Non-adherence to Medication Withdrawal Reduces Diagnostic Accuracy of MSLT in Narcolepsy but not Idiopathic Hypersomnia
Katherine Beshears1, Jinu Johnson2, James Bena3, Shannon Morrison3, Matheus Lima Diniz Araujo2, Sikawat Thanaviratananich2, Nancy Foldvary-Schaefer2
1Cleveland Clinic Lerner College of Medicine, 2Sleep Disorders Center, Neurological Institute, Cleveland Clinic, 3Department of Quantitative Health Sciences, Cleveland Clinic
Objective:
To assess how diagnostic accuracy of the polysomnogram-multiple sleep latency test (PSG-MSLT), the objective test for central disorders of hypersomnolence (CDH), varies by adherence to medication/substance use protocol (Krahn 2021).
Background:

PSG-MSLT for CDH diagnosis requires discontinuation of sleep-modifying medications at least 14 days (6 weeks for fluoxetine) prior and urine toxicology (UTOX) at testing. International Classification of Sleep Disorders-3 TR (ICSD-3-TR) criteria require mean sleep latency (MSL) <8 minutes to diagnose idiopathic hypersomnia (IH), plus >2 sleep-onset REM periods (SOREMPs) for narcolepsy type 1/2 (NT1/NT2). Accurate diagnosis is critical for treatment outcomes and to reduce diagnostic delay.

Design/Methods:
Retrospective analysis was conducted of 1550 PSG-MSLT studies for diagnosis of CDH January 2012 to February 2025 at a single academic medical center. Clinical phenotyping was performed by sleep medicine specialists independent of MSLT results. Patients were counselled regarding type/timing of medication withdrawal and avoidance of confounding substances prior to testing. 
Results:

Of 786 cases, 90 (11.4%) met clinical criteria for NT1, 155 (19.7%) NT2, 314 (39.9%) IH, and 227 (28.9%) were undifferentiated (UH). Medication withdrawal at appropriate time occurred in 342/786 (43.5%) overall, 51/90 NT1 (56.7%), 70/155 NT2 (45.2%), 119/314 IH (37.9%), and 102/227 UH (44.9%). REM-suppressing medications had the highest non-adherence, representing 54.7% non-adherent cases (243/444). Adherence to medication guidelines for NT1/NT2 was associated with satisfying ICSD-3-TR diagnostic criteria vs those non-adherent (NT1: 86.3% vs 64.1% (P=0.014); NT2: 75.7% vs 60.0% (P=0.038)). No difference was found for IH (56.3% vs 60.0% (P=0.52)). Cannabinoids were the most common substance detected by UTOX, found in 33/786 (4.20%), but did not affect ICSD-3-TR criteria satisfaction.

Conclusions:

Recommended medication withdrawal increased diagnostic accuracy of the PSG-MSLT for NT1 and NT2 but not IH. Diagnostic accuracy in IH was poor.  The impact of cannabinoids on MSLT outcomes warrants further investigation given their use in patients with suspected CDH.

10.1212/WNL.0000000000217535
Disclaimer: Abstracts were not reviewed by Neurology® and do not reflect the views of Neurology® editors or staff.