Older adults with no cognitive symptoms whose baseline plasma p-tau217 fell in the very high category had an estimated 38 percent absolute risk (95 percent CI, 33 percent to 43 percent) of progressing to cognitive impairment within five years, and about 78 percent within ten, according to a longitudinal cohort study by Rachel F. Buckley of Mass General Brigham and colleagues, published online in JAMA on 14 July 2026 and presented the following day at the Alzheimer's Association International Conference in London. Participants in the high but not very high category had an estimated 24 percent five-year risk (95 percent CI, 20 percent to 28 percent). Over two years, estimated risk was low across every category, in the range of 1 percent to 4 percent.
The analysis harmonised data from 2,684 cognitively unimpaired adults across six observational and clinical trial cohorts in North America, Japan and Australia (median age 69.6 years, 63 percent female), with the earliest enrolment in 2004 and the most recent follow-up in 2025. There were 478 events of progression, defined as mild cognitive impairment, dementia, or two consecutive global Clinical Dementia Rating scores of 0.5 or higher, over a median follow-up of 5.4 years and a maximum of 13.5 years. Each one standard deviation increase in baseline p-tau217 was associated with a 38 percent higher rate of progression (hazard ratio 1.38; 95 percent CI, 1.30 to 1.46), and the association held after adjustment that included amyloid PET Centiloids (hazard ratio 1.32; 95 percent CI, 1.24 to 1.41). This is an observational association, not proof that the protein causes decline.
The question the paper answers is narrower than the one families are likely to ask. Dementias Today has previously reported on plasma p-tau217 being tested outside Alzheimer's disease, including in Parkinson's disease, and on assay performance and regulatory clearance. Here the marker is being asked to forecast, in people who feel well. The Alzheimer's Association clinical practice guideline on blood-based biomarkers, published in Alzheimer's and Dementia in July 2025, applies only to people with objective cognitive impairment in specialised care and explicitly does not extend to cognitively unimpaired individuals. The first FDA-cleared plasma p-tau217 test, cleared in May 2025, is likewise intended for patients being evaluated for cognitive decline, not for screening. The authors themselves wrote that "these findings should not be taken as support for routine p-tau217 screening in cognitively unimpaired older adults".
The estimates are model based and cohort dependent. The six cohorts differed in recruitment, assay platform and how outcomes were decided, and were not demographically representative; the category cut-offs are expressed in standard deviations from a cohort-specific reference rather than a single laboratory number a reader could look up. An earlier preprint of the same analysis, posted in February 2026 with 2,705 participants, gave the ten-year figure for the very high group as 61 percent (95 percent CI, 53 percent to 68 percent), which shows how much these numbers can move with modelling choices. The published paper calls for validation in unselected populations before the estimates inform individual prognosis.