Risk Factors for Spinal Cord Compression on MRI in Oncology: Enhancing Diagnostic Yield

Epidural spinal cord compression (ESCC) is an important cause of disability among patients with cancer. Early detection is crucial for optimizing clinical outcomes. MRI is the preferred imaging technique for ruling out ESCC and is frequently requested in radiology departments, particularly in the emergent setting. However, data on the efficacy and diagnostic yield of total spine MRI for the diagnosis of ESCC in oncology patients remain limited. This study evaluates the frequency of positive findings and associated risk factors in a tertiary cancer center.

This retrospective study included patients who underwent total spine MRI for the assessment of ESCC during a 3-year period. A standardized noncontrast MRI protocol was used. Clinical and imaging data, including patient demographics (sex, age); tumor pathology; tumor, node, metastasis stage; ESCC grade; symptoms; prior treatments (radiation therapy, surgery, chemotherapy); and ordering physician/department, were retrospectively reviewed. Patients were categorized into 2 groups on the basis of the presence or absence of cord compression (ESCC 2 or 3). Associations between ESCC and other variables were assessed via the Wilcoxon rank-sum test, Pearson χ² test, and Fisher exact test. Statistical significance was defined as P < .05.

Among 289 patients (median age, 66 years; 148 women) and 300 total spine MRI examinations, ESCC was detected in 18 cases (6.0%). Significant associations with ESCC included advanced tumor, node, metastasis stage (P = .028) and prior treatments, such as radiation to the site of compression (P = .002), decompression surgery (P = .011), and recent systemic chemotherapy (P < .001). Bone metastases to the spine on body CT examinations performed within 2 weeks before MRI also correlated with ESCC (P < .001). Notably, no ESCC cases occurred in patients without spine bone metastases on recent body CT or in those with less than stage IV disease. Patient symptoms did not correlate with ESCC presence (P = .3).

This study suggests that the diagnostic yield of total spine MRI for ESCC in oncology patients is relatively low and may be improved by refining the selection criteria. Patients with advanced-stage disease, prior spinal interventions, and bone metastases on recent body CT may be at higher risk.