Background: Neurodegenerative brain disorders are chronic, progressive, and incurable diseases that clinically manifest with cognitive decline and/or movement disorders. Early clinical diagnostic accuracy is relatively low. In recent years, various imaging biomarkers were studied to gain better diagnostic accuracy in early disease stages and to distinguish among similar syndromes. Functional brain imaging with 18F-FDG PET measures regional glucose metabolism, which reflects the synaptic activity of neurons. Brain glucose metabolism is altered already at early stages of neurodegenerative diseases. 18F-FDG PET brain analysis can help us to differentiate between different neurodegenerative syndromes and to learn about pathophysiological processes in the brain. Furthermore, multivariate statistical analysis of 18F-FDG PET images enable us to identify syndrome-specific metabolic patterns of functionally interconnected brain areas that serve as a metabolic biomarker of a particular neurodegenerative syndrome. Aims of this dissertation were (1) to identify and validate a metabolic pattern specific for sporadic Creutzfeldt-Jakob disease (CJD) and to study its properties; (2) analysis of the diagnostic accuracy of the clinical diagnosis of neurodegenerative parkinsonisms in comparison to automated 18F-FDG PET based diagnosis; and (3) analysis of motor and cognitive metabolic patterns characteristic for Parkinson's disease (PD) and relationship between them at different stages of the disease and in comparison to other parkinsonian syndromes.
Methods: In a retrospective study, we analyzed clinical data and 18F-FDG PET brain images of 784 subjects: 29 patients with CJD, 26 patients with Alzheimer’s dementia (AD), 20 patients with behavioral variant of frontotemporal dementia (bvFTD), 356 patients with PD, 125 patients with multisystem atrophy, 113 patients with progressive supranuclear paresis (the latter two entities combined into an atypical parkinsonian syndrome (APS) group) and 115 healthy subjects (HS). All subjects underwent 18F-FDG PET brain imaging. After image preprocessing, we (1) analyzed the 18F-FDG PET brain images of patients with CJD and HS using a scaled subprofile modelling – principal component analysis (SSM-PCA) method to identify and validate a metabolic pattern specific for CJD; (2) determined the accuracy of the clinical diagnosis of parkinsonisms 12 months after imaging and compared it with the diagnosis calculated by an automatic logistic differential diagnostic algorithm based on 18F-FDG PET images which was considered gold standard; (3) analyzed the difference in motor and cognitive brain metabolic patterns specific for PD in different stages of PD, compared it with other parkinsonian syndromes and evaluated its role in differential diagnosis.
Results: (1.) CJDRP was identified and validated and its clinical significance was proven by significant correlations with patients’ clinical measures. It was found to be specific for CJD in contrast to AD and bvFTD. (2.) A comparison of real-life clinical diagnosis with automated diagnosis based on 18F-FDG PET as the gold standard showed 94.7% sensitivity, 83.3% specificity, 81.8% positive predictive value (PPV), and 95.2% negative predictive value (NPV) for PD and 88.2%, 76.9%, 71.4%, and 90.9% for APS diagnosis made by clinical neurologists. (3.) A consistent relationship between motor and cognitive patterns in PD patients was found. The delta, the difference between motor and cognitive PD patterns, steadily progressed with disease progression in several cross-sectional and longitudinal cohorts. Moreover, combination of motor pattern and the delta accurately discriminated PD patients from APS.
Conclusion: In this research project, we provided a novel multivariate CJD specific brain pattern with potential use in basic neurodegenerative research. Further, superiority of 18F-FDG PET based diagnosis against clinical one has been demonstrated in real clinical practice and a study of metabolic brain patterns in PD associated with motor symptoms and cognition provided clues of disease/neurodegeneration progression consistent with the Braak hypothesis.
To conclude, with this research, we provided novel insights into the pathophysiology of neurodegeneration in CJD, PD and APS and laid the foundation for the introduction of algorithm-based diagnosis established on the basis of multivariate brain patterns into clinical practice. Pathophysiological knowledge and accurate diagnosis are prerequisites for successful disease modifying trials in the future.
The above-mentioned network analytical methods are increasingly being used in brain network research and in the clinical diagnosis of neurodegenerative diseases. This PhD thesis has contributed to the general knowledge of functional brain networks, an emerging scientific field, which is in detail reviewed in the last chapter of dissertation.
|
