Antiseizure medications for post‐stroke epilepsy: A real‐world prospective cohort study

Abstract Background and purpose The management of post‐stroke epilepsy (PSE) should ideally include prevention of both seizure and adverse effects; however, an optimal antiseizure medications (ASM) regimen has yet been established. The purpose of this study is to assess seizure recurrence, retention, and tolerability of older‐generation and newer‐generation ASM for PSE. Methods This prospective multicenter cohort study (PROgnosis of Post‐Stroke Epilepsy [PROPOSE] study) was conducted from November 2014 to September 2019 at eight hospitals. A total of 372 patients admitted and treated with ASM at discharge were recruited. Due to the non‐interventional nature of the study, ASM regimen was not adjusted and followed standard hospital practices. The primary outcome was seizure recurrence in patients receiving older‐generation and newer‐generation ASM. The secondary outcomes were the retention and tolerability of ASM regimens. Results Of the 372 PSE patients with ASM at discharge (median [IQR] age, 73 [64–81] years; 139 women [37.4%]), 36 were treated with older‐generation, 286 with newer‐generation, and 50 with mixed‐generation ASM. In older‐ and newer‐generation ASM groups (n = 322), 98 patients (30.4%) had recurrent seizures and 91 patients (28.3%) switched ASM regimen during the follow‐up (371 [347–420] days). Seizure recurrence was lower in newer‐generation, compared with the older‐generation, ASM (hazard ratio [HR], 0.42, 95%CI 0.27–0.70; p = .0013). ASM regimen withdrawal and change of dosages were lower in newer‐generation ASM (HR, 0.34, 95% CI 0.21–0.56, p < .0001). Conclusions Newer‐generation ASM possess advantages over older‐generation ASM for secondary prophylaxis of post‐stroke seizures in clinical practice.


INTRODUCTION
Post-stroke epilepsy (PSE), one of the major sequelae of stroke, is the most common cause of epilepsy in the elderly (Sen et al., 2020). Stroke survivors with potential risk of PSE have been increasing  as recent advances in acute stroke treatments have dramatically improved mortality rate. Antiseizure medications (ASMs) are the mainstay in seizure control in PSE, with most patients generally well controlled by a single dose of ASM (Ryvlin et al., 2006); however, approximately one-third of PSE patients experience seizure recurrence under ASM treatment within 1 year (Tanaka et al., 2015), and another study also demonstrated that approximately 20% of patients with PSE developed pharmaco-resistance and exhibited associations with younger age at stroke onset, stroke type and severity, status epilepticus occurrence, and seizure type (Lattanzi et al., 2021). These results may lead to heightened anxiety regarding seizure recurrence and suppressed social activities in stroke survivors.
According to American and European guidelines (Holtkamp et al., 2017;Winstein et al., 2016 ), use of ASM for secondary prophylaxis of post-stroke seizures has been recommended and newer-generation ASM, levetiracetam (LEV) and lamotrigine (LTG), have been suggested as first-line treatments due to fewer adverse effects (AE) (Ferlazzo et al., 2016;Feyissa et al., 2019).
Despite such recommendations, there is currently no established evidence on whether newer-generation ASM are suitable for prevention of PSE . To date, two small randomized control studies (RCT) have specifically evaluated ASM in PSE (Brigo et al., 2018). In a nationwide, population-based study, Huang et al. reported valproic acid (VPA) and newer-generation ASM have superior seizure control compared to phenytoin (PHT) (Huang et al., 2015); however, this study was retrospective and was based on information from a health insurance database. Furthermore, a randomized control study conducted by Werhann et al. found elderly patients with newonset focal epilepsy (originated from stroke in 65.9%), showed superior tolerability and higher drug retention of LEV and LTG compared to controlled-released carbamazepine (CBZ) (Werhahn et al., 2015); however, no significant differences were found in seizure freedom rates.
The objectives of the current study were to assess whether newergeneration ASM have advantages over older-generation ASM in terms of seizure control, retention, and tolerability in PSE over a 1-year period in a real-world clinical practice.

Study design and patients
The PROgnosis of Post Stroke Epilepsy (PROPOSE) study was a multicenter, noninterventional, prospective observational cohort study.
Patients hospitalized with PSE between November 2014 and September 2018 were recruited from eight hospitals in Japan. According to the new clinical definition from the International League Against Epilepsy (ILAE) (Fisher et al., 2014), one unprovoked seizure more than 7 days (late seizure) after index stroke was diagnosed as PSE.  (Fisher et al., 2017).
Semiological information of the ictal and post-ictal phase, such as motor seizure, ictal paresis, aphasia, amnesia, and the other semiologies, were collected. Standard scalp EEG was performed for more than 20 min after admission and analyzed for focal and rhythmic slow wave, periodic discharge, and paroxysmal activity. EEG terminologies were based on the Salzburg EEG criteria (Hirsch et al., 2013). Ictal or postictal SPECT and diffusion-weighted MRI (DWI) with apparent diffusion coefficient (ADC) were performed for identifying a hyperperfusion area in endorsing the PSE diagnosis, based on our previous report (Fukuma et al., 2020).

ASM seizure control, retention, and tolerability assessments
We evaluated the seizure control, retention, and tolerability of ASM treatment in the secondary prevention of seizure after discharge. We These definitions were determined regardless of the number of ASM and selection of regimen was not altered due to the non-interventional nature of the current study, which followed standard hospital practices. The dosage of each ASM, as well as information on whether the serum concentration of older-generation ASM reached an adequate therapeutic range at discharge, were collected. After a baseline dataset was collected during admission period, each participant was followedup according to usual clinical practice in each hospital over a 1-year period. The mixed-generation ASM group was excluded from the main analyses.
The primary outcome was seizure freedom rate over the complete follow-up period since discharge. The secondary outcomes were the time to discontinuation or dosage adjustment of initial ASM regimen due to ineffectiveness or intolerable ASM-related AE (defined as ASM retention) and time to discontinuation of initial ASM regimen due to intolerable ASM-related AE (defined as ASM tolerability, a subgroup of ASM retention outcome). The incidence of discontinuation or dosage adjustment due to poor adherence or other reasons were not considered outcomes. Board-certified epileptologists or neurologists closely assessed episodes of first unprovoked seizure recurrence, ASM regimens, and reasons behind ASM discontinuation or dosage adjustment from hospitalization and/or outpatient visit records and seizure diaries with face-to-face consultations. If outpatient visits were not feasible, assessment was made through telephone interview with patients, relatives, or general physicians at 6 and 12 months. Patients who died or were lost to follow-up were assessed as of the last visit. The maximum period of observation until the last follow-up was 800 days.

Verification
In the current prospective study, the older-generation ASM group comprised just 36 patients, presumably owing to the trend of using newergeneration ASM. LEV was predominantly prescribed in preference to older-generation ASM in Japan (Nakamura et al., 2020), as typically also found in other countries (Glauser et al., 2013). To compensate for the imbalance in the number between groups and to verify the results, we added older-generation ASM cases (n = 69) from the retrospective PSE cohort database of our institution between January 2011 to October 2014, comprising 15 newer-generation, 18 mixed-generation, and 69 older-generation ASM. Retrospective data were obtained from registered PSE patients treated and followed up after discharge completely at our hospitals. During this period, most newer-generation ASM monotherapies were not covered by Japanese public health insurance.

Statistical analyses
Data are presented as median (interquartile range [IQR]) or number (%). Variables were compared using Wilcoxon's test for continuous F I G U R E 1 Participant flowchart of study cohorts. After exclusion of cases without antiseizure medication (ASM), 372 post-stroke epilepsy patients were prospectively followed up for incident seizure recurrence, withdrawal, and change of dosages due to adverse effects or inadequate seizure control (retention) and withdrawal due to adverse effects (tolerability) (See Figure 2). For verification analysis, the retrospective older-generation ASM cohort (n = 69) was obtained from the retrospective cohort database (See eFigure 2) variables and Pearson's chi-square test for categorical variables. Time to first seizure recurrence (seizure freedom rates), discontinuation or dosage adjustment (ASM retention rates) and discontinuation (ASM tolerability rates) between older-and newer-generation ASM groups was analyzed by the Kaplan-Meier method and compared by log-rank test. Follow-up time was defined as the period from discharge until death, loss to follow-up or final follow-up visit. Patients who withdrew due to the above reasons were censored at the time of withdrawal, and those who did not withdraw were censored at their last visit. We also determined cumulative seizure freedom rates by 90 days, 180 days, and 1 year after discharge.
Cox proportional hazards modeling was used to calculate hazard ratios (HR) with 95% confidence intervals (CI). Potential confounding factors such as sex (Kim et al., 2016), age (Kim et al., 2016;Tanaka et al., 2015;Tomari et al., 2017), and other variables with a p value < .10 between the ASM groups were adjusted in a multivariate model (model 1: age and sex; model 2: age, sex, dyslipidemia, dementia, temporal lobe stroke lesion, and spike or sharp wave on EEG). In addition, we conducted additional analyses by eliminating data on seizure outcomes of PSE patients with inadequate or unknown concentrations of oldergeneration ASM (n = 18).
As with verification analyses, prospective newer-generation ASM were compared against the retrospective older-generation ASM group, or the combination of prospective and retrospective older-generation ASM groups, using the same methods. Two-sided values of p < .05 were considered significant. All statistical analyses were performed with the JMP 12.2.0 software package (SAS Institute Inc., Cary, NC, USA). All data were anonymized, stored, and fixed in an electronic data capture system by the data manager prior to analysis.

RESULTS
From November 2014 to September 2018, a total of 392 patients with a diagnosis of PSE in the eight sampled hospitals met the eligibility criteria and were prospectively enrolled into the PROPOSE study ( Figure 1). Of these, we excluded 20 patients not receiving ASM at discharge. Of the remaining 372 patients, 36 were treated with oldergeneration, 286 with newer-generation, and 50 with mixed-generation, ASM as a secondary prophylaxis.
The demographics and clinical course characteristics of older-and newer-generation ASM groups are shown in   (Table 3). Even after PSE patients with inadequate or unknown concentration of older-generation ASM (n = 18) were excluded from analysis, newer-generation ASM retained a lower risk of seizure recurrence than older-generation ASM (p < .0001) (eFigure 1).

Secondary outcomes: Retention and tolerability of ASM
During the same period, 36.5% patients in the older-generation, and 72.0% patients in the newer-generation ASM group maintained the initial ASM regimens. Withdrawal from the study was due to drug ineffectiveness and AE: 68.2% and 31.8% in the older-generation, and 69.6% and 30.4% in the newer-generation ASM. Compared with older-generation, retention rate was significantly higher in the newergeneration ASM (p < .0001, log-rank) (Figure 2b). In the Cox proportional hazards models, adjusted HRs for newer-generation ASM were 0.32 compared with older-generation (95%CI, 0.20−0.53) in model 1 and 0.35 (95%CI, 0.21−0.61) in model 2 ( Table 3).
The ASM tolerability rate tended to be higher in the newergeneration ASM (Figure 2c), though there were no significant differences between the groups (p = .12). The most common AE leading to discontinuation or dosage adjustment were skin eruptions TA B L E 2 Basis of diagnosis for PSE in prospective cohort (n = 322) Focal or generalized motor seizure a n = 236  (Hirsch et al., 2013). c MRI + means the presence of hyperintensities that are cortically based with an atypical vascular distribution and mostly reversible on DWI (not stroke) (Koksel et al., 2018). d SPECT + means the presence of ictal or postictal hyperperfusion area on SPECT. The details of identification of hyperperfusion are described in our previous report (Fukuma et al., 2020).

Verification analysis: Prospective and retrospective data
To validate the results of this prospective study, verification analysis was conducted using a retrospective older-generation ASM cohort Abbreviations: ASM, antiepileptic drug.; CI, confidence interval; HR, hazard ratio. a Model 1: age, sex. b Model 2: age, sex, dyslipidemia, dementia, temporal lobe stroke lesion, and spike or sharp wave on electroencephalogram.
cWithdrawal of initial ASM regimens (discontinue or change the dosage of ASM) due to ineffectiveness or ASM-related adverse effect. d Discontinuation of initial ASM due to intolerable ASM-related adverse effect.

DISCUSSION
The present study demonstrated advantages of newer-generation ASM in the secondary prevention of PSE. The latest guidelines recommend ASM treatment for secondary PSE prophylaxis (Holtkamp et al., 2017), as more than 70% of patients with a first unprovoked late seizure after stroke experience a subsequent seizure over the 10- year follow-up period (Hesdorffer et al., 2009). When considering initial ASM selection, however, there is no current consensus regarding the effectiveness of newer-generation ASM for seizure control in PSE, largely due to insufficient evidence (Gilad, 2012;Xu, 2019). In a recent meta-analysis (Brigo et al., 2018), only two randomized control trials specifically targeting PSE were included, one comparing LTG to CBZ (Gilad et al., 2007) and the other LEV to controlled-release CBZ (Consoli et al., 2012). These studies showed lower incidence of AE with LTG and LEV than CBZ; however, any differences in seizure freedom were not identified, likely due to small sample sizes. While a large nationwide retrospective study from Taiwan demonstrated that VPA and newer-generation ASM had fewer recurrences of PSE than PHT (Huang et al., 2015), it possessed inherent issues in diagnostic accuracy emanating from the use of health insurance database as the primary data source. Although the PROPOSE study is not a randomized control trial, it is based on accurate diagnoses, with established examinations by board-certified specialists and reliable prospective follow-up information.
Regarding focal seizure in adults (mainly due to stroke), the ILAE report recommends CBZ, LEV, PHT, and ZNS as initial monotherapy based on level A evidence (Glauser et al., 2013). In recent clinical studies, older-generation ASM, such as CBZ, VPA, and PHT, were still being used (Hsieh & Huang, 2011;Johnell & Fastbom, 2011;Larsson et al., 2019;Pugh et al., 2008). The standard and new antiepileptic drugs (SANAD) trial comparing newer-generation ASM with CBZ in patients with focal epilepsy demonstrated LTG was better tolerated, though seizure control was not different than other drugs (Marson et al., 2007). In the subgroup analysis of Keppra versus Older Monotherapy in Epilepsy Trial (KOMET), comparing the effectiveness of LEV with extended-release VPA and controlled-release CBZ in elderly patients with newly diagnosed epilepsy, LEV had more favorable tolerability profiles (Pohlmann-Eden et al., 2016). These studies did not demonstrate superiority of seizure control using newergeneration ASM; however, a survey of expert opinion across the United States reported preference for LTG and LEV over PHT, GBP, PH, and CBZ in the elderly population (Shih et al., 2017).
The current PROPOSE study showed that use of newer-generation ASM was associated with approximately 50% reduction of seizure recurrence, compared to older-generation ASM. Similar results were obtained in verification analysis using our retrospective cohort data.
Previous meta-analyses have revealed no significant superiority of seizure control in newly diagnosed focal epilepsy  and epilepsy in elderly patients  Polypharmacy may also interfere with the efficacy of older-generation ASMs, which are known to have drug-drug interactions. Moreover, CBZ, a CYP-inducing ASM, could interfere with the efficacy of statins, consequently increasing lipid levels (Mintzer et al., 2018). Several studies have also demonstrated the risk of subsequent stroke was increased after the onset of epilepsy (Chang et al., 2014;Cleary et al., 2004). This may lie in the use of older-generation ASM, which may affect the lipid profiles or drug metabolism of antithrombotic drugs. Older-generation ASM may interfere with stroke prophylaxis and affect cerebrovascular risks (Zelano, 2016). Such effects may be important in the selection of ASM regimen in clinical practice in stroke survivors.
Our study has several limitations. First, as this was a prospective observational study, ASM regimen could not be controlled, meaning the older-generation ASM were used less frequently and in smaller dosage than newer-generation ASM. Second, the PROPOSE study was not a like-for-like comparison of each ASM, which may have resulted in insufficient statistical power. However, RCTs may potentially limit generalizability, which cannot be necessarily regarded as general principles valid for all subjects, some of whom may have had difficulties answering more complex questions. In the field of epilepsy, for instance, the various combinations of ASMs are used in a wide variety of clinical, social, and economic conditions. It is impractical to conduct separate RCTs to study the effect of each specific ASM regimen. Observational studies based on routinely collected data have generally a greater external validity than RCTs (Carlson & Morrison, 2009). Third, the diagnosis of PSE may be difficult due to the diverse semiology of PSE (Bentes et al., 2017). In order to increase reliability, we performed brain MRI, SPECT, and EEGs in the acute stage under careful consideration by certified epileptologists or neurologists.

CONCLUSIONS
The current study is a real-world prospective cohort study to evaluate seizure control, retention, and tolerability of older or newergeneration ASM treatment in PSE. These findings suggest potential for newer-generation ASM as the primary choice in the secondary prophylaxis of PSE. However, further studies are needed to confirm ASM drug interactions and stroke risks and to find the optimal regimen for ASM in PSE.

CONFLICT OF INTEREST
Akio Ikeda belongs to The Department of Epilepsy, Movement Disorders, and Physiology, an Industry-Academia Collaboration Course, supported by a grant from Eisai Corporation, Nihon Kohden Corporation, Otsuka Pharmaceutical Co., and UCB Japan Co.
All other authors report no conflict of interest specifically related to this manuscript.

AUTHOR CONTRIBUTIONS
Study concept and design was provided by TT, KF, KK, AS, KN, RM, AI, and MI. Acquisition, analysis, or interpretation of data was performed by TT, KF, and MI. Drafting the manuscript and figures was performed by TT and MI. All authors participated in critical revision of the article for intellectual content. KN and DO were the statisticians. AI and MI provided study supervision.

DATA AVAILABILITY STATEMENT
Analyses for PROPOSE study are ongoing; however, once completed, the data generated from this work will be available upon reasonable request.

PEER REVIEW
The peer review history for this article is available at https://publons.