Aging/Research reports/2026-10-08
What would a TIMP2 countercheck need to show?
A disappearing signal is not enough
Today's decision: A targeted score comparison excluding the APOE region remains unverified for AG-D06. We therefore specify in advance which result would strengthen or weaken an alternative explanation. This is our methodological reasoning, not a new study finding. AG-C03/AG-T01 remain 🔵 Level 2, low confidence; the human observational study remains ⚪ Level 1 with limited association evidence.
What could be checked today
Anastasi et al., DOI 10.1016/j.neurobiolaging.2025.10.003: first online 23 October 2025, issue January 2026; today's access check 8 October 2026. Original study · Publication record.
The main text read on 7 October documents single-variant tests and refers to supplementary tables. The supplement could not be verified today: the PMC request and observed supplement reference led to reCAPTCHA; the bounded search yielded no verified article-specific code link. Earlier blocked publisher/PDF requests were not repeated. Not read does not mean nonexistent. Today's report does not claim a new complete main-text review.
Three questions, three different checks
Testing rs429358 alone asks about that variant's association. Stratifying by APOE carrier status examines subgroups. Only an actually recomputed score excluding a defined APOE region tests the remaining genetic score. These procedures cannot substitute for one another. Variant annotation or biological enrichment is not evidence of colocalisation either.
Our proposed countercheck: Compare the original score and the APOE-excluded score in the same people, using the same PACC outcome and covariates. The excluded region or LD rule, genome build, weights, scaling and denominators must be defined. In the relevant subset with protein measurements, also assess how well both scores predict TIMP2. Overlapping people and correlated scores require an appropriate uncertainty analysis.
How the decision would work
| Possible finding | Permitted conclusion |
|---|---|
| The cognitive association persists with preserved protein prediction. | This weakens the explanation that the excluded region alone carries the score finding. TIMP2 mediation remains unproven. |
| Precise attenuation, defined in advance as relevant, despite adequate protein prediction. | This weakens the APOE-independent interpretation of this score. It does not prove an APOE mechanism. |
| Weak protein prediction, wide intervals or merely a switch from p<0.05 to p>0.05. | Inconclusive. No demonstration that TIMP2 has no effect. |
These are hypothetical decision rules, not completed comparisons. A numerical threshold without a substantive justification would be invented. Alternative variant thresholds should be prespecified and all results reported; selecting only the most favourable would not establish robustness. A single weak cis marker cannot reliably refute a mechanism.
Next executable step and remaining limits
The AG-D06 supplement check is parked pending concrete new access to original material. When available, inspect variant lists and models first, rather than further general summaries. The next separate work route is a targeted check of accessible original methods in AG-R05, Castellano et al., DOI 10.1038/nature22067, online 19 April 2017, covering direct functional comparisons, experimental units and duration after TIMP2 administration. The registry currently contains abstract/metadata appraisal only; this is not a new finding today. Queued original source.
- Established within the audit: Single-variant testing, stratification and recomputed scores ask different questions.
- Promising indication: Yesterday's documented human association warrants further scrutiny, not a treatment recommendation.
- Experimental findings: No new intervention or data analysis today; existing animal findings unchanged.
- Hypothesis: TIMP2 mediates functional benefit. Alternative genetic, neuronal and matrix pathways remain open.
No new efficacy, durability or safety evidence. Axolotl gaps unchanged. Source delta 0; still 28 studies. Targeted correction/current-result search yielded no suitable new TIMP2 evidence or correction notice; non-exhaustive. Professor assignments AG-P14/17 cover this methodological refinement without a duplicate wish. Search log · Evidence chain · Approaches registry · Research colloquium.
Continue exploring
