HRCT Patterns Of Usual Interstitial Pneumonia In Rheumatoid Lung

RATIONALE. Interstitial lung disease (ILD) is a well-recognized complication of rheumatoid arthritis (RA) and can present with different patterns on high-resolution computed tomography (HRCT) of the chest. It has been recently shown that a definite HRCT usual interstitial pneumonia (UIP) pattern is highly specific and moderately sensitive for a histopathologic UIP pattern. The aims of the present study were: i) to evaluate the prevalence of the UIP pattern on HRCT in patients with RA-ILD, as compared with patients with idiopathic pulmonary fibrosis (IPF) and ii) to assess the level of agreement between two experienced chest radiologists in detecting the UIP pattern in the two groups of patients. METHODS. 30 patients with RA and at least one chest HRCT showing interstitial changes were retrospectively identified from a single-center cohort of RA patients. Fifty-two patients with IPF based on current diagnostic criteria served as diseased controls. For patients who had more than one HRCT the more recent HRCT was selected. Two experienced thoracic radiologists (radiologist A and B) blinded to patient diagnosis scored all HRCT images independently. Radiologic patterns were categorized as definite UIP, possible UIP or inconsistent with UIP according to the most current international guidelines. The prevalence of the different patterns was assessed for both groups and compared by using the chi square test. The concordance between radiologists was determined using the Cohen kappa score. RESULTS. Radiologist A detected 4 definite UIP (13%), 10 possible UIP (33%) and 16 inconsistent with UIP patterns (53%) among patients with RA-ILD and 16 definite UIP (30%), 24 possible UIP (46%) and 12 inconsistent with UIP patterns (23%) among patients with IPF (chi square 0.016). Radiologist B identified 6 definite UIP (20%), 9 possible UIP (30%) and 15 inconsistent with UIP patterns (50%) in the RA-ILD group and 23 definite UIP (44%), 16 possible UIP (30%) and 13 inconsistent with UIP patterns (25%) in the IPF group (chi square 0.036). Diagnostic agreement in UIP pattern detection was 71.2% (κ= 0.57) for the IPF group and 76.7% (κ =0.62) for the RA-ILD group. CONCLUSIONS. A definite UIP pattern can be identified in chest HRCT of a sizeable fraction of patients with RA-ILD. Moreover, the possible UIP pattern appears to be almost equally distributed in patients with RA-ILD and IPF. Of note, the level of agreement between two experienced chest radiologists in detecting the UIP pattern is higher in patients with RA-ILD than in IPF.