Clinical utility of principal components analysis on PET data in the prediction of Alzheimer’s disease dementia

Abstract: In this study, the voxel-wise principal components analysis (PCA) was applied to [18F]FDG PET and [18F]AV-45 PET data to identify metabolic patterns related to conversion from mild cognitive impairment (MCI) to Alzheimer’s disease (AD) dementia.

Methods: [18F]FDG PET scans of 544 MCI patients were included for analysis. Voxel-based PCA was implemented to identify AD dementia conversion-related pattern (FDG-ADCRP). PCA was additionally applied to the amyloid PET data to construct the Aβ-ADCRP. Cox proportional hazard regression models assessed the predictive value of the pattern expression score (PES) of FDG-ADCRP and Aβ-ADCRP alone and in combination with non-imaging variables. The models were compared in their abilities to stratify subjects according to their conversion risks employing Kaplan-Meier survival analyses. Those subjects who had also CSF measures of phosphorylated tau available were categorised following the AT(N) classification scheme. PES of FDG-ADCRP was compared among groups and its prognostic value was assessed within the group of subjects with biologically defined AD. Then, constructed patterns were validated against neuropathological staging schemes of Braak and Thal.

Results: PCA applied to [18F]FDG PET revealed the FDG-ADCRP that involved regions with a relative decrease in metabolism (temporoparietal, frontal, posterior cingulate, and precuneus) and a relative increase in metabolism (sensorimotor and occipital, cerebellum, and putamen). The constructed Aβ-ADCRP showed high amyloid load in the posterior cingulate cortex and precuneus, the mesial frontal, the insular and ventral striatum. The PES of Aβ-ADCRP yielded significantly lower predictive value than PES of FDG-ADCRP, while both were improved when combined with non-imaging variables. Best prediction accuracy was reached when the PES of Aβ-ADCRP, the PES of FDG-ADCRP, and non-imaging variables were combined into one model. In subjects categorised according to the AT(N) classification scheme, PES of FDG-ADCRP was significantly higher only in the group with biologically defined AD. Finally, the PES of FDG-ADCRP and Aβ-ADCRP showed a highly significant association with the neuropathological examinations.

Conclusions: The PCA applied to [18F]FDG PET resulted in FDG-ADCRP, which is a valuable biomarker of conversion in subjects with MCI and biologically defined AD. It shows great potential for stratifying subjects according to the predicted risk. The strong association with post-mortem data promotes its use in research and clinical settings

Angaben zum Objekt

• Standort: Deutsche Nationalbibliothek Frankfurt am Main
• Umfang: Online-Ressource
• Sprache: Englisch
• Anmerkungen: Universität Freiburg, Dissertation, 2021

Klassifikation und Themen

• Schlagwort: Dementia ; Principal components analysis ; Alzheimerkrankheit ; Nuklearmedizin

Beteiligte, Orts- und Zeitangaben

• Ereignis: Veröffentlichung
• (wo): Freiburg
• (wer): Universität
• (wann): 2021
• Urheber: Blazhenets, Ganna
• Beteiligte Personen und Organisationen: Meyer, Philipp Tobias ; Urbach, Horst ; Albert-Ludwigs-Universität Freiburg. Medizinische Fakultät

Weitere Informationen

• DOI: 10.6094/UNIFR/193982
• URN: urn:nbn:de:bsz:25-freidok-1939827
• Rechteinformation: Der Zugriff auf das Objekt ist unbeschränkt möglich.
• Letzte Aktualisierung: 25.03.2025, 13:50 MEZ