Alzheimer's Markers Are Moving Detection Years Earlier — With One Cohort Warning

A blood marker that tracks the stages

In a preprint on plasma eMTBR-tau, levels of the marker show significant stage-related differences across the Alzheimer's disease continuum, with 6.55-fold higher levels in Alzheimer's dementia compared with subjective cognitive decline. A marker that rises with stage is useful in a different way than a marker that merely separates sick from healthy: it can, in principle, place a person along the continuum rather than only on one side of a line. m:…0110

The preprint's conversion result is the stronger claim. Plasma eMTBR-tau243 provides superior discrimination of conversion from mild cognitive impairment to Alzheimer's disease, with an area under the curve of 0.936, compared with 0.882 for BD-pTau217, over a two-year follow-up. The head-to-head framing matters: eMTBR-tau243 is not presented as the first tau marker, but as one that outperforms an existing tau measure at the specific task of predicting who converts within two years. m:…0110

An imaging index that sees conversion coming

A second preprint takes the imaging route. A digital biomarker of white-matter routing from diffusion MRI, called the Route Concordance Index, detects conversion to mild cognitive impairment or Alzheimer's disease up to four years before clinical diagnosis, with an area under the curve of 0.831. Four years of lead time changes what early detection means: it moves the question from diagnosing disease that is already clinically apparent to identifying people while intervention trials and planning could still matter. m:…9659

The preprint benchmarks the index against the usual imaging suspects, and reports that it outperforms hippocampal volume, entorhinal thickness, global conductance, and DTI-ALPS in preclinical discrimination of Alzheimer's disease. Replication is reported in the ADNI cohort of 609 participants, with an area under the curve of 0.797 — lower than the headline figure, as replication figures usually are, but still in the range that keeps the index credible rather than collapsing it to a single-cohort finding. That the replication figure sits below the headline figure is expected when a measure moves to a new cohort, and the preprint reports both rather than only the stronger one. m:…9659

The cohort that pushes back

A third preprint complicates any simple story about blood biomarkers traveling well. In older adults in the Democratic Republic of the Congo, higher plasma MME levels are associated with better language, global cognition, and delayed memory, while higher plasma NEFL levels are associated with poorer global cognition and delayed memory. Those are proteomic signals of their own, tied to cognition in this cohort on their own terms. The two directions matter: one protein rises with better cognition, the other with poorer cognition, so the preprint is not reporting a single proteomic direction. m:…8250

The warning sits in the same preprint: plasma p-tau217-based measures did not remain significantly associated with cognitive outcomes after Benjamini-Hochberg false discovery rate correction in this cohort. A marker that performs well in one population cannot be assumed to carry the same association everywhere, and a correction for multiple testing is exactly the step that separates a robust association from a suggestive one. The honest reading of the three preprints together is therefore two-sided: detection is moving earlier, in blood and in imaging, but each marker still needs to prove itself cohort by cohort before earlier means better for any given patient. m:…8250 m:…9659