Brain MRI to pre-symptomatically predict seizure onset for Sturge-Weber Syndrome

摘要 斯特奇-韦伯综合征是一种罕见的神经系统疾病，其最大的问题是神经认知 按学龄(6-10岁)划分的残疾。为了改善神经认知结果，Kenndy Krieger研究所(Dr. COMI，这一R21的联合PI)和波士顿儿童医院(平托医生，这一R21的联合PI)一直是领先的或 各种临床试验的关键地点，以测试新的治疗方法(NCT02332655(2014-19年)；NCT0304980(2017-19年)； NCT04447846(2019-21))，或开发可以选择高危患者的神经成像生物标志物 (NCT01345305(2010-2012年)；NCT01425944(2010-2020年)；NCT04717427(2021-2024年))。然而，所有这些审判 重点放在症状后阶段--癫痫症状出现后。我们最近的证据表明 症状前治疗-在癫痫症状出现之前对患者进行治疗，理想情况是在2年前 年龄--可延缓或避免癫痫症状。这一点很重要，因为那些在2点之前没有癫痫症状的人 适龄儿童(占SWS患者的10%-25%)通常在学龄前具有良好的神经认知效果。动机是 对此，多学科专家齐聚一堂，在2018年、2019年、2021年达成共识，呼吁立即 对症状前治疗的调查。作为对这一呼吁的及时回应，KKI和BCH这两家最大的 对SWS进行症状前治疗的国家中心正计划进行一项试验，以综合评估 抗癫痫药物左乙拉西坦在两个治疗组中的作用(使用小剂量左乙拉西坦与不使用小剂量左乙拉西坦 阿司匹林)用于症状前治疗。然而，这项计划中的试验的瓶颈问题是缺乏 准确、有症状地识别有癫痫风险的SWS患者的生物标记物 2岁时出现症状。这些高危患者应该是我们计划中的理想候选人 审判。这款R21旨在解决这一瓶颈生物标记物问题。我们计划追溯建造最大的 来自KKI和BCH临床数据的多点症状前数据库(目标1)。我们将彻底评估两个 临床和脑MRI生物标记物，用于在症状前识别高危患者(目标2)。中环 假设人工智能(AI)算法从临床和大脑中提取的复杂特征 MRI可以作为一种生物标志物，在症状前识别有癫痫风险的SWS患者 2岁时出现症状。这是第一个人工智能支持的、大数据集驱动的针对症状前的严格研究 这种罕见疾病的临床和核磁共振生物标志物。这样的生物标记物将立即用于我们计划的 评价左乙拉西坦加或不加小剂量阿司匹林用于症状前治疗的临床试验。

Abstract Sturge-Weber syndrome (SWS) is a rare neurological disease, and its biggest concern is neurocognitive impairments by school age (6-10 years). To improve neurocognitive outcomes, Kenndy Krieger Institute (Dr. Comi, co-PI of this R21) and Boston Children’s Hospital (Dr. Pinto, co-PI of this R21) have been the leading or key sites in various clinical trials, to test new treatment (NCT02332655 (2014-19); NCT0304980 (2017-19); NCT04447846 (2019-21)), or to develop neuroimaging biomarkers that can select at-risk patients (NCT01345305 (2010-2012); NCT01425944 (2010-2020); NCT04717427 (2021-2024)). However, all these trials focus on the post-symptomatic phase – after seizure symptoms have occurred. Our recent evidence suggested that pre-symptomatic treatment – treating patients before seizure symptoms occur, ideally before 2 years of age – may delay or avoid seizure symptoms. This is important, because those without seizure symptoms by 2 years of age (10-25% of SWS patients) often enjoy good neurocognitive outcomes by school age. Motivated by this, multidisciplinary experts gathered and reached a consensus in 2018, 2019, and 2021, calling for immediate investigations of pre-symptomatic treatments. In a timely response to this call, KKI and BCH, the two largest national centers that treat SWS pre-symptomatically, are planning for a trial to comprehensively evaluate the effect of anti-epilepsy drugs Levetiracetam, in two treatment arms (Levetiracetam with versus without low-dose aspirin) for pre-symptomatic treatment. The bottleneck issue for this planned trial, though, is the lack of a biomarker to accurately and pre-symptomatically identify SWS patients who are at risk to develop seizure symptoms by 2 years of age. Those at-risk patients should be ideal candidates to be included in our planned trial. This R21 aims to address this bottleneck biomarker problem. We plan to retrospectively build the largest multi-site presymptomatic database from clinical data in KKI and BCH (Aim 1). We will thoroughly evaluate two clinical and brain MRI biomarkers to identify at-risk patients pre-symptomatically (Aim 2). The central hypothesis is that sophisticated features that artificial intelligence (AI) algorithms extract from clinical and brain MRI could serve as a biomarker to pre-symptomatically identify SWS patients at risk of developing seizure symptoms by 2 years of age. This is the first AI-powered, large-dataset-driven rigorous study for presymptomatic clinical and MRI biomarkers for this rare disease. Such a biomarker will be immediately used in our planned clinical trials to evaluate Levetiracetam with or without low-dose aspirin for pre-symptomatic treatment.