Introduction: Cervical Spondylotic Myelopathy (CSM) is the commonest cause of spinal cord dysfunction in the elderly worldwide. Though Magnetic Resonance Imaging (MRI) is the chief image modality for confirming the diagnosis of CSM, its ability to predict postsurgical outcome remains unclear.
Methods: Of 278 adults with =1 clinical signs of CSM, 114 had complete data available for assessment at baseline and 6-months postsurgical follow-up. MRIs were reviewed by 3 investigators for presence or absence (+/-) of signal change. Further quantitative analysis of T2 signal intensity was conducted. A dichotomized modified Japanese Orthopedic Association measure (mJOA) was used to discriminate between a good (=16) or poor (<16) postsurgical outcome. Manual backward stepwise logistic regression identified the best combination of imaging predictors and subsequent addition of clinical predictors using manual forward stepwise logistic regression identified the best combined model.
Results: Univariate analysis identified 2 imaging parameters that were significantly associated with our binary mJOA variable: spinal canal compromise (OR: 0.965, 95% CI: 0.934, 0.997, p=0.0322) and T2-WI signal change height (OR: 0.459, 95% CI: 0.222, 0.948, p=0.0354). The final image-based model consisted of 2 predictors: +/- T1-WI signal hypointensity (p=0.0675) and canal compromise (p=0.0136). The area under the curve (AUC) was 0.687 (95% CI: 0.576, 0.799). The combined clinical-MRI prediction model included baseline mJOA (p<0.0001), +/- psychiatric history (p=0.0149), canal compromise (p=0.0022) and +/- T1-WI signal hypointensity (p=0.0068). This combined model had an AUC of 0.872 (95% CI: 0.801, 0.943), which indicates better predictive performance than either clinical or imaging prediction models independently.
Conclusions: Canal compromise and +/- T1-WI signal hypointensity are the strongest MRI predictors of outcome. Combining MRI and clinical predictors yields a prediction model with 87% accuracy and has better performance than either one independently. It is thus recommended that comprehensive MRI analysis accompany clinical assessment of surgical candidates.
Patient Care: The combined imaging and clinical prediction model provides a tool for surgeons to assess how patients can expect to fare after surgical intervention and provides guidance for managing patient expectations.
Learning Objectives: By the conclusion of this session, participants should be able to: 1) Understand the importance of baseline MRI analysis in predicting postsurgical outcome 2) Describe which MRI and clinical factors are important clinical predictors of postsurgical outcome 3) Discuss how this information can be useful in clinical practice.