Retention of brivaracetam in adults with drug-resistant epilepsy at a single tertiary care center

INTRODUCTION
Brivaracetam (BRV) is licensed as an adjunctive treatment for focal epilepsy. We describe our clinical experience with BRV at a large UK tertiary center.


METHODS
Adults initiated on BRV between July 2015 and July 2020 were followed up until they discontinued BRV or September 2021. Data on epilepsy syndrome, duration, seizure types, concomitant and previous antiseizure medication (ASM) use, BRV dosing, efficacy, and side effects were recorded. Efficacy was categorized as temporary (minimum three months) or ongoing (at last follow-up) seizure freedom, ≥50% seizure reduction, or other benefits (e.g., no convulsions or daytime seizures). Brivaracetam retention was estimated using Kaplan-Meier survival analysis.


RESULTS
Two-hundred people were treated with BRV, of whom 81% had focal epilepsy. The mean (interquartile range [IQR]) follow-up time was 707 (688) days, and the dose range was 50-600 mg daily. The mean (IQR) of the previous number of used ASMs was 6.9 (6.0), and concomitant use was 2.2 (1.0). One-hundred and eighty-eight people (94%) had previously discontinued levetiracetam (LEV), mainly due to side effects. 13/200 (6.5%) were seizure free for a minimum of six months during treatment, and 46/200 (23%) had a ≥50% reduction in seizure frequency for six months or more. Retention rates were 83% at six months, 71% at 12 months, and 57% at 36 months. Brivaracetam was mostly discontinued due to side effects (38/75, 51%) or lack of efficacy (28/75, 37%). Concomitant use of carbamazepine significantly increased the hazard ratio of discontinuing BRV due to side effects (p = 0.006). The most commonly reported side effects were low mood (20.5%), fatigue (18%) and aggressive behavior (8.5%). These side effects were less prevalent than when the same individuals took LEV (low mood, 59%; aggressive behavior, 43%). Intellectual disability was a risk factor for behavioral side effects (p = 0.004), and a pre-existing mood disorder significantly increased the likelihood of further episodes of low mood (p = 0.019).


CONCLUSIONS
Brivaracetam was effective at a broad range of doses in managing drug-resistant epilepsy across various phenotypes, but less effective than LEV in those who switched due to poor tolerability on LEV. There were no new tolerability issues, but 77% of the individuals experiencing side effects on BRV also experienced similar side effects on LEV.


Introduction
Approximately 1 in 3 people with epilepsy fail to become seizure free after treatment with two appropriate antiseizure medications (ASMs) [1], resulting in significant morbidity, mortality, and lower quality of life [2]. Brivaracetam (BRV), an analog of levetiracetam (LEV), is licensed as an adjunctive treatment for focal epilepsy since 2015. Brivaracetam binds to the synaptic vesicle protein 2A via a different target site with a higher affinity than LEV [3][4][5]. Phase III trials suggested efficacy in people with drug-resistant focal epilepsy [6][7][8] with a lower incidence of neuropsychiatric side effects than LEV [9].
Data on ''real world" experiences are an essential complement to regulatory trials. Our group has extensive experience assessing https the retention of new ASMs as we have carried out these assessments since the 1990s [10][11][12][13][14][15]. We have now evaluated the effectiveness, retention, and tolerability of BRV in adults with various epilepsy syndromes followed up for up to 6 years, including its use at higher doses and those with intellectual disabilities.

Material and methods
This was an observational study of adults 17 years and older with epilepsy who received their first prescription for BRV at the epilepsy specialist clinics at the National Hospital for Neurology and Neurosurgery (Queen Square and Chalfont sites) between July 2015 and July 2020. They were followed up until they discontinued BRV or until September 2021. They were identified in the central pharmacy prescription records and the departmental clinical database. Using the electronic health record system, we collected: demographic and baseline clinic information (including epilepsy syndrome and seizure type, any previous epilepsy surgery or vagal nerve stimulator, the presence/absence of intellectual disability or coexistent mental health conditions, and previous and current ASMs), response to LEV (including the effect on seizure frequency and side effects), BRV dose, duration of treatment, impact on seizure frequency, other clinical benefits (improvement in concentration, mood, seizure severity), reported adverse events (general physical and neuropsychiatric), reasons for discontinuing BRV, and concomitant medication changes made while on BRV. Seizure frequency was binarized into these mutually inclusive categories: seizure freedom (for a minimum of 6 months), 50% reduction, any seizure reduction, and seizure aggravation during treatment. Individuals were asked about side effects in the clinic, and responses were qualitatively collected and categorized according to themes. One condition of BRV prescription in our center was that LEV had previously been tried with some benefits being observed, but the individual had discontinued it due to tolerability issues, with some exceptions. SPSS (SPSS, Inc.) was used for statistical analysis, and SPSS and GraphPad Prism (GraphPad, Inc.) were used to produce figures. We used Kaplan-Meier survival analysis to estimate retention rates of BRV. We used a binary logistic regression model to ascertain how increasing drug dose (independent continuous variable) affected side-effect prevalence (dependent binary variable). Two binary logistic regression models were also used to assess whether certain covariates were associated with behavioral change (model one) and low mood (model two, both binary dependent variables) on BRV. We chose categorical covariates as independent variables for these two models. We thought they would have the strongest influence on behavioral change and low mood: sex, presence of intellectual disability, having an established psychiatric disorder, and previously experiencing behavioral change or low mood on LEV. We used Cox regression analysis to assess the impact of different covariates on drug retention. We used two models, one investigating the effects of concomitant medication (the most prescribed medications were chosen, all were coded as binary independent variables) and another exploring the impact of clinical and demographic features (continuous independent variable: age; categorical independent variables: sex, presence of intellectual disability, experiencing side effects, and experiencing seizure freedom on BRV) on the hazard ratio of discontinuing BRV. We chose these independent variables because we thought they were most likely to be associated with drug retention.
Our Clinical Audit and Quality Improvement Committee approved this project as an audit of outcomes to inform the place of BRV in the antiseizure armamentarium.

Effectiveness of BRV
The effectiveness of BRV on seizure frequency is shown in Table 2. Two of the four seizure-free people on LEV suffered a seizure relapse after switching to BRV. While 44 of 200 (22%) reported a period of seizure aggravation during treatment, this was sustained and led to drug discontinuation in 16 (8%). Individuals described improvements other than improved seizure control when taking BRV compared to LEV, including improved mood (25/200, 12.5%), less challenging behavior (15/200, 7.5%), having more energy (15/200, 7.5%), and improved cognition (8/200, 4%). Of the 31 individuals with genetic generalized epilepsy, four were seizure free, with seven having a >50% reduction in seizures and 21 reporting a decrease in seizure frequency.

Retention and side effects from BRV
The duration of BRV treatment ranged from 2 days to 6.2 years. Retention rates during the first 36 months of treatment are shown in Fig. 2 and Table 3. Reasons for stopping BRV included side effects (38/75, 51%), no benefit (28/75, 37.3%), and seizure aggravation (16/75, 21.3%). A range of general physical and neuropsychiatric side effects was observed (Table 4). We used a binomial logistic regression to investigate how increasing total daily doses of BRV affected the likelihood of experiencing drug side effects. This suggested that increasing doses were associated with a small but significant reduction in side effect prevalence (Exp (B) 0.996 [0.993-0.999], p = 0.004). Lastly, we tested the effects of several covariates on the retention of BRV (Table 5).

Neuropsychiatric side effects of BRV
Using binary logistic regression, we assessed whether several covariates affected the likelihood of developing low mood or  (1) For all percentages, the denominator was 200 (our full sample size). SD = Standard Deviation.  behavioral side effects (Table 6). Overall model fit (Hosmer-Lemeshow test for goodness-of-fit) was significant when low mood (chisquare df 14.645 5 , p = 0.012) and behavioral side effects (chisquare df 11.649 5 , p = 0.040) were used as dependent variables.
Having an intellectual disability significantly increased the likelihood of becoming irritable or aggressive on BRV (hazard ratio (95% CI) 3.62 (1.50-8.74), p = 0.004). An established mood or anxiety disorder significantly increased the likelihood of developing low mood on BRV (hazard ratio (95% CI) 2.02 (1.12-3.65), p = 0.019). We also found that people who complained of low mood from LEV treatment were less likely to complain of low mood on BRV (hazard ratio (95% CI) 0.56 (0.31-1.00), p = 0.048). To investigate drug cross-over effects, we compared side-effect prevalence between people who had switched directly from LEV to BRV because of side effects (n = 76) versus those who had previously taken LEV but not within one month of starting BRV (n = 112). There was no significant difference between these groups (26.7% vs 38.9%, chi-square df 3.023 1 , p = 0.087). Therefore, no evidence of a cross-over effect for mood or behavioral side effects was observed. In those directly switching from LEV, the prevalence of general physical side effects was significantly lower (20.0% vs 51.3%, chi-square df 18.626 1 , p = <0.001). There was no difference in the likelihood of discontinuing BRV between direct and non-direct switchers (chi-square df 0.053 1 , p = 0.818).

Discussion
We assessed outcomes of BRV treatment for refractory epilepsy at a single tertiary center. A broad range of clinical phenotypes were evaluated, including people with generalized epilepsy, on more concomitant medications and over more extended followup periods, making this cohort more heterogeneous than previous observational studies [16][17][18][19][20][21][22][23][24][25]. Most of the cohort had previously discontinued LEV due to tolerability issues.
Brivaracetam was effective and well-tolerated in this drugresistant cohort, with acceptable retention rates over 36 months of treatment. Our responder rate is comparable to previous cohort studies [16][17][18][19][20][21][22][23][24][25]. For instance, a study reported a responder rate of 53.7% at 36 months [25], comparable to our rate of 57%. The retention rates in this cohort, of 57% at 36 months, were higher than those observed in our historical cohort of people on perampanel (43%) and lacosamide (35%) [26,27]. The comparable rate at 36 months for LEV in our series was 58% [28]; however, our current cohort was smaller and consisted almost entirely of people who had previously discontinued LEV.
The similarity in responder rates between focal and generalized epilepsy syndromes is in keeping with a recent report [21]. In contrast to previous studies, we observed similar responder rates between people with and without intellectual disability [21,22].
During the first six-month follow-up, almost a fifth discontinued BRV, most commonly due to side effects, no improvement in seizure activity, and seizure exacerbation. Two of the four seizure-free people on LEV who were switched over to BRV experienced further seizures. We found that concomitant use of carbamazepine was significantly associated with discontinuation of BRV while clinical outcomes, demographics, and other concomitant ASM use were not. Interestingly oxcarbazepine, a structural derivative of carbamazepine, did not share this association. Carbamazepine and oxcarbazepine are metabolized differently by the cytochrome P450 system to their primary active metabolites (carbamazepine-10,11-epoxide for carbamazepine and 10monohydroxy derivative carbazepine); [29,30]). The link between carbamazepine use and discontinuation of BRV could relate to the adverse neurological effects associated with carbamazepine-10,11-epoxide [31].
Side-effect profiles were comparable to previous reports [9]. Skin rash during BRV treatment has been previously reported in a cohort study, with similar prevalence [21], but not in an extensive, pooled analysis of clinical trials [9]. Skin rashes observed included maculopapular rash, some requiring dermatological review. We did not find that side-effect prevalence increased with higher doses of BRV, suggesting that higher drug doses within our treatment range are tolerable.
Levetiracetam is associated with neuropsychiatric side effects, including low mood, aggression, and irritability. A frequency of 10% was reported [32], which can be severe enough to lead to dose reduction or drug discontinuation [33]. We cannot comment on differences in the prevalence of neuropsychiatric side effects in people naïve to BRV and LEV. Neuropsychiatric side effects seem less prevalent on BRV than when the same individuals took LEV. This is consistent with recent observational cohort studies [34][35][36], including data from people who had directly switched drugs or had previous LEV exposure, suggesting that BRV has a favorable side-effect profile. These improvements may reflect reductions in seizure frequency and off-target pharmacodynamic effects. We also observed that those reporting low mood on LEV were significantly less likely to report this on BRV. This may be due to different affinities between these medications at various central neurotransmitter receptors [37]. Alternatively, this may reflect confirmation bias since BRV was associated with less mood disturbance than  LEV before starting treatment. Lastly, we found that people with intellectual disabilities were significantly more likely to have behavioral disturbances on BRV. This reflects the increased prevalence of challenging behaviors among people with an intellectual disability [38] and underscores the importance of training in managing challenging behavior [39]. One possible reason for the high retention of BRV could be that no new ASM had been licensed since BRV until the study ended. The lack of further pharmacological options may have contributed to the duration of BRV therapy. Furthermore, most individuals switching from LEV to BRV did so due to tolerability issues instead of a lack of efficacy with the former. Therefore, those in whom the adverse effects were mitigated were more likely to persevere on BRV.
Our report provides one of the most extended follow-up periods and the widest dose range compared to previous reports. Almost all people in our cohort had previously trialed and discontinued LEV. We, therefore, assessed a group with relatively high sideeffect prevalence, who are more likely to reflect real-life BRV candidates compared with LEV-naïve groups. This also allowed us to make within-subject observations on the relative effectiveness and side-effect profile between BRV and LEV.
Our study has limitations. As deriving from an observational cohort, our data are not directly powered to compare BRV and LEV. This should be a subject for future investigation, especially comparing their efficacy and tolerability at varying doses, seizure freedom rates, and effect on specific epilepsy syndromes. Retrospective data on seizure count are also subject to recall bias. Due to the high proportion of concomitant treatment changes, perhaps not unexpected considering the length of follow-up and the refractory nature of epilepsy in this cohort, it is also difficult to directly ascribe all changes in seizure frequency and side effects to BRV alone. Lastly, while we investigated how drug dose was associated with side-effect prevalence, we did not grade side-effect severity and therefore did not investigate the relationship between drug dose and side-effect severity.
In conclusion, we provide observational evidence for good retention, responder rates, and neuropsychiatric side-effect prevalence of variable BRV treatment doses in a heterogenous, pharmacoresistant cohort. Our real-world data supplement trial data and could be generalized to a clinic setting, especially for people with epilepsy who have not tolerated LEV therapy.

Funding
This work did not receive any specific grant from funding agencies in the public or charitable sectors. It was performed at the UCLH/UCL Comprehensive Biomedical Research Centre, which receives a proportion of funding from the UK Department of Health Research Centres' funding scheme. SFG is funded by a National Institute for Health Research (NIHR) Academic Clinical Fellowship. MJK, FRG, and JWS are supported by the Epilepsy Society, which has received funding from UCB, the manufacturers of brivaracetam (UCB has had no involvement in the design, conduct, analysis or interpretation of the data for the work nor drafting or approving any associated publications). SR receives support from the NIHR. JWS receives support from the Dr. Marvin Weil Epilepsy Research Fund and the Christelijke Verenigingvoor de Verpleging van Lijdersaan Epilepsie, Netherlands.

Declarations of interest
JWS reports personal fees as speaker or consultant from Arvelle, Eisai, GW Pharmaceuticals, UCB Pharma, and Zogenix. MJK reports personal fees as speaker or consultant from Arvelle, BIAL, Eisai, GW Pharmaceuticals, Novartis and UCB Pharma.