Clinical Knowledge Summary: Blood Age Calculation (Longevity Medicine)

1. Scope

Covered:

– Estimation of biological age from blood: DNA methylation (DNAm) clocks (Horvath, Hannum, PhenoAge, GrimAge, DunedinPACE and their principal-component [PC] versions), leukocyte telomere length, and blood-chemistry/proteomic/inflammatory composite ages.[1][2][3][4]

– Evidence for these measures as predictors of mortality/morbidity, their reliability, patient selection, baseline work-up, interpretation, monitoring, and management scenarios in a private longevity setting.

Not covered:

– Detailed protocols for specific geroprotective drugs (metformin, rapamycin, senolytics, NAD⁺ precursors) — referenced only where they bear on interpreting or acting on a blood-age result.[5][3]

– Non-blood organ-specific clocks (e.g. brain-MRI age) except in brief context.[3]

– Paediatric use and gestational/epigenetic-of-pregnancy clocks.

Positioning: Blood age calculation is adjunctive to, not a replacement for, guideline-based cardiovascular, metabolic and cancer risk assessment. No NICE, MHRA, FDA or EMA guidance currently endorses blood-age testing for routine clinical decision-making; all clinical use is off-guideline and most is best regarded as research or expert-consensus practice with informed consent.[6][7]

2. Background and pathophysiology

Biological age quantifies how far an individual’s physiology deviates from that expected for their chronological age; age advancement (biological minus chronological age) predicts mortality and age-related disease independently of chronological age.[3] The geroscience hypothesis holds that shared ageing mechanisms drive multiple chronic diseases, and that a composite ageing metric may identify “fast agers” earlier than any single disease-risk score.[2]

Mechanisms captured by blood-based measures (hallmarks of ageing):

– Epigenetic alterations — age-related DNA methylation change at specific CpG sites is the substrate of epigenetic clocks.[1][8]

– Telomere attrition — leukocyte telomere shortening associates with CVD and mortality, though causality is unresolved.[8]

– Inflammaging — chronic low-grade inflammation (CRP, IL-6, CXCL9) underlies inflammatory clocks such as iAge.[2][8]

– Clonal haematopoiesis — somatic-mutation clonal expansion in blood, linked to CVD and malignancy risk.[8]

Generations of clocks:

– First-generation (Horvath, Hannum): trained to predict chronological age; proved that “age acceleration” tracks mortality risk.[1][9]

– Second-generation (PhenoAge, GrimAge): trained on clinical phenotypes/mortality; more predictive of morbidity and mortality.[1][10]

– Pace-of-aging (DunedinPACE): trained on longitudinal multi-organ decline; behaves as a “speedometer” rather than an “odometer” and is the measure most consistently responsive to intervention in RCTs.[2][11][12]

Preclinical (clearly separated — mechanistic/animal only, not a basis for clinical action): Caloric restriction, metformin, rapamycin/rapalogs and senolytics (dasatinib+quercetin, fisetin) extend lifespan/healthspan and reduce senescent-cell burden and epigenetic ageing in model organisms; long-term metformin slowed multi-organ ageing in primates. These data establish plausibility only and do not support human clinical benefit.[5][3][13][14]

3. Evidence base and grading

Available human evidence comprises: large prospective cohorts and cohort meta-analyses (predictive validity, longitudinal behaviour), a small number of RCTs and post-hoc RCT analyses (intervention responsiveness — chiefly CALERIE), systematic reviews of methodology, and reliability/technical-validation studies.[10][15][7][11][16][17] Key outcomes studied are all-cause and cause-specific mortality, incident CVD/cancer, frailty and functional/cognitive phenotypes, and the clocks’ own reliability as surrogate endpoints.

Outcome 1 — Blood-based biological age predicts all-cause and cardiovascular mortality

– Evidence statement: Moderate-certainty evidence from multiple large prospective cohorts and a systematic review (e.g. TILDA n≈490; NHANES n≈2,105 followed to 2019) shows GrimAge (and to a lesser degree PhenoAge, Hannum, Horvath) age acceleration predicts all-cause, CVD and cancer mortality independently of chronological age and conventional risk factors.[10][15][9][18]

– GRADE assessment: Downgraded for indirectness (observational associations in specific populations; predictive utility differs by ancestry — Horvath/Hannum/GrimAge were less predictive in Hispanic participants); some inconsistency between clocks (first-generation clocks often not predictive after adjustment). Not downgraded for imprecision (large samples, consistent direction).[15][10]

– Strength: Conditional recommendation to use second-generation/pace-of-aging clocks (not first-generation) if a blood-age measure is used for prognostic enrichment — as an adjunct, not a stand-alone risk tool.

Outcome 2 — Blood-based biological age predicts frailty, functional decline and intrinsic-capacity decline

– Evidence statement: Low-to-moderate certainty from cohort data; GrimAge acceleration associates with walking speed, frailty, polypharmacy and declining intrinsic capacity, with effects strengthening at older ages.[2][10][19]

– GRADE assessment: Downgraded for risk of bias/residual confounding and indirectness (surrogate functional endpoints).

– Strength: Conditional recommendation / only as adjunct to validated frailty assessment.

Outcome 3 — Blood-age measures respond to intervention (as a surrogate endpoint)

– Evidence statement: Low certainty. In the CALERIE RCT (n=220, 2 years, 25% prescribed CR; ~12% achieved), CR slowed DunedinPACE by ~2–3% but did not significantly change PhenoAge or GrimAge; effect sizes were small and the trial’s primary ageing outcome was not met. Small RCTs of multimodal lifestyle interventions show similar ~2% DunedinPACE deceleration. Diet-quality and physical-activity associations are consistent in cohorts but observational.[12][11][20][21][22][23]

– GRADE assessment: Downgraded for imprecision (small effect, single pivotal trial, post-hoc analyses), indirectness (surrogate endpoint), and inconsistency across clocks.

– Strength: Only in research. There is no evidence that a change in blood age translates into reduced disease or mortality in individuals; the linkage from surrogate change to hard outcome is inferred, not proven.[3][11]

Outcome 4 — Measurement reliability of blood-age tests

– Evidence statement: Moderate certainty that original (non-PC) clocks have poor technical reliability — replicate deviations of up to ~9 years and, for some clocks, up to ~15–20 years across platforms — whereas PC-based clocks (PCGrimAge, PC PhenoAge) and SystemsAge are substantially more reproducible (replicate agreement often within ~1–1.5 years). Biological (within-person, short-interval) reliability remains only low-to-moderate even for good clocks.[16][24][25][26]

– Strength: Strong recommendation to use PC-based/reliability-optimised clocks and to avoid interpreting single-timepoint original-clock results, or small between-visit changes, as clinically meaningful.

Overall: there is currently no consensus gold-standard measure of biological age; systematic review supports Klemera–Doubal-type composite methods as reliable, but standardised reference ranges and clinical decision thresholds do not exist.[6][7]

4. Patient selection and indications

Who might reasonably be offered blood age calculation (as adjunctive, consented, mostly research-grade testing):

– Middle-aged adults (approximately 40–70) with elevated cardiometabolic risk (obesity, prediabetes/T2DM, hypertension, dyslipidaemia, smoking history) seeking risk stratification and motivation for lifestyle change, alongside conventional QRISK/lipid/HbA1c assessment.[2][18][3]

– Older adults where an objective ageing/frailty adjunct may inform shared decisions, understanding limited actionability.[2][19]

– Individuals already committed to a structured lifestyle programme, where a reliability-optimised measure (PC clock/DunedinPACE) is tracked within a defined protocol or registry.[11][21]

Clinical scenarios where value is lowest / caution highest:

– Using a single result to diagnose disease, to over-ride guideline-based prevention, or to justify unproven pharmacological “anti-ageing” therapy — not supported.[6][3]

– Athletic/”optimisation” clients seeking reassurance: reasonable for engagement but must be framed as research-grade with wide measurement uncertainty.[16][24]

Exclusions / interpret with caution:

– Acute illness, recent infection, acute stress, recent major dietary change or non-fasting state — short-term biological variation reduces reliability.[16]

– Active malignancy, haematological disease or recent chemotherapy/transfusion — alters leukocyte composition and methylation.

– Non-white ancestry — reduced/uncertain predictive validity for several clocks.[15]

– Pregnancy — adult clocks not validated; pregnancy alters methylation.

Regulatory/ethical status: Off-label/off-guideline; no regulator endorses blood-age testing for clinical decisions. Best practice is adjunctive care with explicit informed consent, ideally within a clinical trial or registry framework, making clear that the test is prognostic and its actionability is unproven.[6][7][3]

5. Assessment and baseline work-up

Pre-test assessment:

– History: cardiometabolic risk factors, smoking/alcohol, physical activity, diet, sleep, medications, family longevity/disease history, current acute illness/stress (to time sampling appropriately).[18][22]

– Examination and validated tools: BP, BMI/waist circumference; grip strength/gait speed and a frailty score in older adults; conventional CVD risk (QRISK3) and metabolic assessment as standard of care.[2][3]

– Sampling standardisation: fasting, avoiding acute illness/intense exercise, consistent laboratory and platform between visits — because meals, stress and pre-analytics measurably shift results.[16]

Baseline investigations tailored to blood age:

– The blood-age assay itself — prefer a PC-based/reliability-optimised DNAm clock, reporting both an “odometer” (PC GrimAge/PhenoAge) and a “speedometer” (DunedinPACE).[2][16][25]

– Supporting clinical labs that also feed composite ages and aid interpretation: FBC (leukocyte composition), HbA1c, lipids, renal/hepatic profile, hs-CRP, cystatin C; consider these as a transparent, lower-cost adjunct/alternative to proprietary clocks.[27][3]

– Telomere length only as a supplementary, lower-certainty marker.[8][28]

Risk stratification: Integrate blood-age acceleration with conventional risk scores rather than in isolation. Treat marked GrimAge/PhenoAge acceleration or high DunedinPACE as a flag prompting intensified conventional risk-factor management, not as an independent diagnosis.[2][10][3]

Baseline documentation: record clock name and version (PC vs original), platform/array (450K/EPIC/sequencing), laboratory, sampling conditions, chronological age, computed age acceleration/pace, and the specific clinical labs used — essential because cross-platform and cross-clock results are not interchangeable.[24][26]

6. Dosing regimens and practical implementation

Blood age calculation is a diagnostic assessment, not a therapeutic — “implementation” concerns testing protocol and any evidence-based interventions triggered.

Testing protocol (supported by reliability data):

– Use PC-based clocks; keep laboratory, platform and sampling conditions constant across serial tests.[16][24][25]

– Do not repeat testing at short intervals: given regression-to-the-mean and small annual change (PC clocks change on the order of ~0.14–0.16 years/year; DunedinPACE shows no significant short-term change in healthy older adults), minimum meaningful re-test intervals are ≥12 months, preferably longer.[24][17]

Interventions that may be offered on the basis of a result (all as general preventive/lifestyle care — robust for health, only surrogate-level evidence for changing blood age):

– Supported by robust human data for health outcomes, and associated with slower epigenetic ageing in RCT/cohort data: aerobic exercise/physical activity; higher diet quality (Mediterranean, AHEI); weight management; smoking cessation; alcohol moderation. These are recommended on their own merits regardless of blood-age result.[8][22][23][21]

– Caloric restriction: the only intervention shown in an RCT to slow a blood-age measure (DunedinPACE), effect small; sustained CR risks loss of lean mass and bone density and long-term survival benefit in non-obese humans is unproven — advise cautiously.[8][11][20]

– Require caution — early-phase/preclinical, not recommended for the purpose of lowering blood age: metformin (TAME ongoing; human data mixed), rapamycin/rapalogs, senolytics, NAD⁺ precursors. Prescribing these to “reverse blood age” is experimental/off-label with no outcome evidence.[5][3][13][14]

7. Monitoring, safety and follow-up

The test itself is low-risk (venepuncture); the principal harms are informational and behavioural — misinterpretation, anxiety, false reassurance, and cascade testing or unproven treatment.

Monitoring plan:

– Track the same PC clock and DunedinPACE on the same platform; interpret only changes exceeding the assay’s known measurement error, not single-point shifts.[16][25]

– Concurrently monitor conventional, actionable parameters (BP, HbA1c, lipids, weight, hs-CRP) and function (gait speed/grip strength in older adults) — these should drive management.[2][3]

Timepoints:

– Short-term (<12 months): generally do not re-test blood age; focus on conventional risk-factor and lifestyle review.

– Medium-term (12–24 months): reasonable interval for a repeat blood-age measure if it is being tracked.[11][24]

– Long-term (≥2 years): assess sustained trajectory alongside clinical endpoints.

Adverse effects/safety:

– Direct: venepuncture-related only.

– Indirect (the main safety issue): anxiety/false reassurance, over-investigation, and initiation of unproven/off-label geroprotective drugs with their own risk profiles. Action for a “worrying” result: reassure regarding measurement uncertainty, and redirect to evidence-based conventional risk-factor optimisation rather than novel therapy.[3]

Interactions/confounders to record and account for: acute illness/infection, recent vaccination, corticosteroids/immunomodulators, chemotherapy, transfusion, pregnancy, and changes in leukocyte composition — all can shift blood-age estimates independent of true ageing.[16]

Special populations: pregnancy/breastfeeding — adult clocks not validated, avoid; significant renal/hepatic impairment — alters cystatin C/creatinine-based and GrimAge surrogate components, interpret cautiously; frailty/extremes of age — predictive value best-established but actionability limited.[2][3][21]

8. Contraindications and cautions

Absolute (do not test, or do not interpret):

– Using the result as a sole basis to diagnose disease or to start/stop a licensed treatment — no validated decision thresholds exist.[6][7]

– Testing to justify unproven anti-ageing pharmacotherapy outside a trial.[3]

Relative / specialist input advised:

– Active malignancy, haematological disease, recent chemotherapy/transfusion, acute severe illness.[8][16]

– Pregnancy/breastfeeding [adult clocks not validated].

– Non-white ancestry (reduced predictive validity).[15]

– Vulnerability to health anxiety.

Harm likely to outweigh benefit: any pathway where a blood-age result triggers unproven, potentially harmful interventions, or displaces guideline-based prevention.[3]

9. Practical management scenarios

Scenario A — Middle-aged adult with multiple cardiometabolic risk factors

– Recommendation: Consider blood-age testing as an adjunct (conditional) — chiefly for risk communication and behaviour change, not to alter guideline-based management.[2][3]

– Assessment: full conventional CVD/metabolic work-up (QRISK3, lipids, HbA1c, BP); PC GrimAge/PhenoAge + DunedinPACE under standardised sampling.

– Shared decision-making/consent: explain the test is prognostic, off-guideline, has measurement uncertainty, and that acting on it is unproven for outcomes.

– Initiation: whatever the result, deliver evidence-based risk-factor optimisation (statin/antihypertensive per NICE where indicated; exercise, diet quality, weight, smoking cessation).[8][22][23]

– Monitoring/follow-up: manage by conventional targets; re-test blood age no sooner than 12–24 months if tracking.[11][24]

– Escalate/refer/stop: refer per standard thresholds (e.g. established CVD, uncontrolled diabetes); stop blood-age monitoring if it drives anxiety or non-evidence-based treatment.

Scenario B — Older, frail patient with multimorbidity

– Recommendation: Restrict largely to research / avoid routine use. Predictive but poorly actionable; validated frailty assessment and CGA are preferred.[2][19][7]

– Assessment: comprehensive geriatric assessment, frailty scoring, functional testing.

– Consent: emphasise limited actionability and potential for distress.

– Management: focus on falls, polypharmacy, nutrition (avoid caloric restriction — risk of sarcopenia), function.[8]

– Escalate/refer: geriatric/specialist input; do not initiate experimental geroprotectors.

Scenario C — Adjunct in a patient already under specialist care

– Recommendation: Consider only with specialist coordination, within a trial/registry where possible; avoid duplicative or contradictory management.[3]

– Assessment: confirm current specialist plan; avoid interference with disease-specific therapy.

– Consent/coordination: share results with the treating team; frame blood age as supplementary.

– Monitoring: align with the specialist’s schedule; act on conventional parameters.

– Stop/refer: discontinue if it introduces confusion or unproven interventions; defer disease decisions to the specialist.

10. Research gaps and future directions

– Surrogate-to-outcome linkage: no RCT has shown that lowering blood age reduces disease incidence or mortality; long-term trials with hard endpoints are the priority.[3][11]

– Standardisation: no consensus gold-standard clock, no reference ranges, no agreed clinically meaningful change; cross-platform/cross-clock non-interchangeability persists.[6][7][24][26]

– Reliability: biological (within-person) reliability remains only low-to-moderate even for PC clocks; improving it is essential before individual-level clinical use.[16]

– Generalisability: predictive validity differs by ancestry and is under-studied in non-white and non-European populations.[15]

– Intervention responsiveness: only DunedinPACE has RCT-level responsiveness (small effect, CALERIE); which clock best serves as a trial endpoint is unresolved.[11][21]

– Ideal current setting: blood age calculation is best confined to well-designed clinical trials and prospective registries rather than as a driver of individual clinical decisions.[3][11]

References

1. Measures of Aging Biology in Saliva and Blood as Novel Biomarkers for Stroke and Heart Disease in Older Adults. Waziry R, Gu Y, Boehme AK, Williams OA. Neurology. 2023;101(23):e2355-e2363. doi:10.1212/WNL.0000000000207909.
2. Impact of Geroscience on Therapeutic Strategies for Older Adults With Cardiovascular Disease: JACC Scientific Statement. Forman DE, Kuchel GA, Newman JC, et al. Journal of the American College of Cardiology. 2023;82(7):631-647. doi:10.1016/j.jacc.2023.05.038.
3. Geroscience. Kritchevsky SB, Cummings SR. JAMA. 2025;334(12):1094-1102. doi:10.1001/jama.2025.11289.
4. Measuring Biological Age: Insights From Omics Studies. Kočar E, Šket R, Vasle AH, et al. Ageing Research Reviews. 2025;:102988. doi:10.1016/j.arr.2025.102988.
5. Drugs Targeting Mechanisms of Aging to Delay Age-Related Disease and Promote Healthspan: Proceedings of a National Institute on Aging Workshop. Espinoza SE, Khosla S, Baur JA, de Cabo R, Musi N. The Journals of Gerontology. Series A, Biological Sciences and Medical Sciences. 2023;78(Suppl 1):53-60. doi:10.1093/gerona/glad034.
6. Integrating Biological Age, Epigenetic Clocks, and Telomere Length in Precision Nutrition Strategies for Chronic Disease Management: Potential Frameworks and Ongoing Challenges. Carvalho BG, Ribeiro AA, da Mota JCNL, Carvalho LM, Nicoletti CF. Nutrition Research (New York, N.Y.). 2025;140:135-160. doi:10.1016/j.nutres.2025.06.010.
7. Methods for the Assessment of Biological Age – A Systematic Review. Zurbuchen R, von Däniken A, Janka H, von Wolff M, Stute P. Maturitas. 2025;195:108215. doi:10.1016/j.maturitas.2025.108215.
8. Biological Versus Chronological Aging: JACC Focus Seminar. Hamczyk MR, Nevado RM, Barettino A, Fuster V, Andrés V. Journal of the American College of Cardiology. 2020;75(8):919-930. doi:10.1016/j.jacc.2019.11.062.
9. “Epigenetic clocks”: Theory and applications in human biology. Ryan CP. American Journal of Human Biology : The Official Journal of the Human Biology Council. 2021;33(3):e23488. doi:10.1002/ajhb.23488.
10. GrimAge Outperforms Other Epigenetic Clocks in the Prediction of Age-Related Clinical Phenotypes and All-Cause Mortality. McCrory C, Fiorito G, Hernandez B, et al. The Journals of Gerontology. Series A, Biological Sciences and Medical Sciences. 2021;76(5):741-749. doi:10.1093/gerona/glaa286.
11. Effect of Long-Term Caloric Restriction on DNA Methylation Measures of Biological Aging in Healthy Adults From the CALERIE Trial. Waziry R, Ryan CP, Corcoran DL, et al. Nature Aging. 2023;3(3):248-257. doi:10.1038/s43587-022-00357-y.
12. Cut Calories, Lengthen Life Span? Randomized Trial Uncovers Evidence That Calorie Restriction Might Slow Aging, but Questions Remain. Rubin R. JAMA. 2023;329(13):1049-1050. doi:10.1001/jama.2023.2437.
13. Cellular Senescence, Inflammaging and Cardiovascular Disease. Zanders L, Arifaj D, Wagner JUG, Dimmeler S. Immunological Reviews. 2026;337(1):e70084. doi:10.1111/imr.70084.
14. New Horizons: Novel Approaches to Enhance Healthspan Through Targeting Cellular Senescence and Related Aging Mechanisms. Tchkonia T, Palmer AK, Kirkland JL. The Journal of Clinical Endocrinology and Metabolism. 2021;106(3):e1481-e1487. doi:10.1210/clinem/dgaa728.
15. Epigenetic Age Acceleration and Mortality Risk Prediction in US Adults. Mendy A, Mersha TB. GeroScience. 2025;47(4):6029-6038. doi:10.1007/s11357-025-01604-x.
16. Biological Versus Technical Reliability of Epigenetic Clocks and Implications for Disease Prognosis and Intervention Response. Sehgal R, S Borrus D, Gonzalez J, et al. Aging Cell. 2026;25(8):e70635. doi:10.1111/acel.70635.
17. Tracking the Epigenetic Clock Across the Human Life Course: A Meta-Analysis of Longitudinal Cohort Data. Marioni RE, Suderman M, Chen BH, et al. The Journals of Gerontology. Series A, Biological Sciences and Medical Sciences. 2019;74(1):57-61. doi:10.1093/gerona/gly060.
18. A Systematic Review of Biological, Social and Environmental Factors Associated With Epigenetic Clock Acceleration. Oblak L, van der Zaag J, Higgins-Chen AT, Levine ME, Boks MP. Ageing Research Reviews. 2021;69:101348. doi:10.1016/j.arr.2021.101348.
19. Accelerated Epigenetic and Inflammatory Aging and Intrinsic Capacity. Rouch L, De Souto Barreto P, Rolland Y, et al. JAMA Network Open. 2026;9(7):e2624102. doi:10.1001/jamanetworkopen.2026.24102.
20. Change in the Rate of Biological Aging in Response to Caloric Restriction: CALERIE Biobank Analysis. Belsky DW, Huffman KM, Pieper CF, Shalev I, Kraus WE. The Journals of Gerontology. Series A, Biological Sciences and Medical Sciences. 2017;73(1):4-10. doi:10.1093/gerona/glx096.
21. Short-Term Responsiveness of DNA Methylation-Based Aging Biomarkers to a Multimodal Intervention Comprising Exercise and Dietary Guidance Involving Daily Consumption of Yogurt Containing Bifidobacterium Longum BB536: An Exploratory Randomized Controlled Trial. Nishimura T, Horigome A, Tanaka M, Hishida Y, Odamaki T. Aging. 2026;18(1):639-655. doi:10.18632/aging.206386.
22. Diet Quality, Physical Activity, and Epigenetic Aging in the Finnish Working-Age Population. Autio I, Saarinen A, Marttila S, et al. The Journal of Nutrition. 2026;156(6):101540. doi:10.1016/j.tjnut.2026.101540.
23. Body Size, Diet Quality, and Epigenetic Aging: Cross-Sectional and Longitudinal Analyses. Li DL, Hodge AM, Cribb L, et al. The Journals of Gerontology. Series A, Biological Sciences and Medical Sciences. 2024;79(4):glae026. doi:10.1093/gerona/glae026.
24. Longitudinal Changes in Epigenetic Measures Over 2 Years: Methodological Implications. Hamaya R, Li S, Chen BH, et al. GeroScience. 2025;:10.1007/s11357-025-01990-2. doi:10.1007/s11357-025-01990-2.
25. A Computational Solution for Bolstering Reliability of Epigenetic Clocks: Implications for Clinical Trials and Longitudinal Tracking. Higgins-Chen AT, Thrush KL, Wang Y, et al. Nature Aging. 2022;2(7):644-661. doi:10.1038/s43587-022-00248-2.
26. Analysis of Variability and Epigenetic Age Prediction Across Microarray and Methylation Sequencing Technologies. Shokhirev MN, Johnson AA. GeroScience. 2025;47(5):6631-6638. doi:10.1007/s11357-025-01824-1.
27. Simplified Assay for Epigenetic Age Estimation in Whole Blood of Adults. Vidal-Bralo L, Lopez-Golan Y, Gonzalez A. Frontiers in Genetics. 2016;7:126. doi:10.3389/fgene.2016.00126.
28. Effect of Long-Term Caloric Restriction on Telomere Length in Healthy Adults: CALERIE™ 2 Trial Analysis. Hastings WJ, Ye Q, Wolf SE, et al. Aging Cell. 2024;23(6):e14149. doi:10.1111/acel.14149.