Key points are not available for this paper at this time.
Abstract Introduction Disambiguation of type 2 narcolepsy (NT2) from idiopathic hypersomnia (IH) is notably challenging, given the absence of biomarkers and testing (MSLT) limitations. The utility of nocturnal sleep features for diagnostic differentiation of NT2 vs. IH has yielded some promising findings. However, data are limited by the reliance on the MSLT to define CNS subtypes, yielding critical weaknesses in generalizability and inference. The goals of this study were to (1) quantify and qualify patterns of sleep disruption in patients being evaluated for hypersomnia, agnostic of MSLT outcomes blinded and (2) evaluate congruence between nocturnal endotypes and unblinded MSLT outcomes/diagnoses. Methods This study used BioSerenity’s archival database of patients being evaluated for hypersomnia with in-lab PSG-MSLT. A total of N=697 PSGs met inclusion criteria. Records were scored by senior RPSGTs and spectral power was computed on raw tracings using FFT with Welch’s method. Data for each 30-s epoch were exported for tabulation summary variables, state sequences, and transition indices. Data were ingested into a statistical program for unsupervised cluster analysis. Results A two-step unsupervised cluster analysis was performed on 18 PSG features; 5 were retained for cohesion: # wake-REM sequences, sleep-wake and REM transition indices, WASO, and arousal index (p.001). The model identified 3 clusters of patients with homogeneous patterns of sleep disruption. Cluster 1 (n=88) had notably disrupted sleep and REM, cluster 2 (n=210) had sleep (not REM) disruption with WASO, and cluster 3 (n=399) had well-consolidated sleep. A multinomial logistic regression indicated similar MSLT outcomes for clusters 1 and 3 (p=.385), whereas cluster 2 had the fewest 2 MSLT REMs (17% vs. 27%; p.001) and longest MSL (8.7 min vs. 7.1 min; p.001). Conclusion This data-driven approach produced 3 distinct groups of nocturnal features that, in theory, were compatible with the conceptual distinction of NT2 from IH. However, MSLT outcomes of NT2 and IH did not reflect a differential pattern of cluster membership. While further data are needed to evaluate the clinical utility of nocturnal sleep/REM features in predicting outcomes, PSG features of sleep/REM disruption, when adjudicated against MSLT outcomes, may provide nuance to guide interpretation and additional testing. Support (if any) Jazz Pharmaceuticals (IST-19-11124)
Cairns et al. (Sat,) studied this question.