Real‐world effectiveness of cenobamate in drug‐resistant and ultra‐refractory epilepsy: A multicenter study Authors: Ondřej Strýček, Jana Amlerová, Martin Pail, Jitka Kočvarová, Jana Slonková, Martin Šimčík, Kateřina Stanzelová, Monika Šibalová, Petr Marusič, Milan Brázdil Published: 2026 ABSTRACT AbstractObjectiveTo evaluate the real‐world effectiveness, safety, and treatment retention of cenobamate (CNB) in patients with drug‐resistant epilepsy (DRE), including ultra‐refractory epilepsy (URE), high seizure burden, and extensive prior treatment history.MethodsWe conducted a retrospective, multicenter study of 500 adults with DRE (focal or combined epilepsy) who were treated with CNB at three Czech tertiary epilepsy centers. Among these patients, 83.4% met criteria for URE. Seizure outcomes, treatment retention, and adverse events were analyzed, with stratification by treatment duration and cumulative ASM exposure (number of previous and concomitant anti‐seizure medications). Primary outcomes were ≥ 50% responder rate and seizure freedom at 12 months.ResultsAmong 436 patients with evaluable seizure data after ≥ 3 months of CNB treatment, 55.7% achieved ≥ 50% seizure reduction and 16.5% became seizure‐free. At 12 months (n = 331), ≥ 50% response was achieved in 61.6%, and 18.1% became seizure‐free. Among patients with focal to bilateral tonic–clonic seizures or bilateral tonic–clonic seizures, 46.1% achieved seizure freedom at 12 months. CNB efficacy persisted across high cumulative ASM burden, with seizure freedom in 22.7%, 20.2%, and 10.4% of patients with ≤ 5, 6–10, and > 10 ASMs, respectively. Multivariable analyses identified high baseline seizure burden, greater cumulative ASM exposure (> 10 ASMs), and concomitant use of sodium channel blockers as independent negative predictors of treatment response. Treatment retention at 12 months was 83.0%.SignificanceCNB was effective and well‐tolerated in a highly treatment‐resistant population. Although seizure freedom rates declined with increasing cumulative ASM burden, CNB retained clinically meaningful efficacy even in highly refractory patients.Plain Language SummaryThe study examined the effectiveness of cenobamate in routine clinical practice in adults with highly treatment‐resistant epilepsy. In a large multicenter cohort, many patients achieved substantial seizure reduction, including complete seizure freedom, despite extensive prior treatment. Most patients remained on cenobamate during follow‐up, consistent with good effectiveness and tolerability in routine clinical practice. These findings indicate that cenobamate can offer meaningful benefits even in very difficult‐to‐treat epilepsy and may inform treatment decisions in clinical practice.Keywords: cenobamate, drug‐resistant epilepsy, real‐world data Key points Cenobamate effectiveness was assessed in a large real‐world multicenter cohort of adults with drug‐resistant epilepsy. Cenobamate effectiveness was assessed in a large real‐world multicenter cohort of adults with drug‐resistant epilepsy. High responder and seizure‐freedom rates were observed, including in patients with a high cumulative seizure burden. High responder and seizure‐freedom rates were observed, including in patients with a high cumulative seizure burden. Among patients with tonic–clonic seizures, 46.1% were seizure‐free from this seizure type at 12 months. Among patients with tonic–clonic seizures, 46.1% were seizure‐free from this seizure type at 12 months. Long‐term retention exceeded 80% at 12 months, indicating durable efficacy and good tolerability. Long‐term retention exceeded 80% at 12 months, indicating durable efficacy and good tolerability. INTRODUCTION Drug‐resistant epilepsy (DRE) remains a major therapeutic challenge, with approximately one‐third of patients continuing to have seizures despite treatment with multiple anti‐seizure medications (ASMs). 1 , 2 While epilepsy surgery and neuromodulation provide important alternatives for selected individuals, many patients are not candidates for these interventions or fail to achieve adequate seizure control. Within the broad DRE population, a particularly severe subgroup—ultra‐refractory epilepsy (URE)—has been defined by persistent seizures despite at least six different therapeutic approaches, including well‐tolerated ASM trials, surgery, and neuromodulation such as vagus nerve stimulation (VNS) or deep brain stimulation (DBS). 3 Cenobamate (CNB) is a novel ASM with dual mechanisms of action that has shown substantial efficacy in randomized controlled trials (RCTs). In pivotal studies, seizure freedom was achieved in 21%–28% of patients after 12 weeks of treatment—considerably higher than rates reported for previous new ASMs. 4 , 5 Open‐label extension studies confirmed sustained seizure control, with retention rates of 67%–71% at 2 years and long‐term seizure freedom maintained in a notable proportion of patients beyond 4 years. 6 , 7 Real‐world evidence has confirmed these findings. Across observational cohorts, ≥ 50% responder rates typically range from 55% to 63%, and seizure freedom is achieved in 13%–16% of patients. When CNB is introduced earlier in the treatment course, seizure freedom rates approach 20%. Importantly, high retention rates (80%–87% at 6–12 months) have been reported even in cohorts with a median of 10–12 previous ASM failures. 3 , 8 , 9 , 10 , 11 , 12 Analyses consistently indicate that fewer previous treatment failures and lower baseline seizure burden predict better outcomes. 10 Nevertheless, CNB has demonstrated clinically relevant efficacy even in ultra‐refractory epilepsy, where ≥ 50% responder rates exceed 70%, and a meaningful portion achieve seizure freedom. 3 , 13 Taken together, evidence from both randomized trials and real‐world studies establishes CNB as one of the most effective currently available therapies for focal DRE. METHODS Study design and patient selection We conducted a retrospective, multicenter observational study of adult patients (≥ 18 years) with DRE who initiated adjunctive CNB treatment between January 2023 and November 2024 across three epilepsy centers in the Czech Republic. Patients were identified through a review of pharmacy records and electronic medical charts. Only patients diagnosed with focal, combined, or unknown‐onset epilepsy were included in the study; no patients with confirmed generalized epilepsy were enrolled. In addition, four patients who had previously participated in CNB clinical trials between 2017 and 2022 and subsequently transitioned to standard clinical follow‐up were included. All patients who initiated CNB were included, regardless of treatment duration or outcomes, to capture real‐world tolerability and effectiveness. All patients had previously been treated with at least two ASMs, either sequentially or concomitantly, prior to CNB initiation, in accordance with current indication criteria for CNB. One patient, however, initiated CNB as monotherapy after withdrawal of the previous ASM; this off‐label use was retained to reflect real‐world practice. CNB was initiated following a standard titration protocol. However, this schedule could have been prolonged, allowing for better tolerability if deemed necessary by treating physicians. This led to individualized maintenance doses (maximum 400 mg/day). All patients provided general written consent for the use of their anonymized medical data for research and publication purposes upon registration for care. Due to the retrospective design of the study and the use of anonymized data, additional study‐specific informed consent was not required. Data collection Baseline demographic and clinical data were collected, including age, sex, epilepsy type and duration, baseline seizure frequency, number of failed and concomitant ASMs, and prior epilepsy surgeries or neuromodulation therapies. Neuromodulation therapy included both active and historical (inactive or explanted) VNS or DBS devices. Patients with ≥ 20 seizures per month at baseline were classified as having highly active epilepsy. In addition to this group, independently of seizure frequency, a subgroup of patients with URE, defined by failure of six or more prior interventions (ASMs, surgery, or VNS/DBS), was analyzed. Moreover, a subgroup of URE, referred to as “stringently defined URE”, defined as failure of ≥ 4 ASMs combined with unsuccessful surgery and neuromodulation, was analyzed. Seizure frequency was determined from patient‐reported counts during the 3‐month retrospective period preceding each clinical visit, including baseline and all follow‐up assessments. Follow‐up visits were scheduled according to routine clinical practice and were not restricted to fixed time points; in most cases, they occurred more frequently than the predefined analysis intervals (3, 6, and 12 months after CNB initiation). For each predefined time point, data were derived from the clinical visit that occurred closest to the respective interval, using seizure counts from the 3 months preceding that visit. In addition, an overall (last follow‐up) analysis used seizure frequency from the final 3‐month period preceding the most recent available visit, irrespective of the exact time from treatment initiation. Seizure outcomes were recorded separately for all seizures and for focal to bilateral tonic–clonic or bilateral tonic–clonic seizures (FBTCS/BTCS). The latter category encompassed both focal to bilateral tonic–clonic seizures observed in patients with focal epilepsy and bilateral tonic–clonic seizures observed in patients classified as having epilepsy of unknown onset. 14 It also encompassed tonic–clonic seizures occurring in patients with combined epilepsy, in whom focal to bilateral and generalized tonic–clonic seizures could not be reliably distinguished. For clarity, we refer to this combined category (FBTCS/BTCS) collectively as tonic–clonic seizures (TCS) throughout the manuscript. Treatment‐emergent adverse events (TEAEs) were extracted from medical records at each visit, with classification of severity and type. Reasons for discontinuation were categorized as related to TEAEs, inefficacy, or both. Retention data were collected based on total treatment duration. Of the 500 included patients, all were included in safety and retention analyses. A total of 459 patients received CNB for at least 3 months; the remaining 41 patients discontinued CNB before 3 months and were therefore excluded from efficacy analyses. Among the 459 efficacy‐eligible patients, 436 had complete seizure frequency data at both baseline and last follow‐up and were included in the primary efficacy analyses; the 23 patients with incomplete seizure data were included only in safety and retention analyses. Data analysis The primary efficacy endpoint was the responder rate, defined as a ≥ 50% reduction in seizure frequency versus baseline. Secondary endpoints included seizure freedom and a ≥ 90% seizure reduction. Seizure worsening was defined as a ≥ 100% increase in seizure frequency compared with baseline (i.e., at least a doubling of the baseline seizure rate). Patients with < 50% reduction and < 100% increase in seizure frequency were classified as having unchanged seizure control. Efficacy was calculated for all seizures across all patients with ≥ 3 months of treatment, regardless of total treatment duration, using seizure frequency from the final 3‐month retrospective period. In addition, efficacy was assessed separately for TCS alone, restricted to patients who experienced these seizure types at baseline and had complete seizure counts at both baseline and follow‐up. To complement the overall efficacy analysis, outcomes at ≥ 12 months were also examined in patients with at least 12 months of follow‐up and complete seizure count data. All subsequent stratified and subgroup analyses used efficacy data for all seizures combined, based on the final available 3‐month period. A dose–response analysis was performed by stratifying patients into mutually exclusive groups according to their final CNB maintenance dose (< 200, 200, 250, 300, 350, and 400 mg). Efficacy categories were identical to those used in the primary analysis, and only patients with complete seizure counts at baseline and at their final dose were included. Additionally, efficacy was stratified by cumulative ASM exposure (≤ 5, 6–10, > 10 ASMs; previous plus concomitant), and subgroup analyses were conducted for URE, stringently defined URE, and highly active epilepsy (≥ 20 seizures/month at baseline). Descriptive statistics included means, medians, standard deviations, and minimum and maximum values for continuous variables, and counts and percentages for categorical variables. To assess group differences by cumulative ASM burden (≤ 5, 6–10, > 10 ASMs), demographic and clinical variables were compared using Kruskal–Wallis tests (for continuous variables) and Fisher's exact tests (for categorical variables), with post hoc Dunn's tests or pairwise Fisher's tests with Bonferroni correction as appropriate. A separate analysis compared efficacy between final CNB doses < 200 mg and ≥ 200 mg across cumulative ASM exposure groups to assess whether lower CNB doses may already provide meaningful seizure reduction in patients with lower treatment refractoriness, as suggested by recent observational studies. 15 , 16 Final CNB maintenance dose was also compared across cumulative ASM exposure groups using the Kruskal–Wallis rank‐sum test. Comparisons of seizure freedom and ≥ 50% responder rates were performed both across ASM burden groups within each final CNB dose and between final CNB doses within each ASM group using Fisher's exact test. Post‐hoc pairwise comparisons with Bonferroni correction were performed when the global test was significant. TEAEs were summarized descriptively across all enrolled patients (n = 500), including frequency, severity, and categorized reasons for discontinuation. Treatment retention was analyzed by duration of CNB therapy. Predictors of seizure response were assessed via logistic regression for seizure freedom and ≥ 50% response at 12 months. Candidate predictors included age, sex, epilepsy localization, presence of TCS, baseline seizure frequency (categorized), cumulative ASM exposure (≤ 5, 6–10, > 10), and concomitant use of sodium channel blockers (SCBs), clobazam (CLB), or brivaracetam — based on previously published evidence. 10 , 17 , 18 , 19 Univariable models were first performed, with predictors at p < 0.10 included in multivariable models. Correlations among predictors were examined using Kendall's tau and Fisher's exact tests, and multicollinearity was excluded. Following the observation that concomitant SCB use was associated with treatment response, an additional exploratory analysis was performed to evaluate the two most commonly co‐administered SCBs in the cohort (lamotrigine – LTG and lacosamide – LCS). Using the same modeling approach, their associations with seizure freedom and ≥ 50% response at 12 months were analyzed. In addition, tolerability and dose achievement were explored according to concomitant SCB use. Patients were first compared according to the presence or absence of any concomitant SCB with respect to final CNB maintenance dose, occurrence of any TEAE, discontinuation due to TEAEs, and CNB dose reduction. The final CNB dose was defined as the dose at treatment discontinuation for patients who discontinued CNB, or the dose at the end of study data collection for those who remained on treatment. Further exploratory analyses were performed in patients without concomitant SCBs, patients receiving LTG as the only concomitant SCB, patients receiving LCS as the only concomitant SCB, and patients receiving concomitant LTG and LCS without other SCBs. These analyses aimed to assess whether the observed association between concomitant SCB use and treatment response or tolerability was driven by a specific SCB or reflected a broader effect of SCB co‐medication. Patients receiving other SCBs or other SCB combinations were excluded. Group comparisons were performed using the Wilcoxon rank‐sum test, Kruskal–Wallis rank‐sum test, and Fisher's exact test, as appropriate. No adjustment for multiple comparisons was applied due to the exploratory nature of these analyses. RESULTS Patient characteristics A total of 500 patients treated with adjunctive CNB were included in the analysis. At data cutoff, the median duration of CNB therapy was 17.4 months (range 0.1–92.2). The mean CNB daily dose at the time of data collection among those still on treatment was 234.9 mg (12.5–400), and the median dose was 200 mg. The cohort exhibited high baseline seizure burden and treatment resistance. The mean monthly frequency of all seizures was 27.5 (69.5), with a median of 8.0 (0.1–1000.0). Tonic–clonic seizures were reported in 46.0% of patients; among these, the mean monthly TCS frequency was 3.6 (5.5), and the median frequency was 2.0 (0.1–40.0). At the time of CNB initiation, patients had been exposed to a median of nine ASMs (min–max 2–21) in total, including both previous and concomitant treatments. CNB was introduced as an add‐on to a median of three concomitant ASMs (min–max 0.0–6.0), defined as medications that were concomitant at the time of CNB initiation. This count excludes ASMs that were intentionally withdrawn as part of a switch strategy at the time of CNB introduction. Most patients (83.4%) met criteria for URE. Additionally, 31.8% had previously undergone resective surgery, 33.0% had received neuromodulation (VNS or DBS), and 11.6% fulfilled a stringent URE definition, i.e. failure of ≥ 4 ASMs, prior resection, and neuromodulation. Demographic and clinical characteristics are summarized in Table 1. TABLE 1Patient demographics and baseline characteristics.CharacteristicValue (N = 500)Female, n (%)242 (48.4)Mean age, years (SD)40.1 (12.1)Median age, years (minimum–maximum)39.0 (18.0–74.0)Mean age at epilepsy onset, years (SD)13.5 (12.2)Median age at epilepsy onset, years (minimum–maximum)11.0 (0.2–71.0)Mean epilepsy duration, years (SD)26.4 (13.4)Median epilepsy duration, years (minimum–maximum)26.0 (1.0–63.0)Seizure type, n (%)Focal seizures only (without evolution to BTCS)270 (54.0)TCS230 (46.0)Epilepsy type, n (%) Focal epilepsy 435 (87.0)Temporal lobe epilepsy, n (% of focal epilepsy)148 (34.0)Extratemporal lobe epilepsy, n (% of focal epilepsy)163 (37.5)Multifocal epilepsy, n (% of focal epilepsy)76 (17.5)Unspecified type of focal epilepsy, n (% of focal epilepsy)48 (11.0) Combined (generalized + focal) epilepsy 7 (1.4) Unknown epilepsy type 58 (11.6)Etiology, n (%)Structural240 (48.0)Genetic22 (4.4)Autoimmune10 (2.0)Infectious4 (0.8)Metabolic1 (0.2)Unknown223 (44.6)Monthly baseline seizure frequencyAll seizure types, mean (SD)27.5 (69.5)All seizure types, median (minimum–maximum)8.0 (0.1–1000.0)TCS (calculated only for patients who were reported to have TCS), mean (SD)3.6 (5.5)Median monthly frequency of TCS (calculated only for patients who were reported to have TCS), median (minimum–maximum)2.0 (0.1–40.0)≥ 20 seizures per month, n (%)150 (30.0)< 20 seizures per month, n (%)350 (70.0)Cumulative ASM exposure at CNB initiation (previous and concomitant ASMs), mean (SD)9.7 (4.2)Cumulative ASM exposure at CNB initiation (previous and concomitant ASMs), median (minimum–maximum)9.0 (2.0–21.0)Number of ASMs on CNB initiation (concomitant), mean (SD)2.8 (1.1)Number of ASMs on CNB initiation (concomitant), median (minimum–maximum)3.0 (0.0–6.0)Number of concomitant ASMs at CNB initiation, n (%)0 ASM1 (0.2)1 ASM55 (11.0)2 ASM154 (30.8)3 ASM174 (34.8)4 ASM90 (18.0)≥ 5 ASM26 (5.2)Most common concomitant ASMs, n (%)Brivaracetam202 (40.4)Lacosamide192 (38.4)Lamotrigine154 (30.8)Levetiracetam112 (22.4)Valproate100 (20.0)Perampanel90 (18.0)Prior non‐pharmacological interventions, n (%)Epilepsy surgery or neuromodulation266 (53.2)Epilepsy surgery (except for neuromodulation)159 (31.8)Resection or disconnection148 (29.6)Thermocoagulation23 (4.6)Neuromodulation therapy (VNS/DBS, active or explanted)165 (33.0)Epilepsy surgery and neuromodulation and ≥ 4 ASMs58 (11.6) Note: Neuromodulation therapy refers to any history of vagus nerve stimulation (VNS) or deep‐brain stimulation (DBS), including both active and inactive (explanted or deactivated) devices. The term epilepsy surgery includes resection, disconnection, or thermocoagulation. For clarity, focal to bilateral tonic–clonic and bilateral tonic–clonic seizures are collectively referred to as tonic–clonic seizures (TCS).Abbreviations: ASM, anti‐seizure medication; DBS, deep‐brain stimulation; ETLE, extratemporal lobe epilepsy; FS, focal seizures; SD, standard deviation; TCS, tonic–clonic seizures; TLE, temporal lobe epilepsy; VNS, vagus nerve stimulation. Patient demographics and baseline characteristics. Note: Neuromodulation therapy refers to any history of vagus nerve stimulation (VNS) or deep‐brain stimulation (DBS), including both active and inactive (explanted or deactivated) devices. The term epilepsy surgery includes resection, disconnection, or thermocoagulation. For clarity, focal to bilateral tonic–clonic and bilateral tonic–clonic seizures are collectively referred to as tonic–clonic seizures (TCS). Abbreviations: ASM, anti‐seizure medication; DBS, deep‐brain stimulation; ETLE, extratemporal lobe epilepsy; FS, focal seizures; SD, standard deviation; TCS, tonic–clonic seizures; TLE, temporal lobe epilepsy; VNS, vagus nerve stimulation. Overall treatment efficacy Seizure outcomes were analyzed in 436 patients with available seizure counts at both baseline and follow‐up. Within this cohort, 123 patients had TCS at baseline and were analyzed separately. At the last follow‐up, a ≥ 50% reduction in overall seizure frequency was achieved in 55.7% of patients. A ≥ 90% reduction was achieved in 24.3% of patients, including 16.5% who became seizure‐free. No significant change was observed in 42.2% of patients, while seizure worsening occurred in only 2.1%. In the subgroup of patients with TCS, a ≥ 50% reduction in seizure frequency was observed in 74.8% of patients. A ≥ 90% reduction was achieved in 52.0%, including 49.6% who became seizure‐free. No significant change occurred in 19.5%, while seizure worsening was observed in 5.7%. Focusing on the 12‐month outcome in patients treated with CNB for ≥ 12 months (n = 331), a ≥ 50% reduction in overall seizure frequency was achieved in 61.6% of patients. A ≥ 90% reduction was observed in 26.9%, including 18.1% who became seizure‐free. No significant change occurred in 36.3% of patients, and worsening was observed in 2.1%. Among patients with TCS treated for ≥ 12 months (n = 102), 75.5% achieved at least a 50% reduction in TCS frequency. A ≥ 90% reduction was achieved in 49.0%, including 46.1% who became seizure‐free. No significant change occurred in 20.6%, and worsening was observed in 3.9%. For seizure outcomes, see Figure 1. Seizure outcomes. Graph showing the proportions of patients achieving seizure freedom, ≥ 90% seizure reduction (RR 90%), and ≥ 50% seizure reduction (RR 50%), presented separately for all seizures and for focal to bilateral tonic–clonic or bilateral tonic–clonic seizures (referred to here, for clarity, as tonic–clonic seizures; TCS). The two sets of columns on the left correspond to overall outcomes at the last available follow‐up (patients with ≥ 3 months of CNB treatment) (all seizures: n = 436; TCS: n = 123), while the two sets of columns on the right correspond to outcomes after 12 months of CNB treatment (all seizures: n = 331; TCS: n = 102). In the TCS analysis, seizure freedom refers specifically to freedom from TCSs, irrespective of the occurrence of other seizure types. TCS = tonic–clonic seizures. Seizure outcomes by maintenance dose Efficacy outcomes varied across final CNB maintenance doses (Figure 2). Seizure freedom ranged from 8.1% to 22.2%, with the highest proportion at 350 mg (22.2%), followed by 200 mg (19.7%) and 400 mg (17.1%). The highest ≥ 90% responder rates were observed at 350 mg (33.3%) and 400 mg (31.7%), whereas lower rates occurred at intermediate doses (20.9% at 250 mg; 18.9% at 300 mg) and at < 200 mg (23.0%) and 200 mg (24.6%). The overall ≥ 50% responder rate was relatively stable for doses ≥ 200 mg: 54.9% at 200 mg, 62.8% at 250 mg, 58.1% at 300 mg, 61.1% at 350 mg, and 58.5% at 400 mg, compared with 49.0% at < 200 mg. Full distributions of seizure outcomes by dose are shown in Figure 2. Seizure outcomes by maintenance dose. The graph shows the percentage distribution of patients across seizure outcome categories at different final maintenance doses of CNB. Each column represents a mutually exclusive dose group and shows the percentage of patients at a given maintenance dose and their seizure outcomes: ≥ 50% seizure reduction (RR 50%), ≥ 90% seizure reduction (RR 90%), and seizure freedom. Only patients with known seizure counts at both baseline and the final dose were included. The number of patients varied by dose: < 200 mg (n = 100), 200 mg (n = 142), 250 mg (n = 43), 300 mg (n = 74), 350 mg (n = 36), and 400 mg (n = 41). Stratified outcomes by cumulative anti‐seizure medication burden When stratifying patients by the total number of ASMs used (both previously and concomitant), CNB showed comparable efficacy in patients with ≤ 5 and 6–10 ASMs, whereas outcomes were less favorable in those with > 10 ASMs. Among patients with ≤ 5 ASMs, 61.3% achieved at least a 50% reduction in seizure frequency. A ≥ 90% reduction was achieved in 28.0%, including 22.7% who became seizure‐free. In the 6–10 ASM group, 57.3% of patients achieved a ≥ 50% reduction, while 29.8% achieved a ≥ 90% reduction, with 20.2% achieving seizure freedom. Among patients with > 10 ASMs, 51.9% achieved at least a 50% reduction, and 17.5% achieved a ≥ 90% reduction, with 10.4% becoming seizure‐free. For details, see Figure 3. Seizure outcomes stratified by cumulative anti‐seizure medication burden. The graph displays outcomes in 436 patients treated with CNB stratified by cumulative ASM exposure (≤ 5, 6–10, > 10; previous and concomitant). Each column shows the percentage of patients at a given ASM exposure and their seizure outcomes: ≥ 50% reduction (RR 50%), ≥ 90% reduction (RR 90%), and seizure freedom. Only patients with known seizure counts at baseline and follow‐up and ≥ 3 months of treatment were included. The number of patients in each cumulative ASM group was as follows: ≤ 5 ASMs (n = 75), 6–10 ASMs (n = 178), and > 10 ASMs (n = 183). ASM = anti‐seizure medication. Statistically significant differences were observed across groups in epilepsy duration, age at onset, and baseline seizure frequency (all p < 0.001, Kruskal–Wallis test). Post hoc comparisons revealed that patients in the > 10 ASM group had significantly longer epilepsy duration, earlier age at onset, and higher baseline seizure burden compared to those with fewer ASMs. The use of SCBs and CLB also varied significantly across groups (p = 0.009 and p < 0.001, respectively, Fisher's exact test). Specifically, SCBs were less frequently prescribed in patients with lower cumulative ASM exposure (≤ 5 ASMs), while CLB was increasingly used in patients with higher treatment burden—reaching the highest proportion in the > 10 ASM group. Full results and post hoc comparisons are detailed in Table 2 and Table S1. TABLE 2Baseline demographic and clinical characteristics stratified by cumulative anti‐seizure medication burden, in the overall cohort.CharacteristicsOverall (N = 500) ≤ 5 ASM (N = 92)6–10 ASM (N = 200)> 10 ASM (N = 208) p*Epilepsy duration; median (IQR) 26.0 (17.0; 34.4) 12.5(b,c) (6.0; 24.0) 24.0(a,c) (16.0; 33.7) 30.0(a,b) (23.0; 37.0) < 0.001 Epilepsy onset age; median (IQR) 11.0 (4.0; 19.0) 17.5(b,c) (11.5; 27.0) 12.0(a,c) (5.0; 19.0) 7.0(a,b) (3.0; 15.0) < 0.001 Baseline seizure frequency (all); median (IQR) 8.0 (3.0; 29.5) 4.0(c) (2.0; 15.0) 6.0(c) (2.0; 17.0) 11.0(a,b) (5.0; 30.0) < 0.001 Sex(−)(−)(−)0.068Male; n (%)258 (51.6)49 (53.3)114 (57.0)95 (45.7)Female; n (%)242 (48.4)43 (46.7)86 (43.0)113 (54.3)Clobazam (b,c) (a,c) (a,b) < 0.001 No; n (%)432 (86.4)90 (97.8)177 (88.5)165 (79.3)Yes; n (%)68 (13.6)2 (2.2)23 (11.5)43 (20.7)Sodium channel blocker (b,c) (a) (a) 0.003 No; n (%)93 (18.6)29 (31.5)31 (15.5)33 (15.9)Yes; n (%)407 (81.4)63 (68.5)169 (84.5)175 (84.1) Note: This table summarizes baseline clinical and seizure characteristics in the overall cohort (N = 500) and in subgroups defined by the cumulative number of ASMs (≤ 5, 6–10, > 10). Values are presented as median (interquartile range) for continuous variables and as number (percentage) for categorical variables. Between‐group comparisons were performed using the Kruskal–Wallis rank‐sum test for continuous variables and Fisher's exact test for categorical variables. Superscript letters (a, b, c) denote statistically significant pairwise differences based on Dunn's post‐hoc test (continuous variables) or Fisher's exact test with Bonferroni correction (categorical variables): a = significantly different from group ≤ 5; b = significantly different from group 6–10; c = significantly different from group > 10. Significant p‐values are shown in bold. Baseline demographic and clinical characteristics stratified by cumulative anti‐seizure medication burden, in the overall cohort. 26.0 (17.0; 34.4) 12.5(b,c) (6.0; 24.0) 24.0(a,c) (16.0; 33.7) 30.0(a,b) (23.0; 37.0) 11.0 (4.0; 19.0) 17.5(b,c) (11.5; 27.0) 12.0(a,c) (5.0; 19.0) 7.0(a,b) (3.0; 15.0) 8.0 (3.0; 29.5) 4.0(c) (2.0; 15.0) 6.0(c) (2.0; 17.0) 11.0(a,b) (5.0; 30.0) Note: This table summarizes baseline clinical and seizure characteristics in the overall cohort (N = 500) and in subgroups defined by the cumulative number of ASMs (≤ 5, 6–10, > 10). Values are presented as median (interquartile range) for continuous variables and as number (percentage) for categorical variables. Between‐group comparisons were performed using the Kruskal–Wallis rank‐sum test for continuous variables and Fisher's exact test for categorical variables. Superscript letters (a, b, c) denote statistically significant pairwise differences based on Dunn's post‐hoc test (continuous variables) or Fisher's exact test with Bonferroni correction (categorical variables): a = significantly different from group ≤ 5; b = significantly different from group 6–10; c = significantly different from group > 10. Significant p‐values are shown in bold. Seizure outcomes by maintenance dose stratified by cumulative anti‐seizure medication burden When stratifying patients by both cumulative ASM exposure and final CNB maintenance dose (< 200 mg vs. ≥ 200 mg), treatment responses were observed across all dose and exposure groups. Among patients with ≤ 5 ASMs, 55.6% of those receiving < 200 mg and 63.2% receiving ≥ 200 mg achieved at least a 50% reduction in seizure frequency. A ≥ 90% reduction occurred in 22.3% and 29.9%, respectively, including 16.7% and 24.6% who became seizure‐free. In patients with 6–10 ASMs, 52.6% of those receiving < 200 mg and 58.6% receiving ≥ 200 mg achieved a ≥ 50% reduction. A ≥ 90% reduction occurred in 28.9% and 30.0%, respectively, including 26.3% and 18.6% who became seizure‐free. Among patients with > 10 ASMs, 43.2% of those receiving < 200 mg and 54.7% receiving ≥ 200 mg achieved at least a 50% reduction. A ≥ 90% reduction occurred in 18.1% and 17.3%, respectively, including 13.6% and 9.4% who became seizure‐free. For details, see Figure 4. Seizure outcomes by maintenance dose stratified by cumulative anti‐seizure medication burden. The graph displays seizure outcomes in patients treated with CNB stratified by cumulative ASM exposure (≤ 5, 6–10, > 10; previous and concomitant) and by final CNB maintenance dose (< 200 mg vs. ≥ 200 mg). Each column shows the percentage of patients within a given ASM exposure and dose group and their seizure outcomes: ≥ 50% seizure reduction (RR 50%), ≥ 90% seizure reduction (RR 90%), and seizure freedom. Only patients with known seizure counts at baseline and at the final dose (or at discontinuation) and a minimum treatment duration of ≥ 3 months were included. The number of patients in each subgroup was as follows: ≤ 5 ASMs: < 200 mg (n = 18) and ≥ 200 mg (n = 57); 6–10 ASMs: < 200 mg (n = 38) and ≥ 200 mg (n = 140); > 10 ASMs: < 200 mg (n = 44) and ≥ 200 mg (n = 139). ASM = anti‐seizure medication. Final CNB maintenance dose did not differ significantly across cumulative ASM exposure groups (≤ 5, 6–10, and > 10 ASMs; p = 0.341), with median maintenance doses of 200 mg in all groups. When outcomes were compared across ASM burden groups within each CNB dose category, seizure freedom differed significantly only among patients receiving ≥ 200 mg CNB (p = 0.013), with higher seizure freedom observed in patients with lower cumulative ASM exposure. No significant differences were observed for seizure freedom among patients receiving < 200 mg CNB, and no significant differences were found for ≥ 50% responder rates at either maintenance dose level. For details, see Figure 4 and Tables S2–S4. Outcomes in ultra‐refractory epilepsy Among patients with URE, 285 patients had evaluable data after 12 months of CNB treatment. Of these, 60.7% achieved at least a 50% reduction in seizure frequency. A ≥ 90% reduction was achieved in 25.6% of patients, including 16.5% who became seizure‐free. In the stringently defined URE subgroup (≥ 4 ASMs, prior resective surgery, and neuromodulation), 41 patients had available 12‐month outcomes. In this group, 73.2% achieved a ≥ 50% reduction, and 29.3% achieved a ≥ 90% reduction, including 17.1% who became seizure‐free. In the subgroup with highly active epilepsy (≥ 20 seizures/month at baseline; n = 105), 61.0% achieved at least a 50% reduction after 12 months, while 28.6% achieved a ≥ 90% reduction, including 10.5% seizure‐free. For complete subgroup results at all time points, see Table S5. Prediction of efficacy To identify factors independently associated with seizure freedom at 12 months, we analyzed 331 patients with known outcomes and ≥ 12 months of CNB therapy. In univariable models, cumulative ASM exposure > 10 ASMs (OR 0.35, 95% CI 0.16–0.79; p = 0.011) and highly active epilepsy at baseline (OR 0.42, 95% CI 0.20–0.82; p = 0.016) were associated with lower odds of seizure freedom. In the multivariable model, both factors remained significant—highly active epilepsy (OR 0.47, 95% CI 0.22–0.92; p = 0.035) and cumulative ASM exposure > 10 ASMs (OR 0.38, 95% CI 0.17–0.86; p = 0.019)—indicating reduced likelihood of achieving seizure freedom in these subgroups. For results, see Tables S6 and S7. To further clarify the outcomes, we also investigated factors associated with achieving a ≥ 50% reduction in seizure frequency (responder status) sustained for at least 12 months at the time of data cutoff. This analysis was based on the same cohort of 331 patients, of whom 204 achieved this response. In the multivariable model, concomitant SCB use was significantly associated with a lower likelihood of achieving ≥ 50% response (OR 0.45, 95% CI 0.25–0.80; p = 0.008). Concomitant CLB use showed reduced odds (OR 0.53, 95% CI 0.28–1.01; p = 0.053) but did not reach statistical significance. For complete results, including univariable and multivariable predictors of ≥ 50% responder status, see Tables S8 and S9. Regarding correlations among categorical predictors, a significant association was observed between concomitant CLB and the number of ASMs (previous and concomitant) (p = 0.004). Correlations among predictors were generally weak to moderate and did not suggest problematic multicollinearity (see Tables S10 and S11 for the full correlation matrix and commentary). In an additional analysis of treatment effectiveness, neither concomitant LTG nor LCS was significantly associated with seizure freedom or ≥ 50% responder status at 12 months in univariable logistic regression models, and these findings remained unchanged after multivariable adjustment. Detailed results are presented in Tables S12–S15. Safety and tolerability CNB was generally well tolerated, even in this highly treatment‐resistant population. Treatment‐emergent adverse events (TEAEs) were reported in 302 patients (60.4%), with the most common being fatigue (30.6%), somnolence (28.2%), dizziness (22.6%), and ataxia (17.0%). Other TEAEs included diplopia (6.4%), nausea (5.6%), behavioral or psychiatric symptoms (4.8%), and speech impairment (4.2%). Most TEAEs were mild (34.4%) or moderate (41.4%); only 23.5% were classified as severe. In two patients (0.7%), TEAE severity was unknown. No cases of drug reaction with eosinophilia and systemic symptoms (DRESS) were observed. Among patients experiencing TEAEs, reduction of concomitant ASMs was required in 138 patients (45.7%), while withdrawal of at least one concomitant ASM occurred in 19 patients (6.3%). CNB dose reduction was performed in 89 patients (29.5%), and CNB withdrawal in 71 patients (23.5%). Importantly, the most common management approach was a combined strategy of maintaining CNB treatment while reducing concomitant ASMs (27.8%), which was more frequent than CNB withdrawal alone (16.6%) or CNB dose reduction alone (10.9%). Patients receiving concomitant SCBs reported TEAEs significantly more frequently than those without SCBs (62.9% vs. 49.5%, p = 0.019). However, discontinuation due to TEAEs did not differ significantly between groups (17.9% vs. 10.8%, p = 0.121). Similarly, the frequency of CNB dose reduction during treatment was comparable between patients with and without concomitant SCBs (17.7% vs. 19.4%, p = 0.765), and no significant difference was observed in the final achieved CNB maintenance dose (median 200 mg in both groups, p = 0.514). In exploratory analyses of predefined mutually exclusive SCB groups—including patients without concomitant SCBs, patients receiving LCS as the only concomitant SCB, patients receiving LTG as the only concomitant SCB, and patients receiving concomitant LCS plus LTG without other SCBs—no significant differences were observed in final CNB maintenance dose, discontinuation due to TEAEs, overall occurrence of TEAEs, or CNB dose reduction during treatment. Detailed results are provided in Tables S16–S24. Treatment retention was high throughout the observation period. Based on Kaplan–Meier analysis of 494 patients with known follow‐up (excluding six with unknown duration due to missing data on treatment discontinuation or last clinical follow‐up), the estimated probability of continued CNB treatment was 88.6% (95% CI, 85.9%–91.5%) at 6 months, 83.0% (95% CI, 79.8%–86.5%) at 12 months, and 80.9% (95% CI, 77.4%–84.6%) at 18 months. At 24 months, retention remained stable at 79.5% (95% CI, 75.9%–83.4%). See Figure 5 for a Kaplan–Meier survival curve illustrating treatment retention rate for patients, with corresponding numerical data available in Table S25. Treatment retention over time. Figure shows treatment retention over time. The time from treatment initiation with CNB until treatment discontinuation was analyzed. For patients who discontinued treatment, the event was recorded at the time of discontinuation. Patients who had not discontinued treatment by the data cutoff date were censored at that date. The analysis included all patients with known follow‐up, except for six patients whose follow‐up duration was unknown. During the study period, a total of 101 patients discontinued CNB treatment. The primary reasons for discontinuation were TEAEs alone, accounting for 50.5% of these discontinuations. Discontinuation due to both side effects and inefficacy occurred in 32.7% of patients. Inefficacy as a sole reason for discontinuation was observed in 11.9% of cases. A smaller proportion of discontinuations were due to death (3.0%) or unknown reasons (2.0%). None of these deaths was considered related to CNB treatment. DISCUSSION This multicenter observational study represents one of the most extensive real‐world evaluations of CNB to date, with 500 adults treated for drug‐resistant epilepsy, including an exceptionally high proportion of URE. Patients were followed for at least 12 months, enabling a detailed analysis of efficacy, predictors of outcome, treatment timing, and retention in a diverse and clinically challenging population. CNB showed long‐term efficacy in this DRE population. After 12 months, 61.6% of patients achieved at least a 50% reduction in overall seizure frequency, including 18.1% who became seizure‐free. These findings align with previously published real‐world studies where 12‐month seizure freedom ranged from 11.9% to 19.5%. 12 , 20 Beyond overall seizure reduction, CNB demonstrated remarkable efficacy in TCS. At 12 months, 75.5% of patients experienced at least a 50% reduction in TCS, and 46.1% became seizure‐free. These rates are comparable to the highest reported in CNB literature, and significantly exceed those analyzing TCS seizure freedom (< 25%). 10 , 12 , 21 Given the known association between TCS and increased risk of sudden unexpected death in epilepsy, this effect may represent an important therapeutic goal. 22 The ≥ 50% responder rate was relatively stable across doses ≥ 200 mg/day. Seizure freedom was highest at 350 mg/day, with similar rates at 200 and 400 mg/day and lower rates at 250–300 mg/day. Rates of ≥ 90% seizure reduction were also higher at 350–400 mg/day than at lower doses. In our cohort, seizure outcomes therefore did not increase linearly with higher CNB doses, with reduced seizure freedom and ≥ 90% responder rates in the intermediate dose groups (250–300 mg) and higher seizure freedom rates again at 350–400 mg/day. A similar non‐linear pattern has been reported by Loushy et al., 10 who attributed the lower efficacy at intermediate doses to clinical selection, as patients titrated beyond 200 mg/day tended to be more treatment resistant. Moreover, the higher seizure freedom rates at 350–400 mg/day may reflect patients who tolerated further titration and remained on treatment long enough to benefit from dose optimization. Collectively, these findings do not demonstrate a clear linear dose–response relationship. Although some variability in seizure outcomes was observed across dose groups, the differences were modest and inconsistent. These patterns may reflect clinical selection and individual tolerability rather than a direct pharmacological dose effect and therefore support individualized titration rather than a uniform dose‐escalation strategy. Stratification by cumulative ASM burden revealed an association with treatment response. Patients with lower cumulative ASM exposure (≤ 10 ASMs) achieved the most favorable outcomes, with seizure freedom rates of 22.7% in the ≤ 5 ASM group and 20.2% in the 6–10 ASM group, and ≥ 50% responder rates of 61.3% and 57.3%, respectively. In contrast, those with cumulative ASM exposure > 10 ASMs still experienced meaningful benefit (10.4% seizure freedom; 50% responder rate reaching 51.9%), despite having more severe baseline characteristics such as earlier epilepsy onset, longer duration, and higher seizure frequency. These patterns are consistent with the multivariable analysis, where cumulative ASM exposure > 10 ASMs independently predicted lower odds of seizure freedom at 12 months. Nevertheless, CNB remained effective even in highly refractory patients, indicating that while earlier introduction enhances outcomes, meaningful efficacy can still be achieved at advanced stages of treatment. 9 , 10 , 12 In an additional analysis stratified by cumulative ASM exposure, CNB demonstrated clinically meaningful efficacy even at doses below 200 mg/day across all seizure burden categories. In patients with ≤ 5 and 6–10 ASMs, overall responder rates at < 200 mg were comparable to those observed at ≥ 200 mg. Notably, in the 6–10 ASM group, seizure freedom was observed in a substantial proportion of patients receiving < 200 mg/day. The results in the ≤ 5 ASM group should be interpreted cautiously due to the smaller sample size, but they are consistent with previous studies suggesting that clinically relevant CNB efficacy may be achieved at relatively low doses, particularly in patients with lower cumulative ASM exposure. 15 , 16 However, when outcomes were compared across ASM burden groups within each dose category, seizure freedom differed significantly only among patients receiving ≥ 200 mg CNB, with the highest rates in patients with ≤ 5 ASMs and the lowest in those with > 10 ASMs, while ≥ 50% responder rates did not differ significantly. The study also focused on URE, a group often considered pharmacoresistant beyond rescue. Among patients with ≥ 6 prior interventions, 12‐month seizure freedom reached 16.5% with responder rates exceeding 60%. These results surpass prior reports from similar URE cohorts. 3 , 13 Notably, in the subgroup with prior resective surgery and neuromodulation failure, 12‐month seizure freedom was still achieved in 17.1%, indicating the potential value of CNB even after failure of invasive therapies. For patients with highly active epilepsy, complete seizure control was lower (10.5% seizure freedom), yet a marked reduction in burden was observed, a clinically meaningful result given this group's increased morbidity and risk profile. These findings demonstrate the efficacy of CNB, even in highly treatment‐resistant patient populations. Based on these outcomes, CNB should be considered prior to referring patients for surgical evaluation, particularly in cases where the likelihood of achieving seizure freedom through resective surgery is limited — such as in non‐lesional epilepsy or when the presumed epileptogenic zone is located within or adjacent to eloquent cortex. Multivariable analysis identified predictors of seizure freedom. In the univariate analysis, patients with lower cumulative ASM exposure had a significantly higher chance of achieving seizure freedom, supporting the principle that earlier introduction in less pharmacoresistant stages enhances efficacy. In the multivariate model, patients with fewer than 20 seizures/month at baseline had a significantly higher chance of achieving seizure freedom compared to patients with highly active epilepsy, which is predictable. However, as we showed in the subanalysis, even in patients with a higher cumulative ASM burden or highly active epilepsy, CNB still showed promising results in terms of achieving complete seizure freedom. Concomitant use of SCBs was associated with a lower likelihood of achieving a ≥ 50% seizure reduction. One possible explanation is that patients who have already failed previous SCB‐based treatments may derive less benefit from another ASM with a partially overlapping mechanism of action, including CNB. In additional analyses focusing on the two most used SCBs in our cohort (LTG and LCS), neither drug was independently associated with seizure freedom or ≥ 50% responder status, suggesting that the lower treatment response does not appear to be driven by a specific SCB but rather reflects a broader effect of SCB‐based polytherapy. Patients receiving concomitant SCBs also experienced more TEAEs more frequently than those without SCBs, indicating a higher overall burden of adverse effects in this combination. However, this was not accompanied by significantly higher rates of TEAE‐related discontinuation, CNB dose reduction, or lower achieved CNB maintenance doses. Similarly, no significant differences in tolerability were observed among patients receiving LTG as the only concomitant SCB, patients receiving LCS as the only concomitant SCB, and patients receiving concomitant LTG plus LCS without other SCBs. These findings suggest that although SCB co‐medication may increase the overall frequency of TEAEs, the lower efficacy associated with SCB use is unlikely to be explained primarily by poorer tolerability. Instead, it may reflect pharmacodynamic overlap between sodium channel–acting agents and the more treatment‐resistant clinical profile of patients requiring SCB‐based polytherapy. Regarding CLB, concomitant use was associated with a lower likelihood of achieving a ≥ 50% response in univariate analysis, although this association became non‐significant after adjustment in the multivariate model. Indeed, our correlation analysis revealed a significant association between CLB use and a higher total number of previous and concomitant ASMs (p = 0.004), suggesting that CLB is frequently prescribed to patients with a higher baseline disease burden. Interestingly, a contrasting finding was reported in another study, where co‐administration with CLB emerged as a significant positive predictor of complete seizure freedom. 10 This discrepancy may reflect differences in the timing of CLB introduction across cohorts — in our setting, CLB is often introduced relatively late, typically after multiple prior ASM failures, which could partly explain the observed associations. These divergent findings highlight the need for further investigation into how the timing and clinical context of CLB use may influence CNB treatment outcomes. Treatment retention with CNB was notably high, with 83.0% of patients remaining on CNB after 12 months and 79.5% after 24 months. These findings consistently align with other real‐world studies, which report high long‐term treatment persistence. 8 , 10 , 11 , 12 Favorable retention reflects sustained efficacy, tolerability, and safety, which were consistent with the growing body of CNB literature. Adverse events were reported in 60.4% of patients, primarily somnolence, dizziness, and fatigue, with most cases being mild or moderate. TEAEs were the main reason for discontinuation, while inefficacy was the reason in only 11.9% of cases. Among patients experiencing TEAEs, reduction of concomitant ASMs was more frequent than CNB dose reduction or withdrawal. Importantly, the most common management approach was a combined strategy of maintaining CNB treatment while reducing concomitant ASMs, which was more frequent than CNB withdrawal alone or CNB dose reduction alone. This suggests that tolerability management in clinical practice often relies on the optimization of accompanying ASMs rather than CNB dose reduction or discontinuation. These adjustments likely reflect both reactive management of adverse events and proactive optimization of ASM polytherapy after CNB initiation. The observed safety profile is consistent with previous large‐cohort studies. 11 , 23 Limitations The study's limitations include its retrospective and observational design, which inherently means a lack of a control group and randomization, as well as reliance on existing clinical records. This can lead to potential confounding factors and bias. Although efficacy was assessed in mutually exclusive cohorts based on treatment duration, this approach resulted in reduced sample sizes for more extended follow‐up periods, which may limit the generalizability of the long‐term findings. Another limitation is that patients assessed after only 3 months of treatment may still have been in the titration phase rather than on a stable maintenance dose, so early outcomes in this subgroup should be interpreted with caution when compared to later follow‐up intervals. CLINICAL RELEVANCE CNB demonstrated meaningful seizure control across diverse and challenging patient cohorts, with particularly notable outcomes in TCS. Notably, the study demonstrated seizure freedom or ≥ 90% seizure reduction even in patients with ultra‐refractory and highly active epilepsy. These findings highlight the therapeutic value of CNB even in DRE, where alternative strategies have failed. The variability in seizure freedom across maintenance doses further supports the need for individualized titration. Importantly, tolerability management often relied on adjustment of concomitant ASMs rather than CNB dose reduction or discontinuation. Given its strong efficacy, favorable tolerability, and high long‐term retention rates, CNB should be considered as a core component of rational polytherapy in patients with epilepsy. Future studies should focus on defining predictors of response to CNB and optimizing titration strategies in different clinical subgroups. AUTHOR CONTRIBUTIONS Ondřej Strýček conceived the study, collected the data, drafted the manuscript, and contributed to the interpretation and discussion of the results. Jana Amlerová conceived the study, collected the data, contributed to manuscript writing, and participated in the interpretation and discussion of the results. Jitka Kočvarová, Martin Pail, and Jana Slonková collected the data; Martin Pail and Jana Slonková also critically revised the manuscript. Martin Šimčík and Kateřina Stanzelová collected the data. Monika Šibalová performed the statistical analyses. Petr Marusič and Milan Brázdil critically reviewed the manuscript and contributed to data interpretation, manuscript writing, and discussion. All authors read and approved the final version of the manuscript and agreed to be accountable for all aspects of the work. FUNDING INFORMATION This work was supported by the Czech Health Research Council (grant No. NW25‐04‐00345). Additional financial support was provided by the IBA Foundation, Czech Republic (Identification No. 21396566). CONFLICT OF INTEREST STATEMENT None of the authors has any conflict of interest to disclose. ETHICS STATEMENT We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. Supporting information Table S1: Baseline demographic and clinical characteristics stratified by cumulative antiseizure medication exposure (previous plus concomitant), in patients treated for ≥ 3 months. Table S2: Final cenobamate maintenance dose across cumulative ASM exposure groups. Table S3: Seizure outcomes by final CNB maintenance dose within cumulative ASM exposure groups. Table S4: Seizure outcomes across cumulative ASM exposure groups stratified by final CNB maintenance dose. Table S5: Seizure outcomes by epilepsy refractoriness (ultra‐refractory, stringently ultra‐refractory, and highly active) and treatment duration (≥ 3 and ≥ 12 months). Table S6: Predictors of complete seizure freedom in patients treated with cenobamate for ≥ 12 months: univariate logistic regression analysis. Table S7: Predictors of complete seizure freedom in patients treated with cenobamate for ≥ 12 months: multivariate logistic regression analysis. Table S8: Predictors of 50% seizure reduction in patients treated with cenobamate for ≥ 12 months: univariate logistic regression analysis. Table S9: Predictors of 50% seizure reduction in patients treated with cenobamate for ≥ 12 months: multivariate logistic regression analysis. Table S10: Correlation between covariates – based on patients with at least 12 months of CNB treatment duration and with known seizure frequency ‐ continuous variables. Table S11: Correlation between covariates – based on patients with at least 12 months of CNB treatment duration and with known seizure frequency ‐ categorical variables. Table S12: Predictors of seizure freedom – concomitant lamotrigine and lacosamide – patients treated with cenobamate for ≥ 12 months – univariate logistic regression. Table S13: Predictors of seizure freedom – concomitant lamotrigine and lacosamide – patients treated with cenobamate for ≥ 12 months – multivariable logistic regression. Table S14: Association of concomitant lamotrigine and lacosamide with ≥ 50% responder status – patients treated with cenobamate for ≥ 12 months – univariable logistic regression. Table S15: Association of concomitant lamotrigine and lacosamide with ≥ 50% responder status – patients treated with cenobamate for ≥ 12 months – multivariable logistic regression. Table S16: Final cenobamate maintenance dose according to concomitant sodium channel blocker use. Table S17: Discontinuation due to treatment‐emergent adverse events according to concomitant sodium channel blocker use. Table S18: Occurrence of any treatment‐emergent adverse event according to concomitant sodium channel blocker use. Table S19: Cenobamate downtitration according to concomitant sodium channel blocker use. Table S20: Distribution of individual sodium channel blocker groups. Table S21: Final cenobamate maintenance dose across individual sodium channel blocker groups. Table S22: Discontinuation due to treatment‐emergent adverse events across individual sodium channel blocker groups. Table S23: Occurrence of any treatment‐emergent adverse event across individual sodium channel blocker groups. Table S24: Cenobamate downtitration across individual sodium channel blocker groups. Table S25: Treatment retention. ACKNOWLEDGMENTS The authors thank all treating physicians of the patients and Barbora Čellár‐Nitschová for her valuable assistance with data collection. Strýček O, Amlerová J, Pail M, Kočvarová J, Slonková J, Šimčík M, et al. Real‐world effectiveness of cenobamate in drug‐resistant and ultra‐refractory epilepsy: A multicenter study. Epilepsia Open. 2026;11:1276–1289. 10.1002/epi4.70298 Petr Marusič and Milan Brázdil share senior authorship. St. Anne's University Hospital and Motol University Hospital are full members of ERN‐EpiCARE. DATA AVAILABILITY STATEMENT The data that support the findings of this study are available from the corresponding author upon reasonable request. The data are not publicly available due to privacy and ethical restrictions. REFERENCES Associated Data Supplementary Materials Table S1: Baseline demographic and clinical characteristics stratified by cumulative antiseizure medication exposure (previous plus concomitant), in patients treated for ≥ 3 months. Table S2: Final cenobamate maintenance dose across cumulative ASM exposure groups. Table S3: Seizure outcomes by final CNB maintenance dose within cumulative ASM exposure groups. Table S4: Seizure outcomes across cumulative ASM exposure groups stratified by final CNB maintenance dose. Table S5: Seizure outcomes by epilepsy refractoriness (ultra‐refractory, stringently ultra‐refractory, and highly active) and treatment duration (≥ 3 and ≥ 12 months). Table S6: Predictors of complete seizure freedom in patients treated with cenobamate for ≥ 12 months: univariate logistic regression analysis. Table S7: Predictors of complete seizure freedom in patients treated with cenobamate for ≥ 12 months: multivariate logistic regression analysis. Table S8: Predictors of 50% seizure reduction in patients treated with cenobamate for ≥ 12 months: univariate logistic regression analysis. Table S9: Predictors of 50% seizure reduction in patients treated with cenobamate for ≥ 12 months: multivariate logistic regression analysis. Table S10: Correlation between covariates – based on patients with at least 12 months of CNB treatment duration and with known seizure frequency ‐ continuous variables. Table S11: Correlation between covariates – based on patients with at least 12 months of CNB treatment duration and with known seizure frequency ‐ categorical variables. Table S12: Predictors of seizure freedom – concomitant lamotrigine and lacosamide – patients treated with cenobamate for ≥ 12 months – univariate logistic regression. Table S13: Predictors of seizure freedom – concomitant lamotrigine and lacosamide – patients treated with cenobamate for ≥ 12 months – multivariable logistic regression. Table S14: Association of concomitant lamotrigine and lacosamide with ≥ 50% responder status – patients treated with cenobamate for ≥ 12 months – univariable logistic regression. Table S15: Association of concomitant lamotrigine and lacosamide with ≥ 50% responder status – patients treated with cenobamate for ≥ 12 months – multivariable logistic regression. Table S16: Final cenobamate maintenance dose according to concomitant sodium channel blocker use. Table S17: Discontinuation due to treatment‐emergent adverse events according to concomitant sodium channel blocker use. Table S18: Occurrence of any treatment‐emergent adverse event according to concomitant sodium channel blocker use. Table S19: Cenobamate downtitration according to concomitant sodium channel blocker use. Table S20: Distribution of individual sodium channel blocker groups. Table S21: Final cenobamate maintenance dose across individual sodium channel blocker groups. Table S22: Discontinuation due to treatment‐emergent adverse events across individual sodium channel blocker groups. Table S23: Occurrence of any treatment‐emergent adverse event across individual sodium channel blocker groups. Table S24: Cenobamate downtitration across individual sodium channel blocker groups. Table S25: Treatment retention. Data Availability Statement The data that support the findings of this study are available from the corresponding author upon reasonable request. The data are not publicly available due to privacy and ethical restrictions.