3.2.1. Description of Studies

Four multi-centre, double-blind, placebo-controlled, parallel-group phase 3 randomized controlled trials (RCTs) were included in the systematic review: Study 1252 (N = 399),^14,22 Study 1253 (N = 400),^15,23 Study 1254 (N = 480),^16,24 and Study 1358 (N = 768).^17,25 All trials investigated the efficacy, safety and tolerability of brivaracetam given as adjunctive therapy (i.e., added on to a background regimen of one to three AEDs) in patients aged 16 years and older for the treatment of uncontrolled POS. Studies 1252 and 1253 were identical in design, with the exception of the brivaracetam doses evaluated (figures 2 and 3). Study 1358 was also of similar design, but included patients up to 80 years of age, whereas all the other trials included only patients up to age 70 (Figure 5). Study 1358 also investigated the highest recommended dosage of brivaracetam (200 mg/day) and excluded patients who had been exposed to levetiracetam within the past 90 days. With the exception of Study 1254, patients were randomized to the full dose of brivaracetam without any up-titration phase. Study 1254 was a flexible-dose trial that included both a dose-finding (up-titration) phase and a maintenance phase during the treatment period (Figure 4). All four trials were published in the peer-reviewed medical literature.^22–25

Figure 2

Design of Study 1252. D = dose; V = visit; W = week. Source: Study 1252 Clinical Study Report.

Figure 3

Design of Study 1253. D = dose; V = visit; W = week. Source: Study 1253 Clinical Study Report.

Figure 5

Design of Study 1358. D = dose; V = visit; W = week. Source: Study 1358 Clinical Study Report.

Figure 4

Design of Study 1254. BRV = brivaracetam; DL = dose level; EDV = early discontinuation visit; LTFU = long-term follow-up; PBO = placebo; V = visit. Source: Study 1254 Clinical Study Report.

Study 1252 was a 24-week, therapeutic confirmatory RCT (Figure 2). Patients were enrolled and entered an eight-week baseline period, at the end of which 399 patients were randomized in a 1:1:1:1 fashion to one of four treatment groups (brivaracetam 20 mg/day, 50 mg/day, 100 mg/day, or matching placebo) Randomization was stratified by geographical region and by use of concomitant levetiracetam at study entry.

Study 1253 was a 23-week RCT that was identical in design to Study 1252, with the exception of the brivaracetam doses studied (Figure 3). Enrolled patients entered an eight-week baseline period, at the end of which 400 patients were randomized in a 1:1:1:1 fashion to one of four treatment groups (brivaracetam 5mg/day, 20 mg/day, 50 mg/day, or matching placebo). Randomization was also stratified by geographical region and by use of concomitant levetiracetam at study entry.

Study 1254 was a 19-week, flexible-dose study conducted to determine the safety and tolerability of brivaracetam in patients with localization-related or generalized epilepsy (Figure 4). A secondary objective was to confirm the efficacy of brivaracetam in reducing POS, and data from the POS population are reported in this review. Patients entered a four-week baseline period after which they were centrally randomized 3:1 to either brivaracetam 20 mg/day or matching placebo. Randomization was stratified by type of epilepsy, geographic region, and concomitant levetiracetam use at study entry. Following the baseline period, patients entered an eight-week dose-finding phase during which brivaracetam was initiated at 20 mg/day and up-titrated in a stepwise manner to either brivaracetam 50 mg/day, 100 mg/day or 150 mg/day based on the investigator’s assessment of efficacy and tolerability, or matching placebo. Patients then entered an eight-week maintenance period at the last dose reached in the dose-finding period.

Study 1358 was a 26-week trial that evaluated two active doses of brivaracetam (100 mg/day or 200 mg/day). Upon entry, patients completed an eight-week baseline period, after which they were centrally randomized 1:1:1 to either a brivaracetam dose or matching placebo. Randomization was stratified by country, levetiracetam use status, and number of AEDs previously used but discontinued prior to study entry. This was followed by a 12-week treatment period as illustrated in Figure 5. Study 1358 differed from the other trials in that patients exposed to levetiracetam for 90 days or more before the first study visit were excluded from entering the study.

Following the treatment periods of studies 1252, 1253, and 1358, patients had the option of continuing in long-term follow-up (LTFU) studies or undergoing a down-titration phase of one to two weeks and a study drug-free period of two weeks. Pooled data from the LTFU studies have been published and the findings are summarized in Appendix 6: Summary of Other Studies. In Study 1254, patients did not enter a LTFU study; instead, they underwent a three-week down-titration period followed by a two-week study drug-free period. Only results from the treatment groups of the trials that correspond with the Health Canada–approved doses for brivaracetam (i.e., 50 mg to 200 mg daily in two divided doses) are reported in this review.

3.2.2. Populations

b. Baseline Characteristics

Patient populations were similar across the four trials and baseline characteristics appear to be well balanced between the treatment groups in individual trials, as detailed in Table 5. There was fairly equal representation of male and female patients in all treatment groups across the trials. The mean age of included patients ranged from 36 to 39 years of age and the majority of included patients were Caucasian (approximately 70%), with the exception of Study 1254 where only approximately 57% of patients were Caucasian, followed by approximately 42% Asian patients.

Table 5

Summary of Baseline Characteristics.

The duration of epilepsy in enrolled patients ranged from 20 to 26 years across the trials. The age of onset of epilepsy was in the teenage years, ranging from 15 to 17 years in studies 1252 and 1358, and 12 to 14 years in studies 1253 and 1254. Across all the trials, enrolled patients had epilepsy for more than half their lives. As per the inclusion criteria, all patients had POS, with the majority of patients having complex partial seizures (type 1B) and partial seizures that secondarily generalize (type 1C). The proportion of patients with GS (i.e., approximately 1.5% to 8%), was low and similar between treatment groups in the trials. Although the exclusion criteria stated that patients with cluster seizures should be excluded, in Study 1358, approximately 3% to 6% of patients in the individual treatment groups had cluster seizures at baseline.

The history of use of previous AEDs (i.e., up to five years prior to baseline) by the number of AEDs used is presented in Table 6. The most common AEDs used (i.e., taken by at least 5% of patients) prior to study entry are summarized by generic drug name and treatment group in Table 7. The majority of patients in all trials (i.e., approximately 70% or more) had historically used two or more AEDs. In studies 1252, 1253, and 1254, approximately 10% to 19% of patients had historically tried five or more AEDs, whereas in Study 1358, approximately 47% of patients had used five or more AEDs. The most common AEDs that had been used were levetiracetam (21.8% to 54.8%), carbamazepine (19.4% to 51.0%), valproic acid (18.0% to 51.0%), topiramate (17.3% to 43.4%), and lamotrigine (13.9% to 43.4%). Of note, prior clobazam use was only reported in Study 1358, where between 16.2% and 18.1% of patients in the treatment groups had previous use.

Table 6

History of Number of Previous Antiepileptic Drugs at Study Entry (ITT Population or ITT POS Population).

Table 7

History of Previous Antiepileptic Drug Use at Study Entry (by at Least 5% of Overall Patients) (ITT Population or ITT POS Population).

▬

All patients were taking at least one concomitant AED at baseline (Table 10). The most frequently used concomitant AEDs were carbamazepine (37.1% to 48.5%), lamotrigine (20.0% to 29.6%), valproic acid (15.3% to 31.9%), oxcarbazepine (9.9% to 22.0%), and topiramate (3.0% to 22.2%). Of note, concomitant levetiracetam was used by 18.0% to 20.4% of patients in studies 1252, 1253, and 1254 (i.e., use of levetiracetam was an exclusion criterion in Study 1358). The frequency of use of these AEDs was generally similar across treatment groups in the individual trials — with the possible exception of more frequent topiramate use in the placebo groups, as compared with the brivaracetam groups — of studies 1252, 1253, and 1358 (Table 10).

Table 10

Summary of Concomitant Antiepileptic Drug Use during Baseline (by at Least 3% of Overall Patients) (ITT Population or ITT POS Population).

Table 8History of Vagal Nerve Stimulation at Study Entry (ITT Population or ITT POS Population)

View in own window

▬	▬			▬		▬		▬
	▬	▬	▬	▬	▬	▬	▬	▬	▬	▬
▬	▬	▬	▬	▬	▬	▬	▬	▬	▬	▬
▬	▬	▬	▬	▬	▬	▬	▬	▬	▬	▬
▬	▬	▬	▬	▬	▬	▬	▬	▬	▬	▬

ITT = intention to treat; POS = partial-onset seizure; VNS = vagal nerve stimulation.

a

Studies 1252, 1253, and 1254 did not collect information on the status of VNS (i.e., active or not active). For these studies, patients with medical history indicating prior VNS implantation are assumed to be active VNS unless they also had medical history of VNS removal, in which case they are included in the no-VNS group. According to the manufacturer, the data are based on data from the Clinical Study Reports as well as pooled data from the long-term follow-up studies, which provided additional information regarding prior VNS surgery.

Source: UCB Canada response to request for additional information.²⁸

Table 9History of Surgical Procedures at Study Entry (ITT Population or ITT POS Population)

View in own window

▬	▬			▬		▬		▬
	▬	▬	▬	▬	▬	▬	▬	▬	▬	▬
▬	▬	▬	▬	▬	▬	▬	▬	▬	▬	▬
▬	▬	▬	▬	▬	▬	▬	▬	▬	▬	▬

ITT = intention to treat; POS = partial-onset seizure.

Source: UCB Canada response to request for additional information.²⁹

3.2.3. Interventions

In all four trials, the intervention was brivaracetam oral tablets (i.e., 5 mg/day to 200 mg/day in two divided doses depending upon the individual trial). As previously stated, only results pertaining to the Health Canada–approved doses for brivaracetam are reported in this review (i.e., 50 mg/day to 200 mg/day). In studies 1252, 1253, and 1358, patients were randomized to the full dose of brivaracetam or matching placebo without an up-titration phase. In Study 1254, during the dose-finding period, patients started on brivaracetam 20 mg/day or matching placebo and then had the dose escalated to the next level (brivaracetam 50 mg/day, 100 mg/day, or 150 mg/day or matching placebo), based on the investigator’s assessment of efficacy and tolerability. At the end of the dose-finding period, patients were maintained on the final dose during the maintenance period. In studies 1252, 1253, and 1254, during the treatment period, the dose could be reduced using a fallback option as follows: brivaracetam 150 mg/day to 100 mg/day, brivaracetam 100 mg/day to 50 mg/day, brivaracetam 50 mg/day to 20 mg/day, brivaracetam 20 mg/day to 5 mg/day or placebo, brivaracetam 5 mg/day to placebo, or from placebo to placebo. The fallback option could be exercised only once and, once it had occurred, the dose was kept stable for the rest of the treatment period. The comparator in all four trials was matched oral placebo tablets.

Patients were permitted to be on at least one, but not more than two (Studies 1252, 1253, and 1358) or three (Study 1254), concomitant AEDs. Provided patients were at a stable dose for at least one month (or three months for phenobarbital or primidone) prior to the first visit, and kept stable during the entire study period, the following concomitant AEDs were allowed during the study: carbamazepine, clobazam, clonazepam, diazepam, ethosuximide, gabapentin, lamotrigine, levetiracetam, oxcarbazepine, phenobarbital, phenytoin, pregabalin, primidone, tiagabine, topiramate, valproic acid, and zonisamide. Prohibited AEDs included felbamate (unless used continuously for more than 18 months before the first visit) and vigabatrin.

Benzodiazepines were considered concomitant AEDs if taken more than once a week, for any indication. Vagal nerve stimulation was allowed if it had been implanted at least nine months prior to entering the study, and if was stable for one month prior to the first study visit and was kept stable during the entire study period.

3.2.4. Outcomes

The primary efficacy outcome was POS (type 1) frequency per week (Studies 1252 and 1253) or per 28 days (Study 1358). In Study 1254, the primary outcome was safety and tolerability; however, the key secondary outcome (or primary efficacy outcome) was the POS (type 1) frequency per week. Other measures of seizure reduction were the seizure-freedom rate (i.e., the proportion of patients who were free of all types of seizures [type 1, 2, and 3]), 50% responder rate (i.e., the proportion of patients with ≥ 50% reduction in POS frequency per week), and the per cent reduction or categorized per cent reduction in POS (type 1) frequency over the treatment period. A 50% reduction in POS frequency was considered to be clinically meaningful, according to the clinical expert consulted for this review. To record seizure frequency, patients filled in a daily record card that was returned at each study visit. The investigator assessed the seizures according to the International League Against Epilepsy codes and recorded the seizure types and frequency on the case report form.

For Study 1358, the primary efficacy outcome for the United States was the per cent reduction in POS (type 1) frequency over placebo based on analysis of covariance. The primary efficacy outcome for the European Union was the 50% responder rate based on per cent reduction in POS (type 1) frequency from baseline to the end of the 12-week treatment period. ▬¹⁷

▬ As detailed in Appendix 5: Validity of Outcome Measures, the Patient-Weighted Quality of Life in Epilepsy Inventory-31 (QOLIE-31-P) is an adaptation of the original QOLIE-31 instrument.³⁰ Scoring requires the conversion of raw data to a scale of 0 to 100 for each sub-scale, with higher scores reflecting higher health-related quality of life (HRQoL), and lower scores reflecting worse HRQoL. The maximum total score is 100 per sub-scale and total score.³⁰ No information on the validity and minimal clinically important difference (MCID) of the QOLIE-31-P in patients with epilepsy was identified. The MCID of the original QOLIE-31 instrument using various methodologies is reported to range from 4.73 to 11.8.^31,32

The EuroQol 5-Dimensions Questionnaire (EQ-5D) is a generic HRQoL instrument that has been applied to a wide range of health conditions and treatments.^33,34 ▬ The EQ-5D visual analogue scale (VAS) comprises a 20 cm scale with end points labelled 0 and 100, with respective anchors of “worst imaginable health state” and “best imaginable health state.” Respondents rate their own health on that day using the EQ-VAS. Thus, lower scores indicate poorer health and higher scores, better health. As per Appendix 5, the validity and MCID of the EQ-5D have not been formally assessed in patients with epilepsy.

▬ The Hospital Anxiety and Depression Scale (HADS) is an instrument that assesses the presence and severity of anxiety and depressed mood. It consists of 14 items that are scored on a four-point severity scale ranging from 0 to 3. A score per dimension (anxiety, depression) can range from 0 to 21, with higher scores indicating higher depression/anxiety. A score of 0 to 7 for either the anxiety or depression sub-scale is suggested as being in the normal range, a score of 8 to 10 suggests the presence of the respective state, and a score of 11 or higher indicates the probable presence of the mood disorder.³⁵ The HADS depression sub-scale was validated and found to be a reliable instrument to screen for depressive disorders in patients with epilepsy.³⁶ No MCID has been identified for the HADS in epilepsy patients.

▬ The Patient Global Evaluation Scale (P-GES) and Investigator Global Evaluation Scale (I-GES) both use a seven-point scale, with the start of the study as the reference time point. The patient completes the P-GES by answering the following question: “Overall, has there been a change in your seizures since the start of the study drug?” The investigator completes the I-GES by responding to the following: “Assess the overall change in the severity of patient’s illness, compared to the start of study drug.” Numerical values are assigned to each response as follows: 1 = marked worsening, 2 = moderate worsening, 3 = slight worsening, 4 = no change, 5 = slight improvement, 6 = moderate improvement, and 7 = marked improvement. As per Appendix 5, no information on the validity and the MCID of the P-GES and I-GES in patients with epilepsy was identified.

▬

Safety variables included AEs, laboratory tests (e.g., blood chemistry, hematology, urinalysis, etc.), electrocardiogram, vital signs, body weight, physical examination, neurological examination, mental status, and psychiatric status.

3.2.5. Statistical Analysis

In all four included trials, sample size considerations were based on the primary efficacy variable. For studies 1252 and 1253, an estimate of the standard deviation for seizure frequency, based on data from phase 2 trials, was ▬. With a power of 90% and a two-sided level of significance of 5%, 87 patients per group were required to detect a treatment difference of −0.223 in log-transformed seizure frequency per week between brivaracetam and placebo. The treatment difference of −0.223 in log-transformed seizure frequency corresponds to a 20% reduction over placebo. Since the various doses of brivaracetam were tested hierarchically at the 5% level of significance, power was lost for lower doses. To compensate for this loss in power, 100 patients per group were included in the trials.

In Study 1254, with a significance level of 5%, a two-sided test with 376 patients randomized in a 3:1 (brivaracetam:placebo) ratio yielded 80% power to detect a difference of −0.174 on the natural log-transformed scale, with an assumed standard deviation of 0.50. Using an approximate formula, this corresponds to a per cent reduction from placebo of 16%. Since up to 20% of patients had generalized epilepsy, in order to secure 376 patients suffering from localization-related (POS) epilepsy, the number of 376 patients had to be adjusted by 1.25. Therefore, the total number of patients that needed to be randomized was 470.

In Study 1358, the sample size was based on the 50% responder rate outcomes, since this yielded the larger sample size across primary outcomes (i.e., the primary outcomes for the United States and European Union differed). Therefore, based on a 50% responder outcome, 231 analyzable patients per treatment group provided 90% power to detect a 15% difference between brivaracetam and placebo at the 0.025 significance level, assuming responder rates of 35% and 20%, respectively. The actual power for the United States’ outcome was 94% based on this sample size. To account for patients who may not have qualified for the primary analysis, 240 patients were randomized in each group, for a total of 720 patients across all three treatment groups.

For all studies, summary statistics were provided for all efficacy, safety, and baseline/demographic variables. Summary statistics consisted of frequency tables for categorical variables. For continuous variables, descriptive statistics (e.g., number of available observations, mean, median, standard deviation, minimum, maximum) were tabulated.

The primary efficacy variable (i.e., POS [type 1] frequency per week or per 28 days) was analyzed according to an analysis of covariance model, including treatment and stratification data (i.e., country/region and concomitant levetiracetam use) as factors, and the log-transformed baseline seizure frequency per week or per 28 days as covariate. In Study 1358, an additional stratification factor was the number of AEDs previously used (≤ 2 or > 2) and discontinued prior to study entry. The primary efficacy variable was log-transformed prior to analysis, as prior data (including results from phase 2 studies) showed this transformation was appropriate to approximate normally distributed data.

In studies 1252, 1253, and 1254, brivaracetam dose groups were tested at the 5% significance level compared with placebo, according to a predefined hierarchical sequentially rejective testing procedure. The highest brivaracetam dose was tested versus placebo and, if the comparison was not statistically significant, the procedure stopped and no groups were declared different from placebo. If the comparison was statistically significant, then that brivaracetam dose group was considered different from placebo and the procedure continued with the next-lower brivaracetam dose and so on. The procedure controlled the overall type 1 error rate at 0.05.

In Study 1358, statistical testing was based on the comparison of each brivaracetam group with placebo, with control of the overall type 1 error rate based on the Hochberg procedure. The Hochberg procedure does not require a pre-specified order of testing for brivaracetam dose groups versus placebo. In a setting with a comparison of two active treatment groups to placebo, the Hochberg procedure was applied by first testing the brivaracetam group with the larger P value with testing at the 0.05 level. If statistical significance was achieved at this step, then the study was positive and both brivaracetam dose groups were declared statistically different from placebo. If the largest P value was not significant at the 0.05 level, then the Hochberg procedure steps to the smaller P value with testing at the 0.025 level. If statistical significance was achieved at this step, then the study was positive and the brivaracetam dose group associated with the smaller P value was declared statistically different from placebo. If the smaller P value was not significant at the 0.025 level, then neither brivaracetam dose group was statistically different from placebo, and the study was not positive.

▬

▬

For the comparison of POS responder rate for each brivaracetam dose and placebo, a logistic-regression model was used, including treatment as a factor and the log-transformed baseline seizure frequency per week as a covariate. An estimate of the treatment odds ratio (OR) and 95% confidence interval (CI) was derived from this model. The per cent reduction from baseline for POS frequency was analyzed by applying the Wilcoxon-Mann-Whitney test to compare brivaracetam dose and placebo. The categorized response in POS frequency per week for brivaracetam compared with placebo was tested using the Cochran-Mantel-Haenszel test. Patients who had no (zero) baseline seizure frequency per week were categorized in the “under −25%” category. Seizure freedom rates for each brivaracetam dose were compared with placebo using the Fisher’s exact test. ▬

▬

▬

3.3.1. Study 1252

A total of 486 patients were screened, and of these, 87 (17.9%) were screening failures. One patient randomized to brivaracetam 50 mg/day was dispensed the drug, but died prior to consuming any study drug and thus was excluded from the ITT population. Across treatment groups, 82.8% to 88.0% of patients completed the study and entered the LTFU, whereas 4.0% to 6.1% completed the study but did not enter the LTFU; rather, they completed the down-titration period and safety follow-up.

3.3.2. Study 1253

A total of 509 patients were screened and, of these, 109 (21.4%) were screening failures. Four patients were excluded from the ITT population. One patient (in the brivaracetam 5 mg/day group) was excluded for failing to take the study drug and three patients randomized to brivaracetam 5 mg/day, placebo and brivaracetam 50 mg/day, respectively, were randomized in error and were not dispensed the study drug. Across treatment groups, 87.8% to 90.1% of patients completed the study and entered the LTFU, whereas 2.0% to 7.1% completed the study but did not enter the LTFU; rather, they completed the down-titration period and safety follow-up.

3.3.3. Study 1254

A total of 543 patients were screened and of these, 63 (11.6%) were screening failures. No patients were excluded from the ITT population. Across treatment groups, 86.1% to 91.7% of patients completed the study and entered the LTFU, whereas 2.5% to 3.9% completed the study but did not enter the LTFU; rather, they completed the down-titration period and safety follow-up. The disposition of patients with POS was similar to the overall population.

3.3.4. Study 1358

A total of 1,045 patients were screened and of these, 277 (26.5%) were screening failures. No patients were excluded from the ITT population. Across treatment groups, 86.2% to 90.1% of patients completed the study and entered the LTFU, whereas 2.0% to 3.4% completed the study but did not enter the LTFU; rather, they completed the down-titration period and safety follow-up.

3.5.1. Internal Validity

All four included trials were prospective, multi-centre, double-blind, randomized, placebo-controlled trials that demonstrated a number of methodological strengths. Treatment was assigned via an Interactive Voice Response System using a central randomization method with stratification as detailed in Section 3.2.1. Allocation concealment methods were appropriate, and matched placebo tablets were used to ensure blinding during the treatment phases. Baseline demographic and disease characteristics were generally well balanced across treatment groups, with no significant baseline imbalances expected to have affected the study outcomes. Discontinuation rates were low in all trials and similar between treatment groups; however, a consideration is that the duration of treatment was only 12 to 16 weeks.

The primary efficacy outcome in all the trials was the reduction in POS frequency; to record seizure frequency, patients filled in a daily record card that was returned at each study visit. Although this is a standard method of reporting seizure frequency related outcomes in clinical trials of AEDs, self-reporting is subject to individual variability in reporting accuracy and completion.

▬

Although the analysis populations in studies 1252, 1254, and 1358 were identified as being ITT populations, due to the definition used (i.e., all randomized patients who received at least one dose of study medication), the study populations are technically a modified ITT population due to the requirement to have had at least one dose of study medication. Nonetheless, the number of randomized patients and patients included in the ITT populations does not differ by more than 2% in any of the trials, so it is unlikely this would have had any effect on treatment outcomes.

With the exception of the three selected scores of the QOLIE-31-P in studies 1252, 1253, and 1254, all secondary outcomes were tested without adjustment for multiplicity. Due to the potential risk of type 1 error, the results of the secondary outcomes should be interpreted with caution.

3.5.2. External Validity

According to the clinical expert consulted for this review, the baseline demographic and disease characteristics of the patients enrolled in the four trials are representative of patients with refractory epilepsy encountered in clinical practice in Canada. Although the pattern of prior AED use was also what would be expected in Canada generally, it was noted that no reported use of clobazam by patients in studies 1252, 1253, and 1254 is not in line with what would be expected in Canada. ▬

As detailed in Table 6, across the treatment groups, from 10.6% to 47.8% of patients had tried five or more AEDs in the past five years prior to entry into the included trials. Due to the large number of prior AEDs tried by these patients, it is likely their epilepsy was more refractory and therefore more difficult to treat than would be found in a less treatment-experienced patient population. This may have an impact on the generalizability of the trial results to patients who have not had as extensive background use of other AEDs and whose epilepsy may not be considered as refractory to treatment.

The included trials enrolled mainly Caucasian patients, followed by Asian patients, which may affect the generalizability of the trial results to a more heterogeneous population with mixed ethnicity, such as the Canadian population. In addition, the results of the trials would not be generalizable to a pediatric or older geriatric population, as these patients were not represented in the trials.

All four trials included stratification based on levetiracetam use (i.e., studies 1252, 1253, and 1254 stratified on concomitant levetiracetam use; Study 1358 stratified based on prior levetiracetam use as the use of levetiracetam ≤ 90 days prior to study entry was an exclusion criterion). This approach has direct clinical relevance because, due to the presumed similar mechanism of action of brivaracetam and levetiracetam (i.e., binding to the SV2A protein), it is important to determine whether or not patients who had previously been uncontrolled on levetiracetam, or a combination of AEDs including levetiracetam, would benefit from brivaracetam. In the included trials, there was a large proportion of patients who used levetiracetam prior to study entry, with the main reason for discontinuation of levetiracetam being due to lack of efficacy. In addition, in studies 1252, 1253, and 1254, approximately 20% of patients (regardless of treatment) used concomitant levetiracetam during baseline. An analysis of the per cent reduction in POS frequency by concomitant levetiracetam use in these studies showed that for patients who received concomitant levetiracetam, the differences between brivaracetam and placebo were not statistically significant, whereas in patients who did not receive concomitant levetiracetam, the differences between brivaracetam and placebo were statistically significant for all comparisons. Study 1358 also included pre-specified subgroup analyses by levetiracetam status; however, no statistical comparisons between subgroups were conducted.

Studies 1252, 1253, and 1358 did not include an up-titration phase, as patients were randomized directly to the full, fixed dose of brivaracetam or matching placebo. Patients remained on the full dose unless they had to exercise the dosing fallback option. In contrast, Study 1254 had a flexible dosing design. This may have complicated the efficacy assessment, because patients who responded well to treatment may have remained on lower brivaracetam doses, whereas in the fixed-dose trials, patients could have been on brivaracetam doses that were higher than required. Study 1254 also had baseline period of only four weeks, whereas the other trials had eight-week baseline periods. It is possible that the shorter baseline period may have led to increased variability in responses, although the treatment period (comprising both the dose-finding and maintenance phases), was longer than the other trials (i.e., 16 weeks versus 12 weeks). There is also a possibility that in Study 1254, due to up-titration relying on the investigator’s assessment of efficacy and tolerability, patients on placebo might have been identifiable due to their need for up-titration, thus potentially compromising the blinding in the study.

All four trials were placebo-controlled, as no active-controlled trials of brivaracetam in patients with uncontrolled POS epilepsy were identified. The use of placebo as a comparator in this refractory patient population does not reflect clinical practice. Such patients would most likely be considered for adjunctive therapy with newer AEDs such as perampanel, lacosamide, or eslicarbazepine. An active comparator trial against one or more of these drugs would have permitted an assessment of the relative benefit-risk profile of brivaracetam compared with another AED currently listed for the treatment of refractory epilepsy in the same target patient population. Of note, the manufacturer did conduct a network meta-analysis of brivaracetam with third-generation AEDs used as adjunctive therapy in patients with refractory POS, including perampanel, lacosamide, and eslicarbazepine. This is summarized in Appendix 7: Summary of Indirect Treatment Comparisons.

The choice of the primary efficacy outcome (i.e., reduction in POS frequency from baseline per week or per 28 days) was clinically relevant, as were the majority of secondary outcomes related to reduction in seizure frequency or HRQoL. Interpretation of the results, however, was limited by the lack of validation of many of the outcomes in patients with epilepsy, as well as the limited identification of MCIDs for the outcomes. The outcomes of the EQ-5D and health care resource utilization parameters could have provided useful information; however, ▬

While the duration of the treatment phases (12 to 16 weeks) in the included trials was sufficient to show early efficacy of brivaracetam, it is inadequate to characterize long-term efficacy and safety. The clinical expert advised this is not unique among trials of AEDs, as the treatment phase in most AED trials is short, even though these therapies are intended for chronic use. Pooled long-term follow-up data from patients who entered the LTFU phases of the included trials are summarized in Appendix 6.

3.6.1. Seizure-Free Status

In all four included trials, seizure-free status was included as a secondary outcome (Table 14). Over the treatment periods, the proportion of patients who were seizure-free was small in all trials, ranging from 0% to 5.2% among the brivaracetam groups compared with 0% to 0.8% among the placebo groups. In Study 1358, a larger proportion of patients treated with brivaracetam 100 mg/day (5.2%) and brivaracetam 200 mg/day (4.0%) were seizure-free compared with placebo-treated patients (0.8%). The differences were statistically significant for both brivaracetam doses relative to placebo. Differences in the proportion of seizure-free patients treated with brivaracetam 50 mg/day to 150 mg/day in the other trials were not statistically significantly different from placebo.

3.6.2. Health-Related Quality of Life

▬

▬

▬

▬

3.6.3. Change in Seizure Frequency

The results for the change from baseline to end of treatment in the POS frequency per week in studies 1252, 1253, and 1254, and per 28 days in Study 1358, are provided in Table 17 and Table 18. At baseline, the median POS frequency per week was similar between all treatment groups, ranging from 1.80 to 2.85 (or 9.3 to 10.0 per 28 days in Study 1358). At the end of treatment, the median POS frequency per week ranged from 1.26 to 1.74 for brivaracetam (or 6.8 to 6.9 per 28 days in Study 1358), compared with 1.75 to 2.15 per week (or 9.2 per 28 days in Study 1358) for placebo. The treatment difference in POS frequency per week was statistically significant for brivaracetam 100 mg/day in Study 1252 and for brivaracetam 50 mg/day in Study 1253, and per 28 days for both brivaracetam 100 mg/day and 200 mg/day in Study 1358.

The median per cent reduction in POS frequency over the treatment period ranged from 26.83% to 37.2% with brivaracetam and 17.03% to 18.93% with placebo (Table 20). The median per cent reduction with brivaracetam over placebo ranged from 9.07% to 19.35% for brivaracetam, and the results were statistically significant for brivaracetam 100 mg/day (−19.35%) in Study 1252, brivaracetam 50 mg/day (−15.69%) in Study 1253, and for both brivaracetam 100 mg/day (−15.8%) and 200 mg/day (−18.1%) in Study 1358. There did not appear to be a dose response demonstrated for brivaracetam.

The median per cent reduction in POS frequency by the stratification factor of concomitant levetiracetam use over the treatment period was investigated for studies 1252, 1253, and 1254 (Table 21). In patients with concomitant levetiracetam use at study entry, the median per cent reduction in POS frequency over the treatment period ranged from 3.16% to 15.93% with brivaracetam compared with 14.18% to 22.11% with placebo. None of the comparisons of the median per cent reduction with brivaracetam over placebo were statistically significant. In patients with no concomitant levetiracetam at study entry, the median per cent reduction in POS frequency over the treatment period ranged from 31.51% to 38.31% with brivaracetam compared with 15.70% to 19.19% with placebo. All comparisons of the median per cent reduction with brivaracetam over placebo were statistically significant.

Pre-specified subgroup analyses of the per cent reduction over placebo in POS frequency per 28 days by levetiracetam status and by number of previous AEDs (≤ 2 or > 2) was conducted in Study 1358. For patients who never used levetiracetam, the per cent reduction in 28-day POS frequency over placebo was 29.5% with brivaracetam 100 mg/day, and 27.1% with brivaracetam 200 mg/day. For patients with prior levetiracetam use, the per cent reduction with brivaracetam over placebo was 15.8% with brivaracetam 100 mg/day, and 19.4% with brivaracetam 200 mg/day. The per cent reduction with brivaracetam over placebo in the 28-day–adjusted POS frequency for patients who used ≤ 2 previous AEDs was 27.1% with brivaracetam 100 mg/day, and 27.3% with brivaracetam 200 mg/day. In comparison, for patients who used > 2 previous AEDs, the per cent reduction over placebo was 20.2% with brivaracetam 100 mg/day, and 21.7% with brivaracetam 200 mg/day. Only descriptive statistics were reported and no statistical comparisons were made between either of the subgroups.

The results of the categorized response in POS seizure frequency over the treatment period are provided in Table 22. Patients who had no (zero) baseline seizure frequency per week were categorized in the “under −25%” category. For all studies, the category with the largest proportion of patients across treatment groups was the −25% to 25% group, which included from 28.6% to 33.3% of patients in the brivaracetam groups compared with 40.5% to 44.8% of placebo patients. Overall, the proportion of patients who achieved 75% or better response ranged from 10.1% to 19.9% in the brivaracetam groups compared with 2.8% to 8.0% in the placebo groups. When the proportions of patients in each category were compared across treatment groups, the differences were statistically significant for all comparisons of brivaracetam versus placebo in all the trials, with the exception of the brivaracetam 50 mg/day group in Study 1252.

3.6.4. Other Efficacy Outcomes

3.7.1. Adverse Events

Across the four included trials, approximately 63% to 75% of patients in the brivaracetam groups compared with 53% to 65% of patients in the placebo groups experienced at least one treatment-emergent adverse event (TEAE) as detailed in Table 13. Overall, the most frequently reported TEAEs were somnolence, dizziness, fatigue and headache. Somnolence occurred more frequently in the brivaracetam groups (6.1% to 19.4%) compared with the placebo groups (4.1% to 7.7%). Similarly, dizziness (5.0% to 15.8%) and fatigue (4.0% to 11.6%) occurred more frequently with brivaracetam compared with placebo (5.0% to 9.2% and 2.0% to 4.1%, respectively). Headache was reported in a similar proportion of brivaracetam (6.7% to 18.2%) and placebo (8.4% to 19.8%) patients. The majority of TEAEs were mild to moderate in intensity.

Table 13

Harms During Treatment Period (Safety Population).

3.7.2. Serious Adverse Events

The incidence of serious adverse events (SAEs) across the four trials was low and similar across treatment groups (Table 13). The proportion of patients with at least one SAE ranged from 2.0% to 5.3% among the brivaracetam groups compared with 0% to 7.4% among the placebo groups of the included trials. ▬

3.7.3. Withdrawals Due to Adverse Events

Withdrawals due to adverse events (WDAEs) were also low and similar across treatment groups (Table 13). The proportion of patients with WDAEs ranged from 5.0% to 8.3% in the brivaracetam groups compared 2.0% to 5.0% in the placebo groups of the included trials. The most common reason for WDAEs in both the brivaracetam and placebo treatment groups was convulsion, which occurred in 0.4% to 2.0% of patients in the brivaracetam groups and 0.4% to 1.0% of patients in the placebo groups.

3.7.4. Mortality

There were five deaths reported in the recommended brivaracetam dose or placebo groups during the treatment periods of the four included trials. Of these, only one death was considered to be possibly related to the study drug; it was due to brain hypoxia in the brivaracetam 50 mg/day group of Study 1253. The other causes of death (i.e., sepsis in one patient receiving placebo in Study 1252, drowning of one patient in Study 1254, and two deaths due to sudden unexplained death in epilepsy and unknown cause, respectively, in Study 1358) were unlikely or not considered to be related to the study drug.

3.7.5. Notable Harms

▬

▬