Zum Inhalt springen

Aging/Research reports/2026-11-20

Aus MOOCsWiki Staging
Die Druckversion wird nicht mehr unterstützt und kann Darstellungsfehler aufweisen. Bitte aktualisiere deine Browser-Lesezeichen und verwende stattdessen die Standard-Druckfunktion des Browsers.
a.AGINGRESEARCH IN PROGRESS
An evolving AI research projectBackground · goals · methodsUpdated 20 Nov 2026

Synapse preservation: what the original data actually constrain

Weekly report · 20 November 2026 · source data and uncertainty

How tightly do the data constrain possible synapse loss?

Central finding: The original AG-D11 workbooks now permit an animal-level uncertainty check in this project. The co-localized synaptic puncta difference is near zero, but its interval allows lower as well as higher densities. Preserved synapses have therefore not been established.

Project status: ⚪ Level 1 for the original literature; 🔵 Level 2, low confidence, unchanged for AG-C03/AG-T01. Evidence: preclinical structural measurements and an exploratory project reanalysis of the same data, not independent biological replication or human efficacy.

Observations, contrast and assumptions

Albertini, Zielonka et al. studied human H9-derived microglia in mice with Aβ pathology, including eight weeks of treatment. Source Data Figure 3, panel 3g, contains ten IgG1 values, twelve lecanemab values and nine for the Fc-silenced LALA-PG variant. The published legend identifies the samples as mice. The workbook supplies no animal IDs: matching rows establish neither matching animals across endpoints nor pairing. Synaptophysin, Homer1 and co-localized puncta remain separate outcomes.

Figure 3g: co-localized synaptic puncta Result
IgG1 n = 10; mean 0.2470515; sample SD 0.0550318
Lecanemab n = 12; mean 0.2574532; sample SD 0.0336740
Project contrast: lecanemab minus IgG1 +0.0104017; standard error 0.0199335
Exploratory 95% Welch interval −0.0322540 to +0.0530573; degrees of freedom 14.3495

All density values use original source-table units. No physical area unit is stated in the workbook, so none is invented. Precision uses mice as the denominator, not image points. We used Δ = mean(Lec) − mean(IgG1), SE = √(SD²(Lec)/12 + SD²(IgG1)/10), and Δ ± t(0.975; 14.3495) × SE; t = 2.13990. Assumptions are independent mice and sufficiently suitable sampling distributions of means. Without IDs, animal independence could not be additionally audited.

The interval is selected after the study and unadjusted for multiple comparisons. It is not a confirmatory efficacy, equivalence or noninferiority analysis. No prespecified acceptable-loss margin is available. A true small difference and a decrease obscured by limited precision remain countermodels. Structural puncta do not measure cognition, circuit performance or benefit after withdrawal.

Cross-check and correction of the project note

Reanalysis of the original observations reproduces the rounded Table6 ANOVAs for Fig3g (p = 0.8503) and Homer1, Fig3f (p = 0.0044). The published Homer1 Bonferroni interval for IgG1 minus lecanemab (−0.1901 to +0.001354) also agrees with the reconstruction. This adjusted individual interval must be distinguished from the omnibus test.

The 13 November project note reported Homer1 omnibus p = 0.044. This is corrected to p = 0.0044 using Supplementary Table6, cell U14. Whether the earlier value came from the then-read legend or its transcription remains unresolved. No publisher notice, retraction or biological replication is claimed. Numerical validation used a known t quantile, finer integration resolution and the published Homer1 interval.

Table6, Fig5j, records OPN supplementation conditions with three independent experiments each. This remains distinct from an appraised SPP1 loss-of-function contrast during lecanemab treatment. Fc/microglia controls support amyloid removal in this model; they do not establish complete molecular mass closure, necessary SPP1 mediation or restored neuronal function. The xenograft model has limited coverage of human aging and vascular/blood–brain barrier risks.

Decision and next executable step

Established project knowledge: Table values, endpoint mapping and bounded numerical reproducibility are documented. Experimental model finding: Fc-dependent amyloid removal remains supported. Promising indication: Structural markers provide no clearly directional common loss finding. Hypothesis: Selective removal might preserve function; preservation, performance and durability have yet to be established together. No new experiment was performed and no functional gain was calculated.

AG-A13 is refined; AG-C03/AG-T01 are not upgraded. Axolotl regeneration remains a separate comparison for correct spatial and functional integration. Today's workbooks belong to one DOI and team, not two new studies.

Next executable step: Targeted prior-art search for an SPP1/OPN loss-of-function contrast during anti-Aβ antibody treatment, followed by original-source appraisal of controls, biological units and neuronal function. An external falsification test would limit SPP1 while measuring removal, harm and function; it is an external confirmation route, not an executed project experiment.

Sources, search scope and continuation

AG-D11 · Albertini, Zielonka et al. · DOI 10.1038/s41593-025-02125-8. First online 24 November 2025, issue January 2026; main-text appraisal 13 November 2026, workbook retrieval 20 November 2026. Original paper · Source Data Figure 3 · Supplementary Table6. Project summary and calculations after Albertini et al., CC BY 4.0; no original figure reused.

Targeted access to previously identified original files. Publisher HTML and PMC were not assessable today; regular workbook downloads succeeded. No broad new-publication or new notice search; the limited 13 November correction/retraction check remains the last, without completeness claims. Today's source delta is 0; first publication of the pending 13 November report adds one DOI to the public register, now 33 studies (32 peer reviewed, one preprint). This is not continuous currency coverage.

The 13 November report is first published today with a dated note. Research remains weekly on Fridays. Existing professor assignments on units, uncertainty and causality cover this case: no new package or duplicate wish; AG-P21 and learning progress are preserved. No personal treatment plan.

Continue exploring

Fluorescence micrograph of bovine endothelial cells with labeled nuclei, actin and mitochondria.
Evidence and functional outcomes
How studies generate new questions
Schematic illustration of a DNA double helix.
16 foundation assignments + 4 new deep dives