Discrimination of tuberculous spondylitis from pyogenic spondylitis on MRI

OBJECTIVE. The purpose of this study was to determine the accuracy of MRI for discrimination between tuberculous spondylitis and pyogenic spondylitis. MATERIALS AND METHODS. MR images of 52 patients who had MRI of the spine and confirmed spondylitis were retrospectively reviewed. After review of medical records, we compared MRI findings in 20 patients with tuberculous spondylitis and 20 patients with pyogenic spondylitis. Statistical analysis was performed with the chi-square test. RESULTS. The reviewer identified tuberculous spondylitis with sensitivity, specificity, and accuracy of 100% (20/20), 80% (16/20), and 90% (36/40), and pyogenic spondylitis with sensitivity, specificity, and accuracy of 80% (16/20), 100% (20/20), and 90% (36/40), respectively. The patients with tuberculous spondylitis had a significantly higher incidence of MRI findings as follows (p < 0.05): a well-defined paraspinal abnormal signal (95% [19/20] in tuberculous vs 25% [5/20] in pyogenic), a thin and smooth abscess wall (95% [19/20] vs 15% [3/20]), combination of both findings (90% [18/20] vs 0% [0/20]), presence of paraspinal or intraosseous abscess (95% [19/20] vs 50% [10/20]), subligamentous spread to three or more vertebral levels (85% [17/20] vs 40% [8/20]), involvement of multiple vertebral bodies (60% [12/20] vs 25% [5/20]), thoracic spine involvement (40% [8/20] vs 10% [2/20]), and hyperintense signal on T2-weighted images (95% [19/20] vs 65% [13/20]). CONCLUSION. MRI was accurate for differentiation of tuberculous spondylitis from pyogenic spondylitis.