Three-dimensional upper airway phenotyping and OSA severity: a CT-based study.

Abstract

Obstructive sleep apnea (OSA) involves recurrent upper airway collapse. While anatomical factors are key, the correlation between three-dimensional (3D) upper airway parameters and OSA severity remains inconsistent, limiting personalized treatment. 69 OSA patients underwent polysomnography and computed tomography. The upper airway was segmented into nasopharynx, retropalatal oropharynx, and retroglossal oropharynx. Parameters, including soft palate length, hyoid-to-mandibular plane distance, subregion volumes, minimum cross-sectional area (MIN-CSA), and the ratio of total volume to MIN-CSA (TV-NP-OP/MIN-CSA), were measured. Spearman's correlation, receiver operating characteristic (ROC) analysis, and logistic regression assessed associations with the apnea-hypopnea index (AHI). Bonferroni correction was applied (p < 0.005). AHI significantly correlated with MIN-CSA (r = -0.478), anteroposterior diameter at MIN-CSA (AP-MIN) (r = -0.399), lateral diameter at MIN-CSA (Lat-MIN) (r = -0.323), and TV-NP-OP/MIN-CSA (r = 0.473) (all p < 0.001). TV-NP-OP/MIN-CSA ("Anatomical Risk Index") showed good diagnostic performance for moderate-to-severe (area under the curve (AUC) = 0.779) and severe OSA (AUC = 0.789), and was an independent predictor across body mass index (BMI) subgroups. No volumetric parameters correlated with AHI. Reliability was excellent (intraclass correlation coefficients (ICCs) > 0.90). TV-NP-OP/MIN-CSA is a robust 3D anatomical parameter associated with OSA severity, offering a potential imaging biomarker for risk stratification. Question What is the most relevant 3D anatomical parameter derived from CT imaging that correlates with OSA severity, and can it serve as a reliable imaging biomarker? Findings The ratio of TV-NP-OP/MIN-CSA was strongly correlated with OSA severity (r = 0.473, p < 0.001) and served as an independent predictor across all BMI subgroups, with excellent measurement reliability. Clinical relevance This CT-based "Anatomical Risk Index" can help identify high-risk OSA phenotypes, guide personalized treatment planning, and improve risk stratification without requiring additional imaging, thereby optimizing therapeutic outcomes.