Evidence review

This evidence summary is based on a phase III trial that assessed the efficacy and safety of adjunctive zonisamide therapy in children and young people with partial epilepsy (Guerrini et al. 2013). This trial was the main efficacy study assessed by the European Medicines Agency (EMA) when a licence extension application was submitted for zonisamide (see the European public assessment report (EPAR) for Zonegran). An open-label extension study (Guerrini et al. 2014) assessing the longer term (45–57 weeks) efficacy and safety of adjunctive zonisamide therapy in children and young people with partial epilepsy has been published and is discussed in the safety section.

  • Design: randomised, double-blind, placebo-controlled trial in 41 centres in Europe and India.

  • Population: 207 children and young people (aged 6–17 years, mean age 11 years) with a clinical diagnosis of epilepsy with partial-onset seizures, with or without secondary generalisation. Participants needed to have experienced at least 4 partial seizures (simple or complex, with or without secondary generalisation) per month during the 8-week baseline period, with at least 1 seizure in each 4-week period and no 21-day period without any seizures. In addition, participants needed to be receiving a stable regimen of 1 or 2 anti-epileptic drugs for at least 1 month before the first visit (1 anti-epileptic drug: 44% in the zonisamide group compared with 39.0% in the placebo group; 2 anti-epileptic drugs: 58.9% in the zonisamide group compared with 60.0% in the placebo group). Exclusion criteria included body weight less than 20 kg at screening, progressive neurological disease, history of idiopathic generalised epilepsy, psychogenic seizures, cluster seizures, history of status epilepticus within the previous year, previous treatment with zonisamide, and concomitant treatment with felbamate, acetazolamide, any carbonic anhydrase inhibitor, or any drug with anticholinergic activity. Baseline and demographic characteristics were similar between the groups apart from mean seizure frequency which was lower in the zonisamide group than the placebo group (mean number of seizures per 28 days: 32.9 in the zonisamide group compared with 43.8 in the placebo group). Median baseline seizure frequency however was similar in the zonisamide and placebo groups (median number of seizures per 28 days: 10.5 in the zonisamide group compared with 10.0 in the placebo group).

  • Intervention and comparison: the study consisted of an 8-week baseline period including a 4-week screening period and historical seizure data from the 4 weeks before screening. Eligible people were then randomised 1:1 to zonisamide (n=107) or placebo (n=100), added to 1 or 2 anti-epileptic drugs. Allocation was appropriately concealed. Zonisamide was started at a dosage of 1 mg/kg/day and increased by 1 mg/kg at weekly intervals, up to a target dose of 8 mg/kg/day over 8 weeks (maximum 500 mg daily). Five participants in the placebo group and 2 in the zonisamide group had a single dose reduction during titration. Participants then remained on the same dose for a 12-week maintenance period, apart from 1 participant in the placebo group who had a dose reduction. Most participants received a dosage of 5 mg/kg/day up to less than 9 mg/kg/day (92.9% in the zonisamide group compared with 83.9% in the placebo group) and almost all other participants received a dosage 9–12 mg/kg/day (7.1% in the zonisamide group compared with 15.1% in the placebo group). At the end of the 12-week maintenance period, participants either entered an extension study, or were down-titrated over 3–4 weeks and withdrawn. Outcomes: the primary efficacy outcome was the proportion of responders (that is the number of participants with a 50% or greater reduction in seizure frequency from baseline) during the 12-week maintenance period in the intention-to-treat (ITT) population using last observation carried forward (LOCF) data. Seizure data were obtained from seizure diaries completed by the child or young person's parent or guardian. Secondary efficacy outcomes included the median percentage change from baseline in 28-day seizure frequency; the proportion of participants with a reduction in seizure frequency of 75% or more; and the proportion of participants with an increase in seizure frequency of 25% or more, and 100% or more. Other efficacy outcomes included the proportion of participants experiencing seizure freedom; the percentage change from baseline in 28-day seizure frequency by seizure type; and the relationship between zonisamide plasma level and responder rate. Sensitivity analyses were performed, including for the ITT-observed case data set. Safety assessments included the incidence of treatment-emergent adverse events, and whether these were serious or resulted in withdrawal of study treatment; clinical laboratory parameters; physical and neurological evaluations; vital signs; height and weight; and electrocardiography.

Table 1 Summary of the trial: Guerrini et al. (2013)

Population

Zonisamide at a target dose of 8 mg/kg/day

Placebo

Analysis

Randomised

n=107

n=100

Efficacy a

n=107

n=100

Primary outcome: proportion of responders during the 12-week maintenance periodb

ITT-LOCF

50.0%

31.0%

p=0.0044

ITT-OC

48.0%

31.0%

p=0.0143

Selected secondary outcomes:

Median percentage reduction from baseline in 28-day seizure frequency during the 12-week maintenance period

ITT-LOCF

50.0%

24.5%

Median difference between the groups: 25.2%, 95% CI 12.2 to 38.7%, p<0.0001

Proportion of participants with seizure frequency reduction of 75% or more from baseline during the 12-week maintenance period

ITT-LOCF

27.0%

12.0%

p=0.0064

Proportion of participants with increase in seizure frequency of 25% or more from baseline during the 12-week maintenance period

ITT-LOCF

10.0%

21.0%

p=0.0330

Proportion of participants with increase in seizure frequency of 100% or more from baseline during the 12-week maintenance period

ITT-LOCF

5.0%

9.0%

p value not reported but stated to be non-significant

Safety

n=107

n=100

Participants reporting treatment-emergent adverse events

Safetyc

55.1% (59/107)

50.0% (50/100)

p value not stated

Participants reporting treatment-related treatment-emergent adverse events

Safety

33.6% (36/107)

24.0% (24/100)

p value not stated

Participants reporting serious treatment-emergent adverse events

Safety

3.7% (4/107)

2.0% (2/100)

p value not stated

Participants reporting serious treatment-related treatment-emergent adverse events

Safety

1.9% (2/107)

1.0% (1/100)

Includes 1 person in the zonisamide group who died due to de novo status epilepticus

Participants with treatment-emergent adverse events leading to withdrawal

Safety

0.9% (1/107)

3.0% (3/100)

Participants with treatment-emergent adverse events leading to dose reduction

Safety

2.8% (3/107)

5.0% (5/100)

Abbreviations: CI, confidence interval; ITT-LOCF, intention-to-treat last observation carried forward population; ITT-OC, intention-to-treat observed case population; p, p value.

a The intention-to-treat population consisted of all randomised participants who received at least 1 dose of double-blind study medication. Analysis of the primary efficacy outcome was performed for the intention-to-treat population using last observation carried forward data and sensitivity analyses were performed using the intention-to-treat observed case, and per-protocol last observation carried forward data.

b Responders were defined as participants with a 50% or more reduction in seizure frequency from baseline.

c The safety population consisted of all randomised participants who received at least 1 dose of double-blind study medication.

Clinical effectiveness

During the maintenance period, in the ITT-LOCF population, zonisamide statistically significantly increased the responder rate (that is, the number of participants who experienced a 50% or greater reduction in seizure frequency from baseline) compared with placebo (responder rate: 50% with zonisamide compared with 31% with placebo; p=0.0044). This finding was supported by sensitivity analyses in the ITT observed cases (responder rate: 48% with zonisamide compared with 31% for placebo; p=0.0143) and per-protocol LOCF populations (responder rate: 51% with zonisamide [n=100] compared with 31% for placebo [n=91]; p=0.0046).

In the ITT-LOCF population, the median reduction in 28-day seizure frequency from baseline during the maintenance period was 50% in zonisamide group compared with 24.5% in the placebo group. The median difference between the groups was 25.2% and was statistically significant (95% confidence interval [CI] 12.2% to 38.7%, p<0.0001). The authors report that these findings were consistent in the ITT observed cases, and per-protocol LOCF populations, however no data were reported.

In the ITT-LOCF population, the proportion of participants who had a reduction in seizure frequency of 75% or more was statistically significantly greater in the zonisamide group compared with the placebo group (27% compared with 12% respectively; p=0.0064). The proportion of participants whose seizures increased in frequency by 25% or more was statistically significantly lower in the zonisamide group than in the placebo group (10% compared with 21% respectively; p=0.0330). However, the proportion of participants whose seizures increased in frequency by 100% or more was not statistically significantly different between the zonisamide and placebo groups (5% compared with 9% respectively; p value not reported but the result was stated to be non-significant).

In the ITT-LOCF population, the number of participants achieving seizure freedom during the maintenance period was statistically significantly greater in the zonisamide group compared with the placebo group (14% compared with 3% respectively; p=0.0049). When comparing the median reduction in 28-day seizure frequency from baseline during the maintenance period by seizure type, reductions were statistically significantly greater in the zonisamide group compared with the placebo group for complex partial seizures, but not for simple partial seizures or secondarily generalised seizures. However, interpretation of these analyses is difficult as the study was not powered to detect differences in efficacy outcomes by seizure type.

Safety

In the Guerrini et al. (2013) trial, the incidence of treatment-emergent adverse events was similar in the zonisamide and placebo groups (55.1% compared with 50.0% respectively). However, the incidence of treatment-emergent adverse events judged to be treatment-related was higher in the zonisamide group than the placebo group (33.6% compared with 24.0% respectively).

Treatment-emergent adverse events that were reported more frequently in the zonisamide group than in the placebo group included decreased appetite (6.5% compared with 4.0%), weight loss (4.7% compared with 3.0%), somnolence (4.7% compared with 2.0%), vomiting (3.7% compared with 2.0%), and diarrhoea (3.7% compared with 1.0%). One participant (0.9%) in the zonisamide group withdrew from the study because of a treatment-emergent adverse event (allergic dermatitis) compared with 3 (3.0%) in the placebo group (1 participant withdrew because of upper abdominal pain, 1 because of aggressive behaviour and sleep disorder, and 1 because of complex partial seizures). Four participants in the zonisamide group and 2 in the placebo group experienced serious treatment-emergent adverse events, including 1 child in the zonisamide group who died. The death was considered to be possibly treatment related: the young person (a 14-year-old male) experienced weight loss and diarrhoea which weakened their general condition to such an extent they were unable to take their anti-epileptic medications. This triggered a fatal episode of status epilepticus. Another participant in the zonisamide group experienced severe weight loss, moderately decreased blood glucose, and moderate dehydration, which were all considered to be possibly treatment related. One participant in the placebo group experienced severe vomiting and moderate somnolence.

Bicarbonate levels decreased in the zonisamide group. The proportion of participants who experienced a reduction in bicarbonate level of at least 3.5 mmol/L from baseline to final visit was greater in the zonisamide group than the placebo group (50.5% compared with 16.0%). The authors report there were no other clinically significant changes in laboratory values or electrocardiography from baseline to final visit.

Guerrini et al. (2014) report an open-label extension study assessing the longer term (45–57 weeks) efficacy and safety of zonisamide in 144 children and young people recruited from the Guerrini et al. (2013) study (72 children who had received zonisamide, and 72 who had previously received placebo). During the first 11 weeks of this study, children who received zonisamide in Guerrini et al. (2013) remained on this treatment and children who received placebo changed their treatment to zonisamide and up-titrated, after which the study became open-label. Treatment-emergent adverse events were reported in 48.6% of the study population (51.4% in children who had previously received zonisamide and 45.8% in children who had previously received placebo). The most commonly reported adverse events were nasopharyngitis, weight loss and headache. Serious treatment-emergent adverse events were reported in 10 (6.9%) children and were considered to be treatment related in 3 cases (renal colic, foot fracture, and abdominal pain). Decreases in bicarbonate level of more than 3.5 mmol/L were observed in 64 (44.4%) children, but there were no reports of metabolic acidosis. These changes in bicarbonate levels were considered to be similar to those described in previous trials of zonisamide.

The EPAR for Zonegran also presents pooled safety data in children and young people from all available completed studies at the time of the licence submission (17 studies; children aged under 12 years: n=191, children and young people aged 12–16 years: n=207). These pooled data included data from the Guerrini et al. (2013) study, but not from the extension study.

Treatment-emergent adverse events were reported in 82.9% of those who received zonisamide, and 55% who received placebo. The most common treatment-emergent adverse events in children aged 6–11 years were pyrexia (21.9%), headache (21.2%), upper respiratory tract infection (21.1%), decreased appetite (20.5%), rash (12.3%), somnolence (18.5%), vomiting (14.4%), fatigue (13.7%), nasopharyngitis (13.7%), sinusitis (11.0%), viral infection (11.0%), upper abdominal pain (10.3%), cough (10.3%), insomnia (10.3%) and nasal congestion (10.3%). In children and young people aged 12–16 years, the most common adverse events included headache (18.4%), decreased appetite (17.4%), fatigue (12.6%), dizziness (12.1%) and somnolence (11.1%).

In the pooled data, serious treatment-emergent adverse events were reported in 57 children (14.3%) who received zonisamide. This included 7 children who died (including the 1 young person in Guerrini et al. 2013), 4 of whom had pre-existing functional neurological conditions or remote symptomatic epilepsy. Of the 7 deaths, 2 were considered to be related to zonisamide. The most frequently reported serious adverse events overall were convulsion (4.3%), status epilepticus (2.5%), and dehydration (1.8%). Apart from those which resulted in death, most serious adverse events resolved.

In the EPAR for Zonegran, the Committee for Medicinal Products for Human Use concluded that although the general safety profile for zonisamide in children is similar to that in adults, there are several important issues that raise concerns because they may have greater implications in children than in adults. The product information has been updated to include warnings about adverse events that may more relevant to children. Within the EPAR, the Committee for Medicinal Products for Human Use state that overall the number of children and young people that died during the studies was concerning, although the number was similar to mortality rates in people taking other anti-epileptic drugs.

The summary of product characteristics for Zonegran warns that because of the potential seriousness of decreased body weight in children, weight should be monitored in this population. In addition, the risk of zonisamide-induced metabolic acidosis appears to be more frequent and severe in children; appropriate evaluation and monitoring of serum bicarbonate levels should be carried out in this population. Cases of decreased sweating and elevated body temperature, leading to heat stroke resulting in hospitalisation and in some cases death, have been reported in children. The summary of product characteristics for Zonegran gives advice on preventing overheating and dehydration in children who are being treated with zonisamide.

Evidence strengths and limitations

The EPAR for Zonegran states that the study design of Guerrini et al. (2013) was broadly in line with the Guideline on Clinical Investigation of Medicinal Products in the Treatment of Epileptic Disorders. It also stated that inclusion and exclusion criteria, participants included in the study and the choice of primary and secondary end points were considered appropriate for the indication being applied for.

The proportion of responders (that is, the number of participants with a 50% or greater reduction in seizure frequency from baseline) in Guerrini et al. (2013) was 50% with zonisamide compared with 31% with placebo. The EPAR for Zonegran states that this is of the same order of magnitude as that seen in trials of other anti-epileptic drugs used as adjunctive treatment in partial epilepsy in children. There are currently no studies comparing zonisamide with an active comparator in adjunctive treatment of partial epilepsy in children.

As the European Medicines Agency (EMA) notes in its guideline on missing data in confirmatory clinical trials, it is unrealistic to expect that all patients in any clinical trial will receive treatment with full compliance to the treatment schedule and with a complete follow-up as per protocol. The study by Guerrini et al. (2013) used the LOCF approach to take account of missing data. The EMA's guideline notes that, because people who do not complete a clinical trial may be more likely to have extreme values than those who do, the loss of these 'non-completers' could artificially narrow the confidence interval for the treatment effect. The EMA guideline advises that it will almost always be necessary to investigate the robustness of trial results through appropriate sensitivity analyses that make different assumptions. In Guerrini et al. (2013), the analyses of ITT-observed cases and per-protocol LOCF populations showed similar results for responder rates (the primary outcome), which provides reassurance about the reliability of the results

Safety data from Guerrini et al. (2013) were limited in that children and young people received zonisamide for a mean of only 131 days (range: 7–156 days). The European Guideline on Clinical Investigation of Medicinal Products in the Treatment of Epileptic Disorders requires a study to include at least 100 children and last for at least 1 year to appropriately assess the long-term safety of anti-epileptic drugs. Therefore, during the licence extension application, the manufacturer submitted data from the longer-term (45–57 weeks) extension study (Guerrini et al. 2014). In addition, the EMA looked at pooled safety data in children from all available completed studies at the time of the licence application.

The Guerrini et al. (2013) study was undertaken in 41 centres in Europe and India meaning some centres would have enrolled fewer than 10 children and young people. It is possible that this may have affected how rigorously and consistently trial outcomes were assessed.