- Harvard Health Publishing’s Pathways to Longevity is its first Special Health Report devoted to longevity, and you served as its medical editor. Why was this the right moment, and which editorial judgments were most difficult?
Harvard Health Publishing does not publish at this scale on subjects supported primarily by marketing claims, so I see the report as an institutional recognition that aging biology has become coherent enough to explain responsibly to a broad audience.
Several developments came together. Geroscience moved from description toward mechanism, giving the field a shared vocabulary through the hallmarks of aging. A credible clinical pipeline also emerged, with therapies such as GLP-1 and SGLT2 agents demonstrating cardiovascular and renal benefits in defined populations, alongside earlier signals in other age-related conditions. At the same time, the public became more engaged and sophisticated. People now arrive with wearable data, laboratory results, and increasingly detailed questions about prevention and long-term health.
That interest has created both an opportunity and a responsibility. Consumers encounter longevity information of widely varying quality, while clinicians, health systems, product developers, and care organizations are trying to determine which approaches are ready for implementation and which still require validation. Translating the field for a broad audience therefore felt timely rather than premature.
The most difficult editorial task was avoiding two equally unhelpful extremes: presenting every emerging intervention as a breakthrough or dismissing promising areas simply because the evidence is incomplete.
Biological age testing was perhaps the clearest example. The underlying science is meaningful and advancing rapidly. These tools may eventually become useful components of individualized risk assessment and longitudinal care. At the same time, a single consumer test cannot yet bear the weight of a definitive personal judgment.
We therefore focused on explaining what the tests measure, where their limitations lie, and what questions readers should ask when evaluating a result. That same approach guided sections on stem cells, hyperbaric oxygen, supplements, and other developing areas. Each includes legitimate applications, promising research, and organizations working carefully to establish stronger evidence. The important distinction is between responsible use within a defined context and claims that extend beyond what current data can support.
Another challenge was resisting recency bias. The frontier is exciting, but it should not displace the foundational interventions that still carry the strongest evidence in medicine. The discipline the report required was to be enthusiastic where the evidence justified it, cautious where it did not, and comfortable saying, “We do not know yet.”
- Which human outcomes and forms of evidence should determine whether a longevity intervention is genuinely beneficial, given that trials using lifespan as an endpoint may take decades?
This is the central argument of my editorial in Aging, “Healthy Life Extension: Geroscience’s North Star.” The objective is not simply to add years. It is to extend the period during which people remain healthy, functional, and independent.
That idea is closely related to James Fries’s concept of compression of morbidity: delaying disease and disability so that the period of decline occupies a smaller portion of life. Health-adjusted life expectancy and quality-adjusted life years are useful because they account for function and quality of life rather than survival alone. I have argued that the field should move beyond treating lifespan and healthspan as competing goals and instead organize itself around healthy life extension.
I use an explicit hierarchy when evaluating outcomes.
Hard human outcomes come first, including all-cause mortality, cardiovascular events, cancer events, incident disease, disability, hospitalization, and loss of independence. Validated functional measures come next. Cardiorespiratory fitness is among the clearest examples. Mandsager’s analysis of 122,007 adults found that low fitness was associated with mortality risk comparable to or greater than several major clinical risk factors. Kodama’s meta-analysis estimated approximately 13 percent lower all-cause mortality and 15 percent fewer cardiovascular events for each one-MET improvement in fitness.
Grip strength, muscle function, cognitive trajectory, frailty, and immune resilience also belong in this tier when they are measured with validated methods and linked to meaningful outcomes. Biomarkers are valuable, particularly for early-stage research and treatment monitoring, but they should be interpreted according to how well they predict or mediate the outcomes that matter to patients.
Because multidecade lifespan trials are rarely practical, the field must triangulate. Strong preclinical evidence includes replication across species, sites, sexes, and research groups. This is why the National Institute on Aging’s Interventions Testing Program is so valuable. Its standardized, multisite approach provides a stronger signal than an isolated result from a single laboratory.
For human evidence, I prioritize prospective designs, clearly defined eligibility criteria, transparent treatment assignment, appropriate comparison groups, meaningful follow-up, and independent replication. Genetic approaches can also help distinguish causation from correlation. These safeguards matter because healthy-user bias, immortal-time bias, and other design problems have influenced many historical claims in nutrition and supplementation.
For companies and clinical organizations developing longevity interventions, this hierarchy also provides a practical development roadmap. Early biomarkers and mechanistic findings may justify continued investment, but recommendations become more credible as they progress toward validated functional and clinical outcomes.
The ultimate question is straightforward: does the intervention help people remain healthier, more capable, and independent for longer, or does it only change a measurement?
- How useful are the hallmarks framework and biological age testing today for personalizing prevention, and where do they remain research tools?
The hallmarks framework, expanded by López-Otín and colleagues in 2023 from nine to twelve hallmarks, is one of the most useful organizing concepts in modern biology. I expect it to become an increasingly important mental model for chronic disease prevention and treatment.
Its primary value today is conceptual. It reframes aging as a network of interconnected and potentially modifiable mechanisms rather than as an indistinct, inevitable process. It also helps explain why an upstream biological pathway may influence multiple age-related diseases.
What the framework does not yet provide is a simple clinical menu in which each hallmark corresponds to a single test or treatment. The hallmarks interact, and changing one pathway may influence several others. Responsible clinical translation therefore requires integrated assessment, careful intervention selection, and monitoring rather than isolated pathway targeting.
Biological age testing deserves similarly precise treatment. My position is not that these tests lack value. Population-level evidence is increasingly compelling, and the technology is improving. The question is how they should be incorporated into individual care.
Mavrommatis and colleagues compared fourteen epigenetic clocks against 174 incident disease outcomes in 18,859 adults from Generation Scotland. Second- and third-generation clocks meaningfully predicted disease incidence and mortality, particularly for respiratory and liver-related conditions. However, no single clock performed best across all diseases. That finding illustrates both the promise and the present limitation: interpretation depends partly on which clock, tissue, platform, and assay were used.
There is also within-person variation across repeated samples. Preliminary work from Higgins-Chen’s group suggests that some clocks with strong technical performance may show less biological stability over time. Because that work remains a preprint, it should be treated as informative but provisional.
The practical risk is not the use of biological age testing itself. It is treating one result as a definitive verdict. That can create false precision, unnecessary anxiety, and interventions aimed primarily at changing a score.
I see greater value when these tools are used longitudinally, with a consistent assay, and interpreted alongside established measures such as ApoB, Lp(a), coronary artery calcium, cardiorespiratory fitness, body composition, metabolic health, and functional status. In that setting, a biological age measure may contribute an additional layer of information.
For clinics and testing companies, the opportunity is to build better interpretation systems around the technology. That includes repeat-testing standards, transparent explanations of variability, clinically appropriate reference populations, and protocols that connect results to measured actions rather than broad promises.
The clock can inform the clinical conversation. It should not be expected to settle it on its own.
- How should we distinguish whole foods, the food matrix, phytonutrients, and purified natural products, and what standards should a botanical or nutraceutical meet before being recommended for longevity?
The strongest nutrition evidence generally supports complete dietary patterns. Mediterranean- and MIND-style approaches have been associated in observational studies with healthier cardiovascular and cognitive aging.
Randomized trials provide a more demanding test. In the three-year MIND diet trial published by Barnes and colleagues in 2023, both the intervention and control groups improved on cognitive testing, with no significant difference between them. That does not make the dietary pattern unimportant. It illustrates why observational associations and randomized treatment effects should be described as different levels of evidence.
The history of isolated compounds has also taught the field to be careful when translating evidence from food into supplements. Beta-carotene appeared protective in dietary studies, but high-dose supplementation in smokers increased lung cancer incidence in the ATBC and CARET trials. High-dose antioxidant studies have frequently produced null or unfavorable findings, while compounds such as resveratrol have generated more compelling mechanistic data than definitive human outcomes.
One explanation is the food matrix. Fiber, polyphenols, fats, micronutrients, and other constituents are consumed together, in physiologic quantities and within structures that influence absorption, metabolism, timing, and biological interaction. An isolated compound delivered at a pharmacologic dose is a different intervention.
For that reason, I distinguish among four separate categories: the whole food, the surrounding food matrix, an individual phytonutrient, and the finished commercial product. Evidence for one category should not automatically be transferred to another.
This distinction creates an important opportunity for responsible nutraceutical companies. As the sector matures, the most credible manufacturers are moving beyond ingredient-level narratives and developing evidence around the actual finished formulation that consumers receive.
Before recommending a botanical or nutraceutical specifically for longevity, I look for several elements:
A plausible mechanism supported by relevant human evidence, not only cellular or animal findings.
A clearly characterized and standardized active constituent or formulation.
Independent identity, purity, potency, and contaminant testing.
A defined dose and, ideally, evidence of a dose-response relationship.
A safety assessment that includes medication interactions, contraindications, and the realities of long-term use.
Manufacturing quality and consistency across batches.
Transparent communication about the strength and limits of the evidence.
Disclosure of commercial relationships and potential conflicts of interest.
These standards are demanding, but they should not be viewed as hostile to the industry. They reward the organizations already investing in quality systems, third-party verification, responsible claims, and meaningful human research. As clinicians and consumers become more discerning, those investments will increasingly differentiate durable companies from short-lived products.
There is a substantial legitimate role for supplementation in clinical care. My general approach is food first, followed by targeted supplementation when there is a documented deficiency, a recognized clinical indication, or credible trial evidence for a specific use. Examples may include vitamin D for appropriate patients, omega-3 fatty acids in defined contexts, creatine to support muscle and strength, and fiber supplementation when dietary intake is insufficient.
Drug-induced nutrient depletion also deserves more attention. Monitoring vitamin B12 in patients taking long-term metformin is one example. Other chronic therapies may affect micronutrients, hydration, or electrolyte balance. Establishing a baseline and monitoring over time is straightforward good practice.
The evidentiary threshold should become higher when a product is marketed as extending life rather than correcting a deficiency or supporting a defined function. That is a claim about long-term clinical outcomes. Companies seeking to make that claim credibly will need a development strategy that connects formulation quality, target engagement, functional measures, and ultimately meaningful human outcomes.
- What evidentiary threshold, monitoring, and informed-consent standards should be required before a putative geroprotector is used outside a clinical trial, particularly in otherwise healthy people?
My starting principles are foundations first, a clear individual rationale, qualified clinical supervision, and honest communication about uncertainty. Long-term safety data in healthy humans remain limited for many proposed geroprotectors, so each intervention should be assessed individually rather than treating the entire category as either proven or inappropriate.
I use a structured evaluation framework. What is the strongest human evidence? Which population was studied? What outcome was measured? What was the effect size? What dose and treatment schedule were used? How closely does the patient resemble the study population? What risks require monitoring? What evidence would justify continuing or stopping?
When ranked this way, the candidates differ substantially.
Rapamycin has a strong cross-species preclinical record, including validation through the NIA Interventions Testing Program, and a well-characterized mTOR target. Human research is still early and has produced mixed findings depending on the population and endpoint. Its use for aging in otherwise healthy people should therefore be described as investigational, with carefully designed protocols and follow-up.
Metformin has extensive evidence for established metabolic indications. Evidence in metabolically healthy people for the specific purpose of slowing aging remains limited.
Senolytics have compelling biological rationale and early human pilot studies in selected disease populations. The next step is larger, controlled research that clarifies dosing, target engagement, patient selection, safety, and clinically relevant benefit.
Peptides are a highly heterogeneous category. Some have well-defined medical uses, while others are supported primarily by mechanistic reasoning or limited human data. Evaluation should occur at the level of the individual molecule, formulation, manufacturing source, indication, and protocol rather than through broad conclusions about “peptides” as a class.
Regenerative medicine also requires careful differentiation. Hematopoietic stem cell transplantation is established for defined indications. Some orthobiologic applications have supportive trial evidence of varying strength. Stem-cell products marketed as general treatments for aging have not reached the same evidentiary standard. Partial cellular reprogramming is at an earlier, largely preclinical stage and represents one of the field’s most interesting research directions.
For otherwise healthy individuals, the threshold for routine use should be reproducible human evidence of meaningful benefit, acceptable risk, and a coherent relationship among dose, biological target, and outcome. Where that threshold has not been reached, an intervention may still be studied or considered in limited circumstances, but it should be clearly described as experimental.
Monitoring is equally important. Any off-label or longevity-directed pharmacologic protocol should have a written plan linked to the intervention’s known and plausible risks.
Rapamycin protocols may include complete blood counts, lipid testing, glucose monitoring, assessment for oral or dermatologic effects, and infection surveillance.
Metformin protocols should account for renal function, vitamin B12 status, gastrointestinal tolerance, and the patient’s metabolic profile.
GLP-1 therapies are best understood as established risk-factor treatments with potential geroscience relevance rather than as universal geroprotectors. Monitoring should include gastrointestinal and gallbladder effects, nutritional adequacy, and preservation of lean mass. Resistance training and sufficient protein intake are particularly important when weight loss could include clinically meaningful muscle loss.
The informed-consent discussion should distinguish among established evidence, reasonable inference, and experimental use. It should address the individual’s baseline risk, expected benefit, known and unknown harms, alternatives, financial cost, monitoring plan, success criteria, and stopping rules.
This is also where well-run longevity clinics can distinguish themselves. A protocol that includes appropriate patient selection, documented rationale, quality sourcing, defined monitoring, and predetermined stopping criteria is fundamentally different from a product-driven model in which treatment begins without a clear evaluation plan.
TAME helps clarify the broader regulatory challenge. Its importance is not limited to metformin. The proposed trial demonstrated that a composite of age-related diseases could be considered as an outcome for testing whether an intervention modifies aging-related risk across conditions. The trial’s funding difficulties reflect the commercial limitations of an off-patent drug more than a rejection of the scientific concept.
Metformin therefore occupies a precise position: it has extensive evidence for approved indications but lacks definitive randomized evidence for use specifically to delay multiple age-related outcomes in healthy or broadly selected older adults. Clinicians and patients should understand that distinction clearly.
- Where can wearables, remote monitoring, digital phenotyping, and AI-supported risk assessment already add real value, and what validation and governance safeguards are needed?
The clearest value of these technologies is that they move measurement beyond the occasional clinic visit and into daily life, where behavior, recovery, symptoms, and adherence actually occur.
Objective activity data can be clinically useful. Ding’s 2025 meta-analysis suggested that approximately 7,000 daily steps was associated with substantially lower all-cause mortality compared with 2,000 steps. That gives clinicians and digital health companies a practical target that is more accessible than an undifferentiated instruction to “exercise more.”
Sleep tracking has also contributed useful insights. Windred and colleagues analyzed accelerometer data from approximately 61,000 UK Biobank participants and found that sleep regularity predicted all-cause, cardiovascular, and cancer mortality more strongly than sleep duration. This turns consistency of bedtime and wake time into a potentially actionable behavior.
Continuous glucose monitoring can reveal postprandial patterns and variability that are not visible in a fasting measurement. Its established value is strongest in diabetes care. In metabolically healthy people, the clinical benefit of responding to every glucose excursion remains uncertain. The opportunity for responsible digital programs is to prevent normal physiologic variation from being framed as pathology and instead connect the data to validated risk patterns and achievable behavioral changes.
Resting heart-rate trends, atrial fibrillation detection, activity patterns, exercise adherence, and selected recovery metrics can also add useful signal.
AI’s most credible near-term role is synthesis. Longitudinal laboratory results, imaging, medication data, wearable streams, and clinical history can overwhelm both patients and clinicians. Well-designed systems can organize this information, identify changes that warrant attention, and support more consistent clinical decision-making. At present, AI should generally augment professional judgment rather than replace it.
The limitations are specific and manageable, but they must be addressed. Consumer devices vary in accuracy, and validation for one metric or population cannot automatically be generalized to another. False alerts can create anxiety and unnecessary testing. Algorithms may perform differently across demographic groups. Pulse oximetry performance on darker skin is a well-known example of why representative validation matters. Privacy, ownership, secondary data use, and cybersecurity also require clear governance.
Effective safeguards include validation against appropriate clinical standards, transparent reporting of device performance, representative training data, ongoing bias audits, clear escalation pathways, and meaningful clinician oversight.
Most importantly, data should be connected to decisions. A digital platform becomes clinically valuable when it helps a person change a behavior, helps a clinician identify a risk, or helps a care team evaluate whether an intervention is working. Accumulating more measurements without a response framework rarely improves care.
For developers and health systems, the design goal should be a small number of reliable, actionable signals supported by predefined clinical pathways rather than an endless dashboard of unprioritized data.
- How should clinicians, digital designers, and health systems help people sustain these practices, integrate longevity medicine into mainstream care, establish professional standards, and make healthspan gains broadly accessible?
Much of the foundational knowledge is already available. The more difficult challenge is helping people sustain beneficial behaviors over years and decades.
The evidence supporting physical activity, nutrition, sleep, stress regulation, social connection, and preventive care is among the strongest in medicine. Emerging therapies may eventually add to those foundations, but they are unlikely to replace them. Lifestyle should not be presented as a preliminary step before the “real” intervention. It is the physiological context that determines how well many medical and pharmacologic interventions work.
Closing the implementation gap is partly a clinical challenge and partly a design challenge.
Clinicians should use motivational interviewing and shared decision-making rather than relying primarily on instruction. Exercise, sleep, nutrition, and recovery plans should be prescribed with the same specificity, personalization, and follow-up that would accompany a medication.
Digital designers should optimize for consistency, usability, and long-term engagement rather than novelty. The best systems reduce friction, support realistic goals, and help users recover from lapses without treating imperfect adherence as failure.
Health systems must also account for structural conditions. Recommendations are less effective when people lack safe places to exercise, access to nutritious food, adequate sleep opportunity, or affordable preventive care. Longevity medicine will have limited population impact unless it addresses those realities.
The strongest version of longevity medicine is not a replacement for primary care. It is excellent preventive medicine applied earlier, more systematically, and with greater attention to function, trajectory, and long-term risk. Its principles therefore belong within mainstream care, even as specialized clinics continue to develop new models more quickly.
Integration will require better reimbursement, professional education, evidence standards, data infrastructure, and incentives that reward prevention rather than primarily paying for acute intervention.
The field also needs meaningful professional standards. Competence should include rigorous evidence appraisal, the ability to use lifestyle interventions as clinical tools, appropriate patient selection, familiarity with emerging therapies, and the discipline to establish monitoring and stopping rules before initiating off-label treatment.
This is an area in which responsible clinics can lead. Organizations that document their protocols, track meaningful outcomes, review adverse events, manage conflicts of interest, and revise their practices when evidence changes can help establish the standards the broader field needs.
I advise longevity clinics and care for private clients, so I see the premium end of the market closely. That experience has reinforced my view that specialized care can play an important role in innovation, protocol development, and early adoption. It can also generate models that later become more scalable.
At the population level, however, the largest gains will still come from physical activity, sleep, diet quality, tobacco avoidance, social connection, vaccination, and appropriate control of blood pressure, lipids, and glucose. These interventions do not require a luxury membership, but delivering them effectively does require better systems.
The longevity dividend will be fully realized only when the field combines innovation at the specialized end with scalable implementation through primary care, public health, employers, communities, and digital delivery.
- How can scientists and clinicians communicate uncertainty without amplifying hype or dismissing promising research, and what principle should readers use in judging future breakthroughs?
The central discipline is to place each claim on an evidence continuum and state clearly where it sits.
A finding may be mechanistically plausible, validated in animals, supported by observational human data, demonstrated in an early clinical trial, or confirmed through replicated outcomes research. Each stage has value, but each justifies a different level of confidence and a different type of claim.
This approach allows scientists, clinicians, and companies to communicate genuine enthusiasm without presenting an emerging result as settled science. It also avoids the opposite mistake of dismissing a promising area simply because it has not yet completed the full development pathway.
Falsifiability is one of the most useful tests of credibility. A researcher, clinician, or company should be able to explain what evidence would strengthen a claim, what evidence would weaken it, and what result would cause the recommendation to change. When a claim relies primarily on mechanism, testimonials, or aspiration, it should be presented as an early hypothesis rather than as an established outcome.
The evidence bar across longevity is rising. That is healthy for the field and advantageous for organizations that are investing in real research, quality systems, transparent reporting, and clinically responsible implementation.
Conventional medicine may still underestimate the geroscience hypothesis articulated by Kennedy and colleagues in 2014: targeting aging biology could influence several chronic diseases because those diseases share upstream mechanisms. Medicine remains largely organized around treating individual conditions after damage has accumulated. Geroscience asks whether intervention can occur earlier and across several risk pathways at once.
Over the coming decade, I expect biological age assessment to enter selected clinical settings, the hallmarks of aging to influence mainstream thinking about chronic disease, and geroscience-informed therapies to become standard for specific populations and indications. Progress will probably occur through defined clinical use cases rather than through a single universal anti-aging treatment.
The principle I would offer readers is to judge every claim according to the evidence in the relevant human population, using outcomes that matter.
Ask four questions:
Who was studied?
What outcome was measured?
What are the known and plausible harms?
What future evidence would change the recommendation?
It is also important to allow conclusions to evolve. Revising a position when better evidence becomes available is not inconsistency. It is the normal process of evidence-based practice.
My own assessment of GLP-1 therapies changed as large outcome trials demonstrated cardiovascular benefits that could not be understood only as weight reduction. Similar changes will occur throughout longevity medicine as more rigorous trials are completed.
For now, investigational therapies should be considered in proportion to their evidence and the needs of the individual. The proven fundamentals remain the appropriate starting point for most people, while carefully selected emerging interventions can be evaluated through a structured, transparent, and well-monitored process.
—
Interviewer: Prof. Atanas G. Atanasov
—
About the author
David Barzilai, MD, PhD, MBA, MS, DipABLM, is a longevity medicine physician and Lecturer at Harvard Medical School. He served as Medical Editor of Pathways to Longevity, Harvard Health Publishing’s first Special Health Report devoted to longevity.
He holds a PhD in health services research, is a Diplomate of the American Board of Lifestyle Medicine, and is an inaugural faculty member and Trustee at the Geneva College of Longevity Science. His editorial “Healthy Life Extension: Geroscience’s North Star” appeared in Aging in 2026, arguing that the field should unify around health-adjusted survival rather than treating lifespan and healthspan as competing goals. He is co-editor of Frontiers of Longevity Science, forthcoming from Springer Nature.
Through Barzilai Longevity Consulting, he advises clinics, health systems, and organizations on evidence-based best practices, protocol design, and whether interventions are producing meaningful outcomes. More at barzilaiconsulting.com/advisory.
References
Barzilai DA. Healthy life extension: geroscience’s north star. Aging (Albany NY). 2026;18:100-101.
López-Otín C, Blasco MA, Partridge L, Serrano M, Kroemer G. Hallmarks of aging: an expanding universe. Cell. 2023;186(2):243-278.
Fries JF. Aging, natural death, and the compression of morbidity. N Engl J Med. 1980;303(3):130-135.
Kennedy BK, Berger SL, Brunet A, et al. Geroscience: linking aging to chronic disease. Cell. 2014;159(4):709-713.
Mandsager K, Harb S, Cremer P, Phelan D, Nissen SE, Jaber W. Association of cardiorespiratory fitness with long-term mortality among adults undergoing exercise treadmill testing. JAMA Netw Open. 2018;1(6):e183605.
Kodama S, Saito K, Tanaka S, et al. Cardiorespiratory fitness as a quantitative predictor of all-cause mortality and cardiovascular events in healthy men and women: a meta-analysis. JAMA. 2009;301(19):2024-2035.
Leong DP, Teo KK, Rangarajan S, et al. Prognostic value of grip strength: findings from the Prospective Urban Rural Epidemiology (PURE) study. Lancet. 2015;386(9990):266-273.
Ngandu T, Lehtisalo J, Solomon A, et al. A 2 year multidomain intervention of diet, exercise, cognitive training, and vascular risk monitoring versus control to prevent cognitive decline in at-risk elderly people (FINGER): a randomised controlled trial. Lancet. 2015;385(9984):2255-2263.
Ding D, Nguyen B, Nau T, et al. Daily steps and health outcomes in adults: a systematic review and dose-response meta-analysis. Lancet Public Health. 2025;10(8):e668-e681.
Windred DP, Burns AC, Lane JM, Saxena R, Rutter MK, Cain SW, Phillips AJK. Sleep regularity is a stronger predictor of mortality risk than sleep duration: a prospective cohort study. Sleep. 2024;47(1):zsad253.
Holt-Lunstad J, Smith TB, Layton JB. Social relationships and mortality risk: a meta-analytic review. PLoS Med. 2010;7(7):e1000316.
Barnes LL, Dhana K, Liu X, et al. Trial of the MIND diet for prevention of cognitive decline in older persons. N Engl J Med. 2023;389(7):602-611.
Alpha-Tocopherol, Beta Carotene Cancer Prevention Study Group. The effect of vitamin E and beta carotene on the incidence of lung cancer and other cancers in male smokers. N Engl J Med. 1994;330(15):1029-1035.
Omenn GS, Goodman GE, Thornquist MD, et al. Effects of a combination of beta carotene and vitamin A on lung cancer and cardiovascular disease. N Engl J Med. 1996;334(18):1150-1155.
Lincoff AM, Brown-Frandsen K, Colhoun HM, et al. Semaglutide and cardiovascular outcomes in obesity without diabetes. N Engl J Med. 2023;389(24):2221-2232.
Mavrommatis C, Belsky DW, Ying K, Moqri M, Campbell A, Richmond A, Gladyshev VN, Chandra T, McCartney DL, Marioni RE. An unbiased comparison of 14 epigenetic clocks in relation to 174 incident disease outcomes. Nat Commun. 2025;16:11164.
Belsky DW, Caspi A, Corcoran DL, et al. DunedinPACE, a DNA methylation biomarker of the pace of aging. eLife. 2022;11:e73420.
Moqri M, Herzog C, Poganik JR, et al. Biomarkers of aging for the identification and evaluation of longevity interventions. Cell. 2023;186(18):3758-3775.
Sehgal R, Borrus D, Gonzalez J, et al. Biological versus technical reliability of epigenetic clocks and implications for disease prognosis and intervention response. bioRxiv. 2025. Preprint, not peer reviewed.
Harrison DE, Strong R, Sharp ZD, et al. Rapamycin fed late in life extends lifespan in genetically heterogeneous mice. Nature. 2009;460(7253):392-395.
Barzilai N, Crandall JP, Kritchevsky SB, Espeland MA. Metformin as a tool to target aging. Cell Metab. 2016;23(6):1060-1065.
Justice JN, Nambiar AM, Tchkonia T, et al. Senolytics in idiopathic pulmonary fibrosis: results from a first-in-human, open-label pilot study. EBioMedicine. 2019;40:554-563.
Ross R, Blair SN, Arena R, et al. Importance of assessing cardiorespiratory fitness in clinical practice. Circulation. 2016;134(24):e653-e699.