Multicenter Study Validates Blood Test Accuracy for Detecting Alzheimer’s Pathology

A prospective multicenter investigation conducted across seven memory clinics in Catalonia, Spain, has demonstrated that measuring phosphorylated tau 217 in blood plasma with a fully automated immunoassay reliably identifies brain amyloid pathology in individuals presenting with cognitive impairment. Evaluating participants undergoing routine diagnostic workups for mild cognitive impairment or mild dementia, researchers found that plasma phosphorylated tau 217, abbreviated as p-tau217, distinguished between individuals with and without cerebrospinal fluid-defined amyloid-beta pathology with high accuracy across both specialized academic medical centers and community-based regional hospitals. The findings provide direct empirical support for emerging clinical diagnostic frameworks that incorporate blood-based biomarkers into clinical pathways when assay accuracy exceeds ninety percent, indicating that scalable blood testing could substantially decrease reliance on invasive lumbar punctures or costly positron emission tomography scans in routine memory assessment workflows.

Biological frameworks define Alzheimer’s disease through the in vivo detection of cerebral amyloidosis and neurofibrillary tau tangles, conventionally assessed via cerebrospinal fluid assays of amyloid-beta 42, amyloid-beta 40, and phosphorylated tau 181, or through molecular positron emission tomography neuroimaging. While these established biological markers exhibit excellent diagnostic precision, their widespread clinical adoption remains constrained by high financial costs, limited geographical availability, procedural invasiveness, and contraindications. Consequently, diagnostic misclassification persists across primary care and community memory clinics, where only a small minority of symptomatic patients currently undergo definitive biomarker verification. Blood-based biomarkers, particularly circulating species of tau phosphorylated at threonine 217, have emerged as minimally invasive alternatives capable of capturing core Alzheimer’s pathophysiology. Prior investigations demonstrated that plasma p-tau217 outperforms other candidate blood markers, such as neurofilament light chain, glial fibrillary acidic protein, and plasma amyloid-beta ratios, yet most published data originated from single-center cohorts or research-enriched registries.

To evaluate whether automated blood biomarker testing performs consistently across varied routine clinical environments, investigators recruited consecutive patients between June 2023 and February 2024 from seven memory clinics within Catalonia. The participating clinical sites comprised one tertiary academic medical center with established experience in biomarker validation, the Hospital Clínic de Barcelona, alongside six non-tertiary community and regional centers: Hospital Universitari Santa Maria de Lleida, Hospital de Figueres, Hospital de la Santa Creu de Vic, Hospital Moisés Broggi–Consorci Sanitari Integral, Hospital Asil de Granollers, and Hospital Sant Jaume de Calella. Eligible participants presented with objective cognitive impairment classified clinically as mild cognitive impairment or mild dementia, and all underwent standardized clinical examinations, comprehensive neuropsychological testing, structural neuroimaging, and lumbar puncture as part of standard diagnostic care. Clinical classifications followed international criteria for Alzheimer’s disease, dementia with Lewy bodies, frontotemporal dementia variants, or suspected non-neurodegenerative cognitive impairment characterized by stable cognitive deficits with normal cerebrospinal fluid biomarkers.

Cerebrospinal fluid samples were gathered via lumbar puncture according to established clinical guidelines, and concentrations of amyloid-beta 1-42, amyloid-beta 1-40, total tau, and phosphorylated tau 181 were quantified locally across centers on the fully automated Lumipulse platform using chemiluminescent enzyme immunoassays. Amyloid status was established on the basis of validated center-specific cutoffs, employing the ratio of amyloid-beta 42 to amyloid-beta 40 or concordant single-analyte thresholds according to routine site workflows. Concurrently, peripheral blood samples were drawn into standardized tubes at enrollment, centrifuged at 1800 times gravity for ten minutes at room temperature within two hours of collection, aliquoted into polypropylene vials, and stored at minus eighty degrees Celsius. The frozen plasma specimens were subsequently shipped to a centralized analytical laboratory at the Hospital Clínic de Barcelona, where p-tau217 concentrations were determined on the automated Lumipulse G system following standardized manufacturer operating protocols.

The initial study enrollment comprised 212 individuals, but 27 participants were subsequently excluded from the primary statistical analysis. One exclusion resulted from insufficient plasma volume, while 26 cases were excluded because of extreme biomarker outliers attributable to site-specific pre-analytical anomalies. Specifically, 21 out of 23 participants from a single center exhibited uniformly elevated plasma p-tau217 concentrations ranging from 0.931 to 7.216 picograms per milliliter, which deviated markedly from the remainder of the cohort. Although the remaining two participants from that center demonstrated plausible concentrations of 0.204 and 0.519 picograms per milliliter, all samples from the site were excluded as a precautionary measure. Additionally, three individuals from two other centers showed values at the assay’s absolute lower or upper limits of detection, measuring 0.03 and 10 picograms per milliliter, and were likewise removed to ensure analytical integrity.

The final analyzed cohort consisted of 185 participants with a mean age of 71.5 years and a standard deviation of 5.9 years, of whom 52.4 percent were female and 66.5 percent were classified as amyloid-positive based on cerebrospinal fluid criteria. The clinical diagnostic spectrum encompassed 118 patients with Alzheimer’s disease, 50 with suspected non-neurodegenerative cognitive impairment, 12 with frontotemporal dementia, and five with dementia with Lewy bodies. Amyloid positivity was present in 25 percent of the frontotemporal dementia cohort and 40 percent of the dementia with Lewy bodies cohort. Amyloid-positive individuals exhibited significantly lower mean Mini-Mental State Examination scores of 23.39 compared to 24.78 in amyloid-negative individuals, as well as significantly lower body mass index values averaging 25.99 kilograms per square meter versus 28.55 kilograms per square meter, while age, onset age, sex, and renal function showed no significant differences.

Receiver operating characteristic curve analysis across the entire cohort demonstrated outstanding diagnostic discrimination for plasma p-tau217 in detecting cerebrospinal fluid-defined amyloid status, yielding an area under the curve of 0.916 with a 95 percent confidence interval between 0.875 and 0.958. Applying the Youden index identified an optimal single cutoff value of 0.336 picograms per milliliter, which yielded a diagnostic sensitivity of 82.9 percent, a specificity of 88.7 percent, and an overall diagnostic accuracy of 84.9 percent. Diagnostic performance demonstrated high consistency between clinical environments: the area under the curve was 0.916 at the tertiary academic center and 0.919 across the six non-tertiary regional centers, showing no statistically significant difference by DeLong’s test. Applying the global threshold of 0.336 picograms per milliliter resulted in comparable diagnostic accuracy of 86.9 percent at the tertiary site and 83.2 percent across the peripheral clinics.

To refine clinical decision-making, the investigators modeled two predefined dual-threshold categorization algorithms designed to divide patients into low, intermediate, or high probability of harboring amyloid pathology. Under a lenient dual-threshold strategy calibrated to achieve 95 percent sensitivity and specificity, cutoffs of 0.207 and 0.397 picograms per milliliter yielded 95.1 percent sensitivity and 95.2 percent specificity, placing 24.3 percent of patients in the low-probability group and 49.7 percent in the high-probability group. The remaining 25.9 percent fell into an indeterminate intermediate zone, achieving an overall diagnostic accuracy of 93.4 percent among conclusive classifications and sparing 74.1 percent of patients from requiring confirmatory cerebrospinal fluid or imaging evaluations. A stricter model targeting 97.5 percent sensitivity and specificity utilized thresholds of 0.146 and 0.486 picograms per milliliter, increasing conclusive accuracy to 95.3 percent but expanding the intermediate zone to 42.7 percent of the cohort, thereby sparing 57.3 percent from invasive procedures. Simulating an analytical variability margin of plus or minus ten percent in plasma measurements retained high accuracy at 93.4 percent.

Analysis of plasma biomarker concentrations across cerebrospinal fluid amyloid and tau biomarker combinations revealed progressive biological gradients. Participants were stratified into four classic profiles: amyloid-negative and tau-negative (56 patients), amyloid-negative and tau-positive (six patients), amyloid-positive and tau-negative (29 patients), and amyloid-positive and tau-positive (94 patients). Plasma p-tau217 concentrations differed significantly across these groups, with the highest concentrations observed in the doubly positive group, which was significantly higher than all other subgroups. Concentrations in the amyloid-positive but tau-negative group were significantly higher than those in doubly negative individuals, whereas no significant difference emerged between amyloid-negative tau-positive cases and doubly negative controls. When classified by dual-threshold models, the majority of doubly positive cases were assigned to the high-probability category, while doubly negative individuals concentrated in the low-probability group. Amyloid-positive tau-negative participants predominantly occupied the intermediate range under strict criteria, consistent with an early transitional phase where soluble phosphorylated tau rises prior to widespread neocortical tau deposition.

Multivariate linear regression modeling demonstrated that lower body mass index and reduced estimated glomerular filtration rate were independently associated with higher baseline plasma p-tau217 concentrations after controlling for age, sex, and amyloid status. However, comparing logistic regression models predicting amyloid status revealed that adjusting plasma p-tau217 for body mass index, renal function, or both yielded areas under the curve between 0.931 and 0.938, which did not significantly improve diagnostic performance over unadjusted plasma p-tau217 alone. The authors concluded that routine demographic or physiological adjustments are unnecessary in routine memory clinic workflows, although renal impairment warrants consideration when measurements fall near decision boundaries. Noted study limitations included the absence of direct amyloid or tau positron emission tomography, lack of postmortem neuropathological verification, a modest non-Alzheimer’s sample size, and a homogeneous European clinic population. The investigators emphasized that establishing standardized pre-analytical sample handling, centralized automated testing, and dual-threshold algorithms provides an equitable and practical framework for expanding biological Alzheimer’s diagnostics.

Subject of Research: Diagnostic accuracy of automated plasma p-tau217 for detecting brain amyloid pathology across multicenter memory clinics

Article Title: Predictive Ability of Plasma p‐tau217 for β‐Amyloid Status: A Prospective Multicenter Study

Article References: Massons, M., Guillen, N., Sarto, J., Falgàs, N., Borrego‐Écija, S., Esteller‐Gauxax, D., Castellví, M., Tort‐Merino, A., Pérez‐Millan, A., Antonell, A., Augè Fradera, J. M., Piñol, G., Riba, I., Carnes‐Vendrell, A., Cullell, M., Osuna, M. T., Bajo, L., Romero, T., Bonjoch, E., … Lladó, A. (2026). Predictive Ability of Plasma p‐tau217 for β‐Amyloid Status: A Prospective Multicenter Study. Annals of Clinical and Translational Neurology, 13(11), 2268-2277. https://doi.org/10.1002/acn3.70387

Image Credits: AI Generated

DOI: 10.1002/acn3.70387

Keywords: plasma p-tau217, Alzheimer's disease, biomarkers, cerebrospinal fluid, amyloid pathology, chemiluminescent enzyme immunoassay, Lumipulse, dual-threshold algorithm, mild cognitive impairment, memory clinics, diagnostic accuracy, receiver operating characteristic, body mass index, estimated glomerular filtration rate, tauopathy, pre-analytical variability, clinical validation