多模态 PET-MRI 特征预测孤立性 REM 睡眠行为障碍向路易体痴呆转化

Despite isolated REM sleep behaviour disorder (iRBD) reliably identifying individuals at high risk of developing Parkinson's disease (PD) or dementia with Lewy bodies (DLB), with more than 90% converting, the neurobiological markers of the conversion phenotype are not fully understood. Most previous studies have found univariate imaging biomarkers of phenoconversion, and most of these lack specificity for PD and DLB. Using a well-characterised iRBD cohort (N=21), receiving multiple PET and MRI scans, we aimed to assess if the inter-modality relationship can specifically predict conversion to PD or DLB. The patients received [ 18 F]-DOPA, [ 11 C]-Donepezil and [ 11 C](R)-PK11195 PET and structural MRI scans to derive grey matter (GM) volume along with dynamic susceptibility contrast MRI to extract measures of microcirculatory dysregulation. We used a multimodal adaptation of the scaled sub-profile model to identify patterns of multimodal covariance across the PET and MRI scans. We identified a multimodal network characterised by high neuroinflammation, cholinergic dysfunction, GM atrophy and microcirculatory dysfunction with the covariance pattern converging in the medial occipito-parietal cortex (OPC). The combined pathology from the OPC and striatal [ 18 F]-DOPA uptake was able to specifically predict conversion into DLB (sub-distribution hazard ratio=38.68, 95% confidence interval: 7.545-198.3), and this survived correction for age and disease duration. These findings demonstrate the feasibility of integrating multiple modalities in risk assessment and suggest that posterior cortical pathology, along with striatal dopaminergic impairment, could hold predictive value for phenotype-specific conversion of iRBD patients, and that multiple pathological processes converge in this brain region.