349 Discussion frequently used in narcolepsy type 1 compared to an external similarly aged general Dutch sample [372]. Data-driven phenotyping of central disorders of hypersomnolence Peaks in clustering evaluation metrics are commonly used to determine the optimal number of clusters [344, 345]. In our case (presented in Appendix E of Chapter 9), these measures did not clearly favour any single number of clusters below 15 which suggests that individuals in the EU-NN database are not organized according to a single number of archetypes but more as a spectrum. This means subdivision can still result in distinct clusters, but in presence of individuals closely bordering different clusters. The spectrum-like organization of phenotypes aligns with the clinical impression that people with central disorders of hypersomnolence may present heterogeneous phenotypes with diverse symptom combinations, intensity and/or frequency. Besides the clear relationship between hypocretin deficiency and presence of cataplexy, for many other symptoms (such as hypnagogic hallucinations, sleep paralysis, disturbed nocturnal sleep, quality of awakening, automatic behaviour and certain cognitive complaints) this relationship is less clear as it remains largely unknown why some individuals with central disorders of hypersomnolence experience them and others do not [452]. Until more objective biomarkers are available, the heterogeneity in clinical presentation will inevitably result in diagnostic uncertainty, regardless of the exact diagnostic criteria. The goal should therefore not be to develop diagnostic criteria that unambiguously subgroup all individuals with central disorders of hypersomnolence but to make sure that future criteria better reflect underlying phenotypes, are reproducible and incorporate levels of diagnostic certainty. In line with current diagnostic criteria, our clusters displayed in Chapter 9 identified narcolepsy type 1 as a clear distinct entity. The two clusters including people without cataplexy, however, evenly mixed people formally diagnosed as narcolepsy type 2 and idiopathic hypersomnia and our results hereby do not support current diagnostic criteria with SOREMPs as the main diagnostic differentiator for those without cataplexy. This is substantiated by the poor test-retest reliability of the MSLT for narcolepsy type 2 and idiopathic hypersomnia (respectively just 47% and 25% retain their diagnoses) [34, 35]. Differentiation among those without cataplexy based on other biomarkers that more closely reflect phenotypes, is warranted. Multiple theories exist on the optimal diagnostic categorization for central disorders of hypersomnolence without cataplexy, which will hereafter be presented in light of our findings of Chapter 9 (Figure 2) [7-9, 27, 28]. 11
RkJQdWJsaXNoZXIy MjY0ODMw