A revisited strategy for antiepileptic drug development in children: designing an initial exploratory step

Randomized controlled trials (RCTs) in refractory paediatric epilepsy usually involve the two main types of epilepsy shared by children and adults, focal epilepsy and Lennox-Gastaut syndrome (LGS). Most other epilepsy syndromes, specifically paediatric, are excluded from drug development. In order to identify among them the candidate(s) for dedicated RCTs with a new drug, the European Medicine Agency (EMA) recently recommended proceeding in two steps: (1) an exploratory (prospective-observational) trial (POT) including a large variety of paediatric epilepsy syndromes and (2) a subsequent RCT in each of those that disclose a signal for possible efficacy.

Our objective was to address the three following issues that have not been addressed by the EMA: (1) to determine a minimal threshold for this signal; (2) to establish a list of epilepsies to evaluate; and (3) to estimate the number of patients to include in such POTs.

We extensively reviewed the POTs (including various syndromes) and RCTs reported in paediatric patients with uncontrolled epilepsy using MEDLINE (from 1990 to 2011) and the Cochrane library. We determined the threshold as the lowest percentage of responders observed in a POT with a positive corresponding RCT. The syndromes that reached this threshold in a POT were those to evaluate in an RCT. The minimal number of patients to include for each syndrome for a POT with a new antiepileptic drug was estimated in order to reach at least this threshold of responders with a 95 % confidence interval.

We found the minimal responder threshold to be 25 %. We identified eight epilepsy types/syndromes reaching this threshold and estimated for each of them the minimal sample needed: refractory focal epilepsy (n = 40), Lennox-Gastaut syndrome (n = 32), infantile spasms (n = 50), Dravet syndrome (n = 32), childhood absence epilepsy (n = 12), other symptomatic generalized epilepsy (n = 38), epileptic encephalopathy with continuous spikes and waves during sleep (n = 7) and epilepsy with myoclonic-astatic seizures (n = 4) [the two last samples may be underestimated due to the lack of RCTs in these conditions].

Among the eight epilepsy types/syndromes that we recommend to systematically include in exploratory trials using the POT procedure, we assume that, for the minimal sample given above, a responder threshold of 25 % will provide a reliable efficacy signal, to be confirmed by a dedicated RCT. This strategy should avoid missing new therapeutic possibilities for children with epilepsy and reduce the off-label use of drugs in paediatric neurology.