The blood draw takes about ninety seconds. A nurse finds the vein, the tube fills, the tourniquet snaps off, and somewhere downstream a machine will count molecules present at concentrations so low that until recently nobody could see them in blood at all. For families who have spent a year in the corridor of not knowing, that small vial has become freighted with meaning. The honest question is what it can carry. A review published in Nature on 4 February 2026 by Henrik Zetterberg and Barbara B. Bendlin (volume 650, pages 49 to 59) is the first thing we have read that answers that question across the whole field rather than one test at a time.
It is a review, which matters: no new participants were enrolled, nobody was scanned, nothing was randomised. What Zetterberg and Bendlin have done is synthesise what is already published and lay it out by biology rather than by brand name. They organise the landscape around eight pathologies and processes, all of them relevant to neurodegenerative dementias: amyloid-beta pathology, tau pathology, neurodegeneration, glial reactivity, alpha-synuclein pathology, TDP-43 pathology, synaptic pathophysiology, and cerebrovascular disease. That list is the point. Dementia is not one process, and a biomarker panel that reads only one of the eight is a map of one street.
The review's framing is that improved technologies now allow molecules linked to neurodegenerative brain changes to be detected at very low concentrations in blood, complementing the longstanding cerebrospinal fluid assays rather than replacing them. The authors describe this as a route to molecular phenotyping: linking a person's clinical symptoms to the pathophysiological processes actually running underneath them. In a memory clinic, that is the difference between a label applied by pattern recognition and a description of what the brain is doing.
One corner of the map has already crossed into regulated practice. On 16 May 2025 the US Food and Drug Administration cleared the Lumipulse G pTau217/beta-Amyloid 1-42 Plasma Ratio through the 510(k) pathway, the first blood test cleared to help diagnose Alzheimer's disease, in adults aged 55 and older who are already showing signs and symptoms. The FDA judged it substantially equivalent to an earlier authorised test that used spinal fluid. Note the fences around it: symptomatic people only, an aid to diagnosis rather than a diagnosis, and amyloid rather than Alzheimer's as such. The other seven processes in the review have no such clearance behind them.
Some of the most interesting territory is exactly where the ground is softest. For TDP-43, the protein behind much frontotemporal dementia and ALS, the review describes how the protein's departure from the cell nucleus lets cryptic exons slip into RNA transcripts, and how a peptide encoded by one of them, from HDGFL2, can be measured in spinal fluid and appears to identify people with TDP-43 pathology, something otherwise establishable only after death. For Lewy body pathology, the authors survey assays aimed at aggregated, oligomerised, truncated and phosphorylated forms of alpha-synuclein. These are research assays described in a research synthesis, not tests a family can ask for by name next month.
So what does a reader do with this? Mostly, recalibrate. The single-test headlines of the past two years, each true in its own frame, have created an impression that a blood test can now tell you which dementia you have. This review, from two of the field's most cited biomarker scientists, sets those pieces into one frame and shows the gaps between them: one cleared blood test for one pathology in one clinical situation, and a broad, fast-moving research effort behind it covering everything else. If your clinician has ordered a biomarker test, or declined to, this is the landscape that decision sits in. The interpretation of any individual result belongs in that consulting room, not in a newspaper.