Objective: To identify the combination of clinical data and high resolution computed tomography (HRCT) features that best identified biopsy verified usual interstitial pneumonia (UIP).
Methods: The study included 91 patients with a tentative diagnosis of interstitial lung disease. All underwent clinical investigation, surgical lung biopsy and HRCT. Two independent readers assessed the HRCT images for the extent and pattern of abnormality. On the basis of the biopsy result the patients were categorized in three groups: 1) Usual interstitial pneumonia, 2) Other idiopathic interstitial pneumonias (IIPs) and hypersensitivity pneumonitis and 3) Other interstitial lung diseases. The diagnostic value of HRCT was investigated using likelihood ratio to estimate the post-test probability of UIP.
Results: We found that UIP was associated with significantly higher scores for reticular pattern and for bronchiectasis than the remaining patients (p < 0.001). Moreover, these scores showed a steeper cranial-caudal increase in patients with histologically verified UIP than in the remaining patients (p < 0.001). UIP was associated with lower scores for ground glass opacities (p < 0.001). Using Bayes theorem and likelihood ratio estimation we found that UIP could be diagnosed with 90% certainty in patients 60 years or older and restrictive pattern in spirometry provided that HRCT demonstrated at least 15% reticular pattern and no ground glass opacities.
Conclusion: In older patients with a restrictive spirometry in whom HRCT demonstrates a reticular pattern without ground glass opacities surgical lung biopsy is not warranted for the diagnosis of UIP.
Keywords: Bayes theorem; High resolution computed tomography; Usual interstitial pneumonia.
Copyright © 2015 Elsevier Ltd. All rights reserved.