Zeqiang Ji,1,2 Wanxing Ye,2 Xinyu Wen,1,2 Xingquan Zhao,1– 4,* Na Li1– 3,* 

1Department of Neurology, Beijing Tiantan Hospital, Capital Medical University, Beijing, People’s Republic of China; 2China National Clinical Research Center for Neurological Diseases, Beijing Tiantan Hospital, Capital Medical University, Beijing, People’s Republic of China; 3Research Unit of Artificial Intelligence in Cerebrovascular Disease, Chinese Academy of Medical Sciences, Beijing, People’s Republic of China; 4Center of Stroke, Beijing Institute for Brain Disorders, Beijing, People’s Republic of China

*These authors contributed equally to this work

Correspondence: Na Li, Department of Neurology, Beijing Tiantan Hospital, Capital Medical University, No. 119 S Fourth Ring West Road, Fengtai District, Beijing, 100070, People’s Republic of China, Tel +010-59975029, Email selina1808@163.com

Aim: We aimed to investigate whether a combination of inflammatory and radiological biomarkers can improve intracerebral hemorrhage (ICH) hematoma expansion (HE) prediction.
Methods: A retrospective analysis was conducted on patients with primary supratentorial ICH within 6 h of symptom onset between September 2021 and April 2022. Predictors were explored using univariate and logistic regression analysis. We compared the discrimination of inflammatory indice-based model 1 with models 2 and 3, which included image biomarkers, using the receiver operating characteristic curve and De Long test for area under the curve comparison.
Results: A total of 205 eligible participants were included, 56 (27.3%) of whom experienced HE. The neutrophil-to-lymphocyte ratio (NLR), black hole sign, BAT score, and computed tomography angiography (CTA) spot sign were independently associated with HE in the logistic regression (P< 0.05). The addition of non-contrast computed tomography (NCCT) signs did not provide significant discrimination improvement (AUC, Model 2 0.875 [95% CI, 0.822– 0.929] versus Model 1. 0.811 [95% CI, 0.747– 0.875], p=0.089), whereas the added value of the CTA spot sign remained statistically significant (AUC, Model 3 0.922 [95% CI, 0.878– 0.966] versus Model 2; p=0.030; Model 3 versus Model 1, p=0.005).
Conclusion: The combination of inflammatory and radiological biomarkers can predict HE with a satisfactory performance.

Keywords: intracerebral hemorrhage, hematoma expansion, computed tomography angiography, spot sign, BAT score, neutrophil to lymphocyte ratio