Introduction
Dravet Syndrome (DS) [
1] is a rare and severe Developmental and Epileptic Encephalopathy (DEE). Patients typically have highly pharmaco-resistant convulsive seizures from the first year of life, mostly generalized clonic or tonic–clonic seizures (GTCS), and likely long-lasting in duration, thus resulting in
status epilepticus (SE), especially in infancy and middle childhood [
2‐
5]. GTCS in DS negatively interfere with further development, strongly impact both patients’ and caregivers’ quality of life and can be life threatening [
4,
6,
7].
The initial signal of stiripentol (STP) effect in DS was discovered in the mid-1990s, in a French prospective exploratory (“basket”) phase 2 study [
8]. At that time, the role of the
SCN1A gene was still unknown, the aggravating effects of sodium channel blockers such as carbamazepine and lamotrigine were just emerging [
9], and the combination of valproate and clobazam was the standard of care antiseizure medication. Pharmacological interactions between these two drugs and stiripentol were already known [
8].
Afterwards two pivotal phase 3 trials, the STICLO studies, comparing STP with placebo as an adjunctive therapy to valproate (maximum dose 30 mg/kg/day) and clobazam (maximum dose 0.5 mg/kg/day) comedication, were initiated in France and Italy. Based on a similar randomized, double-blind, placebo-controlled, design, the STICLO studies evaluated the efficacy of a 2-month STP therapy at 50 mg/kg/day on GTCS; then, patients were proposed to enter an additional 1-month open-label extension (OLE) period. Both studies clearly demonstrated STP efficacy [
10,
11] and served as the basis for marketing authorization for STP in DS—the very first one in a DEE—in Europe (2007), in Japan and Canada (2012), and in the US (2018). Since then, the efficacy of stiripentol has been confirmed in numerous prospective and retrospective real-world evidence studies worldwide [
12‐
15], and more than 56,759 patient-year with DS and, with off-label prescriptions, other forms of severe epilepsies have benefited from this treatment as of October 31, 2022 [
16].
Some data from the STICLO trials were not published in extenso at the time, notably the detailed efficacy results of STICLO Italy and data from the OLE period. Also, since the STICLO studies, new relevant outcomes have now emerged in the field of drug trials, as evidenced by the two treatments recently approved in DS, cannabidiol (2018) and fenfluramine (2020). The objective of the present report is to conduct a post-hoc analysis of data from both Phase 3 STICLO studies, which includes the rate of patients with ≥ 75% decrease in seizure frequency, the number of convulsive seizure-free days, and the time to onset of STP efficacy, all these parameters having significant impact on patients and caregivers’ quality of life.
Methods
Study Design and Patients
STICLO studies were conducted with similar protocols and performed at multiple centers in France from October 1996 to August 1998, and in Italy from April 1999 to October 2000. Inclusion of patients diagnosed as DS was made according to the clinical criteria established by Dravet [
17], i.e., onset of epilepsy in the first year of life with generalized clonic (or tonic–clonic) seizures, normal psychomotor development and normal electroencephalogram (EEG) before seizure onset, and, after 1 year of age, occurrence of atypical absences and myoclonia, generalized spikes and waves on EEG, and slowing of psychomotor development. Patients were aged between 3 and 18 years and had to have experienced at least four GTCS a month despite optimized therapy with valproate and clobazam. Patients receiving other drugs (except diazepam as a rescue medication and progabide), and those whose parents were unable to comply regularly with drug delivery and daily seizure diary, were excluded. According to local regulations, the design and conduct of the STICLO studies were approved by the Ethics Committee of the coordinating center in France, and by the Ethics Committees of each center in Italy. Written informed consent was obtained from the parents or guardian of all patients. Both studies were performed in accordance with the Helsinki Declaration of 1964 and its later amendments, Good Clinical Practices, and according to the local regulatory requirements.
After a 1-month baseline period, patients who continued to have at least 4 GTCS per month were randomly assigned to receive placebo or stiripentol at the dose of 50 mg/kg/day (maximum dose of 3000 mg/day) with no titration. Study treatment was given twice or three times daily, added on to valproate and clobazam during the 2-month double-blind period. To minimize adverse effects, the maximum dose of valproate was limited to 30 mg/kg/day and that of clobazam to 0.5 mg/kg/day. Doses could be decreased by 10 mg/kg/day for valproate in cases of loss of appetite and by 25% for clobazam in cases of drowsiness or hyperexcitability.
After the 2-month double-blind period, patients were allowed to enter an additional 1-month open-label extension period where they all received STP, with the blind maintained on treatment previously received (Figure
S1). Patients who benefited from stiripentol treatment were then allowed to enter an open-label observational long-term follow-up study, followed by a temporary authorization for stiripentol use.
Outcome Measures
In both studies, response to treatment was defined by the reduction in the number of convulsive seizures compared to baseline. Response rate was calculated based on the decrease in monthly GTCS frequency, and it was considered as clinically meaningful (≥ 50% reduction), profound (≥ 75% reduction), or as seizure freedom (= 100% reduction).
Study investigators reported seizure frequency monthly by using the diary completed by parents and caregivers. In the present analysis, data from the diaries were entered for the first time and the seizure occurrence was captured daily, allowing the evaluation of additional parameters, such as the total number of seizure-free days and the longest period of consecutive days free of seizures. All types of seizures were reported in the diaries, but only GTCS were taken into account for the assessment of treatment efficacy, since these are the most disabling seizure types, are observed in all patients, and are easily identified. Conversely, counting other types of seizures, such as absence seizures or myoclonus, is less reliable and difficult to obtain.
At the end of the OLE period, investigators completed a specific follow-up form where they reported information regarding STP dosage, seizure frequency, and adverse events, if any. When available, data from the follow-up forms were captured for the present analysis.
Statistical Analyses
Data poolabilty from both STICLO-France and STICLO-Italy studies was assessed based on the demographic data and the primary endpoint results using t tests and chi-squared tests (Type I Error rate of 0.10). Studies were considered poolable if the data from the two studies were not statistically significantly different.
Baseline characteristics and seizure frequency were reported by treatment groups with numbers and proportions for categorical variables, as well as mean standard deviation and the range of values for continuous variables. Seizure frequencies were adjusted for 30-day period to compare the corresponding frequencies. Percentage changes in seizure frequency was also represented on an individual basis by a waterfall chart, where patients were ordered by the magnitude of the change between baseline and 1st month under treatment, 2nd month under treatment, or OLE period.
Responder rates were reported by treatment group at 1-month, 2-month, and 3-month OLE. Considering the small sample sizes, confidence intervals at 95% (CI95%) were computed by the Wilson method with continuity correction [
18]. In addition, the mean percentage change and CI95% from baseline in seizure frequency, and the cumulative proportion of responders as a function of the percentage reduction in the number of seizures, were shown graphically for each treatment group. A Sankey diagram was also used to describe and analyze longitudinal patient data during treatment period and OLE period.
Regarding patients’ diaries, the total number of seizure-free days was calculated and was adjusted for a 60-day period to make the 2-month double-blind period comparable between patients. Three patients with missing data from diaries during the 2-month treatment period were excluded. The longest period of consecutive seizure-free days during the double-blind period was also determined. A conservative approach was used as patients with missing diaries were not excluded and the interval was stopped if data were missing.
Moreover, a time-to-onset analysis was conducted using negative binomial regressions to model the cumulative number of seizures by treatment group during the follow-up from day 1 to day 30. Models were estimated to test the hypothesis that mean cumulative number of seizures is not different between the placebo and the STP group. At day n, the cumulative number of seizures from day 1 to day n during the treatment period were modeled by an intercept, the treatment group effect, and adjusted on the corresponding cumulative number of seizures from day 1 to day n during the baseline period. Logarithm of the period (n days) was added as an offset term. Least squares (LS) mean number of seizures count for STP and placebo groups were reported for each day n and ratios of LS means were tested to see if they were significantly different from 1. LS means and CI95% were reported graphically for each day and treatment group.
Discussion
These new, more comprehensive analyses of the STICLO data reinforce and complete the evidence for a marked efficacy of STP in Dravet syndrome. STP demonstrated a clear superiority over placebo, not only in the ≥ 50% decrease in convulsive seizures but also considering patients with a profound (≥ 75%) reduction in seizure frequency, and the number of patients seizure-free. These results also highlight the rapid onset of STP efficacy, detectable from the fourth day of treatment, and its maintenance during the OLE period. Importantly, longitudinal data evidenced that patients who do not respond initially can achieve a significant seizure reduction and become responder after 3 months of treatment. Finally, both the number or seizure-free days and the longest period of consecutive seizure-free days were significantly higher in the STP group. Beyond the seizures themselves, all these parameters are of critical importance for the child's quality of life and cognitive functioning.
The STICLO France study was the first-ever double-blind, placebo-controlled trial, that evaluated the efficacy of an antiseizure medication in Dravet syndrome. As no data were available in a reference population with the same type of epilepsy, the statistical analysis plan stated that a difference in percentage of responders of at least 25% was required between STP and placebo, and that an interim analysis without unblinding would be done after 20 patients in each group were enrolled. At the time of this interim analysis, in cases of significant difference between the two groups (at the reduced degree of significance, α′ = 2.5%), the study would be terminated; otherwise, it would continue up to a maximum of 100 patients. The interim analysis, performed as planned on 41 evaluable patients, showed a CI95% true difference between 42.2% and 85.7%, thus demonstrating a significant superiority of STP over placebo. Therefore, recommendation was made to stop the study as it had already met the primary efficacy criterion with 20 patients in the placebo group and 21 in the stiripentol group.
In the subsequent STICLO Italy study, the number of patients to be enrolled was extrapolated from STICLO France efficacy results, and it was determined that 20 patients (10 in each treatment group) would suffice to demonstrate a significant difference. Accordingly, 23 patients were enrolled, 12 on STP (mean dose 51.1 mg/kg/day) and 11 on placebo. STICLO trials followed a rigorous methodology, leading to approval of stiripentol for DS in Europe in 2007, then in 2018 in the US after both were validated by an FDA audit in 2016. Their originality lies in the limited number of patients needed that were thoroughly characterized, allowing the exposure of as few children as possible to the trial condition while preserving its quality. This approach praised by the community for “how best to serve children” [
20], paved the way for robust trials that seemed impossible until then for rare pediatric diseases.
Pooled data from the two STICLO studies are presented here for the first time. Although meta-analyses were published previously, they only included partial data, notably for STICLO Italy [
21‐
23]. Efficacy results of the STICLO Italy are presented extensively here, together with efficacy criteria that were not detailed at the time the studies were conducted. Notably, similarly to the seizure reduction thresholds used for recently approved Dravet syndrome medications [
24,
25], the rate of profoundly responder patients was calculated. At the end of the 2-month double-blind treatment period, 56% of the patients on STP experienced a profound (≥ 75%) decrease in convulsive seizures versus 3% of those on placebo. This observation, in line with the high number of patients on STP who were free of convulsive seizures, supports a high efficacy of this drug in DS. Moreover, a reduction in monthly convulsive seizure frequency is associated with clinically meaningful levels of improvement in everyday executive functions in these patients [
26].
The analysis of the time-to-onset to efficacy evidenced a significant and permanent difference in favor of STP with respect to placebo from day 4, which was maintained afterwards. This rapid effect is clinically important given the severity of DS, notably in young patients who are more likely to experience long-lasting seizures often evolving into SE, considering its negative impact on immediate mortality and long-term outcome [
14,
27].
Dravet syndrome imposes a considerable burden on patients’ and caregivers’ quality of life [
4,
6,
7]. Recently, the number of seizure-free days was proposed as a marker of patient and caregiver quality of life [
28‐
30], and it was estimated that each additional patient–seizure-free days increased utility by 0.005, and that increasing seizure-free days per 28 days was a significant predictor of improved quality of life [
30]. In the STICLO studies, both the total number of seizure-free days and the maximum number of consecutive seizure-free days were significantly higher in the STP group than in the placebo group (
P < 0.001 in both cases). Despite no validated scales being applied, this result may indicate that STP could improve patients’ and caregivers’ quality of life by decreasing the burden of seizures and their consequences. This observation is in line with a recent report [
15] which showed that initiating STP in infants with DS significantly reduced long-lasting seizures, including SE, and hospitalizations in the critical first years of life.
From a safety perspective, side effects often led to decrease comedication doses (in 17 patients) as planned by the protocol, attenuating adverse events in all these patients. This observation is in accordance with the Stiripentol Summary of Product Characteristics, in which it is indicated that many of the adverse reactions are often due to an increase in plasma levels of associated anticonvulsant medicinal products and may regress when their dose is reduced.
As diaries or follow-up forms were missing for some patients, the amount of data available for analyses during the OLE period was incomplete. Also, patients and physicians were aware of the treatment administered during the OLE period, making the analyses less robust with respect to the double-blind period. Despite these caveats, efficacy and safety results in the OLE period are fully aligned with those collected at months 1 and 2, altogether corroborating STP efficacy. Follow-up data were not collected further except in the Paris center (STICLO France) [
11]: among the 16 patients included, 12 were still on STP in adulthood. They were satisfactorily controlled with good tolerability 26–28 years later, as recently reported [
14].
Since the STICLO studies have been completed, 30 years of clinical practice and extensive real-world use of STP have confirmed and refined its efficacy in DS, including its use in the long and very long term, with up to 24 years of exposure [
12‐
15,
31‐
34] in adults and in infants [
34‐
36]. Of particular clinical benefit is its effectiveness in reducing long-lasting seizures and episodes of SE, thereby reducing emergency hospitalizations and the need for rescue medication, and in improving the quality of life of patients and their families [
13,
15,
33]. Also, prolonged STP therapy in DS tends to positively impact the longer-term prognosis of epilepsy, especially if initiated before adolescence [
12,
14].