Om een adequate keuze te maken voor de medicamenteuze behandeling bij epilepsie, is het eerst van belang om het aanvalstype vast te stellen. Hierbij wordt onderscheid gemaakt tussen epilepsie met een focaal of gegeneraliseerd begin. Als het aanvalstype onduidelijk is dan wordt gesproken van een epilepsie met onbekend begin. Epileptische aanvallen met een gegeneraliseerd begin zijn aanvallen waarbij vanaf het begin bilateraal verdeelde netwerken betrokken zijn (zie module ‘Classificatie’).

Als er sprake is van epileptische aanvallen met een gegeneraliseerd begin dan zijn er verschillende middelen beschikbaar. Valproaat is van oudsher het middel van eerste keus. Het doel van deze module is om antwoord te geven op 2 vragen:

1. Met welk anti-aanvalsmedicijn kan het beste worden gestart bij een nieuwe diagnose van epileptische aanvallen met een gegeneraliseerd begin
2. Welk middel zou moeten worden toegevoegd wanneer monotherapie onvoldoende is?

Voor absence epilepsie, zie de module ‘Absence epilepsie’.

PICO 1: Which antiseizure monotherapy is the most effective and safe in patients with newly diagnosed generalized onset seizures?

1a. Seizure incidence (critical)

1b. Time to seizure recurrence (critical)

1c. Retention rate (important)

1d. Adverse events (important)

PICO 2: Which antiseizure add-on therapy is the most effective and safe in patients with treatment resistant generalized onset seizures?

2a. Seizure incidence (critical)

2b. Time to seizure recurrence (critical)

2c. Retention rate (important)

2d. Adverse events (important)

Description of studies

PICO 1

The Cochrane review by Nevitt (2022) was used in the NICE guidelines for the recommendations with regard to monotherapy in patients with epilepsy. After the search on April 12^th, 2021, this systematic review included 89 trials (n = 22,040 participants), of which 30 trials recruited individuals with focal seizures only, 5 recruited individuals with generalized tonic-clonic seizures with or without other generalized seizure types or unclassified seizure types only, and the remaining 54 trials recruited both. A total of 5268 patients were classified as experiencing generalized onset seizures. For 4270 of these, individual participant data was available for analysis out of 34 trials, and sensitivity analyses were performed for potential misclassification. A table of characteristics of the included trials and their risk of bias assessment is presented in the “Characteristics of studies section” in the article by Nevitt (2022).

For generalized onset seizures, the drugs carbamazepine, sodium valproate, phenobarbitone, phenytoin, lamotrigine, oxcarbazepine, topiramate, gabapentin, levetiracetam, and lacosamide were assessed. The level of evidence was assessed with the Confidence in Network Meta-Analysis (CINeMA) approach (Nikolakopoulou, 2020), based on the GRADE framework.

The primary outcome of interest was time to treatment failure (retention time), either due to lack of efficacy (i.e. recurrent seizures), poor tolerability (i.e. adverse events) or a combination of both (any treatment-related reason). Secondary outcomes were time to achieve 6- and 12-month seizure-free period (remission) after randomisation, time to first seizure post-randomization, and the occurrence of adverse events.

Results

1a. Seizure incidence (critical)

Nevitt (2022) reports time to achieve 6- and 12-month seizure remission through hazard ratios instead of a relative risk or patient proportion of seizure freedom at 6 or 12 months. The results per drug for time to 12-month seizure remission, compared to sodium valproate as reference, are shown in figure 1a. The conclusion for all comparisons was that no difference between drugs with regard to time to 6- and 12-month remission was found. This conclusion did not change when only considering direct evidence.

Figure 1a. Time to 12-month remission for individuals with generalized seizures (reference sodium valproate). Adapted from Nevitt (2022).

1b. Time to seizure recurrence (critical)

Time to first seizure post-randomisation was also reported through HRs by Nevitt (2022). In Figure 1b, the comparison of different antiseizure medications to sodium valproate is depicted. Lamotrigine is the only ASM showing a significantly worse effect than valproate, yet the confidence interval crosses the border of clinical importance as defined under the heading ‘search and select’.

Figure 1b. Time to first seizure for individuals with generalized epilepsy (reference sodium valproate). Adapted from Nevitt (2022).

1c. Retention rate (important)

Treatment failure can be due to either lack of efficacy (continued seizures or the initiation of add-on treatment) or tolerability issues (side effects or non-compliance). Time to treatment failure is therefore a combined measure of efficacy and tolerability. Sodium valproate was taken as reference; results are depicted in figure 1c. From the NMA, sodium valproate performs significantly better than topiramate, carbamazepine, phenobarbitone, and lacosamide, yet this difference was not found from the direct evidence. No statistically significant differences (not in NMA, nor in direct evidence) were found between sodium valproate and lamotrigine, levetiracetam, gabapentin, phenytoin, and oxcarbazepine.

Figure 1c. Time to treatment failure for individuals with generalized seizures (reference sodium valproate, SVP). Adapted from Nevitt (2022).

1d. Adverse events (important)

Adverse events were reported narratively. A subdivision of adverse events per drug used for focal onset seizures or generalized onset seizures was not made. Adverse event data were available for 20,275 participants (92% of total included patients in the review). Adverse events were reported in 62% of these participants. The most common adverse events across all drugs were drowsiness or fatigue, headache or migraine, gastrointestinal disturbances, dizziness, and rash or skin disorders.

The two most commonly occurring adverse events per studied ASM are summarised in table 3. Comparison of adverse events between different medications should be made with caution, as more detailed information on side effects is available for newer ASMs compared to older ASMs.

Table 3. The two most frequently occurring adverse events for the researched antiseizure medications (used for both focal or generalized onset seizures).

Abbreviations: ASM: antiseizure medication

An elaborate overview of the number and type of reported events per medication for all different medications can be found in Table 21 of the original article by Nevitt (2022).

Level of evidence:

As the evidence is based on RCTs, the level of evidence for all outcome measures started at high.

1a. seizure incidence; 1b. time to seizure recurrence; 1c. retention rate

For these outcome measures no downgrading of the level of evidence took place for risk of bias, publication bias, inconsistency, or indirectness. However, downgrading in varying degrees for imprecision took place for different ASM.

Rationale:

For all three outcome measures above applies: some RCTs were at high risk of bias for at least one domain, yet sensitivity analyses showed similar numerical results and no changes to conclusions. Therefore, the authors judged that risk of bias within the trials had not influenced the overall results. In addition, publication bias was considered through the applicability of the individual participant data used. The authors noted a high risk of misclassification of epilepsy type in several studies, yet investigation of misclassification within the review did not show any important changes to treatment effect sizes and no changes to conclusions. No indication of important inconsistency between direct evidence and NMA results was found, except for:

• Seizure incidence (1a): large heterogeneity was present in the direct comparison of levetiracetam to sodium valproate (likely due to difference in trial designs), yet numerical results with NMA were similar.
• Time to seizure recurrence (1b): The direct evidence showed an advantage to sodium valproate compared to topiramate, yet the NMA showed no significant difference.

For none of these three issues has the level of evidence been downgraded.

For imprecision, downgrading took place by -1, -2, or -3 levels. Table 4 provides overview of the degree of downgrading per outcome measure and per ASM.

1d. Adverse events

The level of evidence regarding the outcome measure adverse events was downgraded by 2 levels because of possibly biased results due to study limitations in some of the included studies (-1, risk of bias); and heterogeneity in assessing and reporting adverse events through different studies (-1, inconsistency). The evidence was not downgraded for indirectness, despite the inclusion of adverse events after medication use for focal seizures, as the expected side effects are similar for both indications (focal and generalized seizures).

Description of studies

PICO 2

For the update of the NICE guidelines, an NMA was performed to answer the question: What add-on antiseizure medications are effective in the treatment of generalized tonic-clonic onset seizures? After the search on February 3^rd, 2021, eight studies were included, with a cumulative of 1,218 participants with generalized onset tonic-clonic seizures for which one or more antiseizure medications had failed (with or without other generalized seizure types such as absence or myoclonus). Of these participants, 599 were in the placebo group (49.2%). In the active treatment groups, patients were treated with one of the following drugs: brivaracetam, lacosamide, lamotrigine, levetiracetam, perampanel or topiramate. All evidence was for active drugs compared to placebo, and no direct evidence was identified between two active treatments.

The outcome measures researched were >50% reduction in seizure frequency and seizure freedom during treatment or maintenance period. In addition, adverse events and health-related quality of life were assessed. Risk of bias was assessed with the Cochrane RoB tool v.2 for RCTs (Higgins, 2019). Of the eight included articles, four were assessed as having low risk of bias, two with some concerns (due to the randomization process or risk of bias in measurement of outcomes), and two with high risk of bias (concerns on multiple risk of bias domains).

Results

2a. Seizure incidence (critical)

Seizure incidence was assessed as >50% reduction in seizure frequency during treatment or maintenance period. The odds ratios (ORs) of ASMs compared to placebo (through a fixed effects NMA) are shown in figure 2a.

However, please note that odds ratio’s are not the desired outcome measure for the outcome 50% seizure reduction, as it can overestimate the risk; risk ratios would have been the adequate reporting measures.  

Figure 2a. The relative effect of antiseizure medicaton on seizure incidence (>50% reduction in seizure frequency) compared to placebo for patients with generalized seizures. Adapted from NICE (2022).

Seizure incidence was also assessed through seizure freedom rates during treatment or maintenance period, expressed in Odds Ratios (ORs), in Table 4.

Table 4. Seizure freedom for patients with generalized seizures, after the use of antiseizure medicaton compared to placebo.

Abbreviations: CI: confidence interval.

2b. Time to seizure recurrence (critical) and 2c. Retention rate (important)

The performed NMA for the update of the NICE guidelines did not include time to seizure recurrence or retention rate as outcome measures.

2d. Adverse events

Side effects, treatment cessation due to adverse events, and mortality were recorded. However, the reporting was inconsistent across studies and the types of adverse events were not reported. The number of patients experiencing adverse events for ASM and placebo, and the number consequentially withdrawing, are presented in table 5. Despite having relatively high adverse event rates, the discontinuation rate due to adverse events was generally low. A relatively large discontinuation rate due to adverse events was seen in patients treated with perampanel.

Table 5. Adverse event rate and consequential withdrawal for antiseizure medications used as add-on medication for patients with generalized seizures.

Abbreviatons: AE: adverse event.

Level of evidence of the literature

As the evidence was based on RCTs, the level of evidence started as high.

For the outcome measure seizure incidence (2a), the level of evidence was downgraded by 3 levels because of study limitations in half of the included studies (-1, risk of bias: in the randomization process in three studies, for deviation from intended interventions in two studies, for missing outcome data in two studies, and in measurement of the outcome in three studies); and because of the use of odds ratios overestimating the risk with the confidence intervals crossing the borders of clinical relevance and/or clinical significance (-2, imprecision). Publication bias was not assessed but might be present, yet was not a reason for downgrading.

The level of evidence regarding the outcomes time to seizure recurrence (2b) and retention rate (2c) could not be assessed with GRADE, as this outcome measure was not studied in the performed NMA.

The level of evidence regarding the outcome measure adverse events (2d) for add-on therapy was downgraded by 3 levels because of study limitations in half of the included studies (-1, risk of bias); heterogeneity in assessing and reporting adverse events through different studies (-1, inconsistency); and large variation in estimates (-1, imprecision).

A systematic review of the literature was performed to answer the following questions:

Subquestion 1: Which antiseizure medication as monotherapy is most effective and safe in patients with newly diagnosed generalized onset seizures?

P: Patients with newly diagnosed generalized onset seizures (with or without myoclonic seizures) and an indication for antiseizure treatment

I: antiseizure medication (ASM) monotherapy

C: other ASMs

O: Seizure incidence (seizure freedom, 50% seizure reduction), time to seizure recurrence, retention or discontinuation rate, adverse events

Subquestion 2: Which antiseizure medication as add-on therapy is the most effective and safe in patients with treatment resistant generalized onset seizures?

P: Patients with treatment resistant generalized onset seizures (with or without myoclonic seizures) and with an indication for antiseizure treatment

I: Add-on ASMs

C: Other ASMs, other doses, placebo, or no treatment

O: Seizure incidence (seizure freedom, 50% seizure reduction), time to seizure recurrence, retention or discontinuation rate, adverse events

Relevant outcome measures

The guideline development group considered seizure incidence and time to seizure recurrence as critical outcome measures for decision making; and retention/discontinuation rate and adverse events as important outcome measures for decision making.

A priori, the working group did not define the outcome measures listed above but used the definitions used in the studies. For instance, for seizure freedom this could mean number of patients being seizure free at a certain time point during follow-up, or for a certain duration during follow-up.

The working group defined a HR or RR ≤0.8 or ≥1.25 (dichotomous outcomes) and 10% (continuous outcomes) as minimal clinically (patient) important difference.

Search and select (Methods)

In 2012, the first version of this module was based on the NICE guideline (NICE, 2012). Studies from before this date were not included in the literature summary. After this date, a yearly search of the literature was conducted to include relevant studies.

For the current revision, the update of the NICE guideline from 2022 was used. In addition, we searched the databases Medline/Pubmed and Embase with relevant search terms from the January 1^st, 2019 (last search) until March 23^rd, 2022. The detailed search strategy is available upon request. The search was limited to systematic reviews (with or without meta-analysis) and randomized controlled trials. The systematic literature search resulted in 1563 hits. Studies were selected based on the following criteria:

• Systematic review (with detailed search strategy, risk of bias assessment and results of individual studies available) or randomized controlled trial;
• Described ASM as monotherapy or add-on therapy as intervention
• Described other ASMs, doses, placebo, or no treatment as comparison
• Described at least one of the outcome measures as described in the PICO
• Full-text English language publication
• Studies including ≥10 participants in each study arm

One single literature search was conducted for the effectiveness and safety of ASMs for focal onset, generalized onset, and unclassified onset seizures. For this module, 10 studies were initially selected based on title and abstract screening (one for monotherapy, and nine for add-on therapy). After reading the full text, one study was found for PICO 1 (monotherapy), and two studies were deemed suitable for inclusion for PICO 2 (add-on therapy). However, the study for PICO 1 was included in a recent Cochrane review from 2022, and the studies for PICO 2 were included in the Network Meta-analysis (NMA) performed in the 2022 update of the NICE guideline. Therefore, all studies were excluded (see the table with reasons for exclusion under the heading Evidence Tables); the Cochrane review and the 2022 update of the NICE guideline was used for the literature analysis.

Results

A recent large Cochrane review (for PICO 1) and the 2022 update of the NICE guidelines (for PICO 2) were used for the literature analysis.