P223 To what extent does demographic, clinical and magnetic resonance imaging (MRI) data provide useful information on the possible treatment duration of spinal tuberculosis (TB)?

Introduction Guidelines for treatment duration for spinal tuberculosis recommend a six-month course of chemotherapy (isoniazid, rifampicin, pyrazinamide and ethambutol [2HRZE+4 hour]), based on early MRC trials in patients with one to two affected vertebrae using conventional spinal X-rays. Many clinicians, however, frequently treat for 9 to 12 months. MRI scans are now considered the best radiological investigation for spinal TB due to their sensitivity, specificity, and ability to precisely delineate soft tissue involvement. Aim This study examined the possible value of MRI, coupled with relevant demographic and clinical data, in facilitating decision-making regarding spinal TB treatment duration. Method Demographic, clinical and imaging data were collected retrospectively from 93 spinal TB patients from two London TB clinics; data from 73 were eligible for further analysis. A linear regression model was developed and applied, and statistically relevant variables identified. We compared treatment duration predictions from the model to actual treatment duration to assess the model’s performance. Results From spinal biopsies alone, 73% were culture positive for M.tuberculosis. Demographic, clinical and imaging results are shown in table 1. All patients received standard quadruple therapy except for three with isoniazid resistance and one with multidrug resistance. Including these, mean treatment duration was 11(range 5–24) months. In general, a clinical decision was made to extend treatment beyond 6 months if symptoms and radiological signs were slow to resolve or there was significant drug resistance. Age (p=0.007, 95% confidence interval (CI): 0.016 to 0.101), cord compression (p=0.019, CI: 0.248 to 2.689), number of vertebrae affected (p=0.010, CI: 0.078 to 0.558) and microbial sensitivity (p=0.052, CI: −11.164 to 0.050), were jointly significant determinants of treatment duration, at the 10% significance level. Surprisingly, delay in time to diagnosis was not (p=0.886, CI: −0.059 to 0.069). Conclusion Young patients and those with fewer affected vertebrae and no cord compression on MRI scan were more likely to recover faster than older patients or those with more extensive disease. Age and MRI scan features in particular, could be used to assist clinical decision-making and help inform patients’ expectations of likely treatment duration at the start of therapy.