On August 24, the FDA cleared the fourth blood-based biomarker test for Alzheimer's disease. This one matters more than the count suggests.
It's the first single-biomarker plasma assay validated to both rule in and rule out amyloid pathology using the same cutoffs in primary care that apply in specialty neurology, and it runs on laboratory infrastructure that already exists in thousands of sites across the country.
For anyone who has spent time in this field, that's a genuine milestone. Confirming amyloid pathology has meant a PET scan or a lumbar puncture, both expensive and invasive, both concentrated in specialty centers, and both often preceded by months on a neurology waitlist. A blood draw ordered during a routine visit takes down these barriers.
But I've watched the reaction to this news over the past few weeks, and a particular misreading keeps surfacing: that a blood test now answers the diagnostic question. It does answer the question about whether they have the biomarker associated with Alzheimer’s Disease, but it does not by itself, constitute a diagnosis. A definitive diagnosis of Alzheimer’s disease requires a cognitive assessment that goes beyond the pathology because clinical impairment is defined cognitively, not biochemically.
A blood-based biomarker test answers a biological question. Is amyloid pathology present, yes or no. In a similar way, having elevated cholesterol does not equate to coronary artery disease, and elevated HbA1c (below 6.4%) does not mean one has diabetes, only increased risk for it.
These markers are valuable, and increasingly reliable. But they do not tell you whether the person sitting in the exam room is cognitively impaired. It doesn't tell you how impaired they are, whether that impairment is progressing, how quickly, if it's affecting daily routines and function,or what should happen next. Those are cognitive and functional questions, and they require objective cognitive and functional measurement.
The FDA clearance language recognizes this fact and is explicit: “results of blood testing are to be interpreted alongside clinical information and other relevant findings as part of the diagnostic pathway.” That isn’t regulatory boilerplate language. It's a description of a successful workflow which most primary care practices have not yet built.
This distinction matters clinically because many cognitively normal older adults have amyloid pathology and will never develop dementia in their remaining lifetime. Pathology is a risk state. Impairment is a clinical state. Treating the first as though it were the second produces two failures, and I'd argue both are serious: telling a healthy person they have a disease they don't have, and stopping the workup on a person with real decline because a single marker came back below threshold.
|
A biomarker tells you what's happening in the brain's chemistry. It doesn't tell you what's happening in a patient's life. |
This isn't my interpretation of an open question. It's the consensus position of the field.
The Alzheimer’s Association International Conference (AAIC) 2026 was the strongest signal yet that blood biomarkers are entering clinical practice, and the message across plenaries, abstracts, and press releases was consistent. Biomarkers improve accuracy, scalability, and access. Cognitive and functional assessment remain essential for diagnosis, staging, and tracking impairment. The conference positioned biomarkers as a complement to cognitive evaluation, not its replacement.
The peer-reviewed literature says the same thing, and it's worth walking through what those studies actually found.
Recent JAMA published work on plasma p-tau217 pooled data across studies like the A4/LEARN, HABS, ADNI, WRAP, and HABS-HD cohorts. The biomarker predicted progression to MCI and dementia more readily when it was paired with longitudinal cognitive composites and clinical staging. In a companion analysis, even where p-tau217 outperformed amyloid PET and APOE genotype as a predictor, progression itself was defined by cognitive and functional endpoints rather than by biomarker thresholds. The biomarker forecasts. Cognition adjudicates.
Across those same cohorts, two findings recur that I think deserve more attention than they've received. Cognitive composites detect subtle decline before the point of clinical impairment. And combined biomarker-plus-cognition models consistently outperform biomarker-only models, because the two data streams are answering different questions and the combination carries more information than either one alone.
Then there's the study that gets cited most often in support of blood testing, and I want to be careful about how it's read. A Swedish primary care study reported diagnostic accuracy rising from roughly 65% to roughly 93% when a blood test was added. That's a striking improvement. But cognitive testing and clinical history remained mandatory components of the workflow that produced it. The 93% is not the accuracy of a blood test. It's the accuracy of a pathway that includes one.
For years, cognitive assessment in primary care was treated as a low-priority task, and honestly, the reasoning behind that was sound. Providers were being asked to screen for a condition with no clear therapeutic target and no defined next step. A positive screen generated work, uncertainty, and a difficult conversation with no good answer at the end of it.
That calculus has changed twice over. There are now two FDA-approved disease-modifying therapies with narrow eligibility windows defined at the MCI and early-stage range. And there are blood tests available in the primary care setting to interrogate the biology behind them.
Which means cognitive assessment moves from optional to obligatory, and it does three distinct jobs in the pathway that's emerging.
It gates the test. The cleared indication covers individuals 55 and older presenting with signs, symptoms, or complaints of cognitive decline. Somebody has to establish that those signs exist. Ordering plasma biomarkers across an unselected panel of older adults isn't what the assay was validated for, and pretest probability drives the predictive value of whatever result comes back.
It interprets the result. A positive biomarker in a cognitively intact patient means something categorically different from a positive biomarker in a patient with measurable decline in delayed recall. Same lab value, two different clinical situations, two different next steps. Without a cognitive measurement, a provider holding a positive result has a finding with no patient attached to it.
It tracks what happens next. Staging, treatment eligibility, treatment response, and care plan adjustment are all cognitive and functional determinations made over time. A blood test isn't designed to be repeated every six months to answer the question families are actually asking, which is whether this is getting worse and what to do about it. Care plans have to be adjusted based on a person’s evolving cognitive status and cognitive tests, not blood tests, tell us what that status is.
If cognitive assessment is the gate, interpreter, and tracker, then the sensitivity of that assessment sets the ceiling on the whole pathway. Pairing an insensitive screener with a highly accurate blood test doesn't produce an accurate workflow. It produces a workflow that loses patients before the blood test is ever ordered.
And that's a real risk, because the tools most widely in use weren't built for the range where these decisions now sit. With paper-based cognitive assessment workflows, up to 61% of people living with dementia remain undiagnosed, and the vast majority of MCI cases go undetected in routine primary care. The traditional screening instruments most clinicians have available were designed to confirm impairment that is already clinically apparent, not to catch subtle decline in the band between normal aging and early MCI. Many individuals with amnestic MCI score within the normal range on the MMSE despite documented decline. That's a structural property of the instrument, not an administration error.
Sensitivity also depends on measuring the right thing. Cognition isn't one quantity that moves up or down. It's a set of integrated systems and disease doesn't touch them equally. Alzheimer's has a well documented early anatomy in the medial temporal lobe structures that support episodic and associative memory, which is why the search for preclinical cognitive markers has concentrated there for decades. A battery that measures those functions weakly, or only as one thin component of a broad composite, will detect the earliest signal weakly. Breadth of coverage is an input. Sensitivity is an output. The two are not the same thing.
This is also where digital administration earns its place, and not for the reason people usually cite. Speed and automatic scoring are real benefits, but they're the least interesting ones. What matters is that a digital assessment captures the process alongside the score: response latency, hesitation, how performance shifts across trials. Timing features from digitally administered tasks have shown associations with amyloid and tau burden in preclinical Alzheimer's disease, in patients whose final scores looked unremarkable. Standardized delivery also makes a result from one clinic comparable to a result from another, and a result today comparable to a result eighteen months from now. For a progressive disease now being evaluated against a biological marker, that feature isn't a convenience. It's a requirement.
The arrival of blood biomarkers has been read in some quarters as a threat to cognitive assessment. I'd argue the evidence points firmly in the other direction.
As biomarker testing scales, more people will be identified as at risk. Every one of them will need their cognition measured, staged, tracked over time, and connected to care. That's not a smaller job than the one we had before. It's a considerably larger one, and it's arriving faster than most health systems are prepared for.
Multimodal assessment, meaning biologic markers, cognitive function and clinical history together, is becoming the standard of care. Neurotrack's Cognitive Care Enablement Platform is built for the cognitive pillar of that standard: sensitive enough to detect decline at the MCI stage, fast enough to fit inside a routine visit, integrated with the EHR, and connected to the care pathways and longitudinal monitoring that give a detection its clinical meaning.
The direction of the field isn't away from cognitive measurement. It's toward it. Biology and cognition are being asked the same clinical question from two sides, and the answers only mean something together. That's what we've spent more than a decade building and I’m encouraged that we’re moving fast toward better, earlier detection.
Ready to build the cognitive pillar of your diagnostic pathway?Neurotrack's Cognitive Care Enablement Platform gives health systems and Medicare Advantage plans the tools to detect cognitive decline at the MCI stage, document with clinical precision, and activate the care pathways that make a biomarker result actionable.
|