Distribution of Airway Findings in ANCA-Associated Vasculitis: A 20-Year Observational Analysis

<b>Objective:</b> Pulmonary involvement is commonly observed in anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV), presenting with manifestations such as diffuse alveolar hemorrhage, inflammatory infiltrates, pulmonary nodules, and tracheobronchial disease. We aimed to identify distinct subgroups of tracheobronchial disease patterns in patients with anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) using latent class analysis (LCA), and to evaluate their clinical characteristics and outcomes. <b>Methods:</b> We conducted a retrospective cohort study using electronic medical records of patients aged >18 years diagnosed with AAV and tracheobronchial disease between 1 January 2002 and 6 September 2022. Patients with follow-up <6 months were excluded. LCA was employed to identify disease subtypes based on 10 pre-defined indicators. Maximum likelihood estimation with 10 repetitions per model ensured robustness in model selection, guided by the Akaike information criterion (AIC). Patient and disease characteristics were summarized and compared across predicted classes. Statistical analyses included Kruskal-Wallis and Fisher's exact tests for continuous and categorical variables, respectively. The primary outcome was time to relapse of the tracheobronchial inflammation after starting immunosuppressive medication, analyzed using the Kaplan-Meier method and log-rank tests. Secondary outcomes included severity of pulmonary disease on pulmonary function tests, endoscopic interventions, tracheostomy, or mortality during follow-up. <b>Results:</b> Among 136 identified AAV patients assessed for tracheobronchial involvement, 111 (81.6%) were included after excluding 25 without tracheal or bronchial disease. Predominant findings included subglottic stenosis (91.0%), lower tracheal stenosis (16.2%), and bronchial stenosis (17.1%). LCA identified a three-class model as optimal: tracheal predominant (<i>n</i> = 94), tracheobronchial (n = 12), and bronchial predominant (n = 5). Tracheal predominant patients showed reduced risk of ear, eye, and lower respiratory manifestations, with milder obstruction on pulmonary function testing (PFT). Tracheobronchial-class patients were prone to saddle nose deformity (50%), extensive lower respiratory involvement (91.7%), and renal disease (66.7%). Bronchial predominant patients exhibited severe obstructive disease (median forced expiratory volume in 1 s (FEV1)% predicted: 58, IQR 34-66; FEV1/forced vital capacity (FVC) ratio: 56.9, interquartile range (IQR) 43-63.3) but lacked systemic AAV manifestations. LCA classes did not predict outcomes such as endoscopic intervention, tracheostomy, recurrent tracheobronchial narrowing, or mortality. <b>Conclusion:</b> LCA shows promise in subtype stratification of AAV patients, yet its utility in predicting outcomes and guiding treatment remains limited based on our analysis. Future studies with enhanced phenotypic data and larger cohorts are warranted to improve predictive accuracy.