Anatomical phenotype of obstructive sleep apnea patients based on cluster analysis

Objectives: To generate a novel subtype of obstructive sleep apnea (OSA) based on anatomical features and verify the differences in the response of different subtypes to orthodontic treatment, thus providing a theoretical reference for clinical decision-making. Materials and Methods: A K -means cluster analysis was performed for this retrospective serial study, which includes 722 OSA patients, aged 44.0 (36.0, 54.0) years, 80.2% male, with apnea-hypopnea index (AHI) of 23.2 (13.4, 39.6) events.h(-1), and body mass index (BMI) of 25.47 +/- 3.00 kg.m(-2). All samples were divided into three subtypes based on AHI, BMI, and five variables of craniofacial measurements. Sixty -seven cases with mandibular advancement devices (MAD) therapeutic results were further applied to validate the efficacy and side effects of this treatment in different subtypes. Results: Two hundred and thirty patients (31.9%) were characterized as cluster 1: AHI of 17.65 (11.80, 30.42) events.h(-1), BMI of 23.65 +/- 2.62 kg.m(-2), with skeletal Class II high -angle shape. Cluster 2 included 278 patients (38.5%): AHI of 17.00 (11.00, 26.48) events.h(-1), BMI of 25.36 +/- 2.53 kg.m(-2), soft palate length (SPL) of 39.25 mm (36.12, 42.20), with basically normal skeleton and normal airway size. Cluster 3, consisting of 214 patients (29.6%), exhibited a combination of anatomical deformity and obesity, with the highest AHI and BMI of 45.35 (30.42, 62.53) events.h(-1) and 27.57 +/- 2.59 kg.m(-2) respectively, but less deformity degree than cluster 1. Cluster 2 had the highest response rate and relatively mild side effects with MAD. Conclusions: Orthodontic treatment based on anatomical morphology could exert a better effect on mild-moderate OSA patients with mild skeletal deformity.