Epilepsy may result from various brain injuries, including stroke (ischemic and hemorrhagic), traumatic brain injury, and infections. Identifying shared common biological pathways and biomarkers of the epileptogenic process initiated by the different injuries may lead to novel targets for preventing the development of epilepsy. We systematically reviewed biofluid biomarkers to test their association with the risk of post–brain injury epilepsy.

We searched articles until January 25, 2022, in MEDLINE, Embase, PsycInfo, Web of Science, and Cochrane. The primary outcome was the difference in mean biomarker levels in patients with and without post–brain injury epilepsy. We used the modified quality score on prognostic studies for risk of bias assessment. We calculated each biomarker's pooled standardized mean difference (SMD) and 95% CI. Molecular interaction network and enrichment analyses were conducted in Cytoscape (PROSPERO CRD42021297110).

We included 22 studies with 1,499 cases with post–brain injury epilepsy and 7,929 controls without post–brain injury epilepsy. Forty-five biomarkers in the blood or CSF were investigated with samples collected at disparate time points. Of 22 studies, 21 had a moderate-to-high risk of bias. Most of the biomarkers (28/45) were investigated in single studies; only 9 provided validation data, and studies used variable definitions for early-onset and late-onset seizures. A meta-analysis was possible for 19 biomarkers. Blood glucose levels in 4 studies were significantly higher in patients with poststroke epilepsy (PSE) than those without PSE (SMD 0.44; CI 0.19–0.69). From individual studies, 15 biomarkers in the blood and 7 in the CSF were significantly associated with post–brain injury epilepsy. Enrichment analysis identified that the significant biomarkers (interleukin [IL]–6, IL-1β]) were predominantly inflammation related.

We cannot yet recommend using the reported biomarkers for designing antiepileptogenesis trials or use in the clinical setting because of methodological heterogeneity, bias in the included studies, and insufficient validation studies. Although our analyses indicate the plausible role of inflammation in epileptogenesis, this is likely not the only mechanism. For example, an individual's genetic susceptibilities might contribute to his/her risk of epileptogenesis after brain injury. Rigorously designed biomarker studies with methods acceptable to the regulatory bodies should be conducted.

癫痫可能由各种脑损伤引起，包括中风（缺血性和出血性）、创伤性脑损伤和感染。识别不同损伤引发的癫痫发生过程的共同生物学途径和生物标志物可能会产生预防癫痫发展的新靶点。我们系统地审查了生物流体生物标志物，以测试它们与脑损伤后癫痫风险的关联。

我们检索了 MEDLINE、Embase、PsycInfo、Web of Science 和 Cochrane 中截至 2022 年 1 月 25 日的文章。主要结果是患有和不患有脑损伤后癫痫患者的平均生物标志物水平的差异。我们在预后研究中使用修改后的质量评分来评估偏倚风险。我们计算了每个生物标志物的汇总标准化平均差 (SMD) 和 95% CI。在 Cytoscape (PROSPERO CRD42021297110) 中进行分子相互作用网络和富集分析。

我们纳入了 22 项研究，其中包括 1,499 例脑损伤后癫痫病例和 7,929 例无脑损伤后癫痫的对照组。通过在不同时间点收集的样本对血液或脑脊液中的 45 种生物标志物进行了研究。在 22 项研究中，21 项存在中度至高度偏倚风险。大多数生物标志物 (28/45) 在单一研究中进行了调查；只有 9 项提供了验证数据，并且研究对早发性和迟发性癫痫发作使用了不同的定义。可以对 19 种生物标志物进行荟萃分析。4 项研究显示，卒中后癫痫 (PSE) 患者的血糖水平显着高于未患 PSE 的患者 (SMD 0.44；CI 0.19–0.69)。根据个别研究，血液中的 15 种生物标志物和脑脊液中的 7 种生物标志物与脑损伤后癫痫显着相关。富集分析发现，重要的生物标志物（白细胞介素 [IL]–6、IL-1β]）主要与炎症相关。

由于方法学异质性、纳入研究的偏倚以及验证研究不足，我们尚不能推荐使用已报告的生物标志物来设计抗癫痫发生试验或在临床环境中使用。尽管我们的分析表明炎症在癫痫发生中可能发挥作用，但这可能不是唯一的机制。例如，个体的遗传易感性可能会增加他/她脑损伤后发生癫痫的风险。应采用监管机构可接受的方法进行严格设计的生物标志物研究。