Proposed section structure
For this topic, a clinically appropriate structure is: Highlights; Clinical background and unmet need; Study design and cohorts; Biomarker rationale; Key results; Clinical interpretation and translational relevance; Strengths and limitations; Conclusion; Funding and trial registration; References.
Highlights
First, a plasma model combining %p-tau217 and eMTBR-tau243 showed high concordance with amyloid and tau PET-based biological staging in both derivation and external validation cohorts, with a C index of 0.91 in each cohort.
Second, the combined model outperformed a strategy based on %p-tau217 alone, particularly for identifying the intermediate biological stage corresponding to A+TMOD+, a clinically important zone in which disease progression is biologically established but not yet maximally advanced.
Third, plasma staging correlated with clinical severity across the Alzheimer disease continuum and aligned strongly with autopsy-confirmed Alzheimer neuropathology in a validation subsample, with an area under the curve of 0.96.
Fourth, the study supports the possibility that minimally invasive blood testing could extend biological staging beyond specialized research centers, with implications for diagnosis, prognosis, and selection for disease-modifying therapy and clinical trials.
Clinical background and unmet need
Alzheimer disease is increasingly defined biologically rather than solely by clinical syndrome. This shift reflects the now well-established dissociation between symptoms and pathology: some cognitively unimpaired individuals already harbor substantial amyloid and tau pathology, whereas some patients with cognitive impairment have non-Alzheimer causes of neurodegeneration. Biological staging can therefore improve diagnostic precision, prognostic counseling, and therapeutic triage.
At present, however, the most robust staging approaches rely on amyloid and tau positron emission tomography or cerebrospinal fluid testing. These methods remain constrained by cost, limited access, radiation exposure in the case of PET, and the need for specialized infrastructure or invasive lumbar puncture. As anti-amyloid and, potentially, future anti-tau therapies enter practice, there is growing demand for scalable biomarker strategies that can identify where a patient lies along the Alzheimer continuum. A blood-based staging model that approximates PET-defined biology would therefore fill an important translational gap.
This study by Salvadó and colleagues addresses that need by testing whether 2 plasma tau-related analytes, phosphorylated tau at threonine 217 expressed relative to its unphosphorylated form (%p-tau217) and endogenously cleaved microtubule-binding region tau containing residue 243 (eMTBR-tau243), can together reproduce a PET-based staging framework. Conceptually, the approach is attractive because the 2 markers may capture different components of disease biology: %p-tau217 is strongly linked to early Alzheimer-type tau phosphorylation and amyloid-associated tau dysregulation, while eMTBR-tau243 may better reflect tau species associated with neurofibrillary tangle burden and later-stage tau pathology.
Study design and cohorts
This was an observational longitudinal study using the BioFINDER-2 cohort as the main derivation cohort and the Knight Alzheimer Disease Research Center cohort as an independent external validation cohort. Data acquisition in BioFINDER-2 spanned November 2019 to January 2025, and in Knight ADRC from September 2007 to March 2020. Analyses were conducted between December 2024 and July 2025.
The main BioFINDER-2 cohort included 872 participants across a broad clinical spectrum: 383 cognitively unimpaired participants, 182 with mild cognitive impairment, 151 with Alzheimer disease dementia, and 156 with non-Alzheimer neurodegenerative diseases. Mean age was 72.8 years, 50.2% were women, and 59.2% were APOE-E4 carriers. The validation cohort included 156 participants from Knight ADRC. A neuropathology subsample from Knight ADRC also enabled comparison against autopsy-confirmed Alzheimer disease neuropathologic change.
The exposure variables were plasma eMTBR-tau243 and %p-tau217. The principal outcome was concordance with PET-based biological stages defined according to revised Alzheimer’s Association criteria. Additional outcomes included relationship to clinical stage and longitudinal biomarker progression. The investigators also compared the 2-marker model against a simpler model using %p-tau217 alone, an important practical comparator because single-marker blood tests are often more attractive from an implementation standpoint.
Biomarker rationale
The biology underlying this model deserves emphasis. %p-tau217 is not simply an absolute concentration but a ratio of phosphorylated tau217 to its unphosphorylated counterpart. This design may reduce interindividual variability related to total analyte abundance and sharpen the signal linked specifically to pathologic phosphorylation. Prior work has shown p-tau217-based measures to be among the strongest plasma indicators of Alzheimer pathology.
eMTBR-tau243 represents a tau fragment derived from the microtubule-binding region, a domain closely connected to aggregated tau species. Because neurofibrillary tangle progression is central to disease staging and symptom emergence, adding eMTBR-tau243 may improve discrimination among stages where tau burden is increasing beyond the earliest amyloid-associated phase. The study’s main hypothesis was therefore mechanistically coherent: a marker sensitive to early Alzheimer-type tau phosphorylation plus a marker related to tangle-associated tau fragments might together map more faithfully onto PET-based biological staging than either alone.
Key findings
Concordance with PET-based biological staging
The central result was the high concordance between the plasma model and the PET-based staging system. In BioFINDER-2, the 2-marker model achieved a C index of 0.91 with a 95% confidence interval of 0.90 to 0.92. That level of agreement is notable for a blood-based model intended to approximate imaging-defined biology across multiple disease stages rather than simply classify amyloid positivity. External validation was equally strong: in Knight ADRC, the C index was also 0.91, with a 95% confidence interval of 0.87 to 0.94. Reproducibility across an independent cohort materially strengthens confidence that the findings are not merely derivation-set overfitting.
Association with clinical stage
The plasma staging model also correlated with clinical severity across the Alzheimer continuum. In the derivation cohort, concordance with clinical stage reached a C index of 0.84 (95% CI, 0.82-0.86), and in the validation cohort a C index of 0.86 (95% CI, 0.82-0.89). These values are lower than concordance with PET, which is expected because clinical phenotype is influenced by cognitive reserve, comorbidity, mixed pathology, and non-neurodegenerative factors. Even so, the results indicate that the plasma stages are not merely laboratory abstractions; they track meaningful differences in disease expression.
Incremental value beyond %p-tau217 alone
A particularly important result is that the combined model performed better than a staging strategy based only on %p-tau217. The advantage was most evident in the intermediate or C stage, designated A+TMOD+. This is clinically relevant because intermediate biological stages may be the hardest to classify yet among the most actionable for prognosis and treatment planning. A marker panel that improves separation in this zone could help distinguish patients who have moved beyond isolated amyloid pathology and are entering a more tau-driven phase of disease.
This point matters in practice. A single-marker blood test may be suitable for screening, but staging often requires more granularity than screening alone. The present findings suggest that eMTBR-tau243 supplies complementary information rather than redundant signal. That raises the possibility of a tiered workflow in which an initial plasma test identifies likely Alzheimer biology and a more refined dual-marker model assigns stage.
Neuropathological validation
In the neuropathology subsample, plasma staging aligned strongly with autopsy-confirmed Alzheimer pathology using the Alzheimer Disease Neuropathologic Change scale, yielding an area under the curve of 0.96 (95% CI, 0.91-1.00). This is one of the most compelling aspects of the study. PET is an excellent in vivo reference standard, but neuropathology remains the ultimate benchmark. Strong autopsy concordance supports the biological validity of the plasma model and reduces concern that the model is only learning PET tracer behavior rather than underlying disease.
Longitudinal implications
The authors also report that the plasma model correlated with longitudinal biomarker progression. Although detailed temporal estimates are not available in the abstract, this suggests the staging model may have value not only for cross-sectional classification but also for disease monitoring. If confirmed, that would expand clinical utility substantially, especially for tracking progression in observational cohorts and enriching clinical trials with participants likely to worsen over a measurable interval.
Clinical interpretation and translational relevance
The study is timely because Alzheimer care is moving toward biology-informed treatment pathways. For anti-amyloid therapy, and eventually combination or tau-directed strategies, clinicians need accessible tools to determine whether a patient has Alzheimer pathology, how advanced that pathology is, and whether the biological stage aligns with expected benefit-risk tradeoffs. PET is not available at sufficient scale to support broad population-level staging. A plasma method with high agreement to PET could therefore serve as a front-line diagnostic and stratification tool.
Several use cases emerge. In memory clinics, a blood-based model could help prioritize who most needs confirmatory imaging or cerebrospinal fluid testing. In health systems with limited PET access, it could reduce delays in diagnosis. In trials, plasma staging could improve screening efficiency and enrich for participants at specific biological phases, which is especially important when therapeutic targets differ by stage. For counseling, biological staging may also sharpen prognosis, although prognostic performance would need more explicit study-level validation than provided in the abstract.
The work also illustrates a broader principle in biomarker development: no single analyte may be sufficient for the full complexity of Alzheimer staging. Amyloid, tau phosphorylation, tangle-associated tau fragments, neurodegeneration, and clinical symptoms evolve on related but nonidentical trajectories. A compact multi-analyte model may therefore strike the best balance between biological fidelity and operational simplicity.
Strengths and limitations
The main strengths are the large derivation cohort, independent external validation, inclusion of participants ranging from cognitively unimpaired through dementia, representation of non-Alzheimer neurodegenerative diseases, and the presence of a neuropathological validation subsample. The direct comparison with a simpler %p-tau217-only model is another strength because it demonstrates incremental value rather than reporting isolated performance metrics.
Important limitations remain. First, this was an observational study, so clinical utility in routine care is inferred rather than proven. Whether plasma staging changes management, improves patient outcomes, or reduces diagnostic cost requires prospective implementation studies. Second, the abstract does not provide detailed information on assay standardization, preanalytical variability, or laboratory interoperability, all of which are critical barriers to real-world adoption. Third, cohort-based validation, though strong, does not automatically establish performance in community populations with greater comorbidity, racial and ethnic heterogeneity, or mixed pathologies. Fourth, the exact thresholds and calibration procedures used to assign stages are not described in the abstract, limiting immediate portability.
There is also a conceptual caution. High concordance with PET does not eliminate the need for confirmatory testing in all cases, especially when treatment decisions carry substantial cost or risk. Rather, the most realistic near-term role is as an accessible first-line staging approach embedded within a multimodal diagnostic pathway.
Conclusion
This study provides persuasive evidence that a plasma model combining %p-tau217 and eMTBR-tau243 can approximate PET-based biological staging of Alzheimer disease with high accuracy, maintain performance in an independent cohort, correlate with clinical stage, and align with neuropathology. The added value of eMTBR-tau243 over %p-tau217 alone is particularly important because it improves classification of intermediate disease biology, where staging can be most challenging and clinically informative.
If assay standardization and prospective implementation studies confirm these findings, this dual-marker strategy could become a practical bridge between research-grade biological staging and everyday clinical care. In an era of biomarker-driven diagnosis and emerging disease-modifying treatment, that is a meaningful advance.
Funding and trial registration
The abstract does not report funding details or a ClinicalTrials.gov registration number. Readers should consult the full JAMA Neurology publication for complete disclosures, funding sources, and any additional protocol information.
References
Salvadó G, Horie K, Barthélemy NR, Schindler SE, Janelidze S, Orduña Dolado A, Bali D, Perrin RJ, Morris JC, Benzinger TLS, Gordon BA, Stomrud E, Mattsson-Carlgren N, Palmqvist S, Vogel JW, Bateman RJ, Ossenkoppele R, Hansson O. Plasma eMTBR-tau243 and %p-tau217 for Biological Staging of Alzheimer Disease. JAMA Neurology. 2026-05-26. PMID: 42189519. https://pubmed.ncbi.nlm.nih.gov/42189519/
