In a human diagnostic-accuracy study, plasma p-tau217 tests performed similarly to cerebrospinal-fluid biomarkers for detecting Alzheimer’s pathology, supporting a less invasive route into specialist assessment.
Two blood-based measures—plasma phosphorylated tau 217 (p-tau217) and the ratio of p-tau217 to amyloid-beta 42 (Aβ42)—performed similarly to established cerebrospinal-fluid biomarkers when researchers classified participants by biological Alzheimer’s disease pathology. A report on the head-to-head findings describes comparable performance in patients, strengthening the case for using blood tests to make specialist assessment less invasive and more accessible.
The evidence comes from a peer-reviewed human diagnostic-accuracy study involving clinical and community cohorts. It evaluated how well the plasma markers identified amyloid and tau abnormalities defined by positron emission tomography (PET), as well as a combined amyloid-positive and tau-positive profile assessed against cerebrospinal-fluid, or CSF, biomarkers.
This was not a treatment trial, and the biomarkers did not measure memory, daily functioning or the pace of cognitive decline. They were evaluated as indicators of the protein pathology that biologically defines Alzheimer’s disease—an increasingly important distinction as clinical practice moves toward biomarker-supported diagnosis and pathology-directed treatment evaluation.
The blood and spinal-fluid results were closely aligned
The peer-reviewed study results showed particularly strong discrimination for the p-tau217/Aβ42 ratio. Across the assessed cohorts, its area under the curve, or AUC, was 0.963–0.966 for identifying abnormal amyloid PET and 0.947–0.974 for abnormal tau PET. An AUC of 1 represents perfect separation between reference-positive and reference-negative participants, while 0.5 indicates performance no better than chance.
In the subset with CSF measurements, plasma p-tau217 classified the combined amyloid-positive and tau-positive profile—often abbreviated A+T+—with 91.0% accuracy. Plasma p-tau217/Aβ42 reached 93.2%. The corresponding figures were 90.2% for the CSF Aβ42/Aβ40 ratio and 93.2% for CSF p-tau181/Aβ42. The reported differences between the blood and CSF measures were not statistically significant.
Predictive values help translate those headline accuracy figures into the study setting. For A+T+ status, p-tau217 had a positive predictive value of 83.6% and a negative predictive value of 97.3%. The p-tau217/Aβ42 ratio had a positive predictive value of 89.2% and a negative predictive value of 96.2%.
Within these cohorts, then, a negative result was especially effective at identifying people who did not have the combined biomarker profile. Positive results were also informative, although a proportion of those classified as positive by blood testing were not A+T+ according to the study’s reference definition.
What the biomarkers capture
P-tau217 is a form of tau protein phosphorylated at the amino acid threonine 217. Its concentration in blood is associated with the amyloid and tau processes characteristic of Alzheimer’s disease. Aβ42 is a form of amyloid-beta that contributes a second signal; expressing p-tau217 relative to Aβ42 can sharpen the distinction between people with and without the target pathology.
The tests do not directly image protein deposits in the brain. Instead, they measure circulating molecules associated with those processes. PET provides a spatial image of amyloid or tau deposition, while CSF obtained by lumbar puncture measures biomarkers in fluid surrounding the central nervous system. Blood sampling is operationally simpler and less invasive, which could allow biomarker assessment to occur earlier in a diagnostic pathway or in settings where PET and lumbar puncture are difficult to provide.
The ratio was not uniformly superior on every reported measure. Its accuracy for A+T+ status was higher than p-tau217 alone in the CSF subset, and its positive predictive value was also higher, but both plasma tests had high negative predictive values. The relevant choice may therefore depend on the intended clinical role, assay availability and threshold selected.
From cohort accuracy to clinical thresholds
Results from a controlled diagnostic study do not create a universal cutoff for every laboratory or patient population. Assay platforms can produce different numerical values, while differences in age, clinical setting and prevalence of Alzheimer’s pathology affect how a test performs in practice. Reference standards also matter: agreement with amyloid PET, tau PET or a combined CSF profile answers related but not identical questions.
A 2026 systematic review of plasma p-tau217 reported pooled sensitivity of 85.4% and specificity of 88.0% for biomarker-defined Alzheimer’s disease. Sensitivity is the proportion of reference-positive cases detected by the blood test; specificity is the proportion of reference-negative cases correctly classified. The review also found heterogeneity in assays, populations, disease definitions and cutoff selection.
That broader evidence supports p-tau217 as a strong biomarker while showing why the exceptionally high discrimination seen in individual cohorts should not be treated as a fixed performance guarantee. AUC values indicate how well a test separates groups across possible thresholds, but clinical use requires choosing a particular threshold and accepting the resulting balance between false-positive and false-negative results.
Positive and negative predictive values also change with the prevalence of pathology in the tested population. A threshold used in a specialist memory service, where Alzheimer’s pathology is relatively common, may not have the same positive predictive value in a lower-prevalence community setting. Laboratories and health systems therefore need validation data tied to the specific assay, population and intended use.
A potential first step in a staged assessment
The most immediate role for these blood biomarkers may be as part of a staged diagnostic pathway rather than as an automatic replacement for PET or lumbar puncture. In a population resembling the study cohorts, a result below a validated threshold could help identify people unlikely to have the combined amyloid-and-tau profile. A result above the threshold could prioritize further specialist evaluation or confirmatory testing when the decision carries substantial consequences.
Blood testing may be particularly useful where access to PET is constrained or where lumbar puncture creates a practical barrier. It could also help clinicians decide more efficiently who is most likely to benefit from further biomarker assessment. Whether confirmation is needed would depend on the assay, the clinical context and the purpose of testing.
An abnormal biomarker result is not, by itself, a complete clinical diagnosis or an explanation for every cognitive symptom. Its value lies in supplying biological evidence that can be interpreted alongside clinical history and cognitive assessment. The head-to-head findings suggest that p-tau217-based blood tests can carry much of the diagnostic information provided by established CSF ratios, potentially moving Alzheimer’s pathology assessment toward a more accessible first-line test.