Clustering-based characterization of clinical phenotypes in obstructive sleep apnoea using severity, obesity, and craniofacial pattern

Abstract

To identify and characterize the phenotypes of adult obstructive sleep apnoea (OSA) patients based on clustering using OSA severity, obesity, and craniofacial pattern. The samples consisted of 89 adult OSA patients whose polysomnography and lateral cephalogram were available. With cluster analysis using apneahypopnea index (AHI, events/hour), body mass index (BMI, kg/m2), ANB (degree), and mandibular plane angle (MPA, degree), three clusters were identified. Cephalometric variables including craniofacial, soft palate, hyoid bone, and pharyngeal space compartments were compared among clusters by one-way analysis of variance or Kruskal-Wallis test. Multivariable linear regression analysis was performed to find contributing factors to OSA severity within each cluster. Cluster-1 (obesity type; 49.4 per cent) exhibited moderate OSA, obesity, and normal sagittal and vertical skeletal pattern (AHI, 22.4; BMI, 25.5; ANB, 3.2 degrees; MPA, 26.3 degrees) without significant upper airway abnormality. Cluster-2 (skeletal type; 33.7 per cent) was characterized by moderate OSA, severe skeletal Class II hyperdivergent pattern with narrow pharyngeal airway spaces, without obesity (AHI, 27.9; BMI, 23.5; ANB, 7.5 degrees; MPA, 36.6 degrees). Cluster-3 (complex type; 16.8 per cent) included severe OSA, obesity, skeletal Class II hyperdivergent pattern (AHI, 52.8; BMI, 28.0; ANB, 4.5 degrees; MPA, 32.2 degrees), with posteriorly displaced hyoid and retroclined soft palate. The main contributing factors to AHI were obesity in Cluster-1; hyperdivergent vertical pattern with narrow pharyngeal space in Cluster-2; and hyperdivergent pattern, obesity, displaced hyoid, and soft palate in Cluster-3. Three OSA phenotypes resulted from this study provide a clinical guideline for differential diagnosis and orthodontic intervention in the interdisciplinary treatment for OSA patients.