Multidimensional Analysis of the Ability of Magnetic Resonance Imaging Variables and Peripheral White Blood Cells to Predict Neurological Impairment Following Acute Traumatic Cervical Spinal Cord Injury.

Abstract

To investigate the association of magnetic resonance imaging (MRI) and peripheral blood inflammatory biomarkers with follow-up American Spinal Injury Association Impairment Scale (AIS) grade in patients with acute traumatic spinal cord injury in the cervical region (TSCI-C) and assess whether combining MRI biomarkers with peripheral white blood cell (WBC) count improves AIS grade prediction. This retrospective cohort study included 307 patients with acute TSCI-C. The patients underwent peripheral blood tests within 12 hours and preoperative MRI within 48 hours of injury. MRI biomarkers included maximum cervical canal compromise, maximum spinal cord compression, intramedullary lesion length, Brain and Spinal Cord Injury Center score, and sagittal grade. Peripheral blood inflammatory biomarkers included WBCs, neutrophils, and neutrophil ratio. Data-driven nonlinear principal component analysis, correlation analysis, and optimal scaling multivariate regression were conducted to predict neurologic impairment at follow-up. Maximum canal compromise and maximum spinal cord compression loaded positively on second principal component, while the Brain and Spinal Cord Injury Center score, sagittal grade, and intramedullary lesion length loaded negatively. The WBC count was modeled quadratically (WBC²), showing an initial increase from grade A to C, followed by a decrease, resembling a quadratic function. Despite negative loadings on first and second principal components, the WBC count had low classification ability for AIS grade. Receiver operating characteristic analysis showed that combining MRI biomarkers with WBC count yielded a higher area under the curve than the MRI-only model. We recommend using a combined approach of MRI biomarkers and peripheral WBC count to improve the prediction of neurological impairment in TSCI-C patients, enhancing the sensitivity and specificity of AIS grade prediction.