Criteria-based probable REM sleep behaviour disorder and risk of Lewy body dementia: analyses of a memory clinic cohort and a community cohort

Abstract

REM sleep behaviour disorder (RBD) is the strongest prodromal marker of alpha-synucleinopathies, but whether a brief questionnaire carries prognostic information for Lewy body dementia is unclear. We tested whether criteria-based probable RBD (pRBD)—a questionnaire algorithm requiring both core clinical features specified in the International Classification of Sleep Disorders, Third Edition (ICSD-3)—shows a subtype-specific association pattern, defined as an association with incident Lewy body dementia (LBD; dementia with Lewy bodies [DLB] or Parkinson’s disease dementia [PDD]) but not with all-cause dementia. Retrospective analysis of prospectively collected data from two independent cohorts—a memory clinic cohort ( n  = 619; median follow-up 3.4 years) and the community-based Korean Longitudinal Study on Cognitive Aging and Dementia (KLOSCAD; n  = 5274; 6.4 years). Criteria-based pRBD required dream-enactment behaviour (RBD Screening Questionnaire item 3) plus at least one complex nocturnal motor behaviour (items 6.1–6.3). Cause-specific hazard models estimated associations with incident LBD (clinic) and all-cause dementia (KLOSCAD), with the competing dementia subtype or death as competing events. Sensitivity analyses included inverse probability of censoring weighting, interval-censoring imputation and Firth-penalised estimation. In the memory clinic, 34 of 619 patients (5.5%) screened positive, of whom 5 (14.7%) developed LBD compared with 21 of 585 (3.6%) who screened negative (cause-specific hazard ratio [HR] 4.82, 95% CI: 1.54–15.10); there was no association with all-cause dementia (HR 1.00, 95% CI: 0.57–1.76). Of the 26 LBD events, 23 were DLB and 3 were PDD. In KLOSCAD, pRBD was not associated with all-cause dementia (HR 0.92, 95% CI: 0.47–1.81). An exploratory KLOSCAD analysis based on only nine LBD events (6 DLB, 3 PDD) produced a large but very imprecise estimate (HR 8.99, 95% CI: 1.71–47.17) that is hypothesis-generating rather than replication. In memory clinic patients, criteria-based pRBD was associated with incident LBD but not with all-cause dementia. The community cohort showed no all-cause association. This subtype-specific profile supports further evaluation of the screen for prioritising alpha-synucleinopathy assessment, pending validation with polysomnography-confirmed RBD. These findings rest on 26 events in a cohort in which only a minority of registered patients returned, and require confirmation.