**Background:** Korsakoff's syndrome (KS) is a neurological complication of chronic alcohol use and thiamine deficiency characterized by severe episodic memory impairment. Up to 50% of Alcohol Use Disorder (AUD) patients without KS also present neuropsychological deficits. Sleep disturbances are common in AUD and may persist after withdrawal, but sleep architecture in KS patients has been poorly studied with mixed results. This study aimed to determine whether specific sleep parameters are impaired in KS compared to AUD patients and whether they relate to cognitive deficits.
**Methods:** Twenty-two AUD patients (all men, mean age 47.6±9.7 years, 10.81±3.80 days abstinent), seven KS patients (all women, mean age 56.1±4 years, all institutionalized >10 years except one at 18 months), and 15 healthy controls (13 men, mean age 48.3±11.3 years) underwent polysomnography using an ambulatory recording device. Sleep was scored in 30-second epochs per AASM criteria. Neuropsychological assessment included processing speed (TMT-A, Stroop naming), executive functions (backward span, Stroop interference, TMT-B minus A), short-term memory (forward span), and episodic memory (delayed free recall of FCSRT). Non-parametric Kruskal-Wallis ANOVAs and Mann-Whitney post-hoc tests were used. Spearman correlations examined relationships between sleep and cognition.
**Key Results:** KS patients slept longer than AUD (465±40.6 vs 376±65.7 min, p=0.002, r=0.77) and HC (p=0.007, r=0.71). Both AUD and KS patients had higher stage shift index (AUD vs HC: p=0.002, r=0.6; KS vs HC: p=0.01, r=0.68) and higher AHI (AUD: 26.2±14.9, KS: 29.5±21.6 vs HC: 9.4±4.3; both p<0.05) compared to HC. AUD patients had lower N3 sleep than HC (14.3±6.6% vs 21.2±7.2%, p=0.007, r=0.53), while KS patients did not differ from HC on N3. KS patients had lower REM sleep than both AUD (14.4±3.4% vs 23.2±4.9%, p=0.001, r=0.84) and HC (p=0.01, r=0.67). Episodic memory showed a graded deficit: HC > AUD > KS (HC: 0±1, AUD: -1.5±1.3, KS: -4.7±1.8; p<0.0001). In the pooled patient group, REM sleep proportion correlated positively with episodic memory performance (r=0.42, p=0.02), remaining significant after controlling for AHI (r=0.52, p=0.006). Subjective sleep complaints (PSQI) showed HC < KS < AUD (2.3±1.4 vs 4.6±2.1 vs 8.1±2.9, p<0.001).
**Clinical Implications:** This study demonstrates that both AUD and KS patients have high rates of sleep-disordered breathing (82% of AUD, 71% of KS with AHI≥15) that may go undetected due to lack of typical symptoms, particularly in women. The distinct REM sleep reduction in KS patients, correlating with episodic memory deficits, suggests REM sleep may contribute to the pathophysiology of alcohol-related amnesia. These findings support systematic screening for sleep disorders in AUD and KS patients using objective measures, as untreated sleep-disordered breathing may worsen cognitive and brain health outcomes. The relative preservation of slow-wave sleep in KS (vs AUD) may reflect recovery with prolonged abstinence. Limitations include small KS sample size, sex confound (all KS women, all AUD men), different recording environments, and psychotropic medication use in KS patients.