Diagnostic accuracy of whole spine magnetic resonance imaging in spinal tuberculosis validated through tissue studies

Abstract

Introduction: Conventional diagnosis of spinal tuberculosis (TB) is based on a combination of clinical features, laboratory tests and imaging studies, since none of these individual diagnostic features are confirmatory. Despite the high sensitivity of MRI findings in evaluating spinal infections, its efficacy in diagnosing spinal TB is less emphasized and remains unvalidated through tissue studies. Methodology: We reviewed consecutive patients evaluated for spondylodiscitis with documented clinical findings, MRI spine, and tissue analysis for histopathology, TB culture and genetic TB PCR. MRI features documented include location, contiguous/non-contiguous skip lesions, para/intraosseous abscess, subligamentous spread, vertebral collapse, abscess size/wall, disc involvement, end plate erosion and epidural abscess. Based on the results, patients were divided into two groups—CONFIRMED TB with positive culture/histopathology and NON-TB. The efficacy of MRI findings in accurately diagnosing spinal TB was compared between the two groups. Results: Among 150 patients, 79 patients were TB positive, and 71 were TB negative. Three MRI parameters showed significant differences (p < 0.001), namely subligamentous spread (67/79, 84.8%), vertebral collapse > 50% (55/79, 69.6%) and large abscess collection with thin abscess wall (72/79, 91.1%) being strongly predictive of TB. Combination of MRI findings had a higher predictive value. 97.5% of TB positive patients had at least one of these three MRI features, 89.8% patients had any two and 58.2% had all three. Conclusion: Our study validated different MRI findings with tissue studies and showed spinal infections with large abscess with thin wall, subligamentous spread of abscess and vertebral collapse were highly suggestive of spinal tuberculosis. Graphical abstract: These slides can be retrieved under Electronic Supplementary Material.[Figure not available: see fulltext.].