Quantitative CT Scan Analysis in Rheumatoid Arthritis-Related Interstitial Lung Disease.

Abstract

Quantitative CT imaging may be a useful predictor of outcome in rheumatoid arthritis-related interstitial lung disease (RA-ILD). What is the utility of deep learning-based lung fibrosis quantitation on CT imaging in assessing disease severity, predicting mortality, and identifying progression in RA-ILD? CT scans on a primary cohort of 289 patients and a validation cohort of 50 individuals with RA-ILD were assessed quantitatively by using the data-driven texture analysis (DTA) method. We examined associations between quantitative scores for extent of lung fibrosis and pulmonary function and survival. DTA fibrosis score at baseline showed moderate negative correlation with FVC percent predicted (primary cohort rho = -0.55; validation cohort rho = -0.50; both, P < .001), and diffusing capacity for carbon monoxide percent predicted (primary cohort rho = -0.67; validation cohort rho = -0.65; both, P < .001). Longitudinal change in DTA fibrosis score was associated with changes in FVC and diffusing capacity for carbon monoxide in the primary cohort (rho = -0.46 and rho = -0.43, respectively; both, P < .001). Cox multivariable models adjusted for potentially influential variables showed that the baseline DTA fibrosis score was significantly associated with mortality risk (primary cohort hazard ratio [HR], 1.04 [95% CI, 1.03-1.05; P < .001]; validation cohort HR, 1.06 [95% CI, 1.01-1.11; P = .026]). In the primary cohort, the increase in DTA fibrosis score on sequential scans was associated with increased risk of mortality (HR, 1.04; 95% CI, 1.01-1.06; P = .003) independent of baseline DTA extent. In 2 cohorts of patients with RA-ILD, quantitative assessment of lung fibrosis on CT imaging was associated with worse lung function at baseline and risk of mortality. Increase in DTA-derived lung fibrosis score on sequential scans was associated with subsequent risk of mortality. Quantitative CT imaging should be considered for use as a clinical and research outcome assessment tool in RA-ILD.