- This is the first study to assess the computerized tomography (CT) test characteristics of the 2018 American Thoracic Society and Fleischner guidelines for diagnosing idiopathic pulmonary fibrosis; histopathology was used as the comparator
- Two CT classification categories, usual interstitial pneumonitis (UIP) and probable UIP, had high specificity for histopathologic UIP (97% and 88%, respectively)
- CT UIP had high positive predictive value for histopathologic UIP, 87%
- The positive predictive value of CT probable UIP was only 46% for combined histopathologic UIP/probable UIP
- Modification or refinement of guideline criteria, specifically feature qualifiers such as "prominent" and "marked," may be helpful
In 2018, the American Thoracic Society, European Respiratory Society, Japanese Respiratory Society and Latin American Thoracic Society (ATS/ERS/JRS/ALAT) and the Fleischner Society separately updated their guidelines for diagnosis of idiopathic pulmonary fibrosis. This form of interstitial lung disease is characterized by a usual interstitial pneumonitis (UIP) pattern on chest computerized tomography (CT). Both guidelines expanded the number of diagnostic categories from three to four: UIP, probable UIP, indeterminate for UIP and alternate diagnosis. The documents give descriptions and criteria for the four categories for both CT scans and histopathology results.
Subscribe to the latest updates from Pulmonary & Critical Care Advances in Motion
Angela R. Shih, MD, pathologist, and Lida P. Hariri, MD, PhD, pulmonary pathologist, both in the Department of Pathology at Massachusetts General Hospital, and colleagues recently assessed the test characteristics of the four CT categories using histopathology as the comparator. In Respiratory Research, they report high specificity and positive predictive value for UIP by CT.
The team retrospectively studied 101 patients who had surgical lung biopsy (85 surgical wedge biopsies and 16 pneumonectomies) to investigate suspected interstitial lung disease at Mass General between 2000 and 2018. Three thoracic radiologists, blinded to clinical and pathological data, re-interpreted the CT scans, and two pathologists, blinded to clinical and radiographic data, re-interpreted the biopsy data.
Working separately, the radiologists and pathologists assigned each patient to one of the four categories specified in the guidelines.
- CT classifications were 15% UIP, 13% probable UIP, 10% indeterminate for UIP and 62% alternative diagnosis
- Histopathology classifications were 34% UIP, 7% probable UIP, 25% indeterminate for UIP and 34% alternative diagnosis, with the most common alternative diagnosis being chronic hypersensitivity pneumonitis
- Patients with CT UIP had histopathologic UIP in 87% of cases (the positive predictive value) with 97% specificity
- Patients with CT probable UIP had histopathologic UIP in 38% of cases; the positive predictive value improved to only 46% for combined histopathologic UIP/probable UIP
- Patients with CT indeterminate for UIP had histopathologic UIP in 27% of cases with a specificity of 90%
- Patients with CT alternative diagnosis had histopathologic UIP in 21% of cases with a specificity of 25%
A Major Limitation
The practicality of the guidelines continues to be limited by user variation in feature recognition. In particular, it is challenging to consistently assess qualifiers such as "prominent," "marked," "extensive" or "mild" and determine how they should influence the categorization of a case.
This problem became apparent because of the interobserver variability reported in studies of earlier guidelines. The variability persisted in the current study: two radiologist readers had very good agreement (kappa, 0.81), but the other two pairs had only fair agreement (kappas, 0.32 and 0.33), more consistent with previous reports.
The creation of a semi-quantitative grading system is worth consideration as a way to improve the uniform application of the guidelines.
Learn more about the Department of Pathology
Learn more about the Division of Pulmonary and Critical Care Medicine