Summary
A bioRxiv preprint describes a continuous score derived from regional amyloid PET scans that captured early-stage variation and was associated with faster subsequent memory decline in two cohorts. The authors present it as a complement to, rather than a replacement for, global amyloid burden.
A bioRxiv preprint describes a continuous amyloid staging score derived from regional PET scans and reports that it was associated with faster subsequent memory decline in two Alzheimer’s disease research cohorts. The authors say the score may add information to global amyloid burden, particularly during early disease stages when the overall measure changes relatively little.
A regional score alongside global amyloid burden
Amyloid PET uses a radioactive tracer to estimate amyloid deposition in the brain. A global amyloid measure summarises signal across the brain, while regional measurements retain information about how that signal is distributed. The researchers developed an end-to-end deep-learning model to estimate a continuous staging score directly from regional florbetapir PET data.
They trained the model using longitudinal scans from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). The authors report that the score preserved the ordering of repeated visits over time better than the amyloid-staging approaches they evaluated. In early stages, it also captured regional amyloid variation while global amyloid burden remained relatively stable.
The distinction mattered in the cognitive analyses. Among participants with similar global amyloid burden at baseline, higher staging scores were associated with faster subsequent memory decline. The authors report the same pattern in analyses from A4+LEARN, providing a second-cohort replication of the cognitive finding.
What the result adds
The findings suggest that a single brain-wide amyloid measure may not capture all the variation relevant to early disease progression. A continuous score that retains regional information could help researchers distinguish people whose global amyloid levels look similar but whose scan patterns differ. In this study, that distinction was associated with different subsequent memory trajectories.
The score is presented as a complement to global amyloid burden, not as a replacement. The reported findings come from a preprint, and its abstract provides no cohort sizes or numerical effect estimates, limiting assessment of the magnitude and precision of the association. The results support further study of regional amyloid staging as a research measure; the supplied report does not establish its role in routine clinical assessment.