C73-11 Deformation-Based Imaging Metrics Associate With Clinical Recovery After Valve Treatment in Chronic Obstructive Pulmonary Disease: Comparison With Current Response Criteria

Abstract Rationale Endoscopic lung volume reduction (ELVR) using one-way valves in patients with chronic obstructive pulmonary disease (COPD) aims to collapse the most diseased lobe, thereby reducing hyperinflation and improving patient lung function. Response is traditionally defined by threshold changes in spirometry or target lobe volume reduction (TLVR), yet TLVR only represents anatomical changes. We hypothesized that deformation-based mechanical metrics derived from image registration could better explain clinical recovery, than TLVR alone. Methods We retrospectively analyzed paired inspiratory and expiratory CT scans acquired at baseline and 90-day follow-up (dfu) from COPD patients who underwent ELVR (N = 147). Each paired scan was processed using a non-rigid registration pipeline to generate voxel-wise displacement fields and corresponding Jacobian determinant maps, quantifying regional lung deformation and expansion. Displacement magnitudes were normalized to account for inter-subject variability. Metrics were averaged within the treated and untreated lungs. TLVR was derived from segmented lobar volumes at baseline and 90 dfu. Emphysema index was quantified using YACTA software. Clinical and physiologic outcomes (FEV1, RV, 6MWD) were recorded. Patients were classified as clinical responders (based on MCID criteria) and/or radiological responders (TLVR 50%). Imaging-derived metrics were compared across responder groups and correlated with clinical changes. Results All clinical parameters improved significantly from baseline to 90 dfu (FEV1 median 0.78 to 0.93 L, RV 5.39 to 4.48 L, 6MWT 279 to 317 m; Wilcoxon all p 0.001). Deformation metrics increased significantly in both treated and untreated lungs, suggesting compensatory expansion after target lobe collapse. In treated lungs, Jacobian (median 1.15 to 1.17; Wilcoxon p0.001) and normalized displacement (mean 0.08 to 0.09; t-test p0.001) both increased, with similar improvements in untreated lungs (all p0.001). Clinical responders (n = 38) exhibited greater anatomic collapse (TLVR median 58% vs 40%; p0.05), higher treated-lung expansion (ΔJacobian 0.08 vs 0.01; p0.05), and increased displacement magnitude (Δdisplacement 0.02 vs 0.01; p0.05). The untreated lung showed parallel but smaller mechanical changes. Among radiological responders (TLVR 50%; n = 66), TLVR itself no longer correlated strongly with ΔFEV1 or ΔRV (all |ρ| 0.05, p 0.1), indicating a saturation effect. In contrast, treated-lung ΔJacobian and untreated lung changes both significantly correlated with ΔFEV1 (ρ = 0.38, p = 0.0017; ρ = 0.30, p = 0.0146), suggesting that lung expansion metrics better reflect treatment-related functional recovery. Conclusion Regional lung expansion quantified by the Jacobian determinant reflects physiological recovery beyond structural volume loss. This abstract is funded by: None