BIOBRIEFPAPER
A 32 patient postmortem study compared a leading Alzheimer's blood test with brain tissue from the same donors. The blood marker, a tau fragment called pTau217, tracked 0.73–0.78 against the cortex.
A 32-patient postmortem study at University Hospitals Leuven ran the same Alzheimer's biomarker through a vial of blood and a chunk of temporal cortex from each donor, and got a partial Spearman correlation of 0.73 to 0.78, the statistical shorthand for a strong, monotonic relationship, not a perfect one. The paper is the rarest kind of validation work in this field: cadaver-to-vial calibration, the same biomarker measured in the same patient on both sides of the skull.
A ρ of 0.73 to 0.78 means the blood reading rose and fell with the brain reading across the cohort, with noise. Diagnostic accuracy for separating Alzheimer's from non-AD dementia held at an AUC of 0.88 to 0.94, depending on the plasma marker. Two of the three markers (plasma pTau217NT and the pTau217Mid/Aβ1-42 ratio) fit the brain data better than pTau217Mid alone (ΔAIC 2.5 and 1.1; ΔBIC 3.6 and 2.1).
The cohort is narrow: 32 patients with dementia, mean age 66 (SD 10), 38% female, all from one Belgian academic memory clinic. Nineteen had Alzheimer's at autopsy; 13 had frontotemporal dementia, a different disease. That mix is the falsifier for any "breakthrough" headline. The next plasma pTau217 headline should say which reference the test was checked against: PET, cerebrospinal fluid, or actual brain tissue.