Double blinded placebo control study of Mesenchymal Intravenous Stromal cell Infusions in children with recessive dystrophic Epidermolysis Bullosa (MissionEB)

Background (condition): Epidermolysis Bullosa (EB) is a heterogeneous group of inherited skin disorders which affects 1 in 17,000 live births and around half a million worldwide. Recessive dystrophic EB (RDEB), a severe subtype, is caused by mutations in the type VII collagen gene leading to reduced or absent type VII collagen. This results in skin blistering following minor mechanical trauma, leading to chronic erosions and extensive scars and contractures. There is no effective treatment for RDEB and management is supportive. It has a significant physical, emotional and socio-economic impact for patients and their families. Best practice treatment involves a multidisciplinary healthcare team, with daily dressings taking up to 4 hours. The average annual cost per child has been estimated at £40,000 (Angelis, 2016). Background (proposed treatment): Mesenchymal stromal cells (MSCs) have been delivered intravenously to 10 children with RDEB in a clinical trial. Nine children went on to have two further infusions. The cells were found to improve wound quality, reduce skin itching and pain with no significant side effects. The beneficial effects lasted for 3-6 months. These cells were derived from bone marrow but there is evidence that MSCs derived from umbilical cord tissue (UC-MSCs) could potentially be more effective. Research question: Are repeated intravenous infusions of allogeneic (donor) UC-MSCs safe and can benefit children with RDEB? Aim: To assess if repeated infusions of MSCs are safe and can benefit children with severe RDEB. Objectives: To assess the: • efficacy of repeated MSCs in improving disease severity, quality of life and symptoms (e.g. pain and itch reduction) in children with RDEB. • safety of repeated infusions of MSCs in this group. • costs and consequences of treatment with UC-MSCs versus usual care • To explore patients and parents views in relation to treatment effectiveness and acceptability. Methods: This is a randomised, placebo controlled, double blinded, crossover trial with an internal phase 1 dose de-escalation trial in the first 3 months and a 12 month continued treatment follow-on open-label study following review of the data. The internal phase 1 dose de-escalation trial is for safety gatekeeping of the proposed dose, with the option of halving the dose if recommended by the data monitoring and ethics committee (DMEC). The open label non-randomised study will go ahead if the treatment is found to be effective during the randomised crossover trial. The trial will be conducted at two sites, Great Ormond Street Hospital (GOSH) and Birmingham Children’s Hospital (BCH), that both specialise in paediatric dermatology and are Nationally Commissioned centres for paediatric EB. Sample size: RDEB is a rare condition and the sample size of 36 (for the crossover trial) is based on feasibility including availability of the patients and not formal power considerations. As such, the statistical analysis focuses on estimation rather than hypothesis testing. Analysis: The study is not formally powered and estimation rather than formal hypothesis testing is the primary aim of the analysis although P values may be provided when frequentist statistical models are fitted. The primary outcome is disease severity as measured by EBDASI at 3 months for groups 1 and 12 months for groups 2. This outcome will be analysed using a linear mixed-effects model that will include treatment, sequence, period, and baseline (for each period) in the model with a random effect on the participant. The difference in means (mean difference) with 95% confidence intervals giving a range of plausible effects will be estimated using restricted maximum likelihood (REML) methods and Satterthwaite degrees of freedom. To aid interpretation and ability to make probabilistic statements about the distribution of the treatment effect, an equivalent Bayesian linear mixed-effects model will be fitted using non-inf