Skip to content

Palmqvist S, et al. Blood biomarkers to detect Alzheimer disease in primary care and secondary care. JAMA. 2024;332:1245-1257. PMID 39068545

One-paragraph summary

Between February 2020 and January 2024, 1,213 patients undergoing clinical evaluation for cognitive symptoms in Sweden were tested with a mass-spectrometry plasma assay measuring the ratio of phosphorylated tau 217 to non-p-tau217 (%p-tau217), alone and combined with the plasma Aβ42:Aβ40 ratio (the amyloid probability score 2, APS2). Cutoffs established in an independent cohort were applied without modification to a primary care cohort (n=307) and a secondary care cohort (n=300) analysed as single batches, and then prospectively to a further primary care cohort (n=208) and secondary care cohort (n=398) with samples sent for analysis within two weeks of collection. The reference standard was AD pathology defined by abnormal CSF Aβ42:Aβ40 and p-tau217. Mean age was 74.2 years; 23% had subjective cognitive decline, 44% MCI and 33% dementia; AD pathology was present in 50% of both care settings. APS2 achieved AUC 0.96–0.97 with positive and negative predictive values of 87–92% across all four cohorts and diagnostic accuracy of 88–92%. Primary care physicians achieved 61% diagnostic accuracy (95% CI 53–69) for clinical AD after clinical examination, cognitive testing and CT, versus 91% (86–96) using APS2; dementia specialists achieved 73% (68–79) versus 91% (88–95). In the overall population, APS2 (90%) and %p-tau217 alone (90%) performed identically (PMID 39068545).

Key findings

  • Predefined, externally derived cutoffs — the design feature that distinguishes this study from most of the plasma-biomarker literature, ~90% of which used data-derived thresholds (Therriault 2025, PMID 40818474).
  • Prospective, real-time analysis reproduced the batched result (AUC 0.96–0.97).
  • Primary care physician accuracy 61% versus 91% with the blood test; specialist accuracy 73% versus 91%.
  • %p-tau217 alone was as accurate as the two-analyte APS2.

Limitations

  • Single country, single laboratory, single assay platform; transportability to other laboratories and assays is untested, and a head-to-head comparison found immunoassays performing at 0.83–0.88 accuracy versus 0.93 for mass-spectrometry %p-tau217 (Warmenhoven 2025, PMID 39468767).
  • The reference standard is CSF, not autopsy.
  • Participants had cognitive symptoms; performance in asymptomatic people — the population any screening programme would test — is not addressed.
  • Prevalence of AD pathology was 50% in both settings, which is high for unselected primary care and inflates predictive values relative to a lower-prevalence population.
  • The study measures diagnostic accuracy, not whether testing improves patient outcomes.

Why it matters

This is the study that moved plasma p-tau217 from "promising biomarker" to "test that outperforms the clinicians currently making the diagnosis," and it did so with the methodological feature the field most lacked — thresholds fixed in advance. It is the empirical basis for the 2025 Alzheimer's Association clinical practice guideline permitting blood tests meeting ≥90% sensitivity and ≥90% specificity to substitute for amyloid PET or CSF in specialised care (Palmqvist 2025, PMID 40729527), and it is the reason the diagnostic bottleneck in Alzheimer's disease is shifting from measurement to capacity: diagnosis is still delayed by 31–44 months on average and differentially by ethnicity (Lin 2021, PMID 34091580).

Cited by wiki pages

  • fluid biomarkers
  • guidelines
  • overview