0708 Identifying Core Sleep Features of Idiopathic Hypersomnia Using a Multimodal Polysomnographic–MSLT Approach and Principal Component Analysis

Résumé fourni par la source

Abstract Introduction Many experts believe that Idiopathic Hypersomnia (IH) should be subtyped based on sleep duration but limited data are available to support such delineation. To characterize physiologic sleep features within the IH diagnosis, we evaluated polysomnography (PSG) and multiple sleep latency test (MSLT) sleep variables in patients with IH compared with clinical controls and applied a data-driven variable reduction strategy to isolate features that explain the most variability within the IH diagnosis. Methods We studied 51 people ages 15-45 years with IH (meeting ICSD-3-TR criteria) and 46 clinical controls (sleepy people not meeting diagnostic criteria for IH or narcolepsy) who underwent a nocturnal PSG followed by MSLT or long sleep PSG study (allowed to sleep till natural waking) at Beth Israel Medical Center or Boston Children’s Hospital. Inclusion criteria specified time in bed of ≥ 9 hours. Candidate sleep features included conventional PSG and MSLT metrics and bout-level analyses of NREM sleep stages. Group comparisons and correlations with the Idiopathic Hypersomnia Severity Scale (IHSS) across groups were used for variable selection (Cohen’s d > 0.40, IHSS correlation r > 0.30 and p < 0.05). Variables were entered into an IH sleep principal component analysis (PCA). Results Three sleep features met inclusion for PCA within the IH group: nocturnal total sleep time (nTST), sleep efficiency, and mean sleep latency (MSL). The Kaiser-Meyer-Olkin measure was acceptable for 3-variable model (KMO = 0.529), and Bartlett’s test indicated factorability (χ² = 58.9, p < .001). PCA yielded one dominant component (eigenvalue = 2.01) with high factor loadings of total sleep time (0.94) and sleep efficiency (0.91) explaining 66.9% of total variance (Component 1). Component 2 explained 27.8% of the variance (eigenvalue = 0.84) predominantly contributed by mean sleep latency (0.84). Conclusion The majority of total explained variance in IH sleep physiology arises from nocturnal sleep duration and consolidation. Data-driven feature reduction supports two IH groups: one defined by high sleep capacity and another by daytime sleepiness. These findings support modified protocols to capture long sleep duration and inclusion of nocturnal sleep metrics—particularly TST and sleep efficiency—in diagnostic criteria for IH. Support (if any) Jazz Pharmaceuticals ISR (Maski)

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.