Predictors of Seizure Freedom in Patients Undergoing Surgery for Central Nervous System Infection-Related Epilepsy: A Systematic Review and Meta-Analysis

Objective: Central nervous system infections (CNSIs), especially viral encephalitis and meningitis, are well-recognized causes of medically refractory epilepsy. Although surgery is an effective and durable intervention against these infections, the seizure control outcomes described in previous surgical series have been variable. Accordingly, it is not clear which variables are most valuable in predicting seizure control following surgery for CNSI. The aim of this meta-analysis was to identify the predictors of favorable surgical outcomes in CNSI-related epilepsy. Methods: The PubMed, EMBASE, Cochrane Library, WANGFANG, VIP, CBM, and CNKI databases were searched for studies according to the inclusion criteria. Prognostic factors, surgical outcomes, and patient characteristics were extracted. Heterogeneity was detected by the I(2) and Q statistics. Results: Seventeen studies were included in our meta-analysis. Eight predictors of favorable outcomes (Engel Class I/II) were determined, including abnormal MRI findings, meningitis, temporal location only, regional ictal pattern, unilateral ictal pattern, older age at epilepsy, longer silent period, and longer time from infection, as follows: OR = 3.34 (95% CI 1.44–7.74), OR = 0.31 (95% CI 0.13–0.70), OR = 0.34 (95% CI 0.16–0.74), OR = 5.65 (95% CI 1.75–18.30), and OR = 9.53 (95% CI 2.36–38.48), respectively, and MD = 2.15 (95% CI 0.20–4.11), MD = 2.40 (95% CI 0.09–4.70), and MD = 8.49 (95% CI 1.50–15.48), respectively. A subgroup analysis found the following associations: regional and unilateral ictal patterns in viral encephalitis, a younger age at infection in parasitic encephalopathy, an older age at surgery, a longer time from onset, and a longer time from infection in unexplained meningitis. A sensitivity analysis restricted to studies that included each variable yielded robust results. Little evidence of publication bias was observed. Conclusions: This meta-analysis suggests that abnormal MRI findings, meningitis, temporal location only, regional and unilateral ictal patterns, older age at epilepsy, longer silent period, and longer time from infection are predictive factors in patients with favorable surgical outcomes in CNSI-related epilepsy. In addition, different infective agents influenced the results in regional and unilateral ictal patterns in ictal electroencephalography, as well as the relationship between age at infection and surgery and the time from epilepsy onset and infection.