Blood Tests for Alzheimer’s Are Finally Real. Here Is What They Do — and Don’t — Change.

For thirty years, the defining bottleneck of Alzheimer’s medicine was diagnostic: amyloid plaques could only be confirmed by PET scan or spinal tap, so most patients waited out their cognitive decline undiagnosed while the therapy pipeline stalled behind them. That bottleneck is breaking. In May 2025 the FDA cleared the first blood test intended to aid in Alzheimer’s diagnosis. By late August 2026 there were four FDA-cleared assays, two of them cleared for use in primary care, with one major laboratory chain planning its own rollout in the final quarter of this year. The distance from “specialist instrument” to “a tube of blood at your family doctor” has been crossed.

This is real progress, and it is also the most misreported medical development of the past year. The clearance headlines have already hardened into folk wisdom — “the Alzheimer’s blood test is 97% accurate,” “anyone can get one,” “Medicare covers it” — and each of those sentences is wrong in a way that matters for real patients. Here is what the approvals actually established.

The four cleared tests, in order

Fujirebio’s Lumipulse G pTau217/β-amyloid 1-42 plasma ratio came first, on May 16, 2025 — the first FDA-cleared blood-based diagnostic to aid Alzheimer’s diagnosis, for symptomatic adults 55 and older. In October 2025, Roche’s Elecsys pTau181 followed with a specific and important niche: it is positioned as a rule-out tool in primary care — a negative result helps exclude amyloid pathology, while a positive one still requires further evaluation. Then, within four days in August 2026, C2N’s PrecivityAD2 extended the cleared population down to symptomatic adults as young as 40, and Roche’s Elecsys pTau217 was cleared to support both rule-in and rule-out assessment across primary and specialty care, with common validated cutoffs in both settings.

Read those indications carefully, because they carry the story. Every cleared test is for people who already have cognitive symptoms. None is approved as a screening test for healthy adults, and the FDA said so explicitly when it cleared the first one. A blood test for pathology is not a blood test for destiny — and the gap between those two ideas is where most of the bad reporting lives.

What the accuracy numbers actually measure

The FDA’s evaluation of Lumipulse compared 499 plasma samples from cognitively impaired adults against amyloid PET or spinal-fluid results. Positive blood results agreed with the reference standard 91.7% of the time; negative results agreed 97.3% of the time; fewer than 20% of results were indeterminate. Those two figures are positive and negative predictive values — the probability that a result is right in a symptomatic population — and they are routinely mislabeled as sensitivity and specificity in coverage. They are also specific to the population tested: they tell you how the test performs in people whose doctors already suspect something, not how it would behave in a worried 50-year-old with no symptoms.

The strongest independent evidence comes from Sweden, where a 2024 JAMA study followed 1,213 symptomatic patients in primary and secondary care using mass-spectrometry %pTau217 and a combined amyloid probability score. Overall diagnostic accuracy ran approximately 88% to 92% against cerebrospinal-fluid confirmation — an excellent result for a blood test — but note what that study did not do: it did not validate any of the four now-cleared commercial assays, each of which uses its own platform, ratios and cutoffs. Performance is not portable across platforms, and no single “the Alzheimer’s blood test is X% accurate” number exists.

The cleared tests answer one question: does this symptomatic person’s blood look like amyloid pathology? They do not answer whether a healthy person will develop dementia.

The guideline position — and its limits

The Alzheimer’s Association’s 2025 clinical practice guideline, published last July, gives the field’s most authoritative answer: in specialized care, after a full clinical assessment, blood assays meeting certain accuracy thresholds can be used to triage patients toward confirmatory PET or spinal fluid, or — at a higher accuracy bar — to substitute for those confirmatory tests. The recommendations are conditional and based on low-certainty evidence, and the Association is explicit that interpretation depends on pretest probability and never replaces clinical evaluation. A primary-care regulatory clearance, meanwhile, is not the same as a primary-care screening guideline: no professional body has endorsed routine blood testing of asymptomatic adults, and the evidence that population-level screening would help rather than harm does not yet exist.

Coverage is the next bottleneck, and it is more tangled than “Medicare covers it.” No nationwide Medicare national coverage determination for Alzheimer’s blood biomarkers has been verified as of October 2026 — individual claims may be paid, but there is no guaranteed entitlement. What has moved is commercial coverage: Anthem’s policy effective October 1, 2026 established medical-necessity criteria for a list of the new assays, an important first domino but a single private payer’s policy, not a national one. Published cash-price estimates for Lumipulse run roughly $500 to $1,000, which a patient without coverage pays out of pocket.

What a positive test would actually set in motion

The treatments the tests feed into are approved but modest. Lecanemab (Leqembi), approved in 2023 for MCI or mild dementia with confirmed amyloid, slowed decline on the trial’s clinical scale by 27% relative to placebo over 18 months — an absolute difference of 0.45 points on an 18-point scale. Slower deterioration, not improvement; not a cure. Donanemab (Kisunla), approved July 2024 in the same population, showed a comparable pattern over 76 weeks in 1,736 patients. Both carry a boxed warning for ARIA — brain edema and bleeding abnormalities that can be serious or fatal — with higher risk in APOE ε4 homozygotes, which is why the labels require APOE testing and serial MRI monitoring before and during treatment. A positive blood biomarker does not substitute for any of that.

There are also ways a blood result can be wrong that have nothing to do with the assay’s quality. Kidney disease can elevate the measured biomarkers; obesity can lower them through dilution; certain medications shift amyloid ratios; and amyloid pathology can coexist with Lewy body and vascular disease, meaning a positive result does not even settle the diagnosis. A January 2026 JAMA Viewpoint laid out the downstream harms clinicians are already bracing for: inappropriate diagnoses from false positives, psychological distress, access problems, and the particular cruelty of telling an asymptomatic person they are “amyloid positive” when the evidence says nothing about whether their future includes dementia.

What a reader should take from this

For a patient with real cognitive concerns, the news is genuinely good: a blood draw at the primary-care office can now start a diagnostic workup that used to require a specialist visit and a $5,000 scan, and the Swedish data shows the pathway works. For everyone else, the honest message is the FDA’s own: these are diagnostic aids for symptomatic people, not screening tools, not destiny tests, and not yet something insurers universally pay for.

The blood test broke a thirty-year bottleneck. It did not break the one that matters most — what to do that meaningfully changes the disease. On that front, the slower-damage therapies are the beginning of an answer, and the work of making early detection translate into benefit rather than anxiety is the next decade’s job.

This article is general information, not medical advice. See our Editorial Standards.