Anna Rozenshtein MD, Abstract Co-Author: Nothing to Disclose

Gregory D. N. Pearson MD, PhD, Presenter: Nothing to Disclose

Belinda D'Souza MD, Abstract Co-Author: Nothing to Disclose

Beth H. Leopold BA, Abstract Co-Author: Nothing to Disclose

David Lederer MD, MS, Abstract Co-Author: Steering Committee, Gilead Sciences, Inc Advisory Board, Gilead Sciences, Inc Steering Committee, InterMune, Inc Advisory Board, InterMune, Inc

We retrospectively evaluated the American Thoracic Society (ATS) 2011 criteria for Computed Tomography (CT) classification of Usual Interstitial Pneumonia (UIP) in a cohort of patients presenting to a tertiary referral center for evaluation of ILD. 

The records of patients presenting to ILD clinic between 2010 and 2012 were reviewed. 3 fellowship trained chest radiologists and a pulmonologist specializing in ILD independently reviewed CT scans in 187 patients, 86 of whom had pathology results available. Based on the ATS 2011 criteria (Raghu et. al (2011) Am J Respir Crit Care Med 183:788-824), patients were classified as Definite UIP, Possible UIP, or Inconsistent with UIP. Consensus was defined as agreement by 3 or more readers. Interobserver agreement was perfomed with Kappa values.

Consensus CT diagnosis was reached independently in 163 of 187 (87%) of cases, and after discussion in an additional 24 (13%) of cases. In 3 cases (2%) no consensus was reached. Interobserver agreement was moderate, with kappa value 0.49. Of the 86 patients with available pathology, 15 (17%) had UIP, 5 (6%) had mixed UIP and Nonspecific Interstitial Pneumonia (NSIP), 31 (36%) had NSIP, 12 (14%) had Hypersensitivity Pneumonia (HP), and 24 (12%) had other diagnoses (totals>100% due to multiple pathologic diagnoses in some biopsies). Sensitivity was low, with only 8 of 20 (40%) of patients with either pathologic UIP or combined UIP/NSIP having a consensus radiologic diagnosis of Definite UIP. Of the 15 patients with UIP on biopsy, 5 were classified on CT as Definite UIP, 4 Possible UIP, 5 Inconsistent, and 1 no consensus. Of the 5 patients with mixed UIP and NSIP on biopsy, 3 were classified on CT as Definite UIP, 1 Inconsistent, and 1 no consensus. Specificity, however, was high (98%), with only 1 of 66 patients without UIP on biopsy classified as Definite UIP on CT.

ATS 2011 criteria for UIP have poor sensitivity but excellent specificity for the diagnosis of UIP in our retrospective cohort. As such, if used to triage patients with ILD for biopsy 40% of patients with pathologic UIP would have avoided biopsy, while 2% of patients without UIP would have been triaged to no biopsy.

In our cohort, ATS 2011 criteria were excellent for triaging patients without UIP to biopsy, but would not have prevented a biopsy recommendation in the majority of patients with pathologic UIP.

Rozenshtein, A, Pearson, G, D'Souza, B, Leopold, B, Lederer, D, Can Chest CT Prevent Unnecessary Biopsy in Patients with Interstitial Lung Disease (ILD)? The Effect of the ATS 2011 ILD Reporting Guidelines on Patient Management .  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14005391.html