[¹⁸F]FDG-PET BIOMARKERS OF RESTING-STATE NETWORK VULNERABILITY IN GBA1 PARKINSON’S DISEASE: AN INDEPENDENT COMPONENT ANALYSIS

Parkinson’s disease (PD) is movement disorder characterised by both motor and non-motor disturbances. Among the genetic risk factors of PD , variations in the GBA1 gene represent a clinically significant association with earlier disease onset and greater cognitive vulnerability. Yet , little is known about the network-level mechanisms underlying the elevated risk associated with GBA1. To address this gap, we examined metabolic brain network alterations across four groups: non-mutated PD (NM-PD), PD with GBA1 variants (GBA-PD), non-manifesting GBA1 carriers (GBA-non-PD), and healthy controls (HC), all of whom underwent [¹⁸F]FDG-PET imaging. Independent component analysis (ICA) was applied and Network Comparison Testing (NCT) was used to identify nine resting-state networks. Age-corrected Euclidean distances were calculated to quantify network-level deviations between each participant’s spatial map and the corresponding group-average map . A LASSO regression was conducted to identify the most discriminative features across groups. Clinical correlations with cognitive and motor measures were further explored .Results revealed that both NM-PD and GBA-PD patients exhibited significantly greater deviations than HC in the Somatomotor and Default Mode Networks. LASSO feature selection consistently highlighted the Somatomotor, Secondary Visual, and Default Mode Networks as the strongest discriminators of disease status, with NM-PD and GBA-PD primarily differentiated by HC with Default Mode Network alterations and GBA-PD showing additional vulnerability in Visual networks. Importantly, GBA non-PD already demonstrated deviations in the Default Mode Network compared with HC, suggesting that alterations in large-scale metabolic networks may serve as prodromal markers of vulnerability. Correlation analyses further showed that deviations in the Cerebellar Networks were associated with poorer cognitive performance and the Default Mode and the Dorsal Attentive networks showed increased motor disturbances .These findings suggest that large-scale metabolic network alterations may emerge prior to clinical diagnosis, providing prodromal markers of cognitive vulnerability in PD. More broadly, they highlight the potential of [¹⁸F]FDG-PET network analysis as a biomarker for early detection and as a framework to support future patient stratification in precision medicine approaches to PD.