In May 2026, early Alzheimer’s diagnosis took a concrete step forward: several blood tests received CE marking in Europe under the IVDR regulation. That approval allows them to run on automated platforms in clinical laboratories, at a scale that wasn’t possible before. As recently as 2024, blood biomarkers were still almost exclusively research tools. «Biomarkers can now genuinely enter the specialist diagnostic pathway», says Massimo Filippi, full professor of Neurology at Vita-Salute San Raffaele University (UniSR) and director of the Alzheimer Centre at the IRCCS San Raffaele Hospital. What’s still unclear is which tests actually work, for whom, and with what limitations.
Why a Timely Diagnosis Matters
An early Alzheimer’s diagnosis makes it possible to start treating symptoms sooner, correct modifiable risk factors and plan follow-up care. «Getting a diagnosis in the early stages matters more than it did just a few years ago», notes Filippi. Therapies that act on the disease’s underlying biological mechanisms are now authorised in the European Union for patients with symptomatic early-stage Alzheimer’s, and their effectiveness is greatest within that early window. Filippi has previously explained why acting as early as possible makes a difference.
Early Diagnosis and Blood Tests: What Has Changed
Until recently, blood biomarkers hadn’t yet entered clinical practice, held back by standardisation issues and limited availability. Clinical and neuropsychological assessment, together with brain MRI, remained the starting point for diagnosis. Confirming the disease’s underlying biology still meant relying on cerebrospinal fluid biomarkers or an amyloid PET scan.
Over the past two years, the evidence on plasma biomarkers has finally consolidated. «Among these, p-tau217 is currently the one that has shown the best performance for identifying Alzheimer’s pathology, with an accuracy that can come close to CSF biomarkers and PET results when properly validated tests are used», says Filippi.
The first clinical guidelines dedicated to blood biomarkers arrived in 2025. In patients with cognitive impairment assessed in a specialist setting, the most accurate blood tests can be used for triage. When sensitivity and specificity are very high, they can even serve as diagnostic confirmation, reducing the need for more invasive procedures.
The regulatory shift happened quickly. «In May 2025, the US FDA authorised the first blood test intended to support the diagnosis of Alzheimer’s disease, based on the combination of p-tau217 and beta-amyloid 1-42. In Europe, the most recent step came in May 2026, with CE marking granted under the IVDR regulation for several automated p-tau217 tests», Filippi recalls.
The Biomarker’s Limits: Why Clinical Context Still Decides
«The recommended diagnostic use applies to people who already show cognitive symptoms. In these cases, a plasma biomarker can help establish whether the underlying biology of Alzheimer’s is actually present», the professor clarifies. In short, it’s a tool for people already showing symptoms, not a population-wide screening test. A positive blood test must always be interpreted alongside clinical assessment. «The biomarker can’t be read independently of the clinical picture: what matters is the specific test’s performance, the cut-offs used, pre-analytical conditions and certain patient characteristics, such as age and kidney function», Filippi adds. Pre-test clinical probability also plays a role: how strongly the assessment before the blood sample was taken already pointed towards an Alzheimer’s diagnosis.
The Role of the APOE Gene
Genetics is among the factors that shape Alzheimer’s risk. The APOE gene exists in several variants, or alleles. People who inherit one or two copies of the ε4 variant have a higher probability of developing the disease than those who don’t carry it. «It’s important to clarify that APOE4 isn’t a diagnostic test for Alzheimer’s», Filippi points out. One or two copies of the ε4 allele raise the risk of developing the disease, but they remain a risk factor, not a certainty.
With the arrival of anti-amyloid therapies, a patient’s genotype has become a safety consideration too. In Europe, these antibodies are authorised for patients with no ε4 allele or a single copy, while those with two copies are excluded from the protocol. «APOE ε4 carriers face a higher risk of developing what are known as ARIA, the radiological changes associated with anti-amyloid therapy, particularly among carriers of two copies of the allele», the neurologist explains. For candidates for these therapies, genetic testing therefore becomes part of the treatment work-up.
«In Italy, the real challenge today is no longer proving that these biomarkers work, but integrating them into standardised diagnostic pathways, ensuring quality control, and defining access and reimbursement within Italy’s national health service (the SSN)», Filippi comments.
In People Without Symptoms: What a Positive Biomarker Means
The early diagnosis described so far concerns people who already have symptoms. The disease’s biological phase begins long before that: the build-up of beta-amyloid and Alzheimer’s other hallmark changes can appear 10 to 20 years before cognitive decline, and in some cases even earlier. In clinical practice, systematic testing for these changes is reserved for people who already show symptoms. «A positive biomarker can indicate a biological process compatible with Alzheimer’s disease many years before symptoms appear, but it doesn’t mean the person already has a cognitive syndrome, nor does it allow us to say with certainty if or when they’ll develop noticeable decline», the professor stresses. This is why screening (testing a cognitively normal person to predict what might happen years later) isn’t currently recommended for the general population.
What the Study Published in JAMA Found
A study published in JAMA followed 2,684 cognitively normal people across six longitudinal cohorts and found that higher p-tau217 values are associated with a greater risk of later cognitive decline. Among participants with very high values, the estimated five-year risk of progression was around 38%, and the association was even stronger over a ten-year horizon. The study marks an interesting shift: it moves the question from “does this person have Alzheimer’s biology?” to “what’s the probability that this person will develop cognitive decline in the coming years?”
An important limitation remains. «We still can’t tell a healthy person whether they’ll develop Alzheimer’s disease within a given number of years. The test expresses a probability, not an individual fate», the neurologist warns, adding that confirming this approach will require validation studies in unselected populations.
The Ethical Implications of a Predictive Test
There’s also an ethical dimension, even before a clinical one. «Telling a healthy person they face an increased risk can have significant psychological, family and social consequences, without being able to offer, at present, a preventive treatment of proven effectiveness», Filippi cautions. «The more capable we become of predicting risk, the more important informed consent, counselling for the people involved, and a correct probabilistic interpretation of the result become», he adds.
Alzheimer’s Research at UniSR: Early, Multimodal Diagnosis
Filippi’s group is also studying how to combine biomarkers linked to Alzheimer’s biology with other sources of data. «We’re working precisely towards a diagnosis that’s increasingly multimodal and personalised», the neurologist observes. The goal is to integrate plasma biomarkers, clinical and neuropsychological data, advanced neuroimaging and genetic information. Filippi’s team wants to find out whether combining them allows for more accurate diagnosis and earlier identification of patients who could benefit most from new therapies.
«We also use artificial intelligence and machine learning methods to identify complex patterns that are difficult to recognise when looking at each biomarker separately», Filippi remarks. «Of course, bearing in mind that an algorithm is only as reliable as the data it was developed and validated on». Research at the Alzheimer Centre at the IRCCS San Raffaele Hospital has one clear aim: making early Alzheimer’s diagnosis increasingly accurate.
Bibliography
https://jamanetwork.com/journals/jama/article-abstract/2851720
