Blood tests move from amyloid detection to tau staging

Clinical takeaway: Tau burden is becoming measurable in blood, but this belongs in research and trial screening for now. Imaging remains the standard where tau load will determine treatment eligibility.

Blood testing has changed how Alzheimer disease gets confirmed, but not how far along it is when it gets identified. That gap matters more now that anti-amyloid therapy is in use, because patients with a low to intermediate tau load appear to gain more from treatment than those whose tangles are already widespread. Tau PET can answer the question. But it is costly, unevenly distributed, and impractical as a routine test. And p-tau217, the marker behind that shift, tracks plaques better than tangles.

Once a patient is known to be amyloid-positive, the useful question becomes how many tangles have formed. p-tau217 answers that only roughly; is an established Alzheimer disease biomarker measured most often in blood plasma. Blood values from patients with limited tangles overlap with values from patients whose tangles are extensive, often too much for the test to tell them apart. That is precisely the population in which staging comes up.

Measured alongside six other blood proteins, p-tau217 identified widespread tangles more accurately than it did on its own. On a 0-to-1 scale measuring how cleanly a test separates two groups, the seven-protein panel scored 0.92 against 0.86 for p-tau217 alone in the first of two patient groups, and 0.94 against 0.88 in the second. The second group was analyzed independently, using the same proteins selected in the first.

The improvement came from catching more cases. At its best cutoff in the second group, the panel correctly flagged 92% of patients with widespread tangles, compared with 66% for p-tau217 alone. It was slightly less precise in the other direction, correctly clearing 85% of patients with limited tangles versus 95%. Fewer patients overall landed in the middle range where a result is too uncertain to act on. p-tau217 still did most of the work. It was by a wide margin the strongest single contributor to the model, and the 6 added proteins sharpened it rather than replaced it.

Researchers measured 125 blood proteins in 560 adults with confirmed amyloid pathology, all of whom had also had a tau PET scan, drawn from two cohort studies: BioFINDER in Sweden and TRIAD in Montreal. Participants ranged from cognitively unimpaired to dementia. Widespread tangles meant tau PET uptake in the Braak V and VI regions. Modeling narrowed the 125 candidates to p-tau217 plus 6 others, which reflect glial activity, synaptic loss, and inflammation, all thought to contribute to tangle formation.

Both cohorts were small, and each used a different tau PET tracer without a shared imaging pipeline, which may be why the proteins selected in one group did not fully match the other.

Treatment staging may be where blood biomarkers are headed. The therapies now in use are sensitive to tangle burden, which makes the amount of pathology a more useful number than simply detecting its presence. A blood test that delivers has the potential to reach far more patients than PET.

Source: Di Molfetta G, et al. (2026 Aug 10) JAMA Neurol. Plasma Biomarkers for Neocortical Tau Burden