Introduction

Historically, idiopathic generalized epilepsies (IGEs) have included juvenile myoclonic epilepsy (JME), childhood absence epilepsy (CAE), juvenile absence epilepsy (JAE), and generalized tonic–clonic seizures alone (GTCA) [1]. The term “idiopathic” refers to “self-originating” or “spontaneously arising” and implies that the condition is genetic [2]. Characterized by 2.5–6 Hz generalized spike waves, IGEs account for approximately 15–20% of all epilepsies, as well as 55% of newly diagnosed generalized epilepsy in children and adolescents [3, 4]. Because IGEs have a strong underlying genetic basis, the updated classification from the International League Against Epilepsy suggested that IGEs should be a subgroup of “genetic generalized epilepsies,” but reserved the term to describe the aforementioned four overlap** syndromes [1].

The diagnosis of IGEs has important implications for their treatment and prognosis. In patients with IGEs, development, neurological examinations, and radiographic results are typically normal [5]. Additionally, because most cases arise in children and adolescents, IGEs are often emphasized to be of pediatric importance only; however, considerable psychosocial symptoms—such as mood disorders, attention deficits, and learning disabilities—can be observed until adulthood [6, 7]. Long-term follow-up studies have revealed correlations between IGEs and outcomes such as poorer employment/financial conditions, decreased interactions with families, and unplanned pregnancies [8]. Thus, although IGEs may seem easier to manage than symptomatic or partial epilepsies, they should not receive less attention than these other epilepsies.

Antiseizure medications (ASMs) are the cornerstone of treatment for IGE syndromes. A good response rate, at 60–80% of seizure control (i.e., more than 1 year without seizure), can be achieved with appropriate ASM selection [9, 10]. First-line monotherapy controls symptoms in the majority of patients with IGEs. Among the first-line treatments, valproate (rather than lamotrigine or topiramate) monotherapy is the recommended first choice for IGEs in boys and men because it was shown to have better efficacy and tolerability in the SANAD study (Level I evidence) [11]. By contrast, levetiracetam monotherapy is favored in women able to have children; it seldom induces drug–drug interactions. Although levetiracetam was inferior to valproate in the SANAD II study [12], it has shown good efficacy in seizure control in cohort studies [13]. The choice of optimal initial monotherapy in IGEs is very important. Management decisions are different for JME, CAE, JAE, and CTSA, and need to be individualized. However, a limited number of randomized controlled trials (RCTs) have compared various ASMs head-to-head as initial monotherapies for IGEs and related subsyndromes. For example, some ASMs, such as carbamazepine or oxcarbazepine, may exacerbate absence seizures, whereas lamotrigine and gabapentin can exacerbate some myoclonic seizures [14]. A comprehensive integration of the evidence is thus needed so that a tailored plan can be developed for each patient.

Adjunctive therapy should be started when two different monotherapies have been unable to successfully control IGEs. The drug of choice generally depends on the main seizure subtype. Lamotrigine and levetiracetam are recommended as adjunctive therapies to valproate, except in JME. Topiramate, zonisamide, and perampanel have also demonstrated efficacy in RCTs or observational studies and are recommended adjunctive options (e.g., in myoclonic seizures). However, the majority of adjunctive medications do not have proof of efficacy in placebo-controlled RCTs. Hence, a comparison of the efficacy and tolerability of adjunctive ASMs remains lacking; clinically, the choice of adjunctive drug often relies on class III or IV evidence. Furthermore, with numerous established and new medications currently available, physicians face difficult decisions when choosing the most appropriate adjunctive drugs because of the limited high-quality evidence [15].

To the best of our knowledge, no previous review has compared the efficacy and tolerability of ASMs for IGEs (neither as monotherapies nor as adjunctive therapies). This network meta-analysis (NMA) aims to provide comprehensive evidence for the relative efficacy and safety of ASMs for controlling IGEs.

Methods

This NMA was conducted following a protocol that was prospectively registered with PROSPERO (No. CRD42022372358) and adhered to the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) statement for network meta-analysis [16].

Search strategy

Reviewers searched PubMed, Embase, and the Cochrane Library for relevant RCTs. Search terms were limited to the type of epilepsy, antiseizure medication, type of study, and year of publication (Appendix Table S1); there were no limitations on language. The reference lists of relevant RCTs and reviews were searched manually. The search procedure was conducted from December 2022 to February 2023, and EndNote X9 was used for reference management.

Study selection

In the first stage of review, two authors independently selected studies by screening the titles, abstracts, and content according to the inclusion and exclusion criteria. Differences in opinion were discussed to obtain consensus, as necessary; disagreements were arbitrated by the senior reviewer.

Inclusion criteria

(1) Population: patients of any age or sex who were diagnosed with IGEs, JME, CAE, JAE, or GTCA. (2) Intervention: monotherapy or adjunctive therapy with ASMs. (3) Outcomes: efficacy outcomes (the proportion of participants with seizure reduction or freedom after 1, 3, 6, or 12 months) and safety outcomes (the proportion of patients who experienced any treatment-emergent adverse event [TEAE], or serious TEAEs leading to discontinuation).

Exclusion criteria

(1) Patients with a diagnosis of another type of epilepsy. (2) Animal or cellular research. (3) Observational study or review article. (4) Incomplete outcomes with no explanation of clinical relevance.

Data extraction and quality evaluation

Data extraction was collected on standardized spreadsheets and double-checked. If multiple articles reported outcomes from the same population, the most comprehensive outcome was noted. When studies reported different terms of follow-up, all non-overlap** information was included. Version 2 of the Cochrane Collaboration Risk of Bias tool (RoB2) for assessing randomized trials was used to evaluate the included studies [17].

Statistical analysis

Statistical analysis was primarily conducted using R software (version 4.2.1, http://www.r-project.org). The gemtc package was used in JAGS 4.3.0 for the analysis (https://CRAN.R-project.org/package=R2jags). Pairwise meta-analyses were performed using a random-effects model for outcomes of the included studies to obtain odds ratios (ORs) and 95% confidence intervals (CIs). The NMA was conducted within a Bayesian framework that assumed a binomial likelihood for the number of events per medication [18, 19]. For outcomes with two or more treatment arms, the arms were pooled to form a single node for the corresponding ASM. The Markov chain Monte Carlo method was used to compare multiple ASMs by synthesizing the results of direct and indirect comparisons [20]. Each model used four Markov chains; the initial interaction value was set to 5000 and the adjusted interaction number was 10,000. The I2 statistic was calculated to quantify heterogeneity; I2 > 50% was defined as high-grade heterogeneity [21]. The local inconsistency model was assessed using a node-splitting method in which significance was set at a two-tailed p value of 0.05. The surface under the cumulative ranking (SUCRA) curves and the mean ranks were used to evaluate different ASMs, with a higher SUCRA representing superior efficacy.

Results

Identification and description of studies

Of the 2790 abstracts that were identified from PubMed, Embase, and the Cochrane Library, 113 were assessed for eligibility by full-text review, of which 87 were excluded. Finally, 28 RCTs containing 4282 patients were included in the NMA (Fig. 1).

Fig. 1
figure 1

Flow chart of data retrieval

Full size image

The demographic characteristics of patients, such as sex and age of entrance and onset, were noted. Detailed information is provided in Table 1. All 28 RCTs assessed the efficacy and safety of ASMs in IGEs or the related entities CAE, JAE, JME, or GTCA (arm of: valproate n = 14, lamotrigine n = 13, levetiracetam n = 6, perampanel n = 5, topiramate n = 3, ethosuximide n = 5, lacosamide n = 1). The number of patients assigned to each ASM, the initial and maximum dose of daily use, and the time of follow-up are shown in Table 2. Study designs—including region, blinding, conflict of interest, and register—were carefully checked to ensure the reliability of results (Table 3). Furthermore, RoB2 assessments were conducted to evaluate bias (Appendix Table S2).

Table 1 Characteristics of the Included Studies
Full size table
Table 2 Description of Intervention in the included studies
Full size table
Table 3 Design information of included studies
Full size table

A favorable consistency of the included studies was identified using the node-splitting method (all p \(>\) 0.05). Moreover, heterogeneity was low in the included studies (all I2 \(<\) 27%).

Efficacy outcomes

The included RCTs provided outcomes regarding the proportion of patients who achieved seizure freedom for 1, 3, 6, and 12 months after ASM treatment. The majority of studies reported the intention-to-treat population; intention-to-treat outcomes were thus analyzed rather than per-protocol outcomes. The network-evidence map plots of seizure-free outcomes for ASMs as monotherapies and adjunctive therapies are shown in Fig. 2A–F. Both short-term seizure-free outcomes (3–6 months) and relatively long-term outcomes (12 months) were analyzed.

Fig. 2
figure 2

Network of treatment comparisons for efficacy and safety. A Seizure free for 3–6 months after monotherapy in overall idiopathic generalized epilepsies (IGEs). B Seizure free for 3–6 months after adjunctive therapy in overall IGEs. C Seizure free for 12 months after monotherapy in overall IGEs (long-term follow-up). D Seizure free for 3–6 months after monotherapy in absence epilepsies (AE). E Seizure free for 3–6 months after adjunctive therapy in myoclonic epilepsies (ME). F Seizure free for 3–6 months after adjunctive therapy in generalized tonic–clonic seizures alone (GTCA). G Any adverse event after adjunctive therapy in overall IGEs. H Any adverse event after monotherapy in overall IGEs. I Adverse events leading to discontinuation after adjunctive therapy in overall IGEs. J Adverse events leading to discontinuation after monotherapy in overall IGEs (part 1). K Adverse events leading to discontinuation after monotherapy in overall IGEs (part 2). AE absence epilepsy, ME myoclonic epilepsy, IGEs idiopathic generalized epilepsies, GTCA generalized tonic–clonic seizures alone, VPA valproate, LTG lamotrigine, TPM topiramate, LEV levetiracetam, ESM ethosuximide, PER perampanel, LCM lacosamide, ad adjunctive

Full size image

The forest plot of the NMA revealed that all ASMs were associated with a higher rate of either short- or long-term seizure-free outcomes compared with placebo (Fig. 3). In the monotherapy analysis for overall IGEs, ethosuximide had a higher 3- to 6-month seizure-free rate than valproate (OR = 1.3, 95% CI = 0.66–2.8), whereas lamotrigine had a significantly lower rate than valproate (OR = 0.40, 95% CI = 0.23–0.77; Fig. 3A).

Fig. 3
figure 3

Forest plots of treatment comparisons for efficacy. A Seizure free for 3–6 months after monotherapy in overall idiopathic generalized epilepsies (IGEs). B Seizure free for 3–6 months after adjunctive therapy in overall IGEs. C Seizure free for 12 months after monotherapy in overall IGEs (long-term follow-up). D Seizure free for 3–6 months after monotherapy in absence epilepsies (AE). E Seizure free for 3–6 months after adjunctive therapy in myoclonic epilepsies (ME). F Seizure free for 3–6 months after adjunctive therapy in generalized tonic–clonic seizures alone (GTCA). AE absence epilepsy, ME myoclonic epilepsy, IGEs idiopathic generalized epilepsies, GTCA generalized tonic–clonic seizures alone, VPA valproate, LTG lamotrigine, TPM topiramate, LEV levetiracetam, ESM ethosuximide, PER perampanel, LCM lacosamide, ad adjunctive

Full size image

In the adjunctive therapy analysis, all ASMs showed superior efficacy to placebo; the effects of levetiracetam (OR = 7, 95% CI = 0.07–14) and topiramate (OR = 8.9, 95% CI = 1.9–39) were significant (Fig. 3B). There were no significant differences in long-term (12-month) follow-up outcomes between adjunctive valproate and adjunctive ethosuximide (OR = 1.3, 95% CI = 0.77–2.1), levetiracetam (OR = 0.82, 95% CI = 0.37–1.8), or topiramate (OR = 0.83, 95% CI = 0.39–1.7); however, adjunctive lamotrigine had significantly lower efficacy than adjunctive valproate (OR = 0.54, 95% CI = 0.37–0.8; Fig. 3C).

Subsyndromes of IGEs were also independently analyzed. In absence epilepsies, ethosuximide (OR = 3.1, 95% CI = 1.4–6.9) and valproate (OR = 2.4, 95% CI = 1.1–4.3) had significantly superior efficacy to lamotrigine as monotherapies (Fig. 3D). However, in the analysis of adjunctive therapies in myoclonic epilepsies (Fig. 3E) and GTCA (Fig. 3F), there were no significant differences between ASMs and placebo, likely because the 95% CIs were very broad.

Safety outcomes

In overall IGEs, adjunctive lamotrigine (OR 4.4, 95% CI = 1.0–24) had a significantly increased risk of any TEAEs compared with adjunctive placebo (Fig. 4A). There were no significant differences in safety between ASMs as either adjunctive therapies (Fig. 4A) or monotherapies (Fig. 4B).

Fig. 4
figure 4

Forest plots of treatment comparisons for safety. A Any adverse event after adjunctive therapy in overall idiopathic generalized epilepsies (IGEs). B Any adverse event after monotherapy in overall IGEs. IGEs idiopathic generalized epilepsies, VPA valproate, LTG lamotrigine, TPM topiramate, LEV levetiracetam, ESM ethosuximide, PER perampanel, LCM lacosamide, ad adjunctive

Full size image

SUCRA

According to the SUCRA, the efficacy ranking for monotherapies was ethosuximide > valproate > topiramate > placebo > lamotrigine in overall IGEs, and the efficacy ranking for adjunctive therapies was topiramate > levetiracetam > lacosamide > perampanel > lamotrigine > placebo. For 12-month seizure-free efficacy, the ranking was ethosuximide > valproate > topiramate > levetiracetam > lamotrigine (Fig. 5A–C and Appendix Table S2A–C). In absence epilepsies, the SUCRA efficacy ranking for monotherapies was ethosuximide > valproate > placebo > lamotrigine. In myoclonic seizures, the efficacy ranking for adjunctive therapies was levetiracetam > lamotrigine > perampanel > placebo. Moreover, for GTCA, the efficacy ranking for adjunctive therapies was topiramate > lacosamide > perampanel > lamotrigine > placebo (Fig. 5D–F and Appendix Table S3D–F).

Fig. 5
figure 5

Ranking of efficacy outcomes according to the surface under the cumulative ranking curve (SUCRA). A Seizure free for 3–6 months after monotherapy in overall idiopathic generalized epilepsies (IGEs). B Seizure free for 3–6 months after adjunctive therapy in overall IGEs. C Seizure free for 12 months after monotherapy in overall IGEs (long-term follow-up). D Seizure free for 3–6 months after monotherapy in absence epilepsies (AE). E Seizure free for 3–6 months after adjunctive therapy in myoclonic epilepsies (ME). F Seizure free for 3–6 months after adjunctive therapy in generalized tonic–clonic seizures alone (GTCA). AE absence epilepsy, ME myoclonic epilepsy, IGEs idiopathic generalized epilepsies, GTCA generalized tonic–clonic seizures alone, VPA valproate, LTG lamotrigine, TPM topiramate, LEV levetiracetam, ESM ethosuximide, PER perampanel, LCM lacosamide, ad adjunctive

Full size image

In overall IGEs, the SUCRA ranking of associations with more total TEAEs for adjunctive therapies was placebo > levetiracetam > perampanel > lamotrigine > lacosamide; for monotherapies, it was perampanel > placebo > lamotrigine > valproate > levetiracetam > ethosuximide (Fig. 6A, B and Appendix Table S3G-H). For serious TEAEs leading to discontinuation, the safety ranking was placebo > perampanel > topiramate > levetiracetam > lacosamide > lamotrigine for adjunctive therapy (Fig. 6C and Appendix Table S3I) and topiramate > valproate > ethosuximide > lamotrigine; placebo > levetiracetam > perampanel for monotherapy (Fig. 6D–E and Appendix Table S3J–K).

Fig. 6
figure 6

Ranking of safety outcomes according to the surface under the cumulative ranking curve (SUCRA). A Any adverse event after adjunctive therapy in overall idiopathic generalized epilepsies (IGEs). B Any adverse event after monotherapy in overall IGEs. C Adverse event leading to discontinuation after adjunctive therapy in overall IGEs. D Adverse event leading to discontinuation after monotherapy in overall IGEs (part 1). E Adverse event leading to discontinuation after monotherapy in overall IGEs (part 2). AE absence epilepsy, ME myoclonic epilepsy, IGEs idiopathic generalized epilepsies, GTCA generalized tonic–clonic seizures alone, VPA valproate, LTG lamotrigine, TPM topiramate, LEV levetiracetam, ESM ethosuximide, PER perampanel, LCM lacosamide, ad adjunctive

Full size image

Discussion

Our NMA indicated that all of the included ASMs were more effective than the placebo. The network forest plots compared monotherapies with valproate and adjunctive therapies with placebo. Significant superiority was identified for adjunctive levetiracetam and topiramate, while inferiority was identified for lamotrigine monotherapy. Non-significant differences were also identified. Using SUCRA, rankings of efficacy and tolerability were summarized.

The analyses of the efficacy outcomes of being seizure free for 3–6 or 12 months did not affect the status of valproate as the first-choice monotherapy for overall IGEs without contraindications. Although ethosuximide ranked first according to SUCRA, its optimal efficacy and tolerability are probably only favorable for absence epilepsies rather than for overall IGEs, and especially tonic–clonic seizures [49]. It is supported as the drug of choice for absence seizures without other seizure types, in accordance with other reviews and guidelines (April 2022, NICE guidelines, https://www.nice.org.uk/guidance/ng217).

Lamotrigine monotherapy unexpectedly had the lowest efficacy in both short- and long-term seizure-free outcomes in the SUCRA analysis and showed significant inferiority in the forest plots for overall IGEs. Although a longer duration of titration may partly account for this short-term result (because lamotrigine must be titrated very slowly to avoid unwanted side effects), this finding is in accordance with a SANAD study suggesting that lamotrigine should not be interpreted as a “broad spectrum” antiseizure medication because it provides worse seizure control than valproate or topiramate in generalized epilepsies [11]. However, these findings should be interpreted with caution. Adjunctive lamotrigine is advantageous in controlling unclassified generalized tonic–clonic seizures (i.e., those unable to be classified as IGEs or partial epilepsy) [50].

In recent years, levetiracetam has been increasingly prescribed and recommended. The present NMA supports its efficacy as a second-line monotherapy and as an ideal adjunctive choice in overall IGEs according to its efficacy ranking. Although it was not found to be a non-inferior monotherapy to valproate in a previous study [12], its favorable efficacy, fast action, and good tolerability (leading to less TEAEs than placebo in our ranking) indicate its considerable potential. However, longitudinal studies are needed in the future (for both levetiracetam and brivaracetam).

In the present study, adjunctive topiramate ranked first in adjunctive therapies according to SUCRA. As a second-generation ASM, topiramate is especially effective in JME and GTCA [51]. In a Cochrane review, the efficacy of topiramate monotherapy in JME was not significantly different from that of valproate (the current drug of choice) [51]. Although topiramate is associated with cognitive TEAEs such as dulling or memory problems [52], which are especially unfavorable in neurodevelopmental disorders, its tolerability was the best ranked in terms of TEAEs leading to discontinuation.

The head-to-head comparison of third-generation ASMs used as adjunctive therapies is of great importance because there is a lack of accumulated evidence, especially for perampanel and lacosamide. In the present analysis, SUCRA demonstrated that their efficacies seem to fall between those of levetiracetam and lamotrigine. In addition, lacosamide may be more effective than perampanel for seizure-free outcomes in GTCA. Although there was previously a lack of high-quality long-term evidence, recent unblinded controlled studies have revealed that perampanel reduces generalized seizures with a median of 90% in 52-week follow-up, and has the potential to increase seizure freedom [42, 53, 54]. Moreover, after 24 weeks of lacosamide adjunctive treatment, the freedom rate from generalized seizures was 27.5% (versus placebo 13.2%) in an RCT [44].

In the current study, perampanel was the best-ranked therapy for tolerability as both a monotherapy and an adjunctive therapy when any adverse event was considered. The characteristic TEAEs of perampanel are irritability and aggression [52, 55]. In contrast, when ranking the therapies in terms of serious TEAEs leading to discontinuation, perampanel was inferior to placebo and levetiracetam. Similarly, the PERMIT study indicated a discontinuation rate of 17.6% at 12 months, in which psychiatric TEAEs were the most common reason for discontinuation [56]. We thus predict a future in which new-generation ASMs, like perampanel, are used to control generalized seizures. However, more high-quality research is warranted to draw stronger conclusions.

We must note that, although all included ASMs significantly improved the seizure-free rate compared with placebo, ASMs can neither cure epilepsies nor treat the underlying pathology that causes them; they merely aim to stop the occurrence of seizure symptoms. A better understanding of the molecular mechanisms underlying the pathogenesis, epileptogenesis, and pharmacoresistance of epilepsies is needed to change our clinical approach. For example, genetic therapies and stem cell therapies will likely cure epilepsies in the future [57].

Despite this, the importance of ASMs should not be ignored, although many patients do not achieve or retain complete seizure freedom. Improved seizure reduction may significantly downgrade the risk of injury and unexpected death [58]; however, in the present review, the NMA of the seizure reduction rate had to be stopped because insufficient data were provided in the included studies. In addition, although some well-controlled complex epilepsies might be disrupted by a single poor night’s sleep or missing dose and breakthrough, substantial improvements have been achieved [59]. Compared with invasive options like vagus nerve stimulation or corpus callosotomies, the use of established and new ASMs may provide more tolerable, incremental benefits. Furthermore, the increase in available ASMs makes it possible to devise more individualized plans, thus benefitting patients. Longitudinal comprehensive studies are therefore warranted to evaluate efficacy in particular populations (or genotypes), more in-detail tolerability, effects on quality of life, and cost-utility for ASMs.

The present study had some limitations. Methodologically, a limited number of outcomes restrained us from analyzing other important efficacy outcomes, such as seizure reduction or electroencephalogram improvements. Furthermore, because specific TEAEs were not evaluated, the tolerability outcome analysis lacked details, and only rough results were obtained because of a lack of information. The search strategy mainly focused on the idiopathic generalized seizure type. Thus, some important ASMs (such as cenobamate, brivaracetam, etc.) most frequently used in focal epilepsies, although recently proven adjunctive use in generalized seizures, were not involved. More meaningful future studies are necessary to elucidate the efficacy of these ASMs.

Although low heterogeneity was identified according to the I2 test, differences between RCTs existed such as the inclusion criteria, time of treatment, and concomitant drugs. Furthermore, although a statistically suitable and well-known analysis was used, certain overestimations or underestimations may still exist. For example, relatively broad 95% CIs were obtained because relatively few RCTs were included. Further studies are therefore required to further confirm our conclusions.

Conclusions

Among the included ASM monotherapies, valproate ranked best for overall IGEs in efficacy and was the third best in tolerability. For the adjunctive therapies, topiramate ranked best for GTCA and overall IGEs, whereas levetiracetam ranked best for myoclonic seizures. Moreover, perampanel ranked best in tolerability measured by any TEAE when used either as a monotherapy or an adjunctive therapy. Overall, valproate is recommended as the monotherapy of choice for overall IGEs without contraindications. However, our results should be interpreted with caution considering the limited available information and the inherent methodological limitations of the NMA.