Development of Noninvasive Biomarkers for Cervical Spondylotic Myelopathy

Cervical spondylotic myelopathy (CSM) represents the most common cause of chronic spinal cord injury (SCI) in adults. Many patients with symptomatic CSM will experience a decline in neurological function and consequently undergo surgical decompression. Unfortunately, surgeons are unable to adequately counsel patients about the benefits of surgery because the natural history of disease and outcome after decompression vary widely among patients. This can hinder the decision-making capacity of patients and physicians. Therefore, we require additional tools to help guide therapy and counsel patients with CSM. Noninvasive biomarkers present valuable potential as predictors of a patient’s recovery in the long term. Due to the limitations of current MRI techniques, there is a lack of clinically useful biomarkers. Thus, there is a critical need for accurate biomarkers to predict the extent of central nervous system (CNS) reorganization, and the extent of recovery following treatment. Advances in imaging have made direct probing of CNS properties a lot easier while acquiring high-resolution data. Many brain networks are connected to spinal cord afferent tracts. Although these afferent circuits have been identified in anatomical circuits, brain imaging biomarkers for CSM remain under-investigated. Heterogeneity in CSM grade, level, time, and techniques have led to incongruent findings and an inadequate understanding of neuroanatomy and temporal progression of cerebral remodeling after CSM. Brain imaging biomarkers could therefore contribute to a multimodal diagnostic algorithm or serve as potential targets for therapy. We applied a Human Connectome Project (HCP) approach which uses multiple imaging modalities to accurately align gray matter to specific brain regions, maximize data quality, minimize technical artifacts, respect cerebral geometry, align topography across subjects and create neuroanatomically accurate maps. We hypothesized that cerebral architecture and connectivity metrics will accurately reflect changes after CSM that can be used to predict clinical course and long-term prognosis. On the CSM spinal imaging forefront, much work remains to be completed. Diffusion tensor imaging (DTI) is a neuroimaging technique widely used to assess CNS tissue pathology and is increasingly used in CSM. However, DTI lacks the needed accuracy, precision, and recall to image pathologies of spinal cord injury as the disease progresses. Thus, we employed diffusion basis spectrum imaging (DBSI) to delineate white matter injury more accurately in the setting of spinal cord compression. We hypothesized that the profiles of multiple DBSI metrics can serve as imaging outcome predictors to accurately predict a patient\u27s response to therapy and long-term prognosis. We tested this hypothesis by using DBSI metrics as input features in a support vector machine (SVM) algorithm. Two non-overlapping patient cohorts were recruited for this study. Nine CSM patients and twenty-seven healthy controls received brain MRI exams. The imaging data was processed using the HCP’s data processing pipelines. This processed data was then parcellated using the HCP’s parcellation. Fifty CSM patients and twenty healthy controls underwent diffusion-weighted spine MRI exams. All spinal cord white matter was identified as the region of interest (ROI). DBSI and DTI metrics were extracted from all voxels in the ROI and the median value of each patient was used in analyses. An SVM with optimized hyperparameters was trained using clinical and imaging metrics separately and collectively to predict patient outcomes. Patient outcome was determined by calculating changes between pre-and postoperative mJOA scores.We found CSM patients showed differences in cortical structural and functional parameters when compared to controls. CSM patients showed decreased cortical thickness within the dorsolateral prefrontal cortex which is implicated in the spatial processing of pain. This change is accompanied by reduced functional connectivity in resting state networks involved in vision and spatial awareness. Furthermore, CSM patients showed altered hub topology with the bilateral primary motor cortex and area 3a appearing as new hubs. When assessing patient outcome (following spinal decompression) using patient mJOA scores, we show that the highest SVM performance was observed when a combination of clinical and spinal cord DBSI metrics were used to train an SVM. We found this to be true for patient outcomes determined using NDI, SF-36 MCS, and DASH patient scores on separately trained SVMs. In conclusion, our goal was to expand our understanding of changes in the brain and spinal cord in the setting of CSM to better predict treatment outcomes. We found that the affected cortical regions are engaged in the coordination of motor control and other sensory processes that facilitate spatial navigation. The alterations in proprioceptive representation, visual perception, and motor activation lead to downstream effects in higher-order processing centers which may directly produce pain and sustain altered motor control strategies. We also found that the accuracy and efficacy of the SVM incorporating clinical and spinal cord DBSI metrics show great promise for clinical applications in predicting patient outcomes. Our results suggest that DBSI metrics along with clinical presentation could serve as a surrogate in prognosticating outcomes of CSM patients