Improved quantification of amyloid burden and associated biomarker cut-off points: results from the first amyloid Singaporean cohort with overlapping cerebrovascular disease

Abstract

Purpose: The analysis of the [¹¹C]PiB-PET amyloid images of a unique Asian cohort of 186 participants featuring overlapping vascular diseases raised the question about the validity of current standards for amyloid quantification under abnormal conditions. In this work, we implemented a novel pipeline for improved amyloid PET quantification of this atypical cohort.

Methods: The investigated data correction and amyloid quantification methods included motion correction, standardized uptake value ratio (SUVr) quantification using the parcellated MRI (standard method) and SUVr quantification without MRI. We introduced a novel amyloid analysis method yielding 2 biomarkers: Aβ_L which quantifies the global Aβ burden and ns that characterizes the non-specific uptake. Cut-off points were first determined using visual assessment as ground truth and then using unsupervised classification techniques.

Results: Subject's motion impacts the accuracy of the measurement outcome but has however a limited effect on the visual rating and cut-off point determination. SUVr computation can be reliably performed for all the subjects without MRI parcellation while, when required, the parcellation failed or was of mediocre quality in 10% of the cases. The novel biomarker Aβ_L showed an association increase of 29.5% with the cognitive tests and increased effect size between positive and negative scans compared with SUVr. ns was found sensitive to cerebral microbleeds, white matter hyperintensity, volume, and age. The cut-off points for SUVr using parcellated MRI, SUVr without parcellation, and Aβ_L were 1.56, 1.39, and 25.5. Finally, k-means produced valid cut-off points without the requirement of visual assessment.

Conclusion: The optimal processing for the amyloid quantification of this atypical cohort allows the quantification of all the subjects, producing SUVr values and two novel biomarkers: Aβ_L, showing important increased in their association with various cognitive tests, and ns, a parameter sensitive to non-specific retention variations caused by age and cerebrovascular diseases.

Keywords: Alzheimer’s disease; Amyloid PET; Biomarkers; Cerebrovascular disease; Cut-off point; Quantification.