To evaluate the computed tomographic (CT) predictors of a clinically significant yield from microbiological tests in patients with a tree-in-bud pattern.
Materials and Methods:
CT examinations in 53 patients (male=34; mean age=52.9±17.3 y) with a tree-in-bud pattern in whom a diagnostic test (sputum analysis, bronchoalveolar lavage or nasopharyngeal aspirates) had been performed within 2 weeks were identified. The following CT patterns were independently quantified by 2 thoracic radiologists: tree-in-bud, bronchiectasis, bronchial wall thickening, consolidation, ground-glass opacification, and nodules. The presence of cavitation (in nodules and/or consolidation) was recorded. Patient charts were reviewed for the presence of a clinically significant positive microbiological result.
A clinically significant causal organism was present in 25/53 (47%) patients. The median extent of a tree-in-bud pattern was 5 [range=1 to 16 (maximum range=0 to 18)], and cavitation was present in 14/53 (26%) patients (cavitating nodules=8, cavitation in consolidation=3, and cavitation in consolidation and nodules=3). There was no independent linkage between the extent of a tree-in-bud pattern and the identification of a clinically significant organism. The microbiological yield was significantly higher if there was coexistent cavitation in nodules or consolidation [11/14 (79%) vs. 14/39 (39%); P=0.005]. On stepwise logistic regression, the only CT predictor of a clinically significant microbiological yield was cavitation on CT (odds ratio=9.7; 95% confidence interval=1.9, 49.9; P<0.01); the extent of a tree-in-bud pattern, concurrent use of antibiotics, age, and sex were not independently linked to a significant microbiological yield.
A specific clinically significant microbiological diagnosis was obtained in approximately 50% of patients with a tree-in-bud pattern. The microbiological yield rises strikingly when a tree-in-bud pattern coexists with cavitation (in nodules or consolidation) but is not predicted by ancillary CT signs or clinical parameters.