C73-24 Novel AI-based Regional Airway Tapering Metrics Reveal Early Structure-Inflammation Relationships in Young Children With Cystic Fibrosis

Rationale

Early cystic fibrosis (CF) lung disease involves inflammatory airway remodeling that precedes clinically overt bronchiectasis and pulmonary exacerbations (PEx). Most existing quantitative CT approaches utilize whole lung metrics, which may obscure early, regionally heterogeneous disease processes in very young children. Recent advances in AI-based airway analysis enable detailed and quantitative, regional characterization of airway geometry, including tapering and shape metrics not captured by conventional methods. We hypothesized that novel, regionally derived quantitative airway tapering features from chest CT scans are associated with bronchoalveolar lavage (BAL) inflammatory markers from the same lung lobes and prior exacerbation history in young children with CF.

Methods

We conducted a retrospective exploratory analysis of inspiratory chest CT scans from 2-year children with CF (N = 11). Our in-house AI-based airway analysis pipeline extracted eight quantitative airway features informed by domain knowledge: tapering angle, intra-branch tapering, inter-branch tapering, cylindricity, carina-to-endpoint distance, average diameter, diameter standard deviation, and residual zero crossing rate. Features were quantified in the right middle lobe (RML), lingula, and across the overall bronchial tree. BAL fluid collected from the RML and lingula was analyzed for IL-8, neutrophil elastase (NE), percent live polymorphonuclear cells (% live PMN), and neutrophil count. Clinical history of pulmonary exacerbations requiring IV antibiotics was dichotomized as 0 PEx versus ≥1 prior PEx. Associations between airway features and BAL markers were assessed using Spearman correlation. Group differences by PEx history were evaluated using Mann-Whitney U tests.

Results

Children with prior PEx demonstrated significantly reduced intra-branch tapering and tapering angle as well as increased cylindricity (all p < 0.05) compared to with those without prior PEx. Significant inverse correlations were observed between median intra-branch tapering and neutrophil percentage in BAL fluid from both lobes: RML (r = −0.64, p < 0.05) and lingula (r = −0.78, p < 0.05). These findings suggest that reduced airway tapering is associated with increased neutrophilic inflammation. Other airway features showed variable, non-significant relationships with BAL inflammatory markers, likely reflecting limited sample size.

Conclusions

AI-derived regional airway tapering metrics from inspiratory CT scans revealed early structural abnormalities associated with neutrophilic inflammation and prior pulmonary exacerbations in very young children with CF. These novel quantitative imaging biomarkers may enable earlier identification of regional airway damage and improve understanding of structure-inflammation relationships in early CF lung disease. Larger longitudinal studies are warranted to validate these findings and assess prognostic value.

This abstract is funded by: None