Blood biomarkers for Alzheimer's pathology have moved from research tool to clinical proposition, and the question now is which assay and which analyte. This meta-analysis pooled 25 diagnostic accuracy studies and 18,073 participants, all using amyloid PET as the reference standard, and stratified by the technical platform used to measure plasma phosphorylated tau 217.
All platforms performed respectably but not identically. Single-molecule array gave sensitivity 0.894 (95% CI 0.861 to 0.920) and specificity 0.875 (0.845 to 0.900). Chemiluminescence immunoassay was similar at 0.886 and 0.865. Electrochemiluminescence had the highest sensitivity at 0.933 (0.918 to 0.946) but the lowest specificity at 0.791, with a very wide interval of 0.571 to 0.915. Mass spectrometry inverted that: sensitivity 0.827 with the highest specificity, 0.938, again very imprecisely estimated at 0.692 to 0.990. Heterogeneity between studies was substantial throughout, with I-squared between 63% and 86%.
The more actionable finding is that the ratio of p-tau217 to amyloid-beta 42 outperformed p-tau217 alone, roughly doubling the diagnostic odds ratio (relative DOR 1.96, 95% CI 1.19 to 3.23), with about 3.6% better sensitivity on average. If a laboratory can report the ratio, it should.
Where this lands clinically is as triage rather than diagnosis. Sensitivity and specificity near 0.88 mean roughly one in eight results will mislead, which is acceptable for deciding who needs a PET or a lumbar puncture and not acceptable for telling someone they have Alzheimer's disease. For Indian practice, where amyloid PET is available in very few centres and rarely funded, a blood test that reliably identifies who is worth referring is more valuable than it is in systems where PET is routine. The condition is knowing which platform your laboratory uses and what threshold it applies, because these results are not interchangeable between assays.
- Use plasma p-tau217 to decide who needs confirmatory amyloid PET or cerebrospinal fluid testing, not to make the diagnosis.
- Ask your laboratory to report the p-tau217 to amyloid-beta 42 ratio where the platform allows it.
- Find out which platform your laboratory uses; sensitivity and specificity differ meaningfully between them.
- Do not compare results across assays or apply a threshold from one platform to another.
- At around 88% accuracy, roughly one in eight results will be wrong — say so when consenting the patient.
The statistics, in plain English
Sensitivity is how often the test catches people who truly have amyloid pathology; specificity is how often it correctly clears those who do not. High sensitivity with lower specificity, as with electrochemiluminescence, means few missed cases but more false alarms — appropriate for a screening step feeding into a confirmatory test. The heterogeneity figures matter here: an I-squared of 63% to 86% means the studies disagreed substantially, so these pooled numbers describe an average across differing populations and thresholds rather than what you would get in your own clinic. The very wide intervals on the mass spectrometry and electrochemiluminescence specificity estimates reflect few contributing studies.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for neurology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free