• Aria Nouri, Lindsay Tetreault, Pierre Côté, Juan J Zamorano, Kristian Dalzell, and Michael G Fehlings.
• *Division of Neurosurgery and Spine Program, Toronto Western Hospital, Toronto, Ontario, Canada; †Toronto Western Research Institute, University Health Network, Toronto, Canada; ‡Institute of Medical Science, University of Toronto, Toronto, Ontario, Canada; §Faculty of Health Sciences, University of Ontario Institute of Technology (UOIT), Director, UOIT-CMCC Centre for the Study of Disability Prevention and Rehabilitation, Toronto, Ontario, Canada; ¶Christchurch Public Hospital, Christchurch, New Zealand; and ‖Burwood Spinal Unit, Christchurch, New Zealand.
• Spine. 2015 Jul 15; 40 (14): 109211001092-100.

Study DesignAmbispective study.ObjectiveTo determine whether MRI parameters improve the predictive performance of a validated clinical prediction rule used to assess functional outcomes in surgical patients with DCM.Summary Of Background DataDegenerative cervical myelopathy (DCM) is the most common cause of spinal cord dysfunction in the elderly worldwide. A clinical prediction rule was developed to discriminate between patients with mild myelopathy postoperatively (mJOA ≥ 16) and those with substantial residual neurological impairment (mJOA < 16). Recently, a separate magnetic resonance imaging (MRI)-based prediction model was created. However, a model exploring the combined predictive value of imaging and clinical variables does not exist.MethodsOne hundred and fourteen patients with MRIs were examined from a cohort of 278 patients enrolled in the AOSpine CSM-North America Study. Ninety-nine patients had complete preoperative imaging and postoperative outcome data. MRIs were evaluated for the presence/absence of signal change on T2- and T1-weighted images. Quantitative analysis of the T2 signal change was conducted and maximum canal compromise and cord compression were calculated. The added predictive performance of each MRI parameter to the clinical model was evaluated using receiver operator characteristic curves.ResultsThe model developed on our subsample yielded an area under the receiver operator curve (AUC) of 0.811 (95% CI: 0.726-0.896). The addition of imaging variables did not significantly improve the predictive performance. Small improvements in prediction were obtained when sagittal extent of T2 hyperintensity (AUC: 0.826, 95% CI: 0.743-0.908, 1.35% increase) or Wang ratio (AUC: 0.823, 95% CI: 0.739-0.907, 1.21%) was added. Anatomic characteristics, such as maximum canal compromise and maximum cord compression, did not improve the discriminative ability of the clinical prediction model.ConclusionIn our sample of surgical patients, with clinical and image-evidence of DCM, MRI parameters do not significantly add to the predictive performance of a previously published clinical prediction rule. It remains plausible that combinations of the strongest clinical and MRI predictors may yield a similar or a superior prediction model.Level Of Evidence3.

31 keywords...

hide…