Abstract
OBJECTIVE: To assess the diagnostic value of extended sleep duration on a controlled 32-hour bed rest protocol in idiopathic hypersomnia (IH). METHODS: One hundred sixteen patients with high suspicion of IH (37 clear-cut IH according to multiple sleep latency test criteria and 79 probable IH), 32 with hypersomnolence associated with a comorbid disorder (non-IH), and 21 controls underwent polysomnography, modified sleep latency tests, and a 32-hour bed rest protocol. Receiver operating characteristic curves were used to find optimal total sleep time (TST) cutoff values on various periods that discriminate patients from controls. RESULTS: TST was longer in patients with clear-cut IH than other groups (probable IH, non-IH, and controls) and in patients with probable IH than non-IH and controls. The TST cutoff best discriminating clear-cut IH and controls was 19 hours for the 32-hour recording (sensitivity = 91.9%, specificity = 85.7%) and 12 hours (100%, 85.7%) for the first 24 hours. The 19-hour cutoff displayed a specificity and sensitivity of 91.9% and 81.2% between IH and non-IH patients. Patients with IH above the 19-hour cutoff were overweight, had more sleep inertia, and had higher TST on all periods compared to patients below 19 hours, whereas no differences were found for the 12-hour cutoff. An inverse correlation was found between the mean sleep latency and TST during 32-hour recording in IH patients. INTERPRETATION: In standardized and controlled stringent conditions, the optimal cutoff best discriminating patients from controls was 19 hours over 32 hours, allowing a clear-cut phenotypical characterization of major interest for research purposes. Sleepier patients on the multiple sleep latency test were also the more severe in terms of extended sleep. Ann Neurol 2018;83:235-247.