A15-03 Quantitative CT Parameters Predict ILA Progression

Abstract Rationale Interstitial lung abnormalities (ILA) are incidentally detected findings on chest CT postulated as precursors to interstitial lung disease (ILD). Predicting which individuals with ILA will progress in the near-term remains challenging. We evaluated a suite of quantitative CT (qCT) airway, vascular, and parenchymal metrics at two centers as prognostic biomarkers for the development of ILD. Methods Deidentified CT datasets from Vanderbilt University Medical Center (VUMC) and Weill Cornell Medical Center (WCMC) were analyzed by FLUIDDA (Kontich, Belgium) using an automated qCT pipeline. The VUMC cohort consisted of first-degree relatives of patients with familial pulmonary fibrosis who received serial research HRCTs as part of a longitudinal cohort study. The WCMC cohort included patients undergoing CT imaging for non-ILD indications who had ILA reported by a radiologist, along with clinically-indicated follow-up scans. Scans from both centers were evaluated for the presence and progression of ILA and ILD. Parameters evaluated included the volumes and morphology of pulmonary vessels and airways, and measurements of parenchymal texture and density. Cox proportional hazard models were fit to each qCT parameter separately to predict risk of ILD, adjusted for age, gender, slice thickness, and category of reconstruction algorithm. Hazard ratios (HR) are reported per 1 standard deviation of each parameter. Results 369 and 82 subjects with 8 and 7 ILD diagnoses were included from VUMC and WCMC, respectively. Mean ages were 53.66 (9.24) and 70.28 (9.21) years for VUMC and WCMC subjects; 62.1% and 47.6% female. Median follow-up was 38 months for VUMC and 18.46 months for WCMC. In the VUMC cohort, parameters which with a significant (p 0.05) or trending (p 0.1) association with ILD-free survival included airway tortuosity, the fractal dimension of the pulmonary vessels, volume of detected fibrosis, the volume of vessels 5-10 mm2 in cross-sectional area, airway wall area, total airway volume normalized to total lung volume, airway radius, and the percentage of lung voxels with attenuation -600 HU. For the WCMC cohort, the quantitative emphysema score was significantly associated with risk of ILD. Point estimates of HR for 8 of 11 features associated with ILD risk in the VUMC cohort were directionally concordant in the WCMC cohort. Figure 1 summarizes the per-parameter HRs and associated p-values. Conclusion Parameters related to vasculature, airway morphology, and parenchyma are associated with likelihood of developing ILD in at-risk individuals. Quantitative CT holds promise as a strategy for risk-stratifying subjects with ILA. This abstract is funded by: ThreeLakes Foundation, NHLBI