Efficacy and safety of eslicarbazepine acetate as a first or later adjunctive therapy in patients with focal seizures

INTRODUCTION
We report outcomes from an open-label, non-randomized, 24-week study of eslicarbazepine acetate (ESL) in adults at earlier and later stages of their treatment history for focal seizures, conducted in a real-world clinical setting.


METHODS
ESL was taken as the first adjunctive therapy to levetiracetam (LEV) or lamotrigine (LTG) monotherapy (Arm 1), or as a later adjunctive therapy in treatment-resistant patients (Arm 2). The primary objective was to evaluate the effectiveness of ESL (by retention rates). Secondary objectives were to evaluate efficacy (seizure frequency), safety, tolerability, behavioral changes, mood, and health-related quality of life (HRQoL) associated with ESL treatment.


RESULTS
The modified intent-to-treat population included 96 patients (Arm 1: n = 41; Arm 2: n = 55) and the safety population included 102 patients (Arm 1: n = 44; Arm 2: n = 58). Overall, 81.8 % of patients in Arm 1 and 63.8 % of patients in Arm 2 completed the 24-week maintenance period. Median reductions in standardized seizure frequency (SSF) were markedly higher in Arm 1 (72.8 %) than Arm 2 (22.8 %), as were responder rates (≥50 % reduction in SSF; Arm 1: 62.5 %; Arm 2: 38.5 %) and rates of seizure freedom (Arm 1: 25.0 %; Arm 2: 9.6 %). Efficacy outcomes were generally more favorable in patients taking ESL in combination with LEV versus other anti-seizure medications (ASMs). Treatment-emergent adverse events (TEAEs; 81 % vs 73 %) and TEAEs leading to discontinuation (16 % vs 2 %) were reported more frequently in Arm 2 than Arm 1, respectively. Serious adverse events were reported infrequently (Arm 1: 0; Arm 2: 7 %). The most common TEAEs were dizziness, nausea, headache, somnolence, fatigue, nasopharyngitis, vomiting, and anxiety. There were no notable changes in depressive symptoms, mood status, or aggression throughout the study. Health and HRQoL scores were generally high at baseline and did not change throughout the study. However, on average, both clinicians and patients perceived improvement in illness over the course of the study.


CONCLUSIONS
ESL was effective and well tolerated both as the first adjunctive therapy to either of the most prescribed first-line ASMs, LEV or LTG, and as a later adjunctive therapy in treatment-resistant patients.


Introduction
Pivotal clinical trials of newer anti-seizure medications (ASMs) are primarily conducted in treatment-resistant adults with a long history of epilepsy. Consequently, uptake of new ASMs as first-or second-line therapies is slow until healthcare professionals have gained substantial clinical experience with the new ASM. Therefore, more efficacy, safety, and tolerability data for third-generation ASMs, in particular, are much needed to inform clinical decision making in patients who are at an earlier stage in their disease trajectory. Examining first adjunctive therapy combinations is of particular interest, due to the potential for drug interactions and the risk of additive side effects when taking ASMs in combination.
Eslicarbazepine acetate (ESL) is a once-daily, oral ASM approved for the treatment of focal (partial-onset) seizures based on clinical studies conducted in both adjunctive and monotherapy settings (BIAL -Portela & C a SA, 2020; Sunovion Pharmaceuticals Inc, 2019). ESL was found to be effective and well tolerated in three pivotal studies of adjunctive ESL and three pivotal studies of ESL monotherapy in adults (Ben-Menachem et al., 2010;Biton et al., 2017;Elger et al., 2009;Jacobson et al., 2015;Sperling et al., 2015a;Sperling et al., 2016;Sperling et al., 2015b;Trinka et al., 2018). In routine clinical practice, healthcare professionals most commonly prescribe ESL as a third-or fourth-line therapy (Mehta et al., 2018).
Pivotal clinical trials are conducted using stringent inclusion and exclusion criteria, and according to strict study protocols (Möller, 2011); these factors reduce bias and facilitate generation of data with high internal validity for regulatory decision making. Consequently, data generated in such clinical trials might not be representative of real-world clinical practice. Therefore, there is a need for additional clinical studies that recruit a full spectrum of patients, with flexible study designs and in a real-world clinical setting. Here, we report outcomes from an open-label, non-randomized, 24-week study in adults at earlier and later stages of their treatment history, conducted in a real-world clinical setting.

Methods
This was an open-label, non-randomized, multicenter, Phase IV study of adjunctive ESL in patients aged ≥18 years (Study 093-701; NCT03116828), conducted at 40 centers in the USA and one in Canada between July 2017 and June 2019. The study was designed, conducted, and monitored in accordance with the principles of the Declaration of Helsinki, the International Conference on Harmonisation guidelines, and relevant national and local laws. The study protocol was approved by the relevant independent ethics committees/institutional review boards, and all patients provided written informed consent.

Patients
Patients were required to have had at least three focal seizures during the previous 6 months. The main inclusion and exclusion criteria are outlined in Table S1. Patients in Arm 1 were required to have been taking levetiracetam (LEV) or lamotrigine (LTG) monotherapy at a stable dose for at least 1 month prior to screening and have no history of current or prior ASM combination therapy. Patients in Arm 2 were required to have been taking stable doses of 1-2 ASMs for at least 1 month prior to screening, with a history (current or previous) of ASM combination therapy. Patients with previous exposure to ESL, current use of oxcarbazepine, or primarily generalized seizure types (e.g., myoclonic, absence, tonic) were excluded.

Study objectives
The primary objective was to evaluate the effectiveness (assessed by retention rate) of once-daily ESL as the first adjunctive therapy to LEV or LTG, or as later adjunctive therapy to 1-2 ASMs in a real-world clinical setting. Secondary objectives were to evaluate efficacy (based on seizure frequency), safety, tolerability, behavioral changes, mood, and healthrelated quality of life (HRQoL) with ESL treatment.

Study design
The study consisted of screening (1-2 weeks), titration (2 weeks), maintenance (24 weeks), and ESL taper/safety follow-up (4 weeks) periods. After screening, oral ESL was initiated at 400 mg/day and titrated to at least 800 mg/day according to the recommendations in the product label, in 400 mg increments per week at most. During the 24-week maintenance period, incremental increases of ESL by 400 mg/day were allowed to the maximum approved dose (1600 mg/day in the USA, 1200 mg/day in Canada) at the discretion of the investigator; one ESL dose reduction was permitted to improve tolerability, to a minimum dose of 800 mg/day. Patients who completed the 24-week maintenance period could choose to remain on treatment with commercial ESL or discontinue ESL (in which case the dose was gradually tapered off); all patients had a 4-week safety follow-up. Patients who prematurely discontinued treatment with ESL were gradually tapered off ESL, unless abrupt discontinuation was required for the treatment of an adverse event (AE), and were withdrawn from the study.
Patients continued to take their baseline ASM(s), generally at the same dose, during the titration and maintenance periods. Patients experiencing intolerable AEs due to a suspected pharmacodynamic interaction between carbamazepine (CBZ) and ESL (in the opinion of the investigator) had the option to reduce the CBZ dose by up to 25 % during the last week of the titration period (week 2) or the first week of the maintenance period (week 3), with the approval of the medical monitor. In addition, patients experiencing intolerable AEs suggestive of phenytoin toxicity had the option to reduce the phenytoin dose by up to 15 %, with the approval of the medical monitor. Patients who became seizure-free during the trial could reduce the dose of baseline ASMs by up to 25 % to improve overall tolerability, at the discretion of the investigator. No other changes in baseline ASM dosing were permitted. After completing the maintenance period, ASM doses could be adjusted at the discretion of the investigator.
It was estimated that if there were a sample size of 100 (approximately evenly split between Arms 1 and 2), and 24-week completion rates were 60 % in Arm 1 and 50 % in Arm 2, the width of the 95 % CIs for completion would be approximately 14 %, based on the normal approximation method.

Assessments and analysis
Patients were evaluated in the clinic at screening and after the start of ESL dosing, at weeks 1, 3, 7, 11, 15, 19, 23, and 27; additionally, there was telephone contact at weeks 9 and 17. The study assessments completed at individual time-points were conducted per protocol. Not all assessments were conducted at all time-points. The safety follow-up was conducted at week 31, either in the clinic or via telephone.
During the screening visit, seizure history was collected using the pre-Human Epilepsy Project (HEP) instrument. As part of the HEP instrument, the Diagnostic Interview for Seizure Classification Outside of Video-Electroencephalogram Recording (DISCOVER) was administered in the absence of a retrospective seizure diary or a baseline period. DISCOVER allowed researchers to ask patients with epilepsy (and/or their caregivers) questions about their seizures, to increase diagnostic accuracy and to standardize seizure classification. This helped the patient/caregiver to use correct and consistent terminology to describe seizure history, and when completing their daily seizure diaries during the study.
During the study, patients/caregivers were required to simultaneously complete a paper seizure diary and an electronic seizure diary (via the Mate App by Empatica Inc.) on a daily basis, regardless of whether a seizure occurred or not. Treatment-emergent AEs (TEAEs) were reported by investigators throughout the study and coded according to the Medical Dictionary for Regulatory Activities.
Demographic and clinical characteristics were recorded at baseline. The primary endpoint was percentage of patients completing the entire 24-week maintenance period (study retention). Secondary efficacy endpoints included standardized seizure frequency (SSF; seizures per 28 days), change from baseline in SSF, responder rate (proportion of patients with ≥50 % reduction in SSF), and seizure freedom rate during the 24-week maintenance period. Efficacy endpoints were evaluated overall and according to concomitant ASM use.
Safety endpoints included overall incidences of TEAEs during the entire study, and incidences according to concomitant ASM use, number of concomitant ASMs, gender, and race. Liverpool AE Profile (LAEP) scores, serious AEs (SAEs), TEAEs leading to discontinuation, and TEAEs of interest for voltage-gated sodium channel (VGSC) blockers (hyponatremia, rash, psychiatric AEs, cognitive-dysfunction related AEs, and suicidality) were also evaluated. To generate LAEP scores, patients assessed the frequency of side effects from treatment by rating 19 items on a four-point Likert scale. Total scores could range from 19 to 76, with higher scores indicating more frequent symptom reporting. Cognitivedysfunction related AEs reported here were identified and selected based on the subset of dose-dependent cognitive AEs that were identified during the pivotal adult studies. Suicidality was assessed using the electronic Columbia-suicide severity rating scale (eC-SSRS).
Assessment of additional endpoints related to mood and behavior used the Neurological Disorders Depression Inventory for Epilepsy (NDDI-E), Profile of Mood States Short-Form (POMS-SF), and Modified Overt Aggression Scale (MOAS) instruments (Curran et al., 1995;Gilliam et al., 2006;Kay et al., 1988). HRQoL was assessed using QoL in Epilepsy Inventory-31 (QOLIE-31), EuroQol five dimensions questionnaire five level (EQ-5D-5L) and visual analogue scale (VAS), Clinical Global Impression of Improvement (CGI-I), and Patient's Global Impression of Change (PGI-C) scores (Cramer et al., 1998;Herdman et al., 2011;Janssen et al., 2013). Scores at baseline and week 27 (the end of the 24-week maintenance period) were summarized according to the scoring rule for each respective instrument, and mean scores were presented. Mood, behavior, and HRQoL scores were evaluated overall and according to concomitant ASM use. Further details of the instruments used to assess mood, behavior and HRQoL are provided in Table S2.
The modified intent-to-treat (mITT) population, defined as all patients who received at least one dose of ESL during the study and had available seizure data, was used for all efficacy analyses. The safety population, defined as all patients who received at least one dose of ESL during the study, was used for all safety analyses. All endpoints were summarized descriptively for Arms 1 and 2 separately, with no statistical comparisons between Arms 1 and 2.

Patients
The mITT population included 96 patients (Arm 1: n = 41; Arm 2: n Fig. 1. Patient flowchart. Arm 1: ESL as first adjunctive therapy to LEV or LTG monotherapy. Arm 2: ESL as later adjunctive therapy. a Two patients had an adverse event leading to discontinuation but 'withdrawal by patient' was listed as the reason for discontinuation. 'Adverse event' was listed as the reason for discontinuation for three patients who had no record of an adverse event leading to discontinuation. ESL: eslicarbazepine acetate; LEV: levetiracetam; LTG: lamotrigine. = 55) and the safety population included 102 patients (Arm 1: n = 44; Arm 2: n = 58; Fig. 1). Baseline demographic and clinical characteristics of the safety population are reported in Table 1; these were comparable in the mITT population. Median age was 40.5 years and 59 % of patients were female. The majority of patients were white (63 %), 24 % were black or African American, and 14 % were Hispanic or Latino. Mean time since epilepsy diagnosis was 9.4 years in Arm 1 and 19.5 years in Arm 2. In Arm 1, 64 % of patients were taking LEV and 34 % were taking LTG, whereas in Arm 2, 41 % of patients were taking LEV and 34 % were taking LTG. In Arm 2, 19 % of patients were taking valproic acid (VPA; all forms) or lacosamide (LCM); 36 % of patients were taking one concomitant ASM, and 64 % were taking >1 concomitant ASM. Few patients used rescue medication at baseline (Arm 1: 4.5 %; Arm 2: 8.6 %). Focal unaware (complex partial) seizures were the most severe seizure type during baseline in the majority of patients (Arm 1: 57 %; Arm 2: 72 %).

Efficacy
Overall, 81.8 % of patients in Arm 1 and 63.8 % of patients in Arm 2 completed the 24-week maintenance period (primary endpoint; Fig. 2A). At 24 weeks, median reductions in SSF were markedly higher in Arm 1 (72.8 %) than Arm 2 (22.8 %; Fig. 2B). At baseline, median SSF was 2.0 in Arm 1 and 2.4 in Arm 2, and mean SSF (standard deviation; SD) was 4.8 (5.2) in Arm 1 and 18.3 (56.4) in Arm 2; at week 24, median SSF was 0.4 in Arm 1 and 2.1 in Arm 2, and mean SSF (SD) was 3.6 (8.6) in Arm 1 and 27.1 (143.5) in Arm 2. Responder rates were higher in Arm 1 (62.5 %) than Arm 2 (38.5 %) during the 24-week maintenance period (Fig. 2C). Rates of 24-week seizure freedom were 25.0 % in Arm 1 and 9.6 % in Arm 2 (Fig. 2D). Fig. 2 also shows the proportion of patients who completed the 24week maintenance period, median changes from baseline in SSF, responder rates, and rates of seizure freedom according to concomitant ASM use. Retention rates with ESL were higher (difference >10 %) in patients taking concomitant LEV than in those taking LTG or any other concomitant ASM, in Arms 1 and 2. In Arm 2, but not Arm 1, the reduction from baseline in SSF was markedly higher (difference >50 %), and the responder rate was higher (difference >20 %), in patients who were taking concomitant LEV than in those who were taking other concomitant ASMs (although some patients taking LEV were also taking other concomitant ASMs). There was little overall change from baseline in SSF in patients taking ESL in combination with VPA or LCM, but there were some responders in these groups (responder rates, VPA: 30 %; LCM: 20 %). In Arm 2, rates of seizure freedom were lower in patients taking concomitant LTG (5.3 %) or LCM (0) than in those taking concomitant LEV or VPA (9.5 % and 10 %, respectively).

Safety and tolerability
TEAEs were reported more frequently in Arm 2 (81 %) than in Arm 1 (73 %). The most frequently reported TEAEs were dizziness, nausea, headache, somnolence, fatigue, nasopharyngitis, vomiting, and anxiety; most were reported more frequently in Arm 2 (Fig. 3). Most TEAEs (>85 %) were mild to moderate in severity. There were no clinically meaningful changes from baseline in total or subscale LAEP scores over the 24-week maintenance period (Fig. S1). There were no deaths during the study.
In Arm 1, TEAEs were reported more frequently in patients who were taking ESL in combination with LTG (93 %) than in those taking ESL in combination with LEV (61 %). In Arm 2, overall TEAE incidence was generally comparable between patients taking ESL in combination with LTG (75 %), LEV (75 %), or VPA (73 %), and somewhat higher in patients taking concomitant LCM (82 %). TEAEs were reported more frequently in patients in Arm 2 who were taking one concomitant ASM (95 %) than in those taking more than one (73 %).
No SAEs were reported in Arm 1 and five SAEs were reported in four patients in Arm 2 (7 %; Table 2). TEAEs leading to discontinuation were reported more frequently in Arm 2 (16 %) than Arm 1 (2 %), and most were reported in one patient each ( Table 2). The one patient with a TEAE leading to discontinuation in Arm 1 was taking concomitant LTG; in Arm 2, TEAEs leading to discontinuation were more frequent in patients taking concomitant LCM (36 %) than in those taking other ASMs (LTG: 15 %; LEV: 13 %; VPA: 9 %). Serum sodium levels below 125 mmol/L were not observed in any patient during the 24-week maintenance period, in either arm; sodium levels decreased by ≥10 mmol/L in two patients in Arm 1 (5 %) and three patients in Arm 2 (6 %). Skin and subcutaneous tissue disorders were reported in 9 % of patients in Arm 1 and 7 % of patients in Arm 2. Rash (Arm 1: 2 %; Arm 2: 3 %) and alopecia (Arm 1: 2 %; Arm 2: 2 %) were reported in more than one patient overall. Psychiatric AEs were reported more frequently in Arm 2 (28 %) than Arm 1 (14 %). Anxiety was the most frequently reported psychiatric AE in Arm 2 (10 %: n = 6); all other psychiatric AEs were reported in ≤1 patient in Arm 1 and ≤2 patients in Arm 2. Two cognitive-dysfunction related AEs occurred in Arm 1 (memory impairment and confusional state, in one patient each [2.3 %]) and Arm 2 (memory impairment, 6.9 %: n = 4; disturbance in attention, 3.4 %: n = 2). There was no evidence of increased suicidality 27 weeks after initiation of treatment with ESL, when assessed using the eC-SSRS (data not shown).

Mood, behavior, and HRQoL
At baseline, mean NDDI-E scores were <16 in both treatment arms, suggesting that most patients did not screen positive for depression (Gilliam et al., 2006). Based on mean NDDI-E total scores, there was no worsening in depressive symptoms between baseline and week 27 in Arm 1 (12 and 10, respectively) or Arm 2 (12 and 12, respectively). According to mean POMS-SF total scores, there were no clinically meaningful changes in mood status between baseline and week 27 in Arm 1 (105 and 108, respectively) or Arm 2 (113 and 120, respectively). NDDI-E and POMS-SF scores were similar across patients taking different baseline ASMs. Median MOAS scores (analysis of aggression) were zero at all visits in Arms 1 and 2. Baseline QOLIE-31 total scores were higher in Arm 1 (69) than in Arm 2 (61; Fig. 4). Changes in mean QOLIE-31 total scores between baseline and week 27 were not clinically meaningful in Arm 1 (69 and 70, respectively) or Arm 2 (61 and 60, respectively; Fig. 4A); changes in total score ≥5.19 are considered clinically meaningful (Borghs et al., 2012). There were clinically meaningful improvements in mean QOLIE-31 total scores between baseline and week 27 in patients who were taking concomitant LEV (60 and 66, respectively) or VPA (61 and 69, respectively) in Arm 2 but not Arm 1 (Fig. 4B). Mean EQ-5D utility was very high at baseline (Arm 1: 0.97; Arm 2: 0.95), week 11 (0.95; 0.95) and week 27 (0.94; 0.96), suggesting near-perfect health. According to patient reporting on EQ-5D VAS, there were no major changes between baseline and week 27 in Arm 1 (83 and 85, respectively) or Arm 2 (80 and 80, respectively).
Through CGI-I assessment, the proportion of patients demonstrating improvement (according to combined scores from the "very much improved", "much improved", and "minimally improved" categories) increased between week 3 and week 27, in Arms 1 and 2 (Arm 1: 53 % and 77 %, respectively; Arm 2: 45 % and 87 %); improvements occurred with all concomitant ASMs, in both arms. Through PGI-C assessment, the proportion of patients who reported an improvement in their illness (according to combined scores from "a great deal better", "better", and "moderately better" categories) increased between week 3 and week 27, in Arms 1 and 2 (Arm 1: 34 % and 48 %; Arm 2: 33 % and 61 %); improvements occurred with all baseline ASMs, except for concomitant LTG in Arm 1.

Discussion
This open-label, non-randomized, 24-week study found that ESL was Fig. 2. (A) Retention rates, a (B) Median change from baseline in SSF, a (C) Responder rates, a,b and (D) Seizure freedom, a overall and according to concomitant ASM use (mITT population). Arm 1: ESL as first adjunctive therapy to LEV or LTG monotherapy. Arm 2: ESL as later adjunctive therapy. a During the 24-week maintenance period. b Responder rate: proportion of patients with ≥50 % reduction in SSF. ASM: anti-seizure medication; ESL: eslicarbazepine acetate; LEV: levetiracetam; LTG: lamotrigine; SSF: standardized seizure frequency.  effective and well tolerated both as the first adjunctive therapy to LEV or LTG, and as a later adjunctive therapy in treatment-resistant adults with focal seizures. This is the first real-world, interventional study to demonstrate that once-daily ESL was effective as the first adjunctive therapy in patients with focal seizures, supporting its consideration in clinical practice. This new study also provides additional supportive evidence that ESL was effective as a later adjunctive therapy in treatment-resistant patients with focal seizures, validating the results of the pivotal studies (Ben-Menachem et al., 2010;Elger et al., 2009;Sperling et al., 2015a) in a patient population representative of US clinical practice and over a longer period of time (6 months). Patient demographics were largely similar between patients taking adjunctive ESL in this trial and the pivotal adult studies (800 and 1200 mg doses) (Biton et al., 2017), with a few exceptions. The current study included a lower proportion of white/Caucasian patients (63 % vs 78 %, respectively) and a higher proportion of black/African American (24 % vs 4 %) and Hispanic or Latino (14 % vs 1 %) patients than the pivotal studies (Biton et al., 2017). More patients were taking baseline LCM (19 % vs 4 %) in the current study, and fewer patients were taking baseline CBZ (9 % vs 49 %) compared with the pivotal studies. These differences in demographics and clinical characteristics likely relate to differences in demographics between Europe and North America (the pivotal studies were largely conducted in Eastern Europe, and the current study primarily in the USA), as well as changes in clinical practice over time; use of CBZ is decreasing over time in the USA and LCM was not commercially available for the majority of the duration of the pivotal studies. Patients taking ESL as a later adjunctive therapy had a longer epilepsy duration than those taking ESL as their first adjunctive therapy. Larger proportions of patients were taking concomitant LEV than LTG in Arms 1 and 2 (64 % vs 34 %, and 41 % vs 34 %, respectively).
The DISCOVER instrument was used to collect seizure history at baseline and aimed to reduce variability in seizure classification across patients and to harmonize terminology across patients/caregivers and investigators. The low median SSF (Arm 1: 2.0; Arm 2: 2.4), low rate of rescue medication use (Arm 1: 4.5 %; Arm 2: 8.6 %), and high mean QOLIE-31 scores (Arm 1: 69; Arm 2: 61) at baseline are strongly suggestive of lesser disease severity than in the pivotal clinical trials (Biton et al., 2017), which is not surprising given that registration trials require a high level of treatment resistance in order for patients to qualify for potential randomization to placebo. In the pivotal studies, baseline median SSF was 8.0-8.5 and mean SSF was 13.1-16.0 across treatment groups (Biton et al., 2017). Due to the availability of numerous alternative therapies and the 6-month duration of the current study, investigators may have been more likely to successfully recruit patients with a relatively lower seizure frequency who did not require urgent intervention. Truly treatment-resistant patients might be expected to use rescue medications more frequently in the clinic than reported here. The baseline seizure frequency in the current study is more representative of a typical clinic population, but may have reduced the potential for observing graded improvement.
Study retention was chosen as the primary endpoint because both efficacy and tolerability/safety typically factor into a patient's willingness to complete a study, and thus would be expected to reflect the effectiveness of ESL in clinical practice. The majority of patients remained in the study in both Arm 1 (82 %) and Arm 2 (64 %); the retention rate was higher in patients taking ESL as their first adjunctive therapy, but the study was not powered to detect a statistical difference between arms. The 82 % 6-month retention rate in Arm 1 is the same as the rate reported during a prospective non-interventional study conducted in Europe in patients taking ESL as their first adjunctive therapy, corroborating this outcome in different patient populations (Holtkamp et al., 2016).
Additional efficacy endpoints demonstrated the effectiveness of ESL as an adjunctive treatment for focal seizures in patients with no history of adjunctive ASM usage. Median reductions from baseline in SSF in Arm 1 (73 %) were more than double those observed in Arm 2 (23 %), and there were more responders in Arm 1 (63 %) than in Arm 2 (39 %). These reductions were observed from already low baseline rates of seizure frequency (SSF: 2.0 in Arm 1; 2.4 in Arm 2), with a final SSF of 0.4 in Arm 1 and 2.1 in Arm 2. The relatively small reduction from baseline in SSF in Arm 2 was likely skewed by a small number of outliers in whom SSF markedly increased from baseline. Responder rates in patients taking ESL as a later adjunctive therapy were comparable to those observed with effective ESL doses in the pivotal studies (39 % vs 32-41 %), whereas median reductions in SSF were slightly lower in the current study (23 % vs 31-33 %), potentially related to the aforementioned differences in baseline seizure frequency between trials (Biton et al., 2017). Rates of seizure freedom were again more than double in Arm 1 (25 %) than in Arm 2 (10 %), compared with 2 % in the pivotal studies (Biton et al., 2017), likely reflecting the type of patients recruited into the trial (those with a relatively low seizure frequency in whom urgent intervention was not required), compared with the treatment-resistant patients recruited into the pivotal clinical trials. Responder and seizure freedom rates were slightly higher in the European prospective observational study than in Arm 1 (82 % vs 63 % and 39 % vs 25 %, respectively) (Holtkamp et al., 2016), potentially due to subtle differences in patient populations between studies.
Efficacy outcomes were generally favorable in patients taking ESL in combination with LEV versus other ASMs. In Arm 1, the retention rate was 89 % in patients taking LEV versus 67 % in patients taking LTG; all other efficacy outcomes were similar regardless of concomitant ASM use. In Arm 2, the retention rate (67 %) was higher, and the reduction from baseline in SSF (69 %) and responder rate (52 %) were substantially higher in patients taking concomitant LEV compared with other ASMs. A potential reason for the increased effectiveness of ESL taken in combination with LEV (vs LTG) could relate to differences in mechanisms of action (MoAs) between ESL and LEV, which are believed to contribute to effects on seizures. ESL and LTG are both thought to exert anti-seizure activity through inhibition of VGSCs (GlaxoSmithKline, 2019; Sunovion Pharmaceuticals Inc, 2019; UCB Inc, 2019); combinations of ASMs with different putative MoAs have been suggested to be more effective than combinations with the same MoA (Margolis et al., 2014;Rogawski et al., 2016). Another finding was that overall SSF did not appear to improve when ESL was taken in combination with LCM (also a VGSC inhibitor) or VPA, although seizure frequency did decrease in some patients, as indicated by responder rates. It is of note that many of the outlier patients mentioned above were taking VPA.
TEAEs and TEAEs leading to discontinuation were more frequent in patients taking ESL as a later adjunctive therapy than in those taking ESL as their first adjunctive therapy (TEAEs: 81 % vs 73 %, respectively; TEAEs leading to discontinuation: 16 % vs 2 %), potentially related to the fact that patients in Arm 2 were taking more concomitant medications (64 % of patients were taking >1 concomitant ASM). Although, curiously, TEAE incidence in Arm 2 was higher in patients taking one concomitant ASM (95 %) than in those taking more than one concomitant ASM (73 %). This could reflect a reduced sensitivity to TEAEs in patients with a higher ASM load; patients are much more likely to report new TEAEs than repetitions of TEAEs that they have already experienced. SAEs were reported infrequently (Arm 1: 0; Arm 2: 7 %). The most frequently reported TEAEs were dizziness, nausea, headache, somnolence, fatigue, nasopharyngitis, vomiting, and anxiety, consistent with the most frequently reported TEAEs during the pivotal studies and with the previously established safety profile of ESL. Most of these TEAEs were mild to moderate in severity and more common in Arm 2 than in Arm 1. In Arm 1, TEAEs were more frequent in patients taking ESL in combination with LTG (93 %) than in those taking ESL in combination with LEV (61 %). This is consistent with a previous analysis which found a potential pharmacodynamic interaction between ESL and LTG (ASMs with a putatively similar MoA) which resulted in a higher incidence of TEAEs typically associated with VGSC-inhibitors (e.g., dizziness, blurred vision, vertigo, diplopia, headache, and vomiting) (Abou-Khalil et al., 2018). In Arm 2, TEAE incidence was comparable between patients taking ESL in combination with LTG and LEV. This could be due to lower doses of LTG in Arm 2 (vs Arm 1), or perhaps related to a reduced sensitivity to TEAEs in patients with a higher ASM load, as discussed above.
Consistent with previous analyses of pivotal study data, hyponatremia and rash occurred in few patients taking ESL (Rogin et al., 2020;Wechsler et al., 2019). A previous analysis of pivotal study data identified that psychiatric TEAEs were also reported infrequently in patients taking ESL, and that there was no clear difference in rates of suicidal ideation and behavior (based on Columbia-Classification Algorithm of Suicide Assessment ratings and C-SSRS scores) between patients taking ESL and placebo (Andermann et al., 2018). In the current study, psychiatric TEAEs were more frequent in Arm 2 (28 %) than in Arm 1 (14 %), perhaps related to the longer duration of disease in Arm 2. It is of note that the current study is twice the duration of the pivotal studies, increasing the likelihood of higher cumulative TEAE rates; in addition, without a placebo arm, it was not possible to determine whether TEAEs were attributable to ESL. Overall, there was no evidence of increased suicidality (assessed via the eC-SSRS) with ESL, although one patient in Arm 2 discontinued treatment due to suicidal thoughts. The patient was taking concomitant LTG and topiramate and was using diazepam as a rescue medication; the event was assessed as moderate in severity and not related to ESL.
Based on NDDI-E scores, patients did not have major depression at baseline and symptoms of depression did not change throughout the study, with the majority of patients remaining below the threshold score of >15 (suggestive of major depression (Gilliam et al., 2006)). Similarly, based on POMS-SF scores, on average, mood status did not change during the study. The MOAS scale was used to evaluate aggression, and scores remained at zero throughout the study.
At baseline, HRQoL measured by QOLIE-31 was higher in Arm 1 (69) than in Arm 2 (61), likely due to patients in Arm 2 having more severe disease and using more concomitant ASMs. Overall, there was no clinically meaningful change in HRQoL (≥5.19 points (Borghs et al., 2012)) over the course of the study in either arm. This was not unexpected due to the high QOLIE-31 scores at baseline; in the pivotal studies of ESL, the average QOLIE-31 score at baseline was 57 (Velez et al., 2017). However, there were clinically meaningful improvements in HRQoL in Arm 2 in patients who were taking concomitant LEV or VPA.
EQ-5D utility scores suggested that patients' quality of life did not significantly change over the course of the study. In the general US population, EQ-5D values range from -0.287 to 0.992 (Craig and Rand, 2018). Baseline EQ-5D utility in the current study was 0.97 in Arm 1 and 0.95 in Arm 2, suggesting near-perfect health on the EQ-5D scale; therefore, there was limited room for improvement during follow-up. However, both clinicians and patients perceived an improvement in illness over the course of the study, when assessed via the CGI-I and PGI-C, in both Arms 1 and 2. Interestingly, both clinician-determined and patient-perceived improvements in illness were more frequent in patients taking ESL as a later versus first adjunctive therapy (CGI-I: 42 % vs 24 %; PGI-C: 28 % vs 14 %), despite the greater improvement in seizure frequency and fewer AEs in Arm 1. This could suggest that patients with a longer duration of epilepsy and a more extensive treatment history, as well as the clinicians treating these patients, are more likely to assess even modest changes in seizure frequency as improving their overall status.
Several limitations of the current study should be noted. Per the study design, all patients were taking ESL with no placebo control. The sample size was small and the study was not powered to detect differences between arms; it was not feasible to recruit the very large sample size that would have been required to power the study. Therefore, data are summarized descriptively, with no statistical comparisons. Disease severity at baseline was generally mild, reducing the potential for observing improvement in seizure control. This is likely a reflection of reduced motivation to participate in a clinical trial when ESL was readily available by prescription. Finally, enrollment was closed early due to challenges with recruitment, limiting overall enrollment to ~50 % of the intended sample size.

Conclusion
In this real-world clinical trial conducted in patients with focal seizures in the USA and Canada, ESL was effective and well tolerated both as the first adjunctive therapy to either of the most commonly prescribed first-line ASMs, LEV or LTG, and as a later adjunctive therapy. This is the first study to assess the clinical effectiveness of ESL in North American patients naïve to adjunct ASM therapy, and supports the use of ESL in patients considering a first add-on therapy for seizures. Assessment of the clinical effectiveness of ESL as a later adjunctive therapy provides additional evidence of clinical effectiveness in treatment-resistant patients in a real-world clinical setting.

Funding
The study and article preparation were funded by Sunovion Pharmaceuticals Inc. Sunovion Pharmaceuticals Inc. was involved in designing the study, collecting, analyzing and interpreting data, writing the report, and the decision to submit the article for publication.

Data availability statement
Sunovion Pharmaceuticals Inc. is part of a clinical trial data-sharing consortium that facilitates access for qualified researchers to selected anonymized clinical trial data. For up-to-date information on data availability please visit: https://www.clinicalstudydatarequest.com/ Study-Sponsors.aspx and click on Sunovion.