Comparison of chest radiographs with minimum intensity projection reconstruction CT scans for the presence of airway stenosis resulting from lymphobronchial TB in children

Abstract

INTRODUCTION: The diagnosis of tuberculosis (TB), and in particular lymphobronchial tuberculosis (LBTB), in children continues to rely on imaging. Modern computed tomography (CT) imaging with coronal minimum intensity projection (MinIP) reconstruction is ideally suited for diagnosing airway compression in LBTB. Resource constraints in developing countries, however, oblige medical practitioners to use plain radiography in the evaluation of children with LBTB. Identification of airway compression on chest X-ray (CXR) is important, as it may be the only diagnostic feature of TB for instituting treatment. AIM: The aim of this study was to demonstrate the sensitivity and specificity of frontal CXR in identifying airway stenosis, at ten specific sites, against the gold standard, CT MinIP and to determine the agreement between the modalities regarding the degree of stenosis. METHOD: This was a retrospective, quantitative cross sectional study comparing CXR and CT MinIP in 37 children with confirmed LBTB at 10 predetermined airway locations. The CXRs were evaluated by one expert reader a paediatric radiologist, the CT MinIP images were evaluated by three general radiologists. RESULTS: Patients age ranged from 2 months 22 days to 147 months 25 days. The median age of the patients was 14.3 months (interquartile range 8.0-23.2 months; range 2.7-147.8months). The original database of patients included a total number of 212 patients, of which only 41 had a CXR available. Of those 41 patients only 37 had a CT MinIP and 4 did not, thus the final study population was 37 in total. Chest X-ray demonstrated that 31 out of the 37 patients in the study group thus 84%, had stenosis at least at one of the ten evaluated sites, confirmed on the CT MinIP images. Computed tomography MinIP, which is taken as the gold standard, showed that all 37 patients, thus 100%, had stenosis at one of the ten evaluated sites. The sensitivity for CXR picking up stenosis at the left main bronchus (LMB) was 92.9%, with a 100% specificity. The sensitivity for stenosis at the bronchus intermedius (BI) was 80% with a specificity of 75%. The kappa agreement between CXR and Minip with regard to the grade of stenosis for each site shows substantial agreement (Kappa0.67) for the LMB and moderate agreement (kappa 0.58) for the BI. CONCLUSIONS: In a paediatric population where the incidence and the severity of airway stenosis from LBTB is high, CXR was shown to be accurate in detecting airway stenosis and in demonstrating severe stenosis at the LMB and BI