Brain tests agree on the level. They disagree on the change.
Plasma assays wobble when tracking one person over time, while other brain-health numbers only make sense in context.
- Plasma assays for pTau217, pTau181, GFAP, and NfL show high cross-sectional agreement but poor longitudinal agreement on individual rates of change, and single-assay slope reliability for pTau217 is at most 0.40. Self-reported m:…7407
- Baseline Syn-SAA positivity is associated with significantly faster motor progression in prodromal individuals, and with a significantly higher hazard of phenoconversion, with an adjusted hazard ratio of 5.72. Self-reported m:…8925
- In the studied DRC cohort, poverty is associated with a 2.34-fold increase in the odds of dementia, while higher education is associated with a 0.51-fold decrease in the odds of dementia. Self-reported m:…8926
- Cannabis use in adolescents is associated with increased odds of suicidal ideation, with an odds ratio of 3.18, and with increased odds of suicide attempts, with an odds ratio of 4.86. Self-reported m:…7405
The level is solid. The change is not.
Plasma biomarker tests have a measurement problem, and a new preprint puts a number on it. Plasma assays for pTau217, pTau181, GFAP, and NfL show high cross-sectional agreement but poor longitudinal agreement on individual rates of change. In other words, the tests agree with each other about where a person stands today, and disagree about how fast that person is changing. Single-assay slope reliability for pTau217 is at most 0.40, which the preprint reads as most measured change in individuals being noise. For anyone using these tests to track change over time, that is the number that matters: detecting a 25% slowing of pTau217 change requires between 924 and 4,030 participants per arm depending on the assay. m:…7407
The implication is about how the tests should be used. If most of the change measured in one person is noise, then a single repeat test should not drive a decision about that person, however precise the assay looks on paper. The preprint's numbers argue for treating these assays as group-level tools for now: reliable for comparing populations and powering trials, shaky for telling one person how fast they are changing. m:…7407
One test that does predict
A second preprint, on the Syn-SAA test, shows the other side of the coin: a baseline result that predicts what happens next. Baseline Syn-SAA positivity is associated with significantly faster motor progression in prodromal individuals compared to Syn-SAA negative individuals, and the association persists after excluding phenoconverters. Baseline positivity is also associated with a significantly higher hazard of phenoconversion, with an adjusted hazard ratio of 5.72. What the test's kinetic parameters do not do is add anything further: they show no significant associations with phenoconversion risk. The signal is in whether the test is positive, not in how strongly it reads. m:…8925
Risk depends on where you live
A third preprint, on dementia in a DRC cohort, shows how much context matters once you leave the lab. Poverty is associated with a 2.34-fold increase in the odds of dementia in the studied DRC cohort, while higher education is associated with a 0.51-fold decrease in the odds of dementia in the same cohort. Adding 16 social, behavioral, and environmental factors significantly improved the model fit for dementia status beyond age and sex, with a Nagelkerke pseudo-R2 increase of 0.262. Age and sex alone do not carry the prediction; the circumstances of a life do. m:…8926
The teenage brain, measured twice
The week's starkest numbers concern adolescents and cannabis. Cannabis use is associated with increased odds of suicidal ideation in adolescents, with an odds ratio of 3.18, and with increased odds of suicide attempts, with an odds ratio of 4.86. Cannabis use is also associated with lower performance across multiple cognitive domains, with beta values of -0.08 to -0.16. One detail points the other way: better episodic memory performance is associated with lower odds of suicidal ideation, at 0.91, and suicide attempts, at 0.88. The preprint does not establish that cannabis causes either outcome, association is all the claims support, but the size of the associations, and the link through memory, is the part that deserves attention. m:…7405
What to do with numbers like these
Taken together, these four preprints argue for the same discipline: ask what a number was measured for. A plasma assay that agrees across tests at one moment may still be the wrong tool for tracking change in one person, while a positive Syn-SAA result carries real prognostic weight. A dementia risk model built on age and sex misses what poverty and education add in the DRC. And an association between cannabis, cognition, and suicide risk in adolescents is a reason to look closer, not a verdict. None of these findings is peer-reviewed yet, and each comes from a single preprint. The numbers are getting better. The judgment about what they mean still has to be done by hand. m:…7407 m:…8925 m:…8926 m:…7405
What is still unresolved
- Whether longer observation makes pTau217 slopes reliable enough to track change in a single person, rather than only in trial arms.
- Whether Syn-SAA positivity predicts progression in broader populations beyond prodromal individuals.
- Whether the link between cannabis use, cognition, and suicide risk in adolescents is causal, and what role episodic memory plays in it.
Provenance by Synorb. Every claim links to its source. This story was written by an agent from the Manifests listed below. Read the methodology and limitations, or send a correction.
Free to reuse, including commercially. Credit “Hanging Context, with provenance by Synorb” and link to this story. Keep the cited Manifest ids. Reuse terms v1.