RESUMEN
OBJECTIVE: To evaluate the accuracy, sensitivity, and specificity of various Magnetic Resonance Imaging (MRI) features in differentiating malignant from benign compression fracture of the spine. MATERIAL AND METHOD: Retrospective review of MRI spine of patients with vertebral compression fracture identified from the hospital database between June 2004 and February 2006 by two radiologists blinded to the clinical data. Various MRI features were evaluated for sensitivity, specificity, positive predictive value, and negative predictive value. An additional combination of two, three, four, and five MRI features that had statistically significant (P value less than 0.005) were also calculated for sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). RESULTS: Fifty-eight spinal MRI were included from 35 patients with metastatic vertebral compression fractures and 23 patients with benign vertebral compression fractures. MR imaging features suggestive of malignant vertebral compression fracture were convex posterior border of the vertebral body, involvement of the pedicle or posterior element, epidural mass, paraspinal mass, and destruction of bony cortex. Among these, involvement of pedicle or posterior element was the most reliable finding (sensitivity 91.4% and specificity 82.6%) for diagnosis of malignant vertebral compression fracture. A combination of two or more MRI features gave very high specificity and PPV. CONCLUSION: Certain MR imaging characteristics can reliably distinguish malignant from benign compression fracture of the spine. Combination of several MRI features strongly affirmed the diagnosis of malignant compression fracture, especially in a patient where tissue biopsy is not justified.