Perturbation Covariance Analysis applied to neuroimaging data in Parkinson's Disease

Parkinson’s disease (PD) is a neurodegenerative disorder characterized by progressive molecular and structural brain alterations. Advanced imaging techniques provide a means to investigate early connectivity deviations and their correlation with clinical symptoms, enabling the identification of reliable diagnostic and prognostic biomarkers. This study applies perturbation network analysis to multimodal neuroimaging data to explore connectivity alterations in PD patients and assess their utility as biomarkers for disease progression. Methods 33 PD patients (27 idiopathic PD (iPD) and 6 LRRK2 mutation carriers) and 25 age matched healthy controls (HC) were included in this study, with longitudinal data collected for 20 iPD patients. Clinical assessments, structural magnetic resonance imaging (MRI), dynamic ¹¹C-UCB-J positron emission tomography (PET), ¹¹C-DASB PET, and DAT single-photon emission computed tomography (SPECT) were performed. These imaging modalities were selected for their ability to capture complementary aspects of PD pathology: structural changes (VBM), synaptic density (¹¹C-UCB-J), and serotonin transporter availability (¹¹C-DASB). Perturbation network analysis was applied to identify deviations in connectivity patterns, statistical analyses were performed to evaluate group differences, longitudinal changes, and correlations with clinical measures. Results PD patients exhibited significant increases in connectivity deviations over time, particularly in the putamen, substantia nigra, and thalamus, regions critically implicated in PD pathology. Distinct deviation patterns emerged across the imaging modalities, highlighting their unique strengths. Among the tracers, VBM demonstrated the strongest associations with clinical symptoms and emerged as the most reliable modality for assessing disease severity and variability. Connectivity deviations correlated with motor and non-motor symptoms, reinforcing the clinical relevance of this approach. Conclusion This study demonstrates that perturbation network analysis effectively captures disease-related connectivity alterations in PD. While each tracer provides complementary insights, VBM stands out as the most effective in tracking disease progression and variability. The strong correlations between connectivity deviations and clinical symptoms underscore the potential of this method for early diagnosis and disease monitoring. Future research should focus on larger cohorts and extended longitudinal assessments to enhance diagnostic accuracy and personalized patient care.