Principal component analysis of FDG PET in amnestic MCI

Flavio Nobili1, Dario Salmaso, Silvia Morbelli

  • 1Clinical Neurophysiology, Department of Endocrinological and Medical Sciences, University of Genoa, Genoa, Italy. flaviomariano.nobili@hsanmartino.it

Abstract

Insights

Combining memory tests and (18)F-FDG PET scans accurately identifies individuals with mild cognitive impairment (MCI) and those progressing to Alzheimer's disease (AD). This combined approach improves diagnostic accuracy for controls, MCI non-converters, and MCI converters.

Area of Science:

  • Neuroimaging
  • Neurology
  • Cognitive Science

Background:

  • Amnestic mild cognitive impairment (aMCI) is a transitional stage between normal aging and Alzheimer's disease (AD).
  • Early and accurate identification of aMCI patients who will convert to AD is crucial for timely intervention.
  • Current diagnostic methods, including memory assessments and neuroimaging, have limitations in predicting conversion.

Purpose of the Study:

  • To evaluate the combined diagnostic accuracy of episodic memory performance and 18F-FDG PET scans.
  • To differentiate between patients with aMCI who convert to AD (MCI/AD), aMCI who do not convert (MCI/MCI), and healthy controls (CTR).

Main Methods:

  • Followed 33 aMCI patients and 15 controls for a mean of 21 months.
  • Utilized 18F-FDG PET with principal component analysis (PCA) to identify patterns of hypometabolism.
  • Performed discriminant analysis using PCA-derived component scores (CCS) and episodic memory test results.

Main Results:

  • PCA identified hypometabolism in specific brain regions (posterior cingulate, temporal pole, orbitofrontal cortex) in both MCI/MCI and MCI/AD groups compared to controls.
  • Distinct patterns of hypometabolism in the posterior cingulate/temporal pole (PC5) and left lateral frontal cortex (PC12) differentiated MCI/AD from MCI/MCI.
  • Combined PET-CCS and memory scores achieved 100% accuracy for CTR, 91% for MCI/MCI, and 82% for MCI/AD.
  • PET improved correct allocations by 25% compared to memory scores alone, and memory scores improved allocations by 15% compared to PET alone.

Conclusions:

  • The combination of episodic memory testing and 18F-FDG PET significantly enhances diagnostic accuracy for CTR, MCI/MCI, and MCI/AD.
  • Specific patterns of hypometabolism, particularly involving the posterior cingulate/temporal pole and left lateral frontal cortex, are key indicators.
  • This multimodal approach offers a more robust method for identifying individuals at risk of AD progression.

Related Concept Videos