Can anatomic level of injury on MRI predict neurological level in acute cervical spinal cord injury?

• Published in: British journal of neurosurgery 2016, Vol.30 (2), p.204-210
• Main Authors: Zohrabian, Vahe M., Parker, Laurence, Harrop, James S., Vaccaro, Alex R., Marino, Ralph J., Flanders, Adam E.
• Format: Article
• Language: English
• Subjects: Adolescent; Adult; Aged; Aged, 80 and over; Cervical Cord - injuries; Cervical Cord - pathology; Cervical Cord - surgery; Cervical Vertebrae - pathology; Cervical Vertebrae - surgery; edema; Female; hemorrhage; Humans; Magnetic Resonance Imaging - methods; Male; Middle Aged; MRI; Neck Injuries - diagnosis; Neck Injuries - surgery; Neurologic Examination - methods; neurological level of injury; Neurology; NMR; Nuclear magnetic resonance; Reliability; Retrospective Studies; Spinal Canal - pathology; Spinal Canal - surgery; Spinal cord injuries; Spinal Cord Injuries - diagnosis; Spinal Cord Injuries - surgery; spinal cord injury; Young Adult
• ISSN: 0268-8697; 1360-046X
• DOI: 10.3109/02688697.2015.1056089

Background. Determining neurological level of injury (NLI) is of paramount importance after spinal cord injury (SCI), although its accuracy depends upon the reliability of the neurologic examination. Here, we determine if anatomic location of cervical cord injury by MRI (MRI level of injury) can predict NLI in the acute traumatic setting. Methods. A retrospective review was undertaken of SCI patients with macroscopic evidence of cervical cord injury from non-penetrating trauma, all of whom had undergone cervical spine MRI and complete neurologic testing. The recorded MRI information included cord lesion type (intra-axial edema, hemorrhage) and MRI locations of upper and lower lesion boundary, as well as lesion epicenter. Pearson correlation and Bland-Altman analyses were used to assess the relationship between MRI levels of injury and NLI. Results. All five MRI parameters, namely (1) upper and (2) lower boundaries of cord edema, (3) lesion epicenter, and (4) upper and (5) lower boundaries of cord hemorrhage demonstrated statistically significant, positive correlations with NLI. The MRI locations of upper and lower boundary of hemorrhage were found to have the strongest correlation with NLI (r = 0.72 and 0.61, respectively; p < 0.01). A weaker (low to moderate) correlation existed between lower boundary of cord edema and NLI (r = 0.30; p < 0.01). Upper boundary of cord hemorrhage on MRI demonstrated the best agreement with NLI (mean difference 0.03 ± 0.73; p < 0.01) by Bland-Altman analysis. Conclusions. MRI level of injury has the potential to serve as a surrogate for NLI in instances where the neurologic examination is either unavailable or unreliable.