Automated Blood Test Detects Alzheimer's Pathology with 93–96% Accuracy
A fully automated plasma p-tau217 test accurately identifies Alzheimer's pathology across 1,767 patients in primary and secondary care settings.
Summary
Researchers validated a fully automated, commercially available blood test measuring plasma phospho-tau217 (p-tau217) using the Lumipulse immunoassay across 1,767 cognitively symptomatic patients in five centers across Sweden, Spain, and Italy. Using predefined cutoffs, the test detected Alzheimer's pathology—defined by abnormal CSF Aβ42:p-tau181—with areas under the ROC curve of 0.93–0.96. Accuracy ranged from 89–91% in secondary care and 85% in primary care. A two-cutoff approach raised accuracy to 92–94% by flagging 12–17% of results as intermediate. The test performed consistently across sex, APOE genotype, diabetes, kidney disease, and cognitive stage, though accuracy dipped in those aged 80 or older.
Detailed Summary
Alzheimer's disease (AD) affects tens of millions globally, yet diagnosis often relies on costly, invasive procedures like cerebrospinal fluid (CSF) analysis or PET imaging. Scalable, accurate blood-based biomarkers could transform how AD is detected—especially in primary care where most patients first present. This large multicenter study evaluated a fully automated plasma phospho-tau217 (p-tau217) immunoassay (Lumipulse, Fujirebio) using predefined biomarker cutoffs, a critical step toward real-world clinical deployment.
The study enrolled 1,767 participants with cognitive symptoms from four secondary care cohorts (Malmö, Sweden n=337; Gothenburg, Sweden n=165; Barcelona, Spain n=487; Brescia, Italy n=230) and one primary care cohort in Sweden (n=548). The primary outcome was AD pathology defined as abnormal CSF Aβ42:p-tau181 ratio, a well-established surrogate for amyloid and tau pathology. Plasma p-tau217 was measured using the fully automated Lumipulse platform, with a subset also tested using a high-performing mass-spectrometry-based percentage p-tau217 assay (C2N Diagnostics) for head-to-head comparison.
Plasma p-tau217 on the Lumipulse platform demonstrated AUROCs of 0.93–0.96 across cohorts. In secondary care, overall accuracy was 89–91%, with positive predictive values (PPVs) of 89–95% and negative predictive values (NPVs) of 77–90%. In primary care, accuracy was 85%, PPV 82%, and NPV 88%. Notably, performance was robust across sex, APOE genotype, chronic kidney disease, diabetes, and cognitive stage. The only subgroup showing reduced accuracy was participants aged ≥80 years (83%), likely due to more mixed pathologies and atypical tau profiles in that population.
A two-cutoff strategy—reporting results as positive, negative, or intermediate—improved accuracy to 92–94% in both secondary and primary care by excluding 12–17% of cases with ambiguous results for further confirmatory testing. Adding plasma Aβ42 to create a p-tau217:Aβ42 ratio did not improve overall accuracy but did reduce the proportion of intermediate results to ≤10%. The mass-spectrometry-based percentage p-tau217 test showed comparable accuracy to Lumipulse in secondary care but outperformed it in primary care and was unaffected by patient age—suggesting complementary roles for these assays depending on clinical context.
These findings are particularly significant given that the Lumipulse platform is already commercially available, standardized, and operable in routine clinical laboratories without specialized expertise. Using predefined cutoffs rather than cohort-derived thresholds strengthens the external validity of the results. The study sets a high bar for what a scalable AD blood test can achieve and provides a roadmap for implementation across diverse healthcare settings, though intermediate-result protocols will need to be integrated into clinical workflows.
Key Findings
- Lumipulse plasma p-tau217 detected Alzheimer's pathology with AUROCs of 0.93–0.96 across all cohorts.
- Accuracy was 89–91% in secondary care and 85% in primary care using predefined cutoffs.
- A two-cutoff approach raised accuracy to 92–94% by flagging 12–17% of results as intermediate.
- Performance was unaffected by sex, APOE genotype, kidney disease, diabetes, or cognitive stage.
- Age ≥80 years reduced accuracy to 83%; mass-spectrometry p-tau217 was unaffected by age.
Methodology
Multicenter observational validation study across 5 cohorts (n=1,767) with cognitive symptoms in Sweden, Spain, and Italy. Plasma p-tau217 was measured using the fully automated Lumipulse immunoassay with predefined cutoffs; the primary reference standard was CSF Aβ42:p-tau181 ratio. A subset underwent mass-spectrometry-based percentage p-tau217 for comparative analysis.
Study Limitations
Performance was reduced in adults aged ≥80, a clinically important group with higher prevalence of mixed pathologies. The reference standard (CSF Aβ42:p-tau181) is a surrogate for amyloid/tau pathology rather than autopsy-confirmed AD. Generalizability beyond European cohorts and to non-symptomatic populations remains to be established.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.