Identifying phenotypes of obstructive sleep apnea using cluster analysis.

Abstract

Over the last decade, advances in understanding thepathophysiology, clinical presentation, systemic consequences and treatment responses in obstructive sleep apnea (OSA) have made individualised OSA management plausible. As the first step in this direction, this study was undertaken to identify OSA phenotypes. Patients diagnosed with OSA on level 1 polysomnography (PSG) were included. Clinical and co-morbidity profile, anthropometry and sleepiness scores were compiled. On PSG, apnea-hypopnea index, positional indices, sleep stages and desaturationindices (T90) were tabulated. Cluster analysis was performed to identify distinct phenotypes among included patients withOSA. One hundred patients (66 males) with a mean age of 49.5 ± 13.3years were included. Snoring was reported by 94% subjects, and 50% were excessively sleepy. Two-thirds of subjects had co-morbidities, the most frequent being hypertension (55%) and dyslipidemia (53%). Severe OSA was diagnosed on PSG in 42%, while 29% each had mild and moderate OSA, respectively. On cluster analysis, 3 distinct clusters emerged. Cluster 1 consisted of older, obese subjects with no gender predilection, higher neck circumference, severe OSA with more co-morbidities and higher T90. Cluster 2 comprised of younger, less obese males with snoring, witnessed apnea, moderate and supine predominant OSA. Cluster 3 consisted of middle-aged, obese males with lesser co-morbidities, mild OSA and lower T90. This study revealed three OSA clusters with distinct demographic, anthropometric and PSG features. Further research with bigger sample size and additional parameters may pave the way for characterising distinct phenotypes and individualising OSA management.