Abstract
BACKGROUND: Bronchial artery embolization (BAE) is the standard intervention for severe hemoptysis, yet predicting which patients will require BAE remains challenging. We evaluated whether multidetector computed tomography angiography (MDCTA) findings can predict the need for BAE. METHODS: We retrospectively reviewed 32 consecutive patients hospitalized for hemoptysis >100 mL/day between April 2012 and March 2015. All patients underwent MDCTA on admission and received standardized conservative management. MDCTA findings were dichotomized as (1) normal or mediastinally confined bronchial arteries and (2) abnormally dilated bronchial arteries extending into the pulmonary parenchyma. Interobserver agreement among three readers was assessed (Fleiss κ). The primary outcome was subsequent BAE. RESULTS: Twenty patients had normal/mediastinally confined arteries and 12 had abnormally dilated parenchymal-extending arteries. None of the former required BAE, whereas 7/12 (58 %) in the latter group underwent BAE (p < 0.001). The imaging criterion yielded sensitivity 100 % (95 % CI 59-100 %) and specificity 80 % (95 % CI 59-93 %). Interobserver agreement for the 3-category MDCTA classification was almost perfect (κ = 0.90). Penalized logistic regression adjusting for age, sex, and infectious etiology retained MDCTA parenchymal extension as the sole predictor (adjusted OR ∼18 (p ≈ 0.002)). CONCLUSIONS: MDCTA identification of an abnormally dilated bronchial artery extending into the pulmonary parenchyma predicts BAE requirement in significant hemoptysis. This readily recognizable imaging criterion may expedite interventional consultation and optimize triage.