Abstract：
BACKGROUND: The diagnosis of fibrotic hypersensitivity pneumonitis (fHP) from other interstitial lung diseases, particularly idiopathic pulmonary fibrosis (IPF), is often difficult. This study aimed to examine computed tomography (CT) findings that were useful for differentiating between fHP and IPF and to develop and validate a radiological diagnostic model.
METHODS: In this study, 246 patients (fHP, n = 104; IPF, n = 142) from two institutions were included and randomly divided into the test (n = 164) and validation (n = 82) groups (at a 2:1 ratio). Three radiologists evaluated CT findings, such as pulmonary fibrosis, small airway disease, and predominant distribution, and compared them between fHP and IPF using binomial logistic regression and multivariate analysis. A prognostic model was developed from the test group and validated with the validation group.
RESULTS: Ground-glass opacity (GGO) with traction bronchiectasis (TB), honeycombing, hypoattenuation area, three-density pattern, diffuse craniocaudal distribution, peribronchovascular opacities in the upper lung, and random distribution were more common in fHP than in IPF. In multivariate analysis, GGO with TB, peribronchovascular opacities in the upper lung, and random distribution were significant features. The area under the curve of the fHP diagnostic model with the three aforementioned CT features was 0.733 (95% confidence interval [CI], 0.655-0.811, p < 0.001) in the test group and 0.630 (95% CI, 0.504-0.755, p < 0.047) in the validation group.
CONCLUSIONS: GGO with TB, peribronchovascular opacities in the upper lung, and random distribution were important CT features for differentiating fHP from IPF.
METHODS: In this study, 246 patients (fHP, n = 104; IPF, n = 142) from two institutions were included and randomly divided into the test (n = 164) and validation (n = 82) groups (at a 2:1 ratio). Three radiologists evaluated CT findings, such as pulmonary fibrosis, small airway disease, and predominant distribution, and compared them between fHP and IPF using binomial logistic regression and multivariate analysis. A prognostic model was developed from the test group and validated with the validation group.
RESULTS: Ground-glass opacity (GGO) with traction bronchiectasis (TB), honeycombing, hypoattenuation area, three-density pattern, diffuse craniocaudal distribution, peribronchovascular opacities in the upper lung, and random distribution were more common in fHP than in IPF. In multivariate analysis, GGO with TB, peribronchovascular opacities in the upper lung, and random distribution were significant features. The area under the curve of the fHP diagnostic model with the three aforementioned CT features was 0.733 (95% confidence interval [CI], 0.655-0.811, p < 0.001) in the test group and 0.630 (95% CI, 0.504-0.755, p < 0.047) in the validation group.
CONCLUSIONS: GGO with TB, peribronchovascular opacities in the upper lung, and random distribution were important CT features for differentiating fHP from IPF.
摘要：
背景：其他间质性肺病对纤维化过敏性肺炎（fHP）的诊断，特别是特发性肺纤维化（IPF），往往很难。这项研究旨在检查计算机断层扫描（CT）的发现，这些发现有助于区分fHP和IPF，并开发和验证放射诊断模型。
方法：在本研究中，246名患者（fHP，n=104；IPF，包括来自两个机构的n=142），并随机分为测试（n=164）和验证（n=82）组（以2：1的比例）。三名放射科医生评估了CT发现，比如肺纤维化，小气道疾病，占主导地位的分布，并使用二项逻辑回归和多变量分析在fHP和IPF之间进行比较。从测试组开发预后模型，并与验证组进行验证。
结果：带有牵引性支气管扩张（TB）的毛玻璃不透明（GGO），蜂窝,低衰减区域，三密度模式,弥漫性头尾分布，上肺支气管血管周围混浊，和随机分布在fHP中比在IPF中更常见。在多变量分析中，带有TB的GGO,上肺支气管血管周围混浊，随机分布特征显著。具有上述三种CT特征的fHP诊断模型的曲线下面积为0.733（95％置信区间[CI]，0.655-0.811，p<0.001）在测试组中，0.630（95％CI，0.504-0.755，p<0.047）在验证组中。
结论：GGO与TB，上肺支气管血管周围混浊，随机分布是区分fHP和IPF的重要CT特征。
方法：在本研究中，246名患者（fHP，n=104；IPF，包括来自两个机构的n=142），并随机分为测试（n=164）和验证（n=82）组（以2：1的比例）。三名放射科医生评估了CT发现，比如肺纤维化，小气道疾病，占主导地位的分布，并使用二项逻辑回归和多变量分析在fHP和IPF之间进行比较。从测试组开发预后模型，并与验证组进行验证。
结果：带有牵引性支气管扩张（TB）的毛玻璃不透明（GGO），蜂窝,低衰减区域，三密度模式,弥漫性头尾分布，上肺支气管血管周围混浊，和随机分布在fHP中比在IPF中更常见。在多变量分析中，带有TB的GGO,上肺支气管血管周围混浊，随机分布特征显著。具有上述三种CT特征的fHP诊断模型的曲线下面积为0.733（95％置信区间[CI]，0.655-0.811，p<0.001）在测试组中，0.630（95％CI，0.504-0.755，p<0.047）在验证组中。
结论：GGO与TB，上肺支气管血管周围混浊，随机分布是区分fHP和IPF的重要CT特征。
