The project proposed is an open-label, randomised controlled, multicentre clinical trial of children aged 3 months to 2 years of age with epilepsy who have failed to respond to two or more pharmacological treatments (AED or corticosteroids), comparing KD to further AED treatment. The study will be conducted in two phases: first, we will carry out a pilot phase in two centres. If the pilot study recruits successfully, we will proceed to the full trial in a further seven centres. See Additional file 1 for a schematic of the trial design.
Eligible children will be consented via their parents. They will undergo baseline assessment including medical and seizure history, neurological and anthropometric examination, administration of quality of life (Infant and Toddler Quality of Life Questionnaire, [33]) and developmental (Vineland Adaptive Behaviour Questionnaire, [34]) questionnaires and biochemical investigations. They will then start a 2-week observation period with documentation of seizure frequency. If the child is prone to particularly frequent seizures in excess of two per day, then a minimum baseline period of 1 week would be considered sufficient. Food diaries required for diet calculation will be returned by post from all enrolled infants 1 week (or sooner for those with an unstable clinical condition) into the observation period. Standardised seizure records will be kept during the observation period and throughout the trial. Randomisation will be conducted using an Internet randomisation system provided by Sealed Envelope™ (Sealed Envelope Ltd.). Randomisation will occur on day 8 or day 15 for participants to receive the KD or a further AED; the allocated treatment will commence following randomisation, with instruction and training.
The randomisation schedule will be independently generated and held by Sealed Envelope™. Allocations will be released by email to the coordinating centres once the investigator or research nurse has entered eligible participant information into the web-based randomisation service. Participants will be allocated to either the KD or further AED arm using a simple, concealed, randomisation method. Randomisation will aim to achieve 92 in the KD group versus 68 participants in the further AED arm (see calculation in ‘Sample size’ section). An Enrollment Log will be maintained to keep records of the screened and randomised patients at each site. Withdrawn patients will not be replaced and replacement numbers will not be issued. Whilst it will not be possible to blind participants to their treatment allocation, efforts will be made to minimise expectation bias by emphasising in the trial literature that the evidence supporting the KD for seizure control is currently limited.
A second assessment (4 weeks after start of treatment) will include clinical review and a tolerability questionnaire (modified Hague Scale of Side Effects – see Additional file 2). Assessments will be repeated at 8 weeks after the start of treatment including clinical review, administration of the Infant and Toddler Quality of Life Questionnaire, the tolerability questionnaire, and biochemical investigations. After the 8-week assessment, according to patient’s clinical response to treatment with regards seizure outcome and tolerability, the KD (diet group) or AED (standard AED group) will then be continued or changed. Those in the AED group who have failed to achieve seizure control at the 8-week assessment will then be offered the KD outside the context of the trial. Those on the KD who have failed to achieve seizure improvement at the 8-week assessment will continue with medical management, as per clinician decision. All participants will be followed up for 12 months following randomisation for retention, seizure outcome and neurodevelopmental status. See Fig. 1 (Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) figure: schedule of enrollment, interventions and assessments) and Additional file 3 for the SPIRIT 2013 Checklist (recommended items to address in a clinical trial protocol and related documents).
At each of the centres, the paediatric neurologist will work with a dietitian for KD implementation. Treatment will be in accordance with the KD Intervention Manual, which will be agreed with reference to a standard text [35] and discussed with the project management team at the outset during the initial workshop to enable standardisation of treatment between different centres. There will be an AED consensus flowchart (Additional file 4) for guidance regarding management of the participant’s epilepsy, written following an initial workshop with paediatric neurologists from all nine centres; this flowchart will be used to create a standardised manual (AED Consensus Document).
Selection of participants
Children aged 3 months to 2 years with an established diagnosis of epilepsy, who continue to have epileptic seizures despite treatment with two AEDs, will be screened for entry into the trial. Recruitment will be from hospital-based paediatric neurology centres, with the additional involvement of a user group, Matthew’s Friends Charity, an organisation set up to raise awareness and availability of the KD in the UK, and which now also supports clinics implementing the KD. Many, if not all of the suitable patients will already be under the care of tertiary paediatric neurology centres according to National Guidelines (https://www.nice.org.uk/guidance/cg137).
Inclusion criteria
-
1.
Age between 3 months and 24 months of age (not beyond second birthday at baseline)
-
2.
Diagnosis of epilepsy confirmed
-
3.
Seizure frequency greater than or equal to four seizures/week on average in the baseline period
-
4.
Failed response to previous trial of two antiepileptic drugs. In the case of infantile spasms, this could include a trial of corticosteroids
-
5.
Children with written informed consent from a parent/guardian
Exclusion criteria
-
1.
Continues on corticosteroids less than 2 weeks prior to randomisation
-
2.
Metabolic disease contraindicating use of the KD, e.g. pyruvate carboxylase deficiency, medium-chain acyl-CoA dehydrogenase (MCAD) deficiency from previous medical investigation and screening at baseline
-
3.
Progressive neurological disease
-
4.
Severe gastroesophageal reflux
-
5.
Previous treatment with the KD
-
6.
Concurrent participation in another clinical trial of an investigational medicinal product (IMP)
-
7.
Patients who are prescribed AEDs not listed in the trial IMPs
-
8.
Patients who have a listed contraindication as per the Summary of Product Characteristics (SmPC) to any of the AEDs listed in the trial IMPs
Sample size
For the primary outcome variable, based on data from Neal et al., [17], we used mean percentage change in seizures from baseline of 62% (SD 45) in the KD group, assuming a change of 90% in the control group (SD 50) (100 = no change in frequency of seizures from baseline) at 90% power and 5% significance. This gives a sample size of 61 in each group (122 in total). Accounting for a 10% dropout rate, gives 68 in each group (136 in total), and inflation for a therapist effect (dietitian) to one group, assuming nine centres, with an average cluster size of eight and an intracluster correlation coefficient (ICC) of 0.05, the inflation factor is 1.35, giving 92 in the KD group and 68 in the control group (160 in total). If dropout was 20%, this sample size would still have 86% power.
Treatment procedures
Trial arm 1
Classical ketogenic diet (KD arm)
The experimental intervention will be an 8-week trial of KD therapy. A KD intervention manual will be created and provided to sites to ensure consistency of the KD implementation across centres. The manual includes basic instructions on how to calculate the classical KD and advice regarding diet implementation such as supplementation, tube feeding, breastfeeding, weaning and fine-tuning the diet. Children allocated to KD therapy will have their diets individually calculated by a paediatric dietitian with consideration of daily calorie requirements, adequate protein intake for growth and vitamin and mineral supplementation. All diets will be implemented according to a classical KD protocol, i.e. based on a ratio of fat to carbohydrate and protein that will usually be between 2:1 and 4:1. In order to achieve a state of ketosis, meal plans have to be accurately calculated for each child individually and recorded in the patient’s medical notes. Breastfeeding can be continued on a KD, in combination with a ketogenic feed which will be given in a prescribed amount before each breastfeed. If breast milk is expressed, this can be mixed with the ketogenic feed to the correct macronutrient ratio. Infants on a KD can be weaned as per Department of Health guidelines [36], with advice given on how to adapt standard weaning foods by addition of extra fat.
Hospital stays will be determined by the clinical team at the treating centre, utilising the ‘Trial Intervention Manual’. A non-fasting initiation protocol will be used for all children. Parents or carers will attend a teaching session prior to diet commencement, including how to manage possible early side effects, such as excess ketosis and hypoglycaemia. Teaching of families in the KD arm will occur following randomisation and prior to starting the diet.
The KD to be implemented will be the classical KD, aiming for at least a 3:1 ratio (fat to carbohydrate and protein). This will be implemented according to a standard text [35]. Further, an initial workshop of dietitians will be led by Elizabeth Neal (co-principal investigator (PI)) in order to ensure consistency of implementation. Cross-site consistency of KD implementation will be monitored after the 8-week and 12-month visits by the dietetic assistant. Details to be monitored include the calculation of energy prescriptions, protein intake, teaching sessions, initiation regimes, supplementation and ketone levels. Monitoring Discrepancy Forms will be created and the Protocol Deviation Log completed, if appropriate.
Trial arm 2
Further antiepileptic drugs (AED arm)
The control intervention will be drug therapy with the most appropriate further AED for a particular child, depending on their presenting seizures, epilepsy syndrome and previous drugs used. This will be chosen by the expert clinician responsible for management of the participant’s epilepsy. Paediatric neurologists will meet at an initial workshop to discuss clinical practice with the aim of forming the basis of a consensus protocol to ensure the consistency of AED treatments delivered. The list of agreed drugs that may be utilised are described in Additional file 5. These are considered IMPs in this trial, irrespective of which arm of the trial the patient is randomised to. This is a pragmatic trial that uses authorised medicinal products for epilepsy within the European Economic Area (EEA). Although the majority of these IMPs are not licensed for paediatric use, or for use in this age group, they are used in routine care as part of established clinical practice. Patients who are prescribed products with no marketing authorisation (‘specials’) for epilepsy or AEDs not listed in the below table will not be eligible for this trial. The KD is not classed as a medicinal product and is not included in Additional file 5. The dietetic assistant will monitor cross-site consistency of IMP prescription according to the protocol.
A discussion about diet and healthy eating will be also be undertaken with families of infants randomised to the AED arm at the randomisation visit. If the participant is already under local dietetic support, it should be ensured that this monitoring continues. If the participant does not have local dietetic support but this is deemed necessary by the ketogenic dietitian, an appropriate referral should be made by the clinician. Otherwise, a very brief discussion about general infant or toddler nutrition will be had, including details such as promotion of breastfeeding, age-appropriate texture progression for weaning, food groups and the important of iron-rich foods.
Statistical analysis
Analysis will be done by intention-to-treat. Baseline characteristics of participants in the control and intervention arms will be summarised. The primary outcome will be seizure count in the final 2 weeks of the intervention period and in the baseline assessment period. Data will be analysed using a Poisson mixed model to account for clustering by centre (synonymous with therapist). The randomised allocation will be entered into the model as a fixed effect as will an indicator of time point (baseline or end of study), whilst the centre will be included as a random effect. Analysis of secondary outcomes (those seizure free and responders) will be analysed using random-effects logistic models – centre being the random effects and randomised group a fixed effect. The process outcomes relating to tolerability and medium-chain fatty acids in the KD group will be analysed using random-effects modelling. Therapist effects will be investigated further in supportive analyses [7, 37].
Discontinuation/withdrawal of participants and ‘stopping rules’
Participants will be withdrawn from the treatment prior to 8 weeks should there be over 50% increase in seizure frequency from baseline or if side effects, such as diarrhoea or constipation, are not resolved by dietary manipulation or medication. Withdrawn patients will not be replaced, but will have scheduled follow-up assessments.
Deliverability and feasibility
An initial pilot study involving two centres over 12 months will aim to recruit approximately 20% of the total sample required: 35 participants over 12 months.
The pilot study will run in two centres in London: Great Ormond Street Hospital for Children and Evelina London Children’s Hospital over the initial 12 months. We will aim to assess approximately 50 eligible patients in this time frame. Additionally, adverse events (AEs) will be monitored and data on safety will be reviewed. We will progress to the full study if the following are achieved:
-
1.
Achieving a 60% recruitment rate; 30 families agreeing to randomisation
-
2.
No more than 10 (29%) failing to complete the 8-week trial period
The main study will proceed to include recruitment from the further seven centres should the above criteria be met. Recruitment will be expected at a rate of 28/year from GOSH, and approximately 7/year from the remaining seven centres (total 84/year). This enables a completed primary outcome in the desired 160 children.
Outcomes
The primary outcome will be the number of seizures experienced during weeks 6–8 compared to the number of seizures in the baseline period.
Secondary outcomes will include (at 8 weeks):
-
Number of children seizure free
-
Responder rate, defined as the number showing more than a 50% improvement in seizure frequency compared to baseline (taken as the mean daily seizure frequency over the observation baseline period immediately preceding the 8-week review)
-
Tolerance to KD as assessed by side effect questionnaire and blood results
-
Relationship between medium-chain fatty acids and seizure control
Secondary outcomes will also include (at 12 months):
-
Retention on treatment (the number of participants who remain on the KD, or the prescribed AED, at 12 months)
-
Quality of life (as measured by the Infant Toddler Quality of Life Questionnaire)
-
Neurodevelopmental outcome (as measured by the Vineland Adaptive Behaviour Scales)
Plasma profiles of medium-chain fatty acids will be evaluated at baseline and at 8 weeks. Assessment of mitochondrial function (respiratory-chain enzymes) and enrichment (citrate synthase) will be determined in white cells and platelets. The effect of specific ratios of medium-chain fatty acids, to mimic patient plasma profiles, upon neuronal mitochondrial function/enrichment (biochemical plus electron microscopy studies) will be documented. Additionally, such fatty acid profiles will be studied, with regards to antiepileptic effect, in an established in vitro (hippocampal slice) model.