www.thelancet.com/neurology Published online July 13, 2017 http://dx.doi.org/10.1016/S1474-4422(17)30169-2 Articles

heparan sulphate assays; the Clinical Core of the Centre for 17 Worgall S, Sondhi D, Hackett NR, et al. Treatment of late infantile Translational Science, Institut Pasteur, Paris, France; the members of neuronal ceroid lipofuscinosis by CNS administration of a the independent data monitoring committee; Cécile Delval and Hélène serotype 2 adeno-associated virus expressing CLN2 cDNA. Lafolly for help at the start of the study; and the patients’ families and Hum Gene Ther 2008; 19: 463–74. the nurses for granting permission for and helping us to treat the 18 Ellinwood NM, Wang P, Skeen T, et al. A model of enrolled patients. mucopolysaccharidosis IIIB (Sanfilippo syndrome type IIIB): N-acetyl-alpha-D-glucosaminidase deficiency in Schipperke dogs. References J Inherit Metab Dis 2003; 26: 489–504. 1 Heron B, Mikaeloff Y, Froissart R, et al. Incidence and natural history 19 Crystal RG, Sondhi D, Hackett NR, et al. Clinical protocol. of mucopolysaccharidosis type III in France and comparison with Administration of a replication-deficient adeno-associated virus United Kingdom and Greece. Am J Med Genet A 2011; 155A: 58–68. gene transfer vector expressing the human CLN2 cDNA to the brain 2 Valstar MJ, Bruggenwirth HT, Olmer R, et al. of children with late infantile neuronal ceroid lipofuscinosis. Mucopolysaccharidosis type IIIB may predominantly present with Hum Gene Ther 2004; 15: 1131–54. an attenuated clinical phenotype. J Inherit Metab Dis 2010; 20 Souweidane MM, Fraser JF, Arkin LM, et al. Gene therapy for late 33: 759–67. infantile neuronal ceroid lipofuscinosis: neurosurgical 3 Weber B, Guo XH, Kleijer WJ, van de Kamp JJ, Poorthuis BJ, considerations. J Neurosurg Pediatr 2010; 6: 115–22. Hopwood JJ. Sanfilippo type B syndrome 21 Zerah M, Piguet F, Colle MA, et al. Intracerebral gene therapy (mucopolysaccharidosis III B): allelic heterogeneity corresponds to using AAVrh.10-hARSA recombinant vector to treat patients with the wide spectrum of clinical phenotypes. Eur J Hum Genet 1999; early-onset forms of metachromatic leukodystrophy: preclinical 7: 34–44. feasibility and safety assessments in nonhuman primates. 4 Beesley CE, Jackson M, Young EP, Vellodi A, Winchester BG. Hum Gene Ther Clin Dev 2015; 26: 113–24. Molecular defects in Sanfilippo syndrome type B 22 Shapiro E, King K, Ahmed A, et al. The neurobehavioral phenotype (mucopolysaccharidosis IIIB). J Inherit Metab Dis 2005; 28: 759–67. in mucopolysaccharidosis type IIIB: an exploratory study. 5 Pollard LM, Jones JR, Wood TC. Molecular characterization of Mol Genet Metab Rep 2016; 6: 41–7. 355 mucopolysaccharidosis patients reveals 104 novel mutations. 23 Truxal KV, Fu H, McCarty DM, et al. A prospective one-year natural J Inherit Metab Dis 2013; 36: 179–87. history study of mucopolysaccharidosis types IIIA and IIIB: 6 Zhao HG, Li HH, Bach G, Schmidtchen A, Neufeld EF. Implications for clinical trial design. Mol Genet Metab 2016; The molecular basis of Sanfilippo syndrome type B. 119: 239–48. Proc Natl Acad Sci USA 1996; 93: 6101–05. 24 Shapiro EG, Nestrasil I, Delaney KA, et al. A prospective natural 7 Muenzer J. Overview of the mucopolysaccharidoses. history study of mucopolysaccharidosis type IIIA. J Pediatr 2016; Rheumatology (Oxford) 2011; 50 (suppl 5): 4–12. 170: 278–87.e1–4. 8 Auclair D, Finnie J, Walkley SU, et al. Intrathecal recombinant 25 Marsh J, Fensom AH. 4-Methylumbelliferyl human 4-sulfatase reduces accumulation of glycosaminoglycans in α-N-acetylglucosaminidase activity for diagnosis of Sanfilippo B dura of mucopolysaccharidosis VI cats. Pediatr Res 2012; 71: 39–45. disease. Clin Genet 1985; 27: 258–62. 9 Cheng SH. Gene therapy for the neurological manifestations in 26 Langereis EJ, Wagemans T, Kulik W, et al. A Multiplex assay for the lysosomal storage disorders. J Lipid Res 2014; 55: 1827–38. diagnosis of mucopolysaccharidoses and mucolipidoses. PLoS One 10 Cressant A, Desmaris N, Verot L, et al. Improved behavior and 2015; 10: e0138622. neuropathology in the mouse model of Sanfilippo type IIIB disease 27 Loison E, Poirier-Beaudouin B, Seffer V, et al. Suppression by after adeno-associated virus-mediated gene transfer in the striatum. thimerosal of ex-vivo CD4+ T cell response to influenza vaccine and J Neurosci 2004; 24: 10229–39. induction of apoptosis in primary memory T cells. PLoS One 2014; 11 Ellinwood NM, Ausseil J, Desmaris N, et al. Safe, efficient, and 9: e92705. reproducible gene therapy of the brain in the dog models of 28 Summerford C, Samulski RJ. Membrane-associated heparan sulfate Sanfilippo and Hurler syndromes. Mol Ther 2011; 19: 251–59. proteoglycan is a receptor for adeno-associated virus type 2 virions. 12 Hwu WL, Muramatsu S, Tseng SH, et al. Gene therapy for aromatic J Virol 1998; 72: 1438–45. L-amino acid decarboxylase deficiency. Sci Transl Med 2012; 4: 134ra61. 29 Selmer KK, Gilfillan GD, Stromme P, et al. A mild form of 13 Kotterman MA, Schaffer DV. Engineering adeno-associated viruses Mucopolysaccharidosis IIIB diagnosed with targeted next-generation for clinical gene therapy. Nat Rev Genet 2014; 15: 445–51. sequencing of linked genomic regions. Eur J Hum Genet 2012; 14 Leone P, Shera D, McPhee SW, et al. Long-term follow-up after 20: 58–63. gene therapy for canavan disease. Sci Transl Med 2012; 4: 165ra3. 30 Yogalingam G, Hopwood JJ. Molecular genetics of 15 LeWitt PA, Rezai AR, Leehey MA, et al. AAV2-GAD gene therapy mucopolysaccharidosis type IIIA and IIIB: diagnostic, clinical, for advanced Parkinson’s disease: a double-blind, sham-surgery and biological implications. Hum Mutat 2001; 18: 264–81. controlled, randomised trial. Lancet Neurol 2011; 10: 309–19. 16 Tardieu M, Zerah M, Husson B, et al. Intracerebral administration of adeno-associated viral vector serotype rh.10 carrying human SGSH and SUMF1 cDNAs in children with mucopolysaccharidosis type IIIA disease: results of a phase I/II trial. Hum Gene Ther 2014; 25: 506–16.

www.thelancet.com/neurology Published online July 13, 2017 http://dx.doi.org/10.1016/S1474-4422(17)30169-2 9