Biomarker-First Alzheimer Diagnosis Arrives: The August FDA Clearance Wave in Blood Tests and Tau PET

Bottom Line: In an 11-day span in August 2026, the FDA approved the tau PET tracer Tauklarify (florquinitau F 18), cleared C2N's PrecivityAD2 blood test for symptomatic adults as young as 40, and cleared Roche's Elecsys pTau217 — the first single-biomarker blood test with both rule-in and rule-out claims. Together these actions make a biomarker-first workup practical for symptomatic patients: plasma p-tau217 as front-line triage, CSF or amyloid PET reserved for indeterminate results (a fraction that varies by assay), and tau PET as a staging tool that calibrates anti-amyloid treatment expectations. None of it licenses testing asymptomatic patients, and blood results still require clinical correlation — and, in many programs, confirmatory testing — before treatment.

What Happened in August

Three regulatory actions landed in rapid succession, and the terminology matters. On August 14, 2026, Lantheus announced FDA approval of Tauklarify (florquinitau F 18, formerly MK-6240) — a drug approval for a radiopharmaceutical. The tracer is indicated for PET imaging in adults with cognitive impairment who are being evaluated for Alzheimer disease (AD), to identify tau neurofibrillary tangle pathology, supported by two blinded-read studies of PET images from more than 500 subjects across three trials. Its label carries an explicit limitation of use: safety and effectiveness have not been established for non-AD tauopathies.

On August 20, the FDA cleared C2N Diagnostics' PrecivityAD2, a mass-spectrometry test combining plasma Aβ42/40 with %p-tau217 (the ratio of phosphorylated to non-phosphorylated tau217) into a single Amyloid Probability Score 2 (APS2). It is the first AD blood test cleared for symptomatic adults as young as 40 — a meaningful extension for early-onset workups.

On August 24, the FDA cleared Roche's Elecsys pTau217, developed with Eli Lilly: the first single-biomarker blood test cleared to support both rule-in and rule-out assessment of amyloid pathology. It is intended for adults 55 and older with signs, symptoms, or complaints of cognitive decline associated with AD, reports three result bands (positive, intermediate, negative) using identical cutoffs in primary and specialty care, and runs on the more than 4,500 cobas analyzers already installed in US clinical laboratories. Roche's own language deserves emphasis: "Elecsys pTau217 is not intended to be used as a stand-alone test. Results should be interpreted in conjunction with other diagnostic tools and clinical information."

For context, this brings the count of FDA-cleared AD blood tests to four — Fujirebio's Lumipulse pTau217/Aβ42 ratio was first, in May 2025 — but Roche's is the first cleared stand-alone p-tau217 measurement, and C2N's now reaches the youngest cleared population.

Why Blood First: The Performance Data Behind the Architecture

The three-band (rule-in / intermediate / rule-out) architecture that now defines cleared blood testing was validated most rigorously in the automated p-tau217 platform study by Palmqvist and colleagues (Nat Med 2025), which evaluated the Lumipulse assay in 1,767 symptomatic patients across five cohorts in Sweden, Spain, and Italy — including 548 primary-care patients. Plasma p-tau217 detected AD pathology with an AUC of 0.93–0.96. A single cutoff (>0.27 pg/mL) achieved 89–91% accuracy in secondary care and 85% in primary care. The pivotal finding was the two-cutoff approach (<0.22 and >0.34 pg/mL): accuracy rose to 92–94% in both settings, at the cost of classifying 12–17% of patients as intermediate. This is the design principle — not the device data — behind the cleared tests' result bands. Palmqvist validates the Fujirebio assay; neither its cutoffs nor its 12–17% intermediate rate can be transposed onto the Elecsys or Precivity platforms, each of which stands on its own validation with its own gray-zone fraction.

For Elecsys pTau217, the published cohort evidence comes from a prototype version of the assay evaluated as a pre-screening tool (Hibar et al., Alz Dement 2026) in 2,148 participants (538 cognitively impaired, 1,610 cognitively unimpaired) drawn from the A4, SKYLINE, AIBL, CREAD, and CREAD2 cohorts. Overall discrimination was AUC 0.878 in cognitively impaired and 0.907 in cognitively unimpaired participants. At the pre-screening rule-out cutoff of <0.189 pg/mL, sensitivity was 98.98% with specificity 29.17% in cognitively impaired participants (NPV 92.51%), and sensitivity 95.54% with specificity 50.72% in cognitively unimpaired participants (NPV 98.60%). Those operating characteristics are exactly what a rule-out threshold should look like — nearly no amyloid-positive patient slips below it, at the deliberate cost of specificity. But the FDA-cleared test's cutoffs, result-band boundaries, and intermediate rate have not been published in press materials; the Hibar figures describe the platform's prototype performance, not the cleared device labeling, and should be quoted as such.

PrecivityAD2's clinical validation reported sensitivity of 88% and specificity of 89% for amyloid status (Meyer et al., Alz Dement 2024). A subsequent validation in an independent cohort of 191 cognitively impaired participants (134 MCI, 57 dementia; amyloid PET positivity prevalence 56%) applied the pre-specified APS2 cut point prospectively and found accuracy of 91% (95% CI 86–94%), sensitivity 90% (95% CI 83–94%), and specificity 92% (95% CI 84–96%) (npj Dementia 2025). Notably, C2N's FDA-clearance validation reports a "Likely Positive" middle category — assigned to 17.3% of subjects — which is a different construct from the Lumipulse intermediate zone and a reminder that each platform's indeterminate band must be understood on its own terms. One caveat of provenance: although the npj Dementia cohort was independent, the paper was authored by C2N-affiliated investigators — the numbers are useful, but readers should know this is manufacturer-conducted validation, not third-party replication.

The economic argument is not trivial. Palmqvist and colleagues modeled cost savings of up to 60% versus CSF-based workups and up to 81% versus amyloid PET, assuming 14.7% of patients fall into the intermediate zone and require confirmatory testing — again, a Lumipulse-based model, but directionally instructive for any blood-first pathway.

First Principle: Objective Impairment and Pretest Probability

Before any algorithm, one discipline must hold: these are tests for symptomatic patients in whom AD is a genuine clinical question. The cleared intended-use populations are anchored to cognitive symptoms — and treatment eligibility for anti-amyloid therapy still generally requires early symptomatic AD (MCI or mild dementia) plus amyloid pathology, not a laboratory value alone. Documenting objective impairment, or at minimum a credible clinical syndrome, before ordering keeps the pretest probability in the range where the published predictive values apply. In low-prevalence populations — worried-well patients, or younger patients in whom amyloid prevalence is substantially lower — positive predictive value falls, and the same numeric result carries less meaning. This is precisely where the PrecivityAD2 age-40 clearance demands the most restraint: it serves genuine early-onset workups and autosomal-dominant families, not reassurance-seeking in the 40s.

The Indeterminate Zone: When You Still Need CSF or Amyloid PET

Every three-band design deliberately trades a gray zone for accuracy, and every local protocol must specify what happens to those patients: CSF Aβ42/p-tau or amyloid PET as tie-breakers. How large that gray zone is depends on the assay — 12–17% in the Palmqvist Lumipulse cohorts, 17.3% "Likely Positive" in C2N's clearance validation, and not yet publicly reported for the cleared Elecsys device — so ask your laboratory for its observed rate rather than assuming a universal figure. Two age-related findings from Palmqvist sharpen test selection. First, single-cutoff accuracy fell to 83% in patients ≥80 years versus 91% in those <73 years (P=0.003) — a decline resolved by the two-cutoff approach. Second, mass-spectrometry %p-tau217 was more accurate than the automated immunoassay in primary care (90% vs 85%, P=0.003) and was age-insensitive — a concrete reason to consider the mass-spec platform for very elderly patients or primary-care-originated referrals. Palmqvist also found performance unaffected by chronic kidney disease, diabetes, sex, APOE genotype, or cognitive stage — reassuring, but assay-specific to Lumipulse; renal function and medical comorbidity remain sensible flags for cautious interpretation until each cleared assay publishes equivalent subgroup data.

Tau PET's Distinct Job: Staging, Not Screening

Tauklarify answers a different question than amyloid status: where the patient is in the disease. The anchor is TRAILBLAZER-ALZ 2 (JAMA 2023), which stratified enrollment by baseline tau PET. In the low/medium-tau population, donanemab slowed iADRS decline by 35.1% (difference 3.25 points; 95% CI 1.88–4.62; P<.001) and CDR-SB decline by 36.0% (−0.67; 95% CI −0.95 to −0.40; P<.001). In the combined population including high-tau participants, the effect was smaller — 22.3% iADRS slowing (difference 2.92; 95% CI 1.51–4.33; P<.001). The careful reading: higher tau burden was associated with smaller expected average benefit in the trial's analyses. The high-tau-only iADRS comparison was a post hoc analysis and was not statistically significant, while CDR-SB still favored donanemab — so the data do not establish that high-tau patients derive no benefit or should be denied treatment. What they do support is that tau stage informs the expected magnitude of benefit, which is exactly the conversation to have before committing a patient to 18 months of infusions.

Practical uses follow: prognostic counseling, adjudicating atypical presentations, and informing borderline treatment decisions. Equally important is what the label forbids by omission — Tauklarify is not validated for non-AD tauopathies and should not be used to sort PSP, CBD, or FTD. Access will be uneven: F-18 chemistry allows regional distribution, but tau PET capacity remains far more limited than the amyloid PET footprint, and availability should be confirmed locally.

The Sequencing Algorithm: A Pragmatic Triage Pathway

What follows is a clinical synthesis, not a cleared or guideline-endorsed protocol: the FDA actions validate each test for use alongside clinical assessment, but no regulator has validated this — or any — system-wide sequencing algorithm. With that caveat, for a symptomatic patient with objective cognitive impairment and a clinical picture in which AD is plausible:

Step 1 — Plasma p-tau217-based testing

Elecsys pTau217 for patients ≥55; PrecivityAD2 for patients 40–54 or where mass-spec %p-tau217 is preferred (very elderly patients, primary-care settings, per the Palmqvist subgroup data above).

Step 2 — Negative result

Amyloid pathology is very unlikely — in the Elecsys platform's prototype cohort study, the rule-out cutoff carried NPVs of 92.51% in cognitively impaired and 98.60% in unimpaired participants (Hibar, Alz Dement 2026). Redirect the workup toward non-AD causes: vascular disease, Lewy body disease, depression, sleep, medications, FTD.

Step 3 — Positive result

In a specialist's hands with a consistent phenotype, a positive result may support moving into the anti-amyloid eligibility pathway — but confirmatory CSF or amyloid PET may still be required by local protocol, payer, or treatment program, and label parity with CSF/PET for treatment qualification is not established. Eligibility screening itself is unchanged: MRI for microhemorrhages and APOE genotyping, because risk stratification depends on it. In Clarity AD, ARIA-E occurred in 12.6% of lecanemab-treated patients overall but 32.6% of APOE ε4 homozygotes (NEJM 2023). In TRAILBLAZER-ALZ 2, ARIA-E occurred in 24.0% of donanemab patients (6.1% symptomatic), with three treatment-related deaths (JAMA 2023). TRAILBLAZER-ALZ 6 showed a modified titration schedule cut ARIA-E to 13.7% versus 23.7% at 24 weeks (Alz Dement 2025) — mitigation strategies now belong in the eligibility conversation.

Step 4 — Indeterminate result (rate is assay-specific)

CSF Aβ42/p-tau or amyloid PET. In the Palmqvist Lumipulse cohorts this was roughly one patient in seven; your assay's rate may differ, and your protocol should be built around the locally observed figure.

Step 5 — Tau PET where staging changes management

Tauklarify — in patients with cognitive impairment under evaluation for AD — when expected magnitude of benefit, atypical or mixed presentations, or borderline treatment candidacy would alter the plan.

What the tests do not license: screening asymptomatic patients. The p-tau217 clock models (Nat Med 2026; Knight ADRC n=258, ADNI n=345) show why the temptation exists — plasma %p-tau217 predicted symptom onset with a median absolute error of 3.0–3.7 years, and the interval from biomarker positivity to symptoms was 20.5 years for those positive at age 60 versus 11.4 years at age 80. Scientifically remarkable; clinically, the authors themselves note that testing unimpaired individuals is not recommended outside research.

Caveats and Open Questions

Clearance is not label parity with CSF or PET for treatment eligibility; payer policies and appropriate-use recommendations are still catching up, and local treatment programs may reasonably continue to require confirmatory testing. Validation cohorts skew non-Hispanic White, and generalizability is under active study. The four cleared blood platforms use different analytes, methods, cutoffs, and indeterminate-band constructs — they are not interchangeable, and a patient should not be tracked longitudinally across assays. Finally, confirm which test your laboratory actually runs, its reported indeterminate rate, and its turnaround — the pathway above is only as good as its local implementation.

What the General Neurologist Needs to Know

• A negative FDA-cleared plasma p-tau217 test in a symptomatic patient ≥55 makes amyloid pathology very unlikely — published prototype/platform data for the Elecsys assay reported NPVs of 92.51–98.60% at its rule-out cutoff (the cleared device's own cutoffs are not yet published) — redirect the workup.
• A positive result in the right clinical context may support entry into the anti-amyloid pathway, but confirmatory CSF/PET may still be required by protocol or payer; reserve those tools routinely for indeterminate results — roughly one in seven in the Lumipulse validation cohorts, with each assay's rate its own.
• Symptomatic patients as young as 40 now have a cleared blood-based entry point (PrecivityAD2) — apply strict pretest-probability discipline in young patients, where lower amyloid prevalence erodes PPV.
• Tau PET is a staging and prognosis tool, not an amyloid substitute: in TRAILBLAZER-ALZ 2, higher tau burden was associated with smaller expected average benefit in trial analyses, and Tauklarify is not validated for non-AD tauopathies.
• None of these clearances licenses testing asymptomatic patients — the p-tau217 clocks are a research instrument, not a screening program.

Conclusion

August 2026 did not end the era of CSF and PET; it reordered it. For symptomatic patients, a well-chosen plasma p-tau217 test is now a defensible first move — cheap enough to deploy broadly, accurate enough to rule out amyloid pathology with confidence, and structured, through the three-band design, to be honest about its own uncertainty. The traditional confirmatory tools shift to the roles they perform best: adjudicating each assay's indeterminate zone and, with tau PET now approved, staging disease to inform what anti-amyloid therapy can realistically offer. The neurologists who benefit most will be those who respect the boundaries — objective impairment before ordering, clinical correlation always, local confirmation requirements and locally observed indeterminate rates honored, and no screening of the asymptomatic. Framed as a pragmatic triage pathway rather than a new universal standard of care, the biomarker-first workup has, at last, arrived.