A75-28 Quantitative CT Analysis Demonstrates Progressive Structural Remodeling in Pulmonary Sarcoidosis: A Longitudinal Case Series

Rationale

Pulmonary sarcoidosis exhibits heterogeneous radiographic manifestations and an unpredictable course, with some patients progressing to fibrotic remodeling despite therapy [1]. Artificial intelligence-based quantitative CT (qCT) enables automated, reproducible assessment of lung texture patterns and may improve detection of longitudinal disease progression and phenotypic variation [2]. We applied AI-driven whole-lung texture analysis to serial CT scans to characterize temporal changes in parenchymal composition, regional distribution, and lung volume in patients with advanced pulmonary sarcoidosis.

Methods

Five patients with biopsy-proven pulmonary sarcoidosis from a tertiary urban lung transplant center who had ≥3 CT chest scans within 5 years prior to transplantation or death were retrospectively analyzed. Lung texture quantification was performed using AView software to measure volumes of normal lung, reticulation, ground-glass opacity, honeycombing, consolidation, and emphysema across lobar segments.

Results

The cohort included five patients (4 female, 1 male; 100% Black and non-Hispanic) with a mean age of 33.3 years at diagnosis (range 22-46). All patients demonstrated progressive loss of normal parenchyma despite treatment (77.8% to 45.2%; −32.6 percentage points [pp]), with the largest individual patient declines of − 57 and −52 pp. Concomitant increases were observed in reticulation (+11.8 pp), emphysema (+13.2 pp), honeycombing (+4.0 pp), consolidation (+1.4 pp), and ground-glass opacity (+2.2 pp). Emphysema progression was variable, with maximal increases up to + 30 pp. Two distinct progression phenotypes emerged: an emphysema-predominant pattern (3/5) and a reticular/ground-glass-predominant pattern (2/5) with minimal emphysema. At the final scan, disease burden showed upper-lobe predominance, most pronounced in the left upper lobe (23.4% average at final scan). Mean whole-lung volume decreased modestly (3352 to 3220 cc; −3.9%), with 3/5 patients exhibiting volume loss while others showed volume gain, likely reflecting competing effects of fibrosis and air trapping. Reduction in normal parenchyma strongly correlated with lung volume decline (r = 0.87).

Conclusions

AI-based qCT demonstrates consistent longitudinal parenchymal loss and identifies distinct radiographic phenotypes in advanced pulmonary sarcoidosis. Quantitative phenotyping may support earlier detection of progressive fibrosis and risk stratification. Larger studies are needed to explore these imaging biomarkers further. References: Desai SR, Nishanth Sivarasan, Johannson KA, George PM, Culver DA, Anand Devaraj, et al. High-resolution CT phenotypes in pulmonary sarcoidosis: a multinational Delphi consensus study. The Lancet Respiratory Medicine. 2023 Dec 15;12(5):409-18. Ledda RE, Roberti C, Sverzellati N. From X-rays to advanced imaging modalities in pulmonary sarcoidosis. Current Opinion in Immunology [Internet]. 2025 Dec 16;98(Special Issue: SARCOIDOSIS 2025):102709. Available from: https://www.sciencedirect.com/science/article/pii/S0952791525001852?casa_token=EMokWMKX87MAAAAA:0KpsJkweSNY0QrkROcRxGwbCM14i3Iz7G_2jBfIYnXmIG0X7_1JqYed2ZzuMTXfcMwS9uGflpsY#bib9

This abstract is funded by: None