Development and Validation of a Novel Mri-based Risk Score (IsCHEMIa) for Prediction of Post-stroke Epilepsy
Abstract number :
1.317
Submission category :
4. Clinical Epilepsy / 4B. Clinical Diagnosis
Year :
2024
Submission ID :
1183
Source :
www.aesnet.org
Presentation date :
12/7/2024 12:00:00 AM
Published date :
Authors :
Presenting Author: William C.Y. Leung, MBBS(HK, MRCP(UK), FHKCP, FHKAM(Med) – Queen Mary Hospital, University of Hong Kong
Gary Kui-Kai Lau, MBBS (HK), MRes (Med) (HK), MRCP (UK), DPhil (Oxon), AFHEA (UK), FHKCP, FHKAM (Medicine), FRCP (Edin, Glasg), FESO – University of Hong Kong
Yuen Kwun Wong, PhD – University of Hong Kong
Ryan Wui-Hang Ho, MBBS – Queen Mary Hospital, University of Hong Kong
Florinda Hui-Ning Chu, MBBS – Queen Mary Hospital, University of Hong Kong
Aneesh B. Singhal, MD – Massachusetts General Hospital, Harvard Medical School
Andrew J. Cole, MD – Massachusetts General Hospital, Harvard Medical School
Rationale: Post-stroke epilepsy (PSE) is an uncommon but important complication of ischaemic stroke. Risk stratification models may identify high-risk patients who may benefit from seizure prophylaxis in the early stages of acute stroke. Existing risk scores for PSE (e.g. SeLECT) are largely derived from clinical and imaging parameters based on computed tomography (CT) scans. With the increasing accessibility of magnetic resonance imaging (MRI), we aimed to develop a novel MRI-based risk model for PSE with external validation in international cohorts.
Methods: We conducted a multi-national, multi-center, retrospective study. The derivative cohort comprised 852 patients consecutively admitted to Queen Mary Hospital, Hong Kong for first-ever acute ischemic stroke from August 2018 to December 2020. Univariate analysis with competing risks regression was performed with all-cause mortality considered as a competing risk. A risk score was derived using demographic, clinical, and imaging variables that reached statistical significance (p < 0.05) on univariate analysis. The risk score was externally validated in an independent cohort of 1438 patients consecutively admitted to Massachusetts General Hospital, Boston, MA, USA for acute ischemic stroke from January 2016 to December 2018. The predictive value of the new IsCHEMIa score was compared with the SeLECT score by the concordance (c) statistic, as illustrated by the area under receiver operating characteristic (ROC) curves.
Results: At a median follow-up duration of 892 and 2131 days, PSE occurred in 5.3% and 6.5% of patients In the Hong Kong and USA cohorts respectively. Mean time to first seizure was 281.1 ± 360.2 days (Hong Kong) and 497.7 ± 619.2 days (USA). All patients underwent a CT scan, while 70.1% and 91.6% had an MRI scan in the respective cohorts. Variables that reached statistical significance in univariate analysis (Table 1) were included in the risk score, which was calculated by dividing the regression coefficients by the median of the lowest three values and rounding to the nearest integer. The resulting IsCHEMIa score comprised of initial severity NIHSS ≥ 11 (Is – 1), cortical involvement (C – 1), hemorrhagic transformation (H – 2), early seizure ≤ 7 days after stroke (E – 7), middle cerebral artery territory (M – 1), and infarct area ≥ 5cm (Ia – 2).
The IsCHEMIa score outperformed the SeLECT score in both the derivative and validation cohorts, with a c statistic of 0.848 (vs. SeLECT 0.774, Z=3.443, p=0.001) in the HK cohort and 0.903 (vs. SeLECT 0.787, Z=7.349, p< 0.0001) in the USA cohort (Figure 1).
Conclusions: The newly proposed IsCHEMIa score has a significant improvement in predictive value for post-stroke epilepsy compared to existing risk scores. Future randomized controlled trials incorporating the IsCHEMIa score may further demonstrate the risk-benefit of seizure prophylaxis among patients with acute ischemic stroke.
Funding: No funding that was received in support of this abstract.
Clinical Epilepsy