B29-22 Airway Tree Characteristics May Influence Bronchoconstriction in Patients With Severe Asthma

Rationale Airway tree size and shape are innately heterogeneous. In patients with asthma, additional heterogeneity may arise from pathological remodelling, characterized by thickened airway walls and narrowed or obstructed lumens. This interpatient variability in airway tree morphology and/or airway structure may influence physiological responses to inhaled stimuli. To investigate this, we characterized the size, shape, and structural features of the airway tree from chest computed tomography (CT) scans in asthma patients with and without saline-induced bronchoconstriction. Methods Forty-nine patients with severe asthma (29F:20M; mean±SD age=54±15) underwent same-day post-bronchodilator chest CT and spirometry before saline inhalation for sputum induction. Patients were stratified by degree of airflow obstruction (post-bronchodilator FEV1≥70%pred (n = 24) or < 70%pred (n = 25)), which informed the standardized saline dosing protocol used for sputum induction. They were subsequently classified based on their decrease in FEV1 during saline inhalation as bronchoconstriction-positive (BC+: ≥10% drop) or -negative (BC-: <10% drop). Patient-specific airway trees were segmented from chest CT scans acquired at full-inspiration using VIDA|vision software (VIDA Diagnostics Inc.), and lumen size (lumen area [LA], airway-to-lung ratio [ALR]), airway wall thickness (wall area percent [WA%], Pi10), and total airway count were measured. Two- and three-dimensional size and shape features of the airway tree were extracted from the segmented airway lumen mask using PyRadiomics, an open-source Python package. Gas trapping (LAA-856HU%) was measured on full-expiration chest CT. Unpaired t-tests were performed to compare CT-derived airway tree features between BC+ and BC- patients within each airflow obstruction cohort. Results 54% of FEV1≥70%pred patients (13/24) and 80% of FEV1<70%pred patients (20/25) bronchoconstricted during saline inhalation. In both cohorts, BC+ and BC- patients did not differ by demographics, FEV1%pred, or classic measures of luminal narrowing (LA, ALR) and wall thickness (WA%, Pi10) (all p > 0.05). However, in the FEV1≥70%pred cohort, BC+ patients had greater airway tree elongation (Figure-1B: 0.71±0.10 vs. 0.80±0.07, p = 0.02), greater airway tree flatness (Figure-1C: 0.25±0.04 vs. 0.30±0.06, p = 0.01), lower luminal surface area (Figure-1D: 353±92cm2 vs. 449±117cm2, p = 0.03), and trends towards lower total airway count (Figure-1E: 187±47 vs. 227±51, p = 0.06) and greater LAA-856HU% (Figure-1F, 14±10% vs. 9±8%, p = 0.18). In the FEV1<70%pred cohort, no differences in airway tree characteristics were observed between BC+ and BC- patients. Conclusion Among patients with severe asthma and mild airflow obstruction, those with saline-induced bronchoconstriction had more elongated and flatter airway trees with lower luminal surface area. Future work is needed to understand the potential interaction between airway tree features and airway responsiveness.