Additive value of early-phase β-Amyloid-PET for the differential diagnosis of non-Alzheimer’s disease dementia

TL;DR

Early-phase β-amyloid-PET scans help distinguish non-Alzheimer’s dementia types by detecting neurodegeneration patterns, improving diagnostic accuracy.

Contribution

Demonstrates the additive diagnostic value of early-phase β-amyloid-PET for non-Alzheimer’s dementia beyond amyloid status.

Findings

Visual assessment of early-phase β-amyloid-PET identified neurodegeneration in 78.8% of amyloid-negative cases.

75.3% of A-/N+ cases were correctly assigned to diagnostic categories using visual assessment.

Logistic regression models provided comparable accuracy but were less effective than visual assessment in assigning diagnoses.

Abstract

•Early-phase β-amyloid-PET reliably detects neurodegenerative perfusion patterns.•The visual read identified neurodegeneration in 78.8% of amyloid-negative cases.•75.3% of A-/N+ cases were correctly assigned to diagnostic categories.•Logistic regression models may aid in cases of uncertainty.

Early-phase β-amyloid-PET reliably detects neurodegenerative perfusion patterns.

The visual read identified neurodegeneration in 78.8% of amyloid-negative cases.

75.3% of A-/N+ cases were correctly assigned to diagnostic categories.

Logistic regression models may aid in cases of uncertainty.

Recent studies demonstrated strong agreement between early-phase β-amyloid-PET perfusion imaging and glucose hypometabolism assessed by [18F]FDG-PET, indicating the potential of early-phase β-amyloid-PET as a surrogate biomarker of neuronal injury. We therefore aimed to investigate the additive value of early-phase β-amyloid-PET for the differential diagnosis of non-Alzheimer’s disease dementia syndromes in clinical routine.

[18F]florbetaben- and [18F]flutemetamol-PET scans (n = 379) performed between July 2013 and July 2021 were analyzed for their amyloid status and the presence of a neurodegenerative hypoperfusion pattern using visual assessment and z-score maps. In patients visually rated as amyloid-negative/neurodegeneration-positive (A-N+), the most likely diagnosis based on perfusion patterns was compared to the final clinical diagnosis, i.e. frontotemporal dementia or psychiatric disorders, suspected 4R-tauopathy, and suspected non-Alzheimer pathophysiology. Logistic regression models based on a data-driven selection of cerebral regions of hypoperfusion by principal component analysis were used to predict neurodegenerative disease and clinical diagnoses. Diagnostic accuracy was compared between visual assessment and the regression models.

Neurodegeneration status was correctly identified in 78.8% (119/151) of amyloid-negative patients through visual rating, compared to 67.5% (102/151) using logistic regression. Visual assessment assigned 75.3% (67/89) of A-N+ patients to the correct diagnostic category. In contrast, the regression model classified 69.7% (62/89) of patients.

The current study demonstrates an additive value of early-phase β-amyloid-PET for the differential diagnosis of dementia syndromes. While visual assessment of early-phase β-amyloid-PET already provides substantial diagnostic accuracy, a data-driven analysis approach could aid in cases of uncertainty.

Figures