259 Clustering in Central Disorders of Hypersomnolence algorithm is able to identify distinct subgroups, principally separating people with cataplexy from those without. Further subdivision of those without cataplexy resulted in 2 clusters evenly mixing individuals with narcolepsy type 2 and idiopathic hypersomnia, which were separated on the basis of variables related to awakening (e.g., sleep drunkenness and subjective difficulty in awakening), sleep need (e.g., weekend-week sleep length difference), and objective biomarkers (HLA-DQB1*0602 positivity and hypocretin-1 level). MSLT parameters were not significantly different between the 2 clusters with individuals without cataplexy. The advanced cluster distinctness and resampling analyses revealed (Appendix A) that the two clusters of people without cataplexy were most distinctly grouped of all clusters (also from each other) and had good cluster reproducibility. People with cataplexy (generally diagnosed as having narcolepsy type 1) were further split into multiple subtypes that likely reflect different disease severities. These subtypes were different mainly in sex distribution and presence and severity of cataplexy, hypnagogic hallucinations, sleep paralysis, and sleep drunkenness. Thanks to the large number of patients and hypersomnolence-related variables, our analyses have produced more reliable and detailed results than two other studies that have previously tried to identify subtypes in central hypersomnolence disorders using agglomerative hierarchical clustering [327, 328]. These studies respectively included only 96 participants or only people with idiopathic hypersomnia [327, 328]. Clustering in both studies was performed on just seven and three variables, respectively. The small number of clustered variables in these studies limited their ability to identify differentiating variables among all hypersomnolence aspects. Multiple narcolepsy experts have advocated revising the current classification, but guidance is lacking in defining new subgroups and corresponding diagnostic criteria [27-29, 164, 317, 318]. Our data-driven approach provides the opportunity to critically evaluate current ICSD-3 classification and shows that refinement of the hypersomnolence without cataplexy criteria is needed that would yield more consistent categorization. The advanced analyses in Appendix A revealed that the two clusters with people without cataplexy had good reproducibility and were most distinctly grouped compared to other clusters. The most prominent differential variables for subgrouping people without cataplexy include the presence of sleep drunkenness, subjective difficulty in awakening, mean weekend-week sleep length difference, and HLA-DQB1*0602 positivity. These differentiating symptoms suggest that certain subtypes/phenotypes of central disorders of hypersomnolence involve neuronal networks different from the cataplectic 9
RkJQdWJsaXNoZXIy MjY0ODMw