The Role of High-Resolution Lung Computed Tomography to Distinguish Between Fibrosing Hypersensitivity Pneumonitis and Usual Interstitial Pneumonia

Background: Hypersensitivity pneumonitis (HP) is an interstitial lung disease (ILD) caused by repeated exposure to inhaled antigens in susceptible subjects. High-resolution computed tomography (HRCT) of the lungs is the leading diagnostic method for ILDs, but in some cases HRCT findings are not sufficient to distinguish HP and other ILDs, particularly, fibrotic HP (fHP) and usual interstitial pneumonia (UIP). Objective: The aim of this study was to develop HRCT criteria to diagnose fHP in patients with a UIP-like pattern. Methods: In this retrospective study, we analyzed HRCT scans of patients with fHP and a UIP-like pattern who underwent lung biopsy, and patients with idiopathic pulmonary fibrosis (IPF) and a UIP pattern in HRCT. Results: We included 51 patients with confirmed fHP and 24 patients with IPF/UIP in the analysis. IPF/UIP patients were older, were prevalently males, and did not have any systemic autoimmune diseases or risk factors for other ILDs. fHP patients were younger, with an equal number of males and females, and were more likely to be exposed to environmental antigens. HRCT abnormalities in the fHP group predominated in the lower lung areas or were diffuse in axial scans, whereas IPF/UIP patients mostly demonstrated a diffuse craniocaudal distribution and subpleural axial predominance. Centrilobular nodules and mosaic attenuation were present significantly more often in the fHP group; honeycombing, traction bronchiectasis, and emphysema prevailed in IPF/UIP patients. In the logistic regression analysis, patients with fHP and IPF/UIP differed in the presence of centrilobular nodules, honeycombing, and in both craniocaudal and axial distributions of HRCT abnormalities. In the ROC analysis, the combination of centrilobular nodules, honeycombing, and diffuse axial and craniocaudal distributions can predict the diagnosis of fHP (AUC, 0.953 ± 0.022; 95%CI, 0.910–0.995; p < 0.001). Mosaic attenuation and reticulation did not change the probability of fHP. Conclusions: The most significant HRCT features of fHP compared to the UIP pattern were centrilobular nodules, honeycombing, and a diffuse axial and craniocaudal distribution of abnormal findings. Reticulation, mosaic attenuation, and GGO do not increase the probability of fHP.