Division of Respiratory Disease, Nihon University School of Medicine, 30-1 Ohyaguchi-Kamicho, Itabashiku, Tokyo, 173-8610, Japan.
Sleep Breath. 2021 Jun;25(2):767-776. doi: 10.1007/s11325-020-02160-8. Epub 2020 Sep 4.
To investigate the different pathophysiologies of obstructive sleep apnea (OSA) phenotypes using cluster analysis. Differences between leptin/adiponectin levels in the resulting OSA phenotypes were also examined.
In total, 1057 OSA patients were selected, and a retrospective survey of clinical records, polysomnography results, and blood gas data was conducted. Patients were grouped into four clusters by their OSA severity, PaCO2, body mass index (BMI), and sleepiness. A k-means cluster analysis was performed, resulting in a division into four subpopulations. The Tukey or Games-Howell tests were used for intergroup comparisons.
Among the 20 clinical OSA items, four common factors (Epworth Sleepiness Scale [ESS], BMI, Apnea-Hypopnea Index [AHI], and PaCO2) were extracted by principal component analysis, and a cluster analysis was performed using the k-means method, resulting in four distinct phenotypes. The Clusters 1 (middle age, symptomatic severe OSA) and 4 (young, obese, symptomatic very severe OSA) exhibited high leptin levels. C-reactive protein levels were also elevated in Cluster 4, indicating a different pathophysiological background. No apparent differences between clusters were observed regarding adiponectin/leptin ratios and adiponectin levels. Classification into groups based on phenotype showed that Epworth Sleepiness Scale [ESS] score and disease severity were not correlated, suggesting that sleepiness is affected by multiple elements.
The existence of multiple clinical phenotypes suggests that different pathophysiological backgrounds exist such as systemic inflammation and metabolic disorder. This classification may be used to determine the efficacy of continuous positive airway pressure treatment that cannot be determined by the AHI.
通过聚类分析研究阻塞性睡眠呼吸暂停（OSA）表型的不同病理生理学。还检查了由此产生的 OSA 表型中瘦素/脂联素水平的差异。
共选择了 1057 例 OSA 患者，对临床记录、多导睡眠图结果和血气数据进行回顾性调查。根据 OSA 严重程度、PaCO2、体重指数（BMI）和嗜睡程度将患者分为四组。进行 k-均值聚类分析，将其分为四个亚群。使用 Tukey 或 Games-Howell 检验进行组间比较。
在 20 项临床 OSA 项目中，通过主成分分析提取了四个共同因素（Epworth 睡眠量表 [ESS]、BMI、呼吸暂停-低通气指数 [AHI]和 PaCO2），并使用 k-均值方法进行聚类分析，得到四个不同的表型。簇 1（中年，症状性严重 OSA）和 4（年轻，肥胖，症状性非常严重 OSA）表现出高瘦素水平。簇 4 的 C 反应蛋白水平也升高，表明存在不同的病理生理背景。在脂联素/瘦素比值和脂联素水平方面，各簇之间没有明显差异。基于表型进行分类表明，Epworth 睡眠量表 [ESS]评分和疾病严重程度没有相关性，这表明嗜睡受多种因素影响。
多种临床表型的存在表明存在不同的病理生理背景，如全身炎症和代谢紊乱。这种分类可用于确定不能通过 AHI 确定的持续气道正压治疗的疗效。
