RIA-01 · Work Package 5 · Aging-Clock Endpoints
v0.1 · 2026-05-12
Aging-clock and rejuvenation-marker endpoint suite
Six clocks, four functional readouts, three molecular panels — mapped to the retina-first program. Distinguishes endpoints that move the FDA from endpoints that move the field.
Computational research use only — not clinical, not therapeutic, not wet-lab instruction.
★ WP5 Headline
For a retina-first program, the right endpoint architecture is:
- FDA-translatable primary: visual function (BCVA, contrast sensitivity, microperimetry). These are accepted ophthalmic endpoints with a precedent (Luxturna 2017).
- Mechanism-confirming secondary: retinal-tissue-derived DNA methylation aging clock + RGC transcriptomic age signature. These confirm partial reprogramming actually happened.
- Field-moving exploratory: blood-based GrimAge / DunedinPACE / CausAge to establish whether a locally-delivered retinal intervention shows systemic age signal (Lu 2020 hinted yes).
The biggest mistake an aging-program study can make is to anchor on epigenetic clocks alone. The FDA does not approve drugs on Horvath-clock movement. The clocks confirm the mechanism; the visual function endpoint approves the drug.
1. The aging-clock landscape relevant to partial reprogramming
RIA-01 distinguishes three families of biological-age estimators by what they measure and what they predict:
| Family | Examples | Measures | Best at | Partial-reprogramming utility |
|---|
| First-gen DNAm clocks |
Horvath (multi-tissue), Hannum (blood) |
CpG methylation patterns |
Estimating chronological age |
Direct mechanistic confirmation (these are exactly what reprogramming resets) |
| Second-gen DNAm clocks |
GrimAge, PhenoAge, DunedinPACE |
CpGs trained on mortality / morbidity / pace-of-aging |
Predicting health outcomes |
Translational endpoint — what matters for healthspan claims |
| Causal / mechanism-aware clocks |
CausAge (Ying 2024), DamAge / AdaptAge |
CpGs filtered for causal vs passenger |
Telling intervention from noise |
Highest signal-to-noise for an intervention study; field-moving |
| Transcriptomic age |
Peters 2015 blood-Tx age, scAge (single-cell) |
RNA expression patterns |
Tissue / cell-type-specific age |
Best for tissue-targeted programs like retina/RGC |
| Proteomic age |
Lehallier 2019 plasma proteome, SomaLogic / Olink clocks |
Plasma protein levels |
Systemic age, organ-of-origin attribution |
Useful for systemic signal from local intervention |
| Telomere length |
qPCR or TeSLA |
Telomere attrition |
Replicative capacity |
Lower priority — OSK doesn't directly target telomerase |
2. Primary functional endpoint: visual function (FDA-translatable)
PRIMARY
NON-INVASIVE
Visual function composite
Modality: BCVA + contrast sensitivity + microperimetry · Precedent: Luxturna (RPE65 gene therapy) approved 2017 on functional vision endpoints
- BCVA (Best-Corrected Visual Acuity, ETDRS letters) — FDA-accepted; reliable
- Contrast sensitivity (Pelli-Robson or qCSF) — sensitive to RGC function specifically
- Microperimetry (MAIA / MP-3) — spatial map of retinal sensitivity; tracks point-by-point recovery
- FST (Full-field Stimulus Threshold) — gold standard for low-vision populations; used in Luxturna trials
PRECEDENT
Luxturna (voretigene neparvovec) was approved on the multi-luminance mobility test plus FST plus BCVA composite. The FDA has accepted functional-vision endpoints for gene therapy in eye since 2017. A partial-reprogramming retinal program inherits this regulatory path.
3. Secondary mechanism-confirming endpoints (aging clocks)
SECONDARY
TISSUE / FLUID
Horvath multi-tissue clock (retinal tissue if obtainable, peripheral if not)
Modality: Illumina EPIC array DNA methylation · First-gen pan-tissue clock
EVIDENCE BASE
Lu et al. 2020 used a retinal-specific DNA methylation age signature to show OSK-mediated rejuvenation in RGCs. The Horvath multi-tissue clock has been applied across mouse and human reprogramming studies and is the most replicated readout in the partial-reprogramming literature.
FEASIBILITY IN HUMAN TRIALS
Retinal tissue is not generally biopsied in human trials. Peripheral DNAm age via blood is the practical surrogate; expect smaller magnitude effect because the intervention is locally delivered. Vitreous tap sampling has been used in research settings and produces RGC-enriched material.
SECONDARY
BLOOD
GrimAge / PhenoAge (mortality- and morbidity-trained clocks)
Modality: Illumina EPIC + serum proteomic (GrimAge) · Trained on mortality outcomes
WHY THESE MATTER
First-gen clocks (Horvath) are validated for age estimation but not for outcome prediction. GrimAge predicts mortality, time-to-disease, and time-to-disability with the largest effect sizes in the clock literature (Lu 2019). For a healthspan-claim program, GrimAge movement is more commercially defensible than Horvath movement.
SECONDARY
BLOOD
DunedinPACE (pace-of-aging clock)
Modality: Illumina EPIC · Trained on longitudinal organ-system decline rate
WHY THIS MATTERS FOR INTERVENTION STUDIES
DunedinPACE measures the *rate* of aging rather than an age estimate. For a single-shot AAV-OSK intervention followed by chemical maintenance, "did we slow the pace?" is more interpretable than "did we move the estimated age by X years?". Belsky et al. 2022 validated DunedinPACE response to caloric restriction (CALERIE trial); precedent for intervention responsiveness exists.
SECONDARY
BLOOD / TISSUE
CausAge / DamAge / AdaptAge (Ying 2024 causal clocks)
Modality: Illumina EPIC, CpGs filtered for causal effect via Mendelian randomization
FIELD-MOVING POTENTIAL
First-gen clocks are saturated with passenger CpGs that move with age but don't cause aging. CausAge attempts to filter for causal CpGs only. If the retina-first program moves CausAge specifically (vs only first-gen clocks), that's the kind of finding that makes a
Cell /
Nature paper.
RISK
Causal clocks are newer and have a thinner replication base. Use as exploratory secondary, not primary mechanism endpoint.
4. Exploratory endpoints (field-moving if positive)
EXPLORATORY
Retinal-specific transcriptomic age (scAge on RGC fraction)
Modality: single-cell RNA-seq on RGC-enriched vitreous samples or post-hoc on enucleation pathology specimens
RATIONALE
scAge (Trapp 2021) computes age estimates per single cell. For a tissue-targeted intervention, asking "did RGCs specifically get younger while non-RGC cells did not" is the cleanest possible mechanism confirmation. RIA-01 considers this the highest-value exploratory endpoint for a Lu-2020-lineage program.
EXPLORATORY
Plasma proteomic age (Lehallier / SomaScan / Olink)
Modality: SomaLogic SomaScan 7K or Olink Explore 3K aptamer / antibody proteomic platforms
SYSTEMIC SIGNAL FROM LOCAL INTERVENTION
Lehallier 2019 demonstrated waves of plasma proteomic age across the lifespan. The interesting exploratory question for retina-first: does a locally-delivered RGC rejuvenation produce a measurable plasma proteomic signal? Lu 2020 hinted at this for vision-related markers. If yes → the program contributes to whole-body aging conversations.
EXPLORATORY
SASP and inflammatory panel
Modality: Olink Inflammation, multiplex serum cytokine panels
RATIONALE
Senescence-associated secretory phenotype is one of the López-Otín hallmarks. Partial reprogramming has been shown to attenuate SASP (Sarkar 2020, Chondronasiou 2022). Worth tracking as part of the mechanism story, especially if combined with senolytic comparisons in any later study extension.
EXPLORATORY
H4K16ac chromatin signature (the WP2.1 mechanism check)
Modality: ChIP-seq or CUT&RUN on retinal tissue (where obtainable); peripheral PBMC ChIP-seq as surrogate
WHY THIS MATTERS
If the maintenance layer becomes KAT8-axis-based in a future protocol, H4K16ac restoration becomes the mechanism-specific confirmatory endpoint. Including PBMC ChIP-seq from baseline establishes the assay infrastructure before the KAT8 program is ready for translation.
5. Time-course design
| Timepoint | Activity | Endpoints |
|---|
| T = –4 wk | Screening + baseline | BCVA, contrast, MP, FST; baseline DNAm (blood + vitreous tap if planned); baseline proteome; baseline H4K16ac (PBMC) |
| T = –4 to 0 wk | Chemical priming phase (Vit C + α-KG oral) | Compliance + tolerability monitoring |
| T = 0 | AAV-OSK intravitreal pulse | Procedure safety endpoints |
| T = 4 wk | Early post-pulse | BCVA, contrast, MP; safety labs; OCT for inflammation |
| T = 12 wk | Mid post-pulse + maintenance start (Tazemetostat cyclic) | Full visual function panel + repeat DNAm (blood); first secondary clock readout |
| T = 26 wk | Primary endpoint window | Full visual function panel; DNAm; SASP; proteome |
| T = 52 wk | One-year durability | Full visual function; repeat clocks; durability assessment; OCT |
| T = 104 wk | Two-year | Sustainability; cumulative safety; clock re-test for durability of methylation shift |
| T = 5 yr | Long-term follow-up | Tumor surveillance; functional vision; final mechanism readouts |
6. Per-candidate clock-readout availability (what published data exists)
| Candidate |
Horvath/Hannum | GrimAge | DunedinPACE | CausAge | scAge / Tx age | SASP |
| OSK (AAV, retinal) |
YES (Lu 2020) |
No |
No |
No |
Partial (transcriptomic shift reported) |
No |
| OSKMLN (mRNA, fibroblast) |
YES (Sarkar 2020) |
No |
No |
No |
YES |
YES (Sarkar 2020) |
| Cyclic OSKM (mouse in vivo) |
YES (Ocampo 2016, Browder 2022) |
No (mouse) |
No |
No |
YES (multi-omics; Chondronasiou 2022) |
YES |
| Vitamin C + α-KG |
Partial (in vitro reports) |
No |
No |
No |
No (direct) |
No (direct) |
| Tazemetostat |
No |
No |
No |
No |
No |
No |
| Chemical cocktails (Yang 2023) |
YES (Horvath reversal claimed) |
No |
No |
No |
No |
No |
| KAT8 activator (TBD) |
UNK |
UNK |
UNK |
UNK |
UNK |
UNK |
The matrix surfaces a critical gap: second-gen and causal clocks (GrimAge, DunedinPACE, CausAge) have essentially no published partial-reprogramming data. The retina-first program collecting these in a Phase 1 / Phase 1b setting would be a first-mover contribution to the field.
7. Endpoint selection summary for the retina-first program
| Tier | Endpoint | Role | Why this one |
|---|
| Primary | BCVA + contrast sensitivity + microperimetry composite | Regulatory | Luxturna precedent; FDA-translatable |
| Co-primary | FST for low-vision populations | Regulatory in subpopulation | Required for severe-baseline patients |
| Key Secondary | Horvath DNAm (blood) and vitreous-derived if obtainable | Mechanism | Most replicated reprogramming readout |
| Secondary | GrimAge + DunedinPACE (blood) | Healthspan claim support | Predict mortality / morbidity; commercial moat |
| Secondary | OCT structural integrity (RNFL thickness) | Tissue-level mechanism | Anatomical confirmation of RGC rescue |
| Secondary | ERG / pattern ERG | Electrophysiology | Functional RGC signal independent of subjective vision |
| Exploratory | CausAge | Field-moving | If positive, distinguishes intervention from passenger CpG drift |
| Exploratory | scAge on RGC fraction (where obtainable) | Field-moving | Cell-type-resolved rejuvenation |
| Exploratory | Plasma proteomic age (Olink / SomaScan) | Systemic signal | Does local treatment produce systemic age effect? |
| Exploratory | H4K16ac ChIP / CUT&RUN (PBMC + tissue where obtainable) | Future KAT8 mechanism | Establishes assay infrastructure for KAT8-maintenance protocol |
| Safety | Standard ophthalmic safety + tumor surveillance | Risk monitoring | Tazemetostat + OSK both require it |
8. Open biomarker questions (for WP-final and beyond)
- Is a vitreous DNAm sampling protocol practical in a Phase 1/1b setting? (Worth a separate feasibility check.)
- Are there validated retinal-specific DNA methylation clocks beyond Lu 2020? (Recent preprint literature should be re-scanned.)
- Should we include CSF DNAm sampling for any CNS-extension protocol (Pairing #2)?
- What is the minimum detectable clock movement at realistic Phase 1 sample sizes? (Power analysis owed.)
- How do clock readouts interact with the Tazemetostat secondary-malignancy surveillance plan? (Some methylation perturbation may confound.)