High-resolution CT predictors of hypersensitivity pneumonitis

Main Article Content

Gilles Rival
Philippe Manzoni
Yves Lacasse
Jean Charles Polio
Virginie Westeel
André Dubiez
Thibaud Soumagne
François Laurent
Jean Charles Dalphin


HRCT, hypersensitivity pneumonitis, diagnostic procedures


Background: The purpose of this study was to evaluate the use of high-resolution chest computed tomography (HRCT) to distinguish hypersensitivity pneumonitis (HP) from other diffuse parenchymal lung diseases (DPLDs). Methods: We examined 130 consecutive patients admitted to our hospital with DPLDs proved by HRCT. Patients underwent clinical and paraclinical examinations. Two readers interpreted 111 HRCT scans using predefined criteria. Results: The findings in patients with HP were compared to those with other DPLDs (non-HP) by univariate and multivariate analyses. Five independent radiological predictors were identified and were given a weight according to their regression coefficient: ground-glass attenuation nodules (4 points), homogeneous ground-glass opacity (3 points), patchy ground-glass opacity (2 points), absence of adenopathy (2 points), and absence of linear/reticular patterns (2 points). A total score (that we called “diagnostic index”) of 5 offered the best trade-off between sensitivity and specificity. At this point of the ROC curve, the sensitivity, specificity, and likelihood ratio were 74%, 90% and 7.7, respectively. Given a pre-test probability of HP of 34% (i.e., 38 HP / 111 patients), the post-test probability was 79%. Conclusion: Our results provide evidence that HRCT can accurately distinguish HP from other DPLDs.

Abstract 261 | PDF Downloads 169