Abstract
Introduction
MRI plays an important role in the diagnosis of spinal tuberculosis with a high specificity and sensitivity. The role of MRI in predicting the healed status of caries spine to establish an end point of treatment is still not clear. In this study, we have proposed few MRI parameters of healed spinal tuberculosis and have attempted to formulate a scoring system to document healing. To find out the reliability of the proposed MRI parameters, we have statistically tested the scores given by three independent observers and assessed the interobserver reliability of the parameters.
Material and Methods
It is a prospective observational study. A total of 20 patients of spinal tuberculosis diagnosed clinico-radiologically and histopathologically after 1 year of ATD were enrolled. Patients younger than 20 years, with multifocal lesions, erratic tubercular treatment, and relapse and immunocompromised states were excluded. The pre and 1-year posttreatment MRIs were collected and a score sheet was used to assess healing. A total of 14 parameters (cord signal intensity, cord compression, epidural space involvement, vertebral body collapse, marrow signal change, posterior element, bony/ligamentous, paravertebral abscess, granulation, disc involvement, endplate changes, kyphosis, dislocation/subluxation, and coronal translation) and 6 MRI sequences (T1 axial and sagittal, T2 axial, sagittal, and coronal, and STIR sagittal) were used. Scoring was done by three independent observers (one radiologist, one neurosurgeon, and one orthopedic spine surgeon) who did not get any clinical detail of the patient. One-tailed t-test was used for statistical significance. A scoring system was proposed with the parameters which maximally predicted the healing status. Interobserver reliability was analyzed using intraclass correlation.
Results
Out of the 14 parameters considered, 4 showed the maximum pre- and posttreatment score difference (CI = 0.01) with excellent interobserver reliability (pre- and posttreatment intraclass correlation value of 0.94 and 0.78). Kyphosis, dislocation/subluxation, and coronal shift were not significant. These parameters were paravertebral abscess (p = 1.43 × 10␂10), cord compression (p = 2.53 × 10␂9), granulation tissue (p = 2.84 × 10␂9), and epidural space (p = 0.0021) in descending order of significance. Each parameter was tested separately and best six sequences which had maximum mean difference in scores were chosen. These were paravertebral abscess (STIR sagittal), cord compression (T2 axial), paravertebral granulation (T1 sagittal and coronal T2), and epidural space involvement (T1 and T2 axial). Each option had a minimum score of 0 and a maximum score of 2 so that the total score ranges from 0 to 12. The mean posttreatment observer score was 6.3. Hence, we propose a scoring system in which healing could be labeled with maximum predictability if the score is below 2.
Conclusion
Paravertebral abscess (STIR sagittal), cord compression (T2 axial), paravertebral granulation (T1 sagittal/coronal T2), and epidural space involvement (T1 axial/T2 axial) are the best healing parameters with excellent interobserver reliability. A proposed score of < 6 at 1-year follow-up is a fair predictor of healing. This being a pilot study needs to be tested in a larger cohort to validate the above data.