Computed tomography in the early detection of asbestosis.

Computed tomography (CT; both conventional (CCT) and high resolution (HRCT)) scans of the thorax were evaluated to detect early asbestosis in 61 subjects exposed to asbestos dust in Québec for an average of 22(3) years and in five controls. The study was limited to consecutive cases with chest radiographs of the International Labour Organisation categories 0 or 1 determined independently. All subjects had a standard high kilovoltage posteroanterior and lateral chest radiograph, a set of 10-15 1 cm collimation CCT scans and a set of three to five 2 mm collimation HRCT scans in the upper, middle, and lower lung fields. Five experienced readers independently read each chest radiograph and sets of CT scans. On the basis of three to five readers agreeing for small opacities of the lung parenchyma, 12/46 (26%) negative chest radiographs were positive on CT scans, but 6/18 (33%) positive chest radiographs were negative on CT scan. On the basis of four to five readers agreeing on a chest radiograph, 36/66 (54%) subjects were normal (group A), 17/66 (26%) were indeterminate (group B), and 13/66 (20%) were abnormal (group C). By the combined readings of CCT and HRCT, 4/31 (13%) asbestos exposed subjects of group A were abnormal (p < 0.001), 6/17 (35%) of group B were abnormal, and in group C, 1/13 (8%) was normal, 2/13 were indeterminate, and 10/13 (77%) were abnormal. Separate readings of CCT and HRCT on distinct films in 14 subjects showed that all cases of asbestosis were abnormal on both CCT and HRCT. Inter-reader analyses by kappa statistics showed significantly better agreement for the readings of CT than the chest radiographs (p < 0.001), and for the reading of CCT than HRCT (p < 0.01). Thus CT scans of the thorax identifies significantly more irregular opacities consistent with the diagnosis of asbestosis than the chest radiograph (20 cases on CT scans v 13 on chest radiographs when four to five readers agreed, 13% of asbestos exposed subjects with normal chest radiographs or 21% of asbestos exposed subjects with normal or near normal chest radiographs. It decreased the number of indeterminate cases significantly from 17 on chest radiographs to 13 on CT scans. All cases of asbestosis detected only on CT scans were similarly seen on CCT and HRCT and did not have significant changes in lung function. The CT scans significantly reduced the inter-reader variability, despite the absence of ILO type reference films for these scans.