Indirect treatment comparison of cenobamate to other ASMs for the treatment of uncontrolled focal seizures

OBJECTIVE
The efficacy and safety of cenobamate relative to other antiseizure medications (ASMs) has not been evaluated. An indirect treatment comparison (network meta-analysis) was performed to determine if adjunctive cenobamate increases the odds ratio (OR) for ≥50% responder rate or for withdrawals due to treatment-emergent adverse events (TEAEs) leading to ASM discontinuation versus adjunctive therapy with other ASMs.


METHODS
A systematic literature review was conducted to identify randomized, double-blind, placebo-controlled trials (maintenance phase ≥ 12 weeks) assessing adjunctive ASMs in adults with uncontrolled focal seizures. Cenobamate was compared to a group of seven other ASMs, and to subgroups of branded (brivaracetam, eslicarbazepine acetate, lacosamide, and perampanel) and older (lamotrigine, levetiracetam, and topiramate) ASMs at FDA-recommended daily maintenance doses (FDA-RDMD), at all doses, and at maximum and minimum daily doses. Statistical significance was set at p < 0.05.


RESULTS
Twenty-one studies were eligible for analysis. The placebo-adjusted ≥ 50% responder rate for FDA-RDMD of cenobamate was superior (OR 4.200; 95% CI 2.279, 7.742) to FDA-RDMD of all seven assessed (OR 2.202 95% CI 1.915, 2.532; p = 0.044) and branded ASMs (OR 2.148; 95% CI 1.849, 2.494; p = 0.037). There was no significant difference for ≥50% responder rate between FDA-RDMD of cenobamate and FDA-RDMD of older ASMs (OR 2.617; 95% CI 1.767, 3.878; p = 0.202). No significant differences were identified for ≥50% responder rate when comparing all doses and maximum/minimum doses of cenobamate to all seven, branded, and older ASMs. Cenobamate demonstrated comparable TEAE withdrawal rates to all seven ASMs, branded ASMs, and older ASMs across each of the four dose ranges (all p > 0.05).


SIGNIFICANCE
Patients receiving FDA-RDMD of cenobamate were more likely to have ≥50% seizure reduction compared with FDA-RDMD of the seven assessed ASMs and branded ASMs, without an increase in treatment discontinuation due to TEAEs.


Introduction
The efficacy of cenobamate, an oral antiseizure medication (ASM) approved in the United States (XCOPRI, SK Life Science, Inc.) for the adjunctive treatment of uncontrolled focal epilepsy, was established in two randomized, double-blind, placebocontrolled studies (Chung SS, et [1][2][3]. Adults with focal seizures despite receiving one to three ASMs had significant median reductions in seizure frequency versus placebo with cenobamate 200 mg/day (C013) and 100, 200, and 400 mg/day (C017) during the double-blind treatment periods (C013: 12 weeks; C017: C017: 12 weeks), 50% responder rates with cenobamate 100, 200, and 400 mg/day and 100% responder rates (seizure freedom) with cenobamate 200 and 400 mg/day were significantly greater than placebo [2][3][4]. Safety data from the above studies and a long-term open-label study (Sperling MR, et al [C021;NCT02535091]) reported the majority of treatment-emergent adverse events (TEAEs) to be mild to moderate in severity, with the most common being central nervous system-related events [2,3,5].
Direct head-to-head comparative efficacy and safety studies of ASMs are rarely conducted [6]. While head-to-head studies are preferred to guide clinical decision-making, these studies are costly and not required for ASM regulatory approval [6,7]. In the absence of direct comparison studies, an indirect treatment comparison (ITC), by means of a meta-analysis based on a network of clinical trials, can identify how ASMs measure up to each other, thereby providing valuable information to guide appropriate prescribing in clinical practice [6][7][8]. This is accomplished by combining data from studies that have at least one treatment arm in common, usually the placebo arm [9].
A properly designed ITC is conducted with an appropriate model, taking into account the heterogeneity among combined studies [9]. Ideally only clinically and methodologically homogenous studies with similar confounding variables affecting the relative treatment effect are included [9]. Odds ratios (OR), relative risks, or risk differences are commonly used as effect measures for ITCs with binary outcomes [9]. In addition, the assumptions made to conduct the ITC are clearly specified and checked with a sensitivity analysis [8,9].
Prior ITCs of adjunctive ASMs for the treatment of uncontrolled focal seizures have evaluated 50% responder rates and discontinuation rates due to TEAEs. A recent review of network metaanalyses that included ITCs and mixed treatment comparisons (studies of ASMs versus placebo and studies of one ASM versus another) found that overall few significant differences have been identified between ASMs for efficacy or safety [7].
A key consideration when comparing ASMs is the influence of ASM dose on the balance between efficacy and safety. Patients may not be able to receive a fully effective dose if an adverse effect leads to ASM discontinuation [10]. Dose-titration is often used to reduce the risk of adverse events and increase tolerability, allowing for the identification of the most appropriate maintenance dose for treatment success [10]. It also permits the evaluation of the doses of concomitant ASMs and the possibility of minimizing the total ASM load of the regimen [11].
The efficacy and safety of cenobamate relative to other ASMs has not been evaluated. This ITC was conducted to determine if treatment with adjunctive cenobamate increases the 50% responder rate or withdrawal due to TEAEs compared to adjunctive therapy with a group of seven ASMs in adults with uncontrolled focal seizures during maintenance therapy. Comparisons were conducted between cenobamate and a group of seven other ASMs, then separate comparisons were conducted for cenobamate versus the subgroups of branded (4/7 medications) and older (3/7 medications) ASMs. Comparisons were based on all daily doses, FDArecommended daily maintenance doses, and maximum and minimum daily doses.

Data sources and search
Trials were retrieved from MEDLINE (PubMed), Cochrane Library, and clinicaltrials.gov. Search terms for PubMed included: epilepsy and randomi* with brivaracetam, cenobamate, eslicarbazepine, lacosamide, lamotrigine, levetiracetam, perampanel, or topiramate. The seven ASMs were selected for the ITC (PROSPERO registration number CRD42021239069) to cenobamate with the purpose of including a combination of branded ASMs (brivaracetam, eslicarbazepine acetate, lacosamide, and perampanel) and older ASMs (lamotrigine, levetiracetam, and topiramate). Search terms for the Cochrane Central Register of Controlled Trials (CEN-TRAL) included: generic or brand names for the ASMs listed above and focal or 'partial-onset' or epilepsy. The search strategy for clinicaltrials.gov included interventional studies (clinical trials) with a status of completed, withdrawn, or unknown for the condition/disease of epilepsy, partial and the other terms were the generic names for the ASMs listed above. Records were loaded and managed with Distiller SR software (Evidence Partners, Ottawa, CA).

Study selection
Two independent reviewers (FFR, VFS) systematically reviewed the clinical trials for eligibility in the study network using PRISMA standards, first based on title review, followed by abstract review, and then by full-text review. Reviewers had to agree on whether a study should be included or excluded. Reviewers were able to link records that referred to the same study during the full-text review. Clinical trials were eligible to be included in the study network for analysis if they met the following criteria: (1) randomized, doubleblind, placebo-controlled trial with a maintenance phase at least 12 weeks in duration; (2) included adult patients diagnosed with focal seizures whose seizures failed at least one prior ASM; (3) had at least one treatment arm including adjunctive therapy with brivaracetam, eslicarbazepine acetate, lacosamide, lamotrigine, levetiracetam, perampanel, topiramate, or cenobamate, as well as a placebo-control treatment arm; (4) reported the 50% responder rate in the full journal publication or on clinicaltrials.gov. The rationale for selecting studies with a maintenance phase of 12 weeks was that payers and health technology assessment agencies generally discount the relevance of randomized, controlled trials with shorter maintenance phases. There were no restrictions based on language or publication year. Flexible-dose trials, trials evaluating monotherapy, trials conducted in pediatric populations, and trials not including at least one site in the Americas or Europe were excluded.

Data extraction
The two reviewers independently extracted the data from each clinical trial and came up with a consensus for any discrepancies. The following data were extracted from each eligible clinical trial: comparator, lead trial author, inclusion and exclusion criteria, titration, and titration phase. Demographics, duration of epilepsy, seizure frequency per 28 days, number of concomitant ASMs, sample size, study drug doses, 50% responder rate, TEAEs, serious TEAEs, and study discontinuation due to TEAEs were extracted for each treatment arm.

Outcomes
Response to treatment during the maintenance phase of therapy was defined as the proportion of patients who demonstrated 50% responder rate in the intention-to-treat (ITT), modified ITT (m-ITT), or per-protocol (PP) population. Safety was determined by withdrawals due to TEAEs during the maintenance phase of therapy.

Statistical analyses
Risk of bias was estimated for each study with the Cochrane Risk of Bias v2.0 Tool [12] available in Distiller SR, using the following independent domains: random sequence generation, allocation concealment, incomplete outcome data, double-blinding of patients and study personnel, blinding of outcome assessors, and selective reporting. Selective publication bias was visually assessed using funnel plots and Egger's method was applied, with adjustment for missing studies, for a quantitative estimate of any potential publication bias [13][14][15].
The OR (95% CI) of achieving a 50% responder rate or withdrawal due to TEAEs for each ASM in each clinical trial was determined and used in the ITC to cenobamate [9]. ASMs were compared using inter-treatment comparison meta-analytic methods. ASM specific estimates, based on combined trial arms, were contrasted with placebo, and combined placebo-adjusted effects for the specific ASM groups and doses were compared using Bucher's method [9]. Placebo responders and nonresponders were split equally among comparator study arms. Only the first sequence of cross-over trials was included.
Random-effects models were used for all comparisons, as a conservative approach, regardless of heterogeneity levels. Random effects estimates were tested at the 95% significance level, twotailed. Inter-group/treatment comparisons were handled using Bucher's method and frequentist heterogeneity testing [8,9].
Statistical significance for the ITC was set at p < 0.05. Sensitivity analyses for the 50% responder rate included meta-regressions or stratified analyses by dose, median seizure frequency during baseline period, mean epilepsy duration, and mean age. All statistical analyses were performed with Comprehensive Meta-Analysis 2.0 (Englewood, NJ, US) and conducted by an expert pharmacoepidemiologist (FFR) with Cochrane Group certification.

Study and patient characteristics
Eight hundred fifty-three records were screened for eligibility after duplicates were removed from the 1042 records identified from the initial search ( Fig. 1). Forty full-text study publications were then evaluated based on the inclusion and exclusion criteria and 21 were eligible to be in the study network. Only one of two randomized, double-blind, placebo-controlled studies evaluating cenobamate met the inclusion criteria for this ITC, as the maintenance phase for one (6 weeks) did not meet criteria.
The funnel plot for the risk of publication bias was symmetrical, but showed some evidence that smaller sized studies demonstrating efficacy versus placebo were over-represented. However, imputing virtually missing studies did not influence the result of the meta-analysis (Fig. 2) [13][14][15].

Safety
The forest plot for withdrawal rates due to TEAEs for all ASMs is depicted in Fig. 3B [3,17,[19][20][21][22]24,25,[30][31][32][33][34][35][36][37][38][39][40][41][42][43]. The combined OR for withdrawal rates due to TEAEs for all ASMs in the study network was 2.335 (95% CI 1.901, 2.869) (Fig. 3B). The combined ORs for the withdrawal rates due to TEAEs were greater than 1 for all doses of cenobamate, all doses of the seven other ASMs, and all doses of the subgroups of branded and older ASMs; however, the 95% CI for FDA-recommended doses of older ASMs, maximum and minimum doses of cenobamate, and minimum doses of other ASMs, branded ASMs, and older ASMs fell below 1. Cenobamate demonstrated comparable TEAE withdrawal rates to all seven ASMs, branded For median baseline seizure frequency, frequency per month was considered to be equal to frequency per 28 days and frequency per week multiplied by 4 to convert to frequency per 28 days. Placebo group responders and nonresponders and patients discontinuing and not discontinuing due to TEAEs were split equally among comparator study arms for the analysis. a Eslicarbazepine acetate 1200 mg/day group had a 12-week maintenance. b 24-weeks total for up-titration and maintenance. ASMs, and older ASMs across each of the four dose ranges (all p > 0.05) (Fig. 4B). All I 2 values were below 25%.

Sensitivity analyses
Sensitivity analyses indicated that the 50% responder rate did not vary significantly among trials based on mean age, seizure frequency during baseline period, or epilepsy duration (p = 0.795, p = 0.753, p = 0.301, respectively). Sensitivity analyses did reveal a dose-dependent effect on the 50% responder rate. Trial arms with maximum ASM doses had higher responder rates (p = 0.019), and arms with minimum doses had lower responder rates (p = 0.010). Trial arms for FDA-recommended maintenance doses did not differ from other arms for responder rate (p = 0.831).

Discussion
FDA-recommended daily maintenance doses of cenobamate were superior in achieving a 50% responder rate for uncontrolled focal seizures compared with FDA-recommended daily maintenance doses of all seven other ASMs and branded ASMs, but not FDA-recommended daily maintenance doses of the subgroup of older ASMs, in this ITC. Cenobamate did not differentiate on 50% responder rate at its lowest or maximum doses (100 or 400 mg/day), or when all study doses were combined, compared to any set of comparators. In general, cenobamate demonstrated comparable withdrawal rates due to TEAEs to all seven other ASMs across most dose ranges. This is the first ITC among the more recent ITCs using ORs and comparing ASMs based on all daily doses, FDA-recommended daily maintenance doses, maximum daily doses, and minimum daily doses to demonstrate such results. Prior examples of ITCs that compared ASMs using ORs to measure the effect sizes had mixed results. Brigo et al. did not find a statistically significant difference between the combined OR for adjunctive brivaracetam compared to the combined OR for lacosamide, eslicarbazepine acetate, or perampanel at maximum daily recommended and minimum doses of these ASMs for the 50% responder rate and withdrawal rates due to TEAEs [44]. Gao et al. also did not find a difference between adjunctive brivaracetam, carisbamate, eslicarbazepine acetate, ezogabine, lacosamide, or perampanel when the combined OR for each ASM was compared against the OR for all the other ASMs combined, without differentiating among the ASM doses, for the 50% responder rate or discontinuation rate due to TEAEs [45].
In contrast, Costa et al. did find differences in these outcomes. In the Costa et al. ITC, like that of Gao et al., differentiation between minimum or maximum daily doses of the ASMs was not carried out. The combined OR for the 50% responder rate for topiramate was significantly greater than the combined ORs for all other ASMs in their analysis (eslicarbazepine acetate, gabapentin, lacosamide, lamotrigine, levetiracetam, oxcarbazepine, pregabalin, tiagabine, and zonisamide) [46]. The combined OR for withdrawals due to TEAEs for levetiracetam was significantly less than the combined ORs for all the other ASMs [46]. Of note the Costa et al. ITC included studies with pediatric patients down to age 2 and at least 8 weeks of follow-up; thus, their analysis included some studies that differed from Gao et al. and Brigo et al [46].
Comparing the ASMs in all four dose categories allowed for a combination of objective and subjective measures. Objectivity was based on all doses, FDA-recommended daily maintenance doses, and maximum and minimum ASM doses based on prescribing information or the network studies. Subjective dosing due to unclear language in the prescribing information (e.g., brivaracetam) or due to dosing recommendations not in alignment with those used in clinical trials or currently used in clinical practice (e.g., lacosamide [23][24][25]47] and topiramate) were also captured in the ITC [29][30][31]42,47].
Strengths of this ITC include the low impact of publication bias and the use of random effects models. The internal validity of an ITC is contingent on several factors, including appropriate identification of studies from a systematic review that comprise the evidence network, the quality of the randomized controlled trials selected, the similarity of the trials' methodologies, the appropriateness of the combination models, and the extent and control of confounding biases [8,9]. All I 2 values were below 25%, which is considered not important, and the funnel plot for publication bias was symmetrical [9]. Moreover, the sensitivity analyses increase our confidence that the ASM comparisons for the 50% responder rate are not due to differences in the study populations across the included trials. Sensitivity analyses based on age, seizure frequency during baseline period, and epilepsy duration were not significant.
Although the ultimate goal of ASM therapy is to achieve seizure freedom, most trials of ASMs have not evaluated this as a primary efficacy endpoint given current regulatory recommendations for study design [7,49]. This creates a challenge for evaluating seizure freedom as an outcome in an ITC because the individual studies are not powered to detect statistical differences in seizure freedom between treatments [7]. Given the scarcity of seizure freedom in this context and the variability in how seizure freedom is defined and evaluated in ASM studies, the ability of an ITC to detect the true effect of this outcome is unclear [9,49]. Numerous trials in this network had zero placebo patients achieving seizure freedom, requiring statistical adjustment, and large confidence intervals, both of which would limit the sensitivity to detect differences among ASMs [7][8][9]. Moreover, only one trial for cenobamate met the inclusion criteria for the ITC (maintenance phase 12 weeks) and had high variability in the seizure freedom estimate among the 4 treatment arms, further limiting the ability to detect differences [9]. A more consistent approach to reporting seizure freedom is needed in order to make precise comparisons among ASMs for this outcome [49].
In closing, this ITC demonstrates that in patients with uncontrolled focal seizures, FDA-recommended daily maintenance doses of cenobamate approximately doubled the odds of 50% seizure reduction compared with FDA-recommended daily maintenance doses of seven other ASMs, and the subgroup of the four ASMs that are branded, with similar withdrawal rates due to TEAEs.