Epigenetic Clocks (DunedinPACE, Horvath, GrimAge)

Navigation & Literature

• Parent Silo: 03 - Diagnostics & Quantified Self MOC
• Literature Dossier: Epigenetic Clocks (DunedinPACE, Horvath, GrimAge) - References

---

1. Epigenetic Clock Generations & Reference Range Comparison

Clock Generation & Metric	Standard Population Baseline	Longevity / Biohacking Optimal Target	Primary Unit	Underlying CpG Architecture & Biological Signal
3rd Gen: DunedinPACE	$1.00\text{ yr/yr}$ (mean pace at age 45)	$<0.80\text{ yr/yr}$ (elite $<0.72\text{ yr/yr}$)	$\text{Bio Yrs / Chrono Yr}$	173 high-reliability CpGs trained on 20-year longitudinal decline of 19 organ-system biomarkers
2nd Gen: DNAm GrimAge / GrimAge2	$\Delta \text{Age}=0.0\text{ yrs}$	$\Delta \text{Age}<-3.0\text{ to }-6.0\text{ yrs}$	$\Delta \text{ Years (AA)}$	1,030 CpGs predicting 10 DNAm plasma proteins (GDF-15, PAI-1, Cystatin C, CRP, A1c, TIMP-1) + pack-years
2nd Gen: DNAm PhenoAge (Levine)	$\Delta \text{Age}=0.0\text{ yrs}$	$\Delta \text{Age}<-4.0\text{ to }-8.0\text{ yrs}$	$\Delta \text{ Years (AA)}$	513 CpGs trained on 9 clinical blood markers (Albumin, Creatinine, Glucose, CRP, Lymphocyte %, MCV, RDW, ALP, WBC)
2nd Gen: OMICmAge	$\Delta \text{Age}=0.0\text{ yrs}$	$\Delta \text{Age}<-3.5\text{ yrs}$	$\Delta \text{ Years (AA)}$	Multi-omic epigenetic surrogate integrating metabolomics, proteomics, and clinical EHR mortality risk
1st Gen: Horvath Pan-Tissue / Hannum	$\Delta \text{Age}=0.0\text{ yrs}$ (353 / 71 CpGs)	$\text{IEAA}<-2.0\text{ yrs}$	$\Delta \text{ Years (AA)}$	Trained solely on chronological age; measures Intrinsic Epigenetic Age Acceleration (IEAA)
Mitotic / Stem Cell Clock (epiTOC2)	Population age-matched percentile	$<25\text{th percentile}$	Cumulative Divisions	Cumulative stem cell mitotic divisions (PRC2 polycomb target CpG hypermethylation)

---

2. Physiological Significance & Hazard Curves

First, Second, and Third-Generation Epigenetic Clocks

DNA methylation (5-methylcytosine at CpG dinucleotides) regulates chromatin accessibility and gene transcription across the lifespan. However, the mathematical training target determines what a clock actually measures:

1. First-Generation Clocks (Horvath 2013, Hannum 2013): Trained via elastic-net regression solely to predict chronological age. Paradoxically, a perfect predictor of chronological age would assign the exact same score to a dying 60-year-old and a marathon-running 60-year-old, making 1st-gen clocks weak predictors of mortality ($\text{HR }\approx 1.02–1.10$ per SD) and unresponsive to lifestyle interventions.
2. Second-Generation Clocks (PhenoAge 2018, GrimAge 2019, GrimAge2 2022): Trained to predict composite clinical phenotypes and time-to-death. GrimAge2 incorporates DNA methylation surrogates for GDF-15 (mitochondrial stress myokine), PAI-1 (plasminogen activator inhibitor-1, a hallmark SASP senescence factor linked to visceral adiposity), Cystatin C, Leptin, TIMP-1, Adrenomedullin, hs-CRP, and HbA1c. A 1-SD increase (~3.5–4 years) in GrimAge acceleration increases all-cause mortality hazard by $35–100\%$ ($\text{HR }1.35–2.04$) and coronary heart disease risk by $40–80\%$.
3. Third-Generation Pace of Aging (DunedinPACE 2022): Rather than estimating static cumulative biological age (an odometer), DunedinPACE acts as a speedometer. Trained on 20 years of longitudinal within-person slopes across 19 cardiovascular, metabolic, renal, hepatic, immune, pulmonary, and dental biomarkers in the Dunedin birth cohort ($n=1,037$, all born in 1972–1973, eliminating birth-cohort and survival bias) and filtered solely for CpG probes with intraclass correlation reliability ($\text{ICC}>0.4$), DunedinPACE shows high test-retest precision ($\text{ICC}>0.87$). Every 1-SD increase in DunedinPACE (~0.10 units, or 10% faster aging) increases all-cause mortality by 56–65% ($\text{HR }1.56–1.65$) and incident chronic disease by 54%.

Intrinsic vs. Extrinsic Epigenetic Age & Immune Confounding

Whole-blood DNA methylation is heavily influenced by leukocyte subset proportions (naive CD8+ T cells decline with age, while exhausted senescent CD8+CD28- T cells and granulocytes rise). Acute viral infections (e.g., SARS-CoV-2, influenza) or high-stress periods transiently shift neutrophil-to-lymphocyte ratios, artificially spiking Extrinsic Epigenetic Age and DunedinPACE for 3–6 weeks. Modern bioinformatic pipelines apply 12-cell-type immune deconvolution (FlowSorted.BloodExtended) to separate immune cell shifts from cell-intrinsic methylation changes.

---

3. Root Causes of Accelerated Epigenetic Aging

• Elevations in DNAm PAI-1 & GDF-15 (GrimAge2 Drivers): Visceral adipose tissue accumulation, hyperinsulinemia, mitochondrial electron transport chain stress, and cellular senescence (SASP).
• Elevations in DunedinPACE ($>0.95\text{ yr/yr}$): Caloric excess, ultra-processed food intake, sleep curtailment ($<6.5\text{ h/night}$), chronic psychological stress (glucocorticoid response elements flank 85 of the 353 Horvath CpGs), smoking/nicotine/pollutant exposure (AHRR cg05575921 hypomethylation), and methyl-donor / alpha-ketoglutarate (TET demethylase cofactor) insufficiency.

---

4. Actionable Interventions

Foundational Protocols (01 - Foundations)

• Caloric Restriction & Fasting Mimetics (CALERIE Validation): Mild sustained caloric restriction (11–15% below maintenance) or structured Time-Restricted Feeding & Fasting slows DunedinPACE by $2–3\%$ ($p=0.0003$), equivalent to a 10–15% reduction in mortality hazard.
• High-Fitness Cardiorespiratory & Resistance Training: Regular Zone 2 & VO2 Max and Resistance Training & Hypertrophy lower DNAm PAI-1, DNAm Leptin, and DNAm GDF-15 while preserving naive T-cell fractions.
• Polyphenol & Methyl-Adaptogen Diet (Fitzgerald Protocol): Combine dietary methyl donors (folate, betaine, choline) with epigenetic methyl-adaptogens (curcumin, EGCG, quercetin, luteolin, rosmarinic acid, sulforaphane) per Macronutrient & Micronutrient Optimization, which prevent hypermethylation of tumor-suppressor/longevity promoter CpG islands while supporting global DNMT maintenance.

Molecular & Pharmacological Interventions (02 - Compounds & Molecules)

• TET Demethylase Cofactors (Alpha-Ketoglutarate + Vitamin C): Ten-Eleven Translocation (TET1/2/3) dioxygenases—which actively remove aberrant DNA methylation marks via 5-hydroxymethylcytosine—require $\alpha$-Ketoglutarate (Ca-AKG $1000\text{ mg/day}$), Vitamin C, and ${\text{Fe}}^{2+}$ as obligate enzymatic cofactors.
• Thymic Regeneration (TRIIM Trial Protocol): In older adults with thymic involution, recombinant human growth hormone (or GH Secretagogues (Ipamorelin, CJC-1295, Tesamorelin)) combined with DHEA and Metformin (AMPK & Glycemic Agents (Metformin, Berberine, Acarbose)) or thymic peptides (Immune & Thymic Peptides (Thymosin Alpha-1, LL-37), Khavinson Bioregulators (Epitalon, Pinealon)) reversed GrimAge by $-1.5\text{ to }-2.5\text{ years}$ over 12 months.
• One-Carbon & NAD+ Homeostasis: Maintain SAMe availability via Methylation & B-Complex and Creatine & Phosphocreatine System.

---

5. Key Scientific Evidence Summary

• Belsky et al. (2022, eLife): Created and validated DunedinPACE, demonstrating superior test-retest reliability ($\text{ICC}\ge 0.87$) and robust prediction of morbidity and mortality ($\text{HR }1.65$ per SD in Framingham).
• Waziry et al. (2023, Nature Aging): Proved in the 2-year CALERIE Phase 2 RCT ($n=220$) that caloric restriction significantly slows the pace of aging measured by DunedinPACE ($p<0.001$).
• Lu et al. (2019 & 2022, Aging): Developed DNAm GrimAge and GrimAge2, establishing DNAm PAI-1 and GDF-15 as top predictors of healthspan and all-cause mortality across $>13,000$ multi-ethnic subjects.
• Fahy et al. (2019, Aging Cell — TRIIM Trial): Demonstrated thymus regeneration and a $-2.5\text{ year}$ reversal across four epigenetic clocks (including GrimAge) following a 12-month rhGH + DHEA + Metformin protocol.

See full bibliographic details and quantitative effect sizes in Epigenetic Clocks (DunedinPACE, Horvath, GrimAge) - References.