rAAV8.hRKp.AIPL1 is a recombinant AAV vector, comprising a human GRK1 promoter region driving the human AIPL1 coding sequence. In the absence of a clinical trial, we made this innovative experimental product available to children with confirmed mutations in AIPL1 under a Specials Licence with the approval of the Paediatric Bioethics Service at Great Ormond Street Hospital for Children (GOSH GMOSC #7).^17,18,19 This non-randomised, single-arm, open-label, first-in-human interventional study was done at Moorfields Eye Hospital (London, UK) and Great Ormond Street Hospital for Children (London, UK).

Children aged 1–3 years with congenital severe retinal dystrophy, biallelic disease-causing variants in AIPL1, and relative preservation of outer retinal structure at the central macula on OCT were considered for treatment. As the ethics approval was limited to four children, we offered treatment to the first four children who satisfied the eligibility criteria. If a difference in visual function between the patients’ eyes could be identified, the better-seeing eye was selected for treatment. The contralateral eye remained untreated for safety. The parents of each child provided fully informed written consent for treatment.

rAAV8.hRKp.AIPL1 was produced at University College London's Wolfson Gene Therapy Unit using three-plasmid transfection in HEK293T cells according to Good Manufacturing Practice guidelines, under a Manufacturer Specials Licence from the UK Medicines and Health products Regulatory Authority. MeiraGTx (holder of a Specials Licence) supported production, storage, quality assurance, and dispensing. 0·1–0·4 mL AAV2/8.hRKp.AIPL1 vector suspension was administered at a titre of 1 × 10¹¹ vector genomes per mL during vitrectomy surgery under general anaesthesia (appendix pp 1–2). The suspension was delivered subretinally to establish a bleb extending across the posterior pole of the retina from the superotemporal vascular arcade to the inferotemporal arcade, encompassing the surviving central macula. To protect against harm from inflammation, children were prescribed oral prednisolone at a dose of 1·0 mg/kg bodyweight daily for 5 days before surgery and 1 mg/kg for the first week after surgery, with tapering of the dose for a further 4 weeks (appendix p 1). We evaluated the children's functional vision, visual acuity, and retinal structure before intervention and twice subsequently. Functional vision was tested by observing and recording the children's visual behaviour and their ability to do simple vision-guided tasks. For example, the children were tasked with locating by sight white objects of a range of sizes in turn, against a dark background under normal office illumination, and with moving crayons between cups; they were also invited to draw on paper. When possible, treated and untreated eyes were tested independently in the same way. Children were also asked to mobilise along a normally lit corridor with the use of both eyes together and to identify doorways. Visual acuity tests were selected according to the age and ability of each child at the time of testing. We assessed visual acuity subjectively with standard age-appropriate recognition tasks (including Cardiff acuity cards, Kay pictures, and the Sonksen logMAR test using Sheridan Gardiner letters); when visual acuity could not be measured using these methods, we tested the children's ability to perceive and follow a pen torch light source at distances ranging from 5 cm to 50 cm. Near visual acuity was also measured in one child with single optotypes of the Sonksen logMAR test. logMAR equivalent values were obtained from the individual tests (eg, Cardiff acuity cards have a logMAR equivalent value, and the Kay picture test is developed using a logMAR scoring system). A logMAR score of 0·0 indicates perfect acuity; 1·3 indicates severe sight impairment; and 2·7 indicates perception of light only. Visual acuity was assessed objectively with a novel contrast sensitivity function (CSF) touchscreen test, the PopCSF test (appendix p 3; figure 1).^20,21 For the PopCSF test, 100% contrast targets were categorised on the basis of their spatial frequency: low (1·5–2·5 cycles per degree [cpd]), medium (2·5–3·5 cpd), and high (>3·5 cpd). Six additional children with AIPL1-associated severe retinal dystrophy (aged 2·9–3·9 years) were recruited and tested monocularly or binocularly with the popCSF test at Moorfields Eye Hospital, offering a benchmark for untreated performance. We evaluated visual signal detection in the visual cortex by recording cortical electrophysiological responses to full-screen black-and-white flickering stimuli and flickering (contrast-reversing) gratings across a range of spatial frequencies (steady-state visually evoked potential [ssVEP] technique; appendix pp 3–5; figure 1). The children's parents were also asked to monitor for signs of discomfort and any changes in the children's functional vision as they occurred; their observations were recorded in the clinical notes. Retinal imaging was done with handheld OCT and widefield fundus imaging (appendix p 5). Ocular examinations, including slit-lamp biomicroscopy and dilated fundoscopy, were done to identify adverse effects such as inflammation and retinal detachment. Safety was assessed by testing of visual acuity (as unexpected deterioration of visual acuity can indicate an adverse event or in itself be an adverse event), ophthalmoscopy, handheld OCT, and widefield fundus imaging.