Skip to content

ITP / CITP bibliography sweep: which publications carry mechanism?

Date: 2026-08-06 Scope: All 20 publications listed on the MPD ITP1 project page and all 24 PubMed-indexed publications listed on the CITP publications page. Question: The hosted data for both programs is overwhelmingly chemical→lifespan association. Where does the mechanism actually live, and which papers would a dismech curator cite?

Why this sweep exists

The NIA Interventions Testing Program (ITP; UM-HET3 mice, 3 sites) and the NIA Caenorhabditis Intervention Testing Program (CITP; 3 Caenorhabditis species, 3 sites) are the two highest-quality replication-controlled geroprotector screens in existence. Their portals are tempting as an evidence source for treatment claims against dismech's hallmark-of-aging modules. But an inspection of the hosted data (recorded below) shows that neither portal supports mechanism directly — the portals are survival statistics. If mechanism is to be curated, it has to come from the publications.

This document classifies every publication in both bibliographies by how much mechanistic content it carries, so that a curator can go straight to the ~9 papers that have any.

What the portals actually contain

Measured from the live data tables, not the portal prose:

ITP1 (MPD) — 74 compound × cohort trials, 62 unique compounds, 21 with a significant lifespan p-value.

  • Only 22 of 74 trials carry any secondary phenotype at all, and all 22 of those are body weight.
  • Beyond body weight, the entire secondary-phenotype layer is two trials: grip strength/duration and rotarod on acarbose C2013 and NDGA C2010.
  • Exactly one trial (acarbose C2013) has the full panel — body composition, fat pads, glucose meter, grip force/duration, rotarod, pathology.
  • The histopathology layer is four organ/condition categories with an odds ratio and p-value per sex (adrenal medullary vasodilation, liver degeneration, lung tumor, renal glomerulosclerosis).
  • No molecular, biochemical, transcriptomic, or proteomic measure anywhere in the portal.

CITP portal — 468 summary rows across 48 compounds.

  • Dose–response: median 5 concentrations per compound, up to 9.
  • Genetic background: 11 strains across 3 species; 9 compounds run across all 9 strain × species combinations.
  • Per-row survival statistics: n dead/censored/total, median + 95% CL, mean ± SE, 90th quantile, Cox PH p-value, Quality flag, lab/site attribution.
  • RNA-seq: 8 experiments, 2 compounds only (sulforaphane 5-point age time course; retinoic acid in N2 + two mutant backgrounds), GEO-linked (GSE289233, GSE272535).
  • The two mutant strains in the RNA-seq arm (PS3551 hsf-1(sy441), RB754 aak-2(ok524)) carry has_ls=False — they exist in the portal only as RNA-seq entries. The lifespan epistasis is in the paper, not the tables.

Tier definitions

Tier Meaning
0 Program design, methods, assay platform, software, or resource paper. No biological claim about an intervention.
1 Lifespan association only. Compound → survival, optionally with body weight. No mechanism.
2 Lifespan + organ, physiology, or healthspan phenotype (pathology, glucose handling, stress resistance, mobility). Localizes the effect to a tissue or function but identifies no molecular pathway.
3 Mechanism-bearing. Molecular or pathway-level evidence: omics, genetic epistasis, target identification, or pharmacokinetics used to explain a differential response.

ITP (mouse) — 20 publications

PMID Year Short title Tier Mechanistic content
17578509 2007 Study design and interim report 0 Design paper; describes planned T-cell subset and activity assays
19424842 2008 Design of aging intervention studies 0 Program design
27923560 2017 NIA ITP: investigating putative agents 0 Program overview (no abstract)
18631321 2008 NDGA and aspirin increase male lifespan 1 Lifespan only
22451473 2013 Resveratrol, GTE, curcumin, OAA, MCT 1 All null; lifespan only
27312235 2016 Protandim, fish oil, UDCA, metformin, 17aE2, NDGA 1 Lifespan; Nrf2 framing asserted, not measured
33788371 2021 17aE2 late-life; NR and 3 others null 1 Lifespan + body weight
36179270 2022 Rapamycin + acarbose combination; captopril 1 Drug-combination potency, no molecular readout
38041783 2024 Astaxanthin, meclizine; 5 null 1 Target class asserted (Nrf2 activator, mTORC1 inhibitor), not assayed
40973907 2026 Epicatechin, halofuginone, mitoglitazone 1 Lifespan only
33145977 2020 Rapamycin late-life dosing regimens 1 Dosing schedule; confounded by diet-supplier weight variation
19587680 2009 Rapamycin fed late in life (Nature) 2 Lifespan + disease patterns at necropsy; mTOR inferred, not measured
20974732 2011 Rapamycin, not resveratrol or simvastatin 2 Lifespan + spontaneous activity + causes of death
22587563 2012 Rapamycin slows aging in mice 2 Multi-tissue aging-rate: heart, liver, adrenal, endometrium, tendon, plus activity; also harms (testicular degeneration, cataracts)
30688027 2019 Acarbose improves health and lifespan 2 Lung tumors, liver degeneration, glomerulosclerosis, refeeding glucose response, rotarod — the C2013 dataset behind the portal's only pathology page
30916479 2019 Glycine supplementation 2 40 necropsy pathology categories; pulmonary adenocarcinoma reduced; methionine-toxicity framing
32990681 2020 Canagliflozin 2 Fasting glucose, glucose tolerance, fat mass; mechanism argued by convergence with acarbose ("blunting of peak glucose levels")
24245565 2014 Acarbose, 17-α-estradiol, NDGA male-preferential 3 PK used to exclude a mechanism: NDGA at a dose producing female blood levels matching males still gave no female benefit
38753230 2024 Sodium thiosulfate, 16-OH-estriol, late canagliflozin 3 "blood levels of Cana were approximately 20-fold higher in aged females than in young males, suggesting a possible mechanism for the sex-specific disparities"
24341993 2014 Rapamycin dose/sex dependent, metabolically distinct from DR 3 Hepatic xenobiotic-metabolism gene expression profiles differ between rapamycin and dietary restriction; drug blood levels differ by sex

ITP totals: 3 Tier-0, 8 Tier-1, 6 Tier-2, 3 Tier-3.

The single most mechanism-bearing ITP paper is PMID:24341993 — it is the one that separates rapamycin from dietary restriction on a molecular readout rather than asserting they differ. The other two Tier-3 entries are pharmacokinetic, not pathway-level: they explain why a response differs by sex, which is mechanistically real but narrow.

Note the striking pattern: ITP's most mechanistic work is about sexual dimorphism, because that is the phenomenon the lifespan data itself forces the investigators to explain.

CITP (nematode) — 24 publications

PMID Year Short title Tier Mechanistic content
28060275 2016 CeleST swim behavior software 0 Methods
28836615 2017 A long journey to reproducible results 0 Nature comment
31042764 2019 The Stress-Chip microfluidic platform 0 Methods
31820364 2019 Automated Lifespan Machines across strains 0 Methods/validation; notes assay-specific intervention effects
33204740 2020 Simplified lifespan machine design 0 Methods
35098051 2022 Genetic diversity estimates for the screening panel 0 Resource; whole-genome-based diversity of the 22-strain panel
40178707 2025 CITP program overview 0 Program design
32010883 2019 Imatinib does not extend lifespan 1 Null
31998863 2020 β-guanidinopropionic acid does not extend lifespan 1 Null
32550518 2020 Obeticholic acid does not robustly extend lifespan 1 Null
34585102 2021 Diuron does not robustly extend lifespan 1 Null
41993912 2026 Levetiracetam does not extend lifespan 1 Null; notes anticonvulsants differ in effect
28220799 2017 Impact of genetic background and reproducibility (Nat Commun) 1 Landmark: 22 strains, 3 species, 10 compounds. Establishes strain/species specificity of DR mimetics vs. robustness of ThioflavinT. Association, but across the genetic axis
34837316 2022 Metformin across diverse Caenorhabditis 2 Lifespan + healthspan; metformin benefit is genetic-background dependent (works in C. elegans strains, not C. briggsae)
37923874 2024 Green tea extract and NDGA 2 Species- and strain-specific lifespan and health effects
38613792 2024 Healthspan–lifespan coupling 2 Swim performance, thermotolerance, oxidative stress resistance across the panel for NP1, propyl gallate, resveratrol; shows the relationships are not simply coupled
40027526 2025 Tamibarotene and bakuchiol do not extend lifespan 2 Structure–activity argument: a potent RAR agonist fails where atRA works, constraining atRA's mechanism
41701440 2026 Male lifespan and reproductive healthspan 2 Sex differences; lifespan and reproductive healthspan decouple (only sulforaphane and metformin improved late-life mating success)
32831297 2020 Insolublome quantification by DIA proteomics 3 Methods paper, but the readout is proteostasis: SDS-insoluble proteome extraction and label-free quantification
32877690 2020 Alpha-ketoglutarate extends lifespan and compresses morbidity in mice (Cell Metab) 3 IL-10 induction suppressing chronic inflammation; systemic inflammatory cytokines; frailty. Mouse, not worm
38753231 2024 Amyloid β accelerates proteome-wide protein insolubility 3 Unbiased proteomics; Aβ drives proteome-wide insolubility in C. elegans even in young animals
40462948 2025 Sulforaphane slows the transcriptional aging clock 3 RNA-seq-derived gene-specific transcriptional aging clock; ~4-day younger transcriptional age (~20% biological-age reduction); detoxification pathways dominant; dose-response shape indicates hormesis
42320027 2026 Translation state modulators extend lifespan 3 4E-BP/eIF4E pathway, 5'-UTR-length-dependent translation; cell-based screen; DR/cold-induced-longevity mimicry; Drosophila + C. elegans
41432067 2025 Retinoic acid modulation drives conserved longevity pathways (eLife) 3 The flagship. Genetic epistasis + RNA-seq

CITP totals: 7 Tier-0, 6 Tier-1, 5 Tier-2, 6 Tier-3.

The atRA paper (PMID:41432067) in detail

This is the only paper in either bibliography that does full pathway dissection, and it is the reason CITP is mechanistically ahead of ITP despite the shorter-lived model.

Epistasis strains and their outcome for atRA lifespan extension:

Strain Gene Human ortholog concept atRA extension
RB754 aak-2(ok524) AMPK catalytic subunit Required
RB759 / VC204 akt-1(ok525) / akt-2(ok393) AKT Required
PS3551 hsf-1(sy441) HSF1 Required
— skn-1 NRF2 Required
CF1038 daf-16(mu86) FOXO Not absolutely required; partial contribution
KU25 pmk-1(km25) p38 MAPK Less critical
IG10 tol-1(nr2033) Toll-like receptor Enhanced response

Transcriptional findings: 17% (2,169) of detected genes differentially expressed; 83% (20/24) of the most-upregulated genes previously IIS-regulated and 71% Nrf2-regulated; sphingolipid metabolism and fatty-acid biosynthesis enriched.

Summary

ITP CITP
Publications swept 20 24
Tier 0 (design/methods) 3 7
Tier 1 (lifespan only) 8 6
Tier 2 (organ/physiology) 6 5
Tier 3 (mechanism) 3 6
Nature of the Tier-3 work Pharmacokinetics + one hepatic expression comparison Genetic epistasis, transcriptomics, proteomics, target pathway

Roughly 20% of the combined bibliography (9 of 44) carries mechanism, and the two programs carry different kinds. ITP's mechanism is pharmacological (why does this drug act differently in males?). CITP's is pathway-level (which conserved longevity pathway does this compound require?).

CITP's Tier-3 papers are also disproportionately recent — five of six are 2024 or later — which suggests the program has shifted from screening toward mechanism as its compound set matured.

Relevance to dismech modules

The Tier-3 set maps onto modules the KB already has. None of this is curatable as a human treatment claim (evidence_source: MODEL_ORGANISM or IN_VITRO throughout, per the evidence policy), but it is legitimate module-level mechanism evidence.

PMID Maps to Note
41432067 deregulated_nutrient_sensing (AMPK, AKT/IIS), loss_of_proteostasis (HSF-1) Epistasis establishes requirement, not correlation
42320027 deregulated_nutrient_sensing, loss_of_proteostasis, mitochondrial_dysfunction 4E-BP/eIF4E translational control
38753231 loss_of_proteostasis, amyloidogenesis Aβ → proteome-wide insolubility; connects two existing modules
32831297 loss_of_proteostasis Assay method for the insolublome
32877690 inflammaging AKG → IL-10 → reduced chronic inflammation; mouse, so the strongest of the set translationally
40462948 epigenetic_alterations (as a transcriptional-age biomarker) Also a hormesis/detoxification claim
24341993 deregulated_nutrient_sensing Separates rapamycin from DR on hepatic gene expression
22587563 Cross-cutting Best available "does it slow aging broadly or just suppress tumors?" multi-tissue evidence
32990681 / 30688027 diabetic_vascular_complications (glucose-excursion arm) Acarbose/canagliflozin converge on postprandial glucose blunting

Caveats

  1. Classification is from abstracts, not full text. A Tier-1 or Tier-2 paper may contain mechanistic figures not described in its abstract. The tiers indicate where mechanism is advertised, which is the right filter for deciding what to read next — not a claim about the paper's total content.
  2. Two CITP-listed papers are mouse or fly work from consortium labs rather than core nematode CITP output (PMID:32877690 alpha-ketoglutarate in mice; PMID:42320027 includes Drosophila). They are listed on the CITP publications page and are included here, but a curator should note the organism explicitly.
  3. PMID:40462948 is a bioRxiv preprint (2025.05.11.653363) at time of sweep. Its RNA-seq is in the portal (GSE289233), but the paper is not peer-reviewed. Treat accordingly.
  4. Tier 3 does not mean "human-relevant." Every mechanism paper here is model-organism or in-vitro. Under the dismech evidence policy these cannot be sole support for a human phenotype claim.
  5. Null results are not failures. Both programs publish negatives by policy, and the Tier-1 nulls (imatinib, obeticholic acid, diuron, β-GPA, levetiracetam, tamibarotene, bakuchiol; resveratrol, curcumin, MCT oil, nicotinamide riboside, fisetin) are among the most valuable content in either bibliography, because they are replication-controlled refutations of widely promoted compounds. In dismech terms these are candidate supports: REFUTE evidence items.

Sources