Mayblyum · Neurology 2021 · prospective cohort study · n=131

Comparing PET and MRI Biomarkers Predicting Cognitive Decline in Preclinical Alzheimer Disease.

Cited 32 times in the scientific literature.

Level 3 - non-randomized controlled study

Prospective cohort prognostic study evaluating imaging biomarkers over longitudinal follow-up

PubMed 33952655 · doi:10.1212/WNL.0000000000012108 · record verified 2026-08-27

What was done

Data from 131 clinically normal older adults in the prospective Harvard Aging Brain Study were analyzed. Participants underwent structural MRI, fluorodeoxyglucose (FDG) PET, flortaucipir (FTP) PET, and Pittsburgh compound B (PiB) PET within a 1-year window, alongside prospective cognitive testing using the Preclinical Alzheimer's Cognitive Composite over a mean 3-year follow-up. Analyses focused on predefined regions of interest (inferior temporal, isthmus cingulate, hippocampus, and entorhinal cortex) using linear mixed-effect models adjusted for demographics, backward stepwise regression, and simulated clinical trial power curves.

What was found

In 131 participants (52 women, mean age 73.98 ± 8.29 years), most biomarkers showed stronger associations with cognitive decline in high-PiB versus low-PiB participants. Stepwise regression containing all candidate biomarkers identified only neocortical PiB, entorhinal FTP, and entorhinal FDG as independent predictors of subsequent cognitive decline. Power simulations indicated that enriching trial enrollment with both high PiB and low entorhinal FDG reduced required participant numbers threefold compared to using high PiB alone.

Why it matters

Entorhinal FDG-PET hypometabolism provides independent prognostic information beyond amyloid and tau imaging in preclinical Alzheimer disease. Combining amyloid and metabolic biomarker criteria can markedly increase statistical power and lower sample size requirements for secondary prevention trials.

Limits

The study is limited by a modest sample size (n = 131) from a single cohort and a relatively brief follow-up period (mean 3 years) for detecting slow preclinical decline. Trial efficiency improvements rely on statistical simulations rather than actual trial validation, and imaging analyses were restricted to predefined regions of interest.

Cited by