Model organism → human extrapolation guide
The wiki uses primary results from model organisms (mouse, worm, fly, yeast, killifish, naked mole rat) to inform claims about human aging. Whether and when these extrapolate is a first-class question. This page defines the rubric.
This page is the data layer: per-organism divergence catalog, the three-dimension rubric, and case-level pitfalls. The companion synthesis is translation-failure-of-aging-interventions (Mode B conceptual frame), which organizes why the pattern of mouse-success / human-modest-or-null persists across cases and what would update the field’s confidence in mouse-validated interventions. Use this page to ask “is this specific extrapolation safe?”; use the hypothesis page to ask “should we expect the next mouse result to translate, and what general barriers apply?”
The three-dimension rubric
For every claim derived from a non-human study, attach a small extrapolation table:
| Dimension | Status |
|---|---|
| Pathway conserved in humans? | yes / partial / no / unknown |
| Phenotype conserved in humans? | yes / partial / no / unknown |
| Replicated in humans? | yes / no / in-progress / contradicted |
What “yes / partial / no” means
Pathway conserved — Are the molecular components (genes, proteins, regulatory motifs) functionally orthologous in humans?
- yes — Direct human ortholog with conserved domain structure and demonstrated equivalent function (e.g., mTOR is mTOR in mouse and human).
- partial — Ortholog exists but with notable functional divergence (e.g., mouse vs human telomerase regulation, mouse vs human p16 dynamics).
- no — No clear ortholog or function is performed by an unrelated mechanism in humans.
- unknown — Ortholog exists, function in humans uncharacterized.
Phenotype conserved — Does the organism-level effect manifest similarly?
- yes — Human aging shows the same phenotype on the same approximate timescale (allowing for lifespan scaling).
- partial — Phenotype occurs in humans but with different magnitude, kinetics, or tissue distribution.
- no — Phenotype absent or qualitatively different in humans (e.g., mouse cardiomyocyte regeneration capacity).
- unknown — Hasn’t been adequately measured in humans.
Replicated in humans — Has the specific intervention or observation been tested in humans?
- yes — Independent human study showed the same direction of effect.
- no — Not yet attempted in humans.
- in-progress — Trial registered (cite NCT number).
- contradicted — Human study failed to replicate or showed opposite effect.
Per-organism extrapolation profiles
Each model-organism page (mus-musculus.md, caenorhabditis-elegans.md, etc.) should include:
- Genome similarity to human (% with one-to-one orthologs)
- Lifespan ratio (their lifespan vs human lifespan, for time-scaling)
- Conserved systems — bullet list of well-conserved pathways relevant to aging
- Divergent systems — bullet list of known major divergences (with refs)
- Strengths — what the model is uniquely good at
- Failure modes — known cases where the model misled aging research
Quick reference (to be expanded per page)
| Organism | Lifespan | % orthologs | Best for | Known divergences from human |
|---|---|---|---|---|
| mus-musculus (mouse) | 2–3 yr | ~85% | mammalian physiology, transgenics, lifespan studies | telomere biology, immune aging, microbiome, drug metabolism |
| rattus-norvegicus (rat) | 2–4 yr | ~85% | cardiovascular, behavioral aging | similar to mouse divergences |
| caenorhabditis-elegans (worm) | 2–3 wk | ~40% | longevity genes, autophagy, IIS pathway | no adaptive immunity, no circulatory system, post-mitotic adult |
| drosophila-melanogaster (fly) | 1–2 mo | ~60% | nutrient sensing, neurodegeneration | open circulatory, no humoral immunity |
| saccharomyces-cerevisiae (yeast) | days–weeks (chronological); ~25 generations (replicative) | ~30% | basic cellular machinery, autophagy, sirtuin discovery | no multicellularity, no tissue context |
| nothobranchius-furzeri (killifish) | GRZ ~9 wk median; wild strains 5–8 mo | vertebrate | short-lived vertebrate, brain aging | aquatic, ectothermic |
| heterocephalus-glaber (naked mole rat) | 30+ yr | mammal, similar to mouse | extreme longevity, cancer resistance | eusocial, hypoxia-adapted, ectothermic-ish |
Common extrapolation pitfalls
These cases historically misled translation; flag them when relevant.
- Telomere biology — Lab mice (C57BL/6) have telomeres 5–10× longer than humans and constitutive telomerase in many somatic tissues; telomere attrition is not a major aging driver in standard mouse strains. Wild-derived strains and telomerase-knockout mice are better models.
- Caloric restriction effect size — CR extends lifespan ~30–40% in mice; the largest controlled human trial (CALERIE) found no lifespan effect at 2 years and modest healthspan markers. Effect-size translation appears poor.
- C. elegans IIS extrapolation —
daf-2mutants live 2× as long; the human IGF1R analog has much smaller effects. The pathway is conserved; the magnitude is not. - Senolytic dosing — Mouse-effective doses for D+Q and fisetin are mg/kg orders of magnitude higher than typical human supplement doses; pharmacokinetic translation is non-trivial.
- Mouse SASP composition differs from human SASP — SASP factor lists from mouse studies should be cross-checked against human cell senescence data.
- Drug metabolism (CYP differences) — mouse CYP3A vs human CYP3A4 substrate specificity differs; some compounds metabolized to active form in mouse aren’t in humans.
When to deprioritize a model-organism finding
Treat a finding as low-confidence-for-human-application if any of these apply:
- Pathway is not conserved in humans (#gap/no-human-ortholog)
- Pathway is conserved but the relevant phenotype is not observed in human aging
- Effect size in the model relies on a feature humans lack (e.g., adult tissue regeneration capacity in fish/amphibians)
- Multiple mammalian models disagree
- Original study is preliminary (single lab, small n, no replication)
Tag with #gap/needs-human-replication and surface in gaps/.