PurposeTo determine whether brivaracetam (BRV) provides an evident improvement in treatment efficacy and a reduction in treatment-emergent adverse events (TEAEs) in patients with refractory epilepsy, who previously failed treatment with levetiracetam (LEV). DesignRetrospective analysis of data extracted from electronic patient files at Epilepsy Centre Kempenhaeghe (Heeze, the Netherlands) from the year 2000 until October 2020. MethodsThe inclusion criteria were met by 407 patients >18 years of age. During data collection, 26 patients were excluded due to too little follow-up information on the use of either LEV or BRV, and two more due to poor medication compliance, leaving a total of 379 patients for further analyses. All had used LEV before they started treatment with BRV. For every patient, data were collected including demographic information, efficacy (positive responder or non-responder) of LEV and BRV, and TEAEs occurring during LEV and BRV treatment. ResultsA total of 121 (29.8%) patients had discontinued BRV treatment before the end of data collection. At time of data collection the mean time since first seizure was 25.4 years. Of the 379 patients, 82.8% were diagnosed with focal epilepsy and 9.8% with generalized epilepsy. The median duration of treatment was 39 months for LEV and 20 months for BRV, the mean maximum dose was 1749.9 mg/day for LEV and 144.2 mg/day for BRV, and the mean number of concomitant AEDs was 1.4 at the start of LEV treatment and 2.0 at the start of BRV treatment. LEV was switched directly to BRV in 208 (54.9%) patients; 171 (45.1%) patients had an interval between discontinuation of LEV and the start of BRV. The mean duration of interval was 77.7 months. Of the patients who discontinued BRV, 30 (24.8%) switched back to LEV. Discontinuation of initial LEV treatment was due to TEAEs in 63.6% of patients, including 55.1% because of behavioural TEAEs. Discontinuation of BRV was due to inadequate efficacy in 24.0% of patients, to TEAEs in 47.1% and to both inadequate efficacy and TEAEs in 22.3%. Concerning efficacy, the analysis showed no significant difference between the positive responder rate of LEV and BRV (72.0% vs 69.1%, p>0.05). Of the patients who were positive responders to LEV treatment, 78.0% also had a positive response to BRV treatment. Of the non-responders to LEV treatment, 46.2% did have a positive response to BRV treatment. In comparison to LEV, patients reported significantly fewer TEAEs during BRV treatment (86.5% vs 61.7%, p<0.05). The most substantial difference was seen in the category ‘behaviour’ (55.1% vs 22.4%, p<0.05). Newly found behavioural TEAEs after switching from LEV to BRV were found in 7.1% of patients. ConclusionOverall BRV was better tolerated than LEV, especially regarding the behavioural TEAEs. Efficacy analyses showed that patients are likely to have a positive response to BRV when they had a positive response to LEV. However, this is not always guaranteed. Lack of response to LEV does not preclude a positive response to BRV. All in all, BRV seems to be an interesting treatment option in patients previously treated with LEV.