BIOANALYSISPAPER
A study of 2,117 people found 44% to 78% of those with borderline results progressed to amyloid positive over five years, reshaping how a single blood draw could steer primary care and anti amyloid therapy selection.
A five-year study of 2,117 people finds a single blood draw flags amyloid risk as reliably as a PET brain scan. The cutoff itself does something else: most people sitting near it cross into amyloid-positive territory over the next five years, while people clearly below it mostly do not.
The study, published this month in Alzheimer's & Dementia, tracked plasma levels of a protein called phosphorylated tau 217, or p-tau217, across multiple cohorts including the Alzheimer's Disease Neuroimaging Initiative (ADNI). A single baseline measurement classified people as amyloid-PET-positive with 86% to 95% accuracy, and that accuracy held for up to five years of follow-up. That is the performance range that gets compared to brain scans, which cost roughly five to ten times more and require a visit to a specialized imaging center (paper).
The researchers split participants into three groups based on their baseline p-tau217 reading. People who started above the cutoff stayed there 93% to 98% of the time, and 90% to 100% of them had amyloid abnormalities on follow-up imaging. People who started clearly below the cutoff stayed there most of the time too, and only 4% to 21% of them crossed into positive territory over five years.
The intermediate zone, people whose first reading sat close to the threshold, behaved like neither group. Between 44% and 78% of them crossed into amyloid-positive status over the follow-up window, depending on the cohort, and 64% to 96% of those converters showed amyloid abnormalities on later scans. They also accumulated tau faster, lost hippocampal volume faster, and reached a dementia diagnosis more often than people whose baseline reading was clearly negative.
That intermediate zone is where the test stops being a yes-or-no answer and starts being a clinical signal. The reading carries prognostic information that a single threshold cannot capture, which matters for two specific use cases: deciding who is a reasonable candidate for anti-amyloid antibody therapy, and choosing participants for clinical trials of those same drugs. Both depend on knowing whether someone is likely to accumulate amyloid in the near term, not just whether they have crossed the line today.
The work was led by a team at Shenzhen Bay Laboratory, with co-authors from the University of Gothenburg, Sun Yat-sen University, Westlake University, and Daping Hospital, and it draws on ADNI's well-characterized reference dataset as one of several cohorts (NCBI abstract).
The caveats matter. This is a risk-stratification tool, not a diagnostic test: a stable-positive result does not mean a person will develop dementia, and a stable-negative result does not mean they will not. The study measures how p-tau217 readings track amyloid biology over time, not whether knowing the result changes what a doctor can do for a patient. That distinction, between the test being technically validated and being clinically useful, depends on access to follow-up, to counseling, and to disease-modifying therapy, all of which are uneven.
The next milestone to watch is whether primary-care pilots using p-tau217 for screening start publishing real-world performance numbers in the next twelve months. The lab numbers are strong. The primary-care numbers decide whether a blood draw becomes a routine part of cognitive risk conversations or stays a research instrument.