Abstract：
BACKGROUND: Seizures are a common complication of subarachnoid hemorrhage (SAH) in both acute and late stages: 10-20 % acute symptomatic seizures, 12-25 % epilepsy rate at five years. Our aim was to identify early electroencephalogram (EEG) and computed tomography (CT) findings that could predict long-term epilepsy after SAH.
METHODS: This is a multicenter, retrospective, longitudinal study of adult patients with aneurysmal SAH admitted to two tertiary care hospitals between January 2011 to December 2022. Routine 30-minute EEG recording was performed in all subjects during admission period. Exclusion criteria were the presence of prior structural brain lesions and/or known epilepsy. We documented the presence of SAH-related cortical involvement in brain CT and focal electrographic abnormalities (epileptiform and non-epileptiform). Post-SAH epilepsy was defined as the occurrence of remote unprovoked seizures ≥ 7 days from the bleeding.
RESULTS: We included 278 patients with a median follow-up of 2.4 years. The mean age was 57 (+/-12) years, 188 (68 %) were female and 49 (17.6 %) developed epilepsy with a median latency of 174 days (IQR 49-479). Cortical brain lesions were present in 189 (68 %) and focal EEG abnormalities were detected in 158 patients (39 epileptiform discharges, 119 non-epileptiform abnormalities). The median delay to the first EEG recording was 6 days (IQR 2-12). Multiple Cox regression analysis showed higher risk of long-term epilepsy in those patients with CT cortical involvement (HR 2.6 [1.3-5.2], p 0.009), EEG focal non-epileptiform abnormalities (HR 3.7 [1.6-8.2], p 0.002) and epileptiform discharges (HR 6.7 [2.8-15.8], p < 0.001). Concomitant use of anesthetics and/or antiseizure medication during EEG recording had no influence over its predictive capacity. ROC-curve analysis of the model showed good predictive capability at 5 years (AUC 0.80, 95 %CI 0.74-0.87).
CONCLUSIONS: Focal electrographic abnormalities (both epileptiform and non-epileptiform abnormalities) and cortical involvement in neuroimaging predict the development of long-term epilepsy. In-patient EEG and CT findings could allow an early risk stratification and facilitate a personalized follow-up and management of SAH patients.
METHODS: This is a multicenter, retrospective, longitudinal study of adult patients with aneurysmal SAH admitted to two tertiary care hospitals between January 2011 to December 2022. Routine 30-minute EEG recording was performed in all subjects during admission period. Exclusion criteria were the presence of prior structural brain lesions and/or known epilepsy. We documented the presence of SAH-related cortical involvement in brain CT and focal electrographic abnormalities (epileptiform and non-epileptiform). Post-SAH epilepsy was defined as the occurrence of remote unprovoked seizures ≥ 7 days from the bleeding.
RESULTS: We included 278 patients with a median follow-up of 2.4 years. The mean age was 57 (+/-12) years, 188 (68 %) were female and 49 (17.6 %) developed epilepsy with a median latency of 174 days (IQR 49-479). Cortical brain lesions were present in 189 (68 %) and focal EEG abnormalities were detected in 158 patients (39 epileptiform discharges, 119 non-epileptiform abnormalities). The median delay to the first EEG recording was 6 days (IQR 2-12). Multiple Cox regression analysis showed higher risk of long-term epilepsy in those patients with CT cortical involvement (HR 2.6 [1.3-5.2], p 0.009), EEG focal non-epileptiform abnormalities (HR 3.7 [1.6-8.2], p 0.002) and epileptiform discharges (HR 6.7 [2.8-15.8], p < 0.001). Concomitant use of anesthetics and/or antiseizure medication during EEG recording had no influence over its predictive capacity. ROC-curve analysis of the model showed good predictive capability at 5 years (AUC 0.80, 95 %CI 0.74-0.87).
CONCLUSIONS: Focal electrographic abnormalities (both epileptiform and non-epileptiform abnormalities) and cortical involvement in neuroimaging predict the development of long-term epilepsy. In-patient EEG and CT findings could allow an early risk stratification and facilitate a personalized follow-up and management of SAH patients.
摘要：
背景：癫痫发作是急性和晚期蛛网膜下腔出血（SAH）的常见并发症：10-20％的急性症状性癫痫发作，五年癫痫发生率为12-25%。我们的目的是确定可以预测SAH后长期癫痫的早期脑电图（EEG）和计算机断层扫描（CT）发现。
方法：这是一个多中心，回顾性,对2011年1月至2022年12月在两家三级医院接受治疗的动脉瘤性SAH成年患者进行纵向研究.在入院期间对所有受试者进行常规30分钟脑电图记录。排除标准是存在先前的结构性脑损伤和/或已知的癫痫。我们记录了脑CT和局灶性电图异常（癫痫样和非癫痫样）中与SAH相关的皮质受累的存在。SAH后癫痫定义为出血后≥7天发生远程无源性癫痫。
结果：我们纳入了278例患者，中位随访时间为2.4年。平均年龄为57(+/-12)岁,188（68％）为女性，49（17.6％）发展为癫痫，中位潜伏期为174天（IQR49-479）。189例（68％）存在皮质脑病变，158例患者发现局灶性EEG异常（39例癫痫样放电，119例非癫痫样异常）。第一次脑电图记录的中位延迟为6天(IQR2-12)。多Cox回归分析显示，CT皮质受累患者发生长期癫痫的风险更高（HR2.6[1.3-5.2]，p0.009)，脑电图局灶性非癫痫样异常（HR3.7[1.6-8.2]，p0.002)和癫痫样放电(HR6.7[2.8-15.8]，p<0.001)。在EEG记录期间同时使用麻醉药和/或抗癫痫药物对其预测能力没有影响。模型的ROC曲线分析在5年显示出良好的预测能力（AUC0.80，95CI0.74-0.87）。
结论：局灶性电图异常（癫痫样和非癫痫样异常）和神经影像学中的皮质受累可预测长期癫痫的发展。患者的EEG和CT发现可以进行早期风险分层，并促进SAH患者的个性化随访和管理。
方法：这是一个多中心，回顾性,对2011年1月至2022年12月在两家三级医院接受治疗的动脉瘤性SAH成年患者进行纵向研究.在入院期间对所有受试者进行常规30分钟脑电图记录。排除标准是存在先前的结构性脑损伤和/或已知的癫痫。我们记录了脑CT和局灶性电图异常（癫痫样和非癫痫样）中与SAH相关的皮质受累的存在。SAH后癫痫定义为出血后≥7天发生远程无源性癫痫。
结果：我们纳入了278例患者，中位随访时间为2.4年。平均年龄为57(+/-12)岁,188（68％）为女性，49（17.6％）发展为癫痫，中位潜伏期为174天（IQR49-479）。189例（68％）存在皮质脑病变，158例患者发现局灶性EEG异常（39例癫痫样放电，119例非癫痫样异常）。第一次脑电图记录的中位延迟为6天(IQR2-12)。多Cox回归分析显示，CT皮质受累患者发生长期癫痫的风险更高（HR2.6[1.3-5.2]，p0.009)，脑电图局灶性非癫痫样异常（HR3.7[1.6-8.2]，p0.002)和癫痫样放电(HR6.7[2.8-15.8]，p<0.001)。在EEG记录期间同时使用麻醉药和/或抗癫痫药物对其预测能力没有影响。模型的ROC曲线分析在5年显示出良好的预测能力（AUC0.80，95CI0.74-0.87）。
结论：局灶性电图异常（癫痫样和非癫痫样异常）和神经影像学中的皮质受累可预测长期癫痫的发展。患者的EEG和CT发现可以进行早期风险分层，并促进SAH患者的个性化随访和管理。
