AI-based HRCT quantification reveals DLCO and TLC as key determinants of ILD severity in connective tissue diseases

Objective Interstitial lung disease (ILD) represents the most common and severe organ manifestation observed in patients diagnosed with connective tissue diseases (CTDs). The aim of this retrospective cross-sectional study was to identify clinical risk factors such as pulmonary symptoms, age, gender, laboratory and pulmonary function test (PFT) parameters associated with the extent of ILD as measured by artificial intelligence-based quantification of pulmonary high-resolution computed tomography (AIqpHRCT). Methods We included patients with a CTD-ILD diagnosis; all underwent PFT and HRCT, and pulmonary symptoms and signs of inflammation were also documented. AIpqHRCT was used to quantify lung volumetry and ILD features including ground glass opacities (GGO), reticulations, high-attenuation lung volume (HAV), emphysema and overall extent of ILD. Finally, 76 CTD-ILD patients were eligible for regression analysis, in order to evaluate the influence of clinical parameters on ILD extent. Results The reduction of diffusing capacity of the lung for carbon monoxide (DLCO), total lung capacity (TLC) and elevated inflammation parameter was significantly associated with the extent of GGO, reticulations, HAV and overall extent of ILD. Pulmonary symptoms, age and forced vital capacity were not associated with the extent of ILD quantified by AIqpHRCT. Conclusion The study presented that DLCO and TLC were predictive for the CTD-ILD severity. Consequently, our findings suggest the performance of PFT, including DLCO for all patients with CTD. In the case of reduced DLCO and TLC, further diagnostics, including HRCT, are necessary.