How to Lower Biological Age According to Clinical Trials

What Lowers Biological Age in Clinical Trials

Calorie restriction, nutritional weight management, and bariatric surgery are the interventions with documented reductions in biological age measurements. In the CALERIE trial analysis (Waziry et al., Nature Aging, 2023), calorie restriction slowed the pace of aging measured by DunedinPACE by 2% to 3%. In a randomized trial by Ho, Qualls and Villareal (J Nutr Health Aging, 2022), behavioral diet therapy decreased Klemera-Doubal Method (KDM) biological age by 2.4 years. Exercise alone changed KDM age by -0.2 years, compared with +0.2 years in controls, and only the diet arms produced the between-group decrease.

A prospective cohort study by Morawitz and colleagues (npj Aging, 2026) observed a 5.55-year drop in phenotypic age twelve months after bariatric surgery. That cohort had no non-surgical comparison group. In the DO-HEALTH trial (Bischoff-Ferrari et al., Nature Aging, 2025), omega-3 supplementation yielded a small slowing effect across three epigenetic clocks over three years. Neither exercise nor vitamin D altered clock readings in that trial. Sleep and smoking have observational associations only on this list. In CALERIE, one clock moved while two did not in the same people, so a retest should use the same test.

Calorie Restriction and Weight Loss

Calorie restriction slowed the pace of biological aging in adults without obesity (mean age 38), whereas weight loss interventions produced multi-year reductions in older adults and bariatric patients. In the CALERIE trial, 220 adults without obesity and with a Body Mass Index (BMI) of 22.0 to 27.9 were randomized to 25% calorie restriction or an ad libitum diet. Blood DNA methylation (DNAm) was evaluated in 197 participants with a mean age of 38 years. The calorie restriction group achieved an average restriction of 11.9%. Restriction slowed the pace of aging measured by the DunedinPACE algorithm, with effect sizes of d = -0.29 at 12 months and d = -0.25 at 24 months. The authors stated that this represents a 2% to 3% reduction in the pace of aging. Treatment effect sizes were small.

Overlap in DunedinPACE trajectories between groups approached 90%, and some restriction participants experienced an increased pace. The intervention did not produce statistically detectable shifts on the PhenoAge or GrimAge clocks. The CALERIE authors wrote that previous studies suggest this 2% to 3% slower pace corresponds to a reduction in mortality risk of as much as 10% to 15%. They stated it remains unclear whether these molecular shifts will translate into fewer clinical diseases. CALERIE, the calorie-restriction trial on this list, moved one measure (DunedinPACE) by 2% to 3% and left two others unchanged, in people who achieved about half their calorie target. Apart from the uncontrolled bariatric cohort, the biggest single-number drops on this list come from an 8-week pilot of 43 men and a 9-person drug trial.

In a one-year trial, Ho, Qualls, and Villareal evaluated 107 older adults with obesity. Ninety-three participants completed the study across the four intervention arms. KDM biological age shifted by -2.4 years in the diet arm and -2.2 years in the diet-plus-exercise arm. The control group gained 0.2 years. Diet and diet-plus-exercise decreased KDM age more than exercise and control (P < 0.05).

Morawitz and colleagues analyzed 505 bariatric surgery patients whose phenotypic age was calculated from routine blood biomarkers. Phenotypic age was 5.55 years younger at 12 months (95% CI -6.12 to -4.97) and remained stable at 24 months. After adjusting for BMI change, phenotypic age was 3.32 years lower. Glucose, C-reactive protein (CRP), and blood cell counts shifted favourably after surgery. Routine lab tests can estimate biological age through the routine blood markers behind phenotypic age.

Exercise Trials and Biological Age Clocks

Exercise programs did not produce measurable reductions in biological age clocks when evaluated in randomized clinical trials. In the DO-HEALTH trial, 777 Swiss adults with a mean age of 75 were randomized in a 2x2x2 design that included a three-year Simple Home Exercise Program (SHEP). Participants performed SHEP three times per week for 30 minutes per session. Bischoff-Ferrari and colleagues reported that SHEP was not associated with changes in any evaluated epigenetic clock. Eighty-eight percent of the Swiss participants were physically active at baseline. In the trial by Ho and Villareal, the exercise-only arm altered KDM biological age by -0.2 years, compared to +0.2 years in controls. The difference between exercise alone and control was not statistically significant.

Cross-sectional epidemiological studies report different observations. Levine and colleagues (Aging, 2018) found physical exercise negatively associated with DNAm PhenoAge. In a study by Lu and colleagues (Aging, 2019), the GrimAge clock detected a negative correlation with physical exercise (r = -0.10, P = 3E-10). An analysis of 4,173 Women's Health Initiative participants by Quach and colleagues (Aging, 2017) observed associations between physical activity and lower epigenetic age.

McGreevy and colleagues (Aging, 2023) developed DNAmFitAge, an epigenetic clock incorporating physical fitness markers. The DNAmFitAge clock includes estimates of gait speed, grip strength, forced expiratory volume in one second (FEV1), and maximal oxygen consumption (VO2max). DNAmFitAge was lower in male bodybuilders than in controls (p = 0.046). Those findings are associations. None came from a randomized intervention trial.

Walking pace was evaluated in 405,981 UK Biobank participants by Dempsey and colleagues (Communications Biology, 2022). Brisk walkers had longer leucocyte telomere length than slow walkers. Mendelian randomisation analysis suggested an association equivalent to 16 years of age-related telomere difference between slow and fast walkers. The adjusted observational difference was equivalent to two years. Telomere length represents a cellular marker proposed as an indicator of biological age. It is not an epigenetic clock. No study reviewed here tested running on a biological age measure. Cardiovascular conditioning and metabolic improvements are discussed in the guide on exercise for insulin resistance.

Sleep Habits and Smoking in Epigenetic Studies

Insomnia symptoms and cigarette smoking correlate with higher biological age in observational studies. No study on this page tested a sleep or quit-smoking intervention on a biological age measure. In a study of 2,078 women from the Women's Health Initiative, Carroll and colleagues (Biological Psychiatry, 2017) examined sleep disturbances. Insomnia symptoms were related to advanced epigenetic age (beta = 1.02, p = 0.005) after adjusting for covariates. Self-reported sleep duration was unrelated to epigenetic age. Factors affecting sleep architecture are reviewed in the guide on poor sleep quality.

Tobacco smoking shows consistent positive associations with epigenetic aging clocks. Levine and colleagues identified smoking as a variable positively associated with DNAm PhenoAge, indicating accelerated biological aging. The GrimAge clock developed by Lu and colleagues incorporates a DNAm-based estimator of smoking pack-years. The CALERIE authors wrote that previous studies suggest the 2% to 3% slower pace corresponds to a mortality risk reduction similar in magnitude to smoking cessation. They also wrote that it is unclear whether the change will translate into less disease.

The Small Trials Behind Age Reversal Claims

Claims of multi-year biological age reversal are derived from small pilot studies that tested multi-component lifestyle protocols or prescription hormone combinations. A pilot randomized trial by Fitzgerald and colleagues (Aging, 2021) tested an eight-week lifestyle intervention in 43 healthy men aged 50 to 72. The protocol included dietary rules, sleep recommendations, exercise guidance, relaxation practices, and supplemental probiotics and phytonutrients. Epigenetic age was measured in saliva using the Horvath DNAmAge clock. The original report noted a 3.23-year decrease in DNAmAge compared with controls (p = 0.018).

In a 2024 correction adjusting for the study timeframe, the treatment group scored an average 2.04 years younger than baseline (p = 0.043). Controls scored an average 1.10 years older (p = 0.191). Among the 18 treatment participants, eight decreased their score, nine were unchanged, and one increased. The authors stated that larger clinical trials of longer duration are necessary to confirm the outcomes.

In the TRIIM trial, Fahy and colleagues (Aging Cell, 2019) evaluated ten men aged 51 to 65. Nine men completed the one-year protocol. Participants received recombinant human growth hormone, dehydroepiandrosterone (DHEA), and metformin intended to regenerate the thymus. Mean epigenetic age was approximately 1.5 years less than baseline after one year, representing a 2.5-year change relative to controls. GrimAge showed a two-year decrease that persisted six months after stopping treatment. The authors stated that these results require verification in an appropriately powered follow-up study. That trial evaluated prescription pharmaceuticals. It did not test lifestyle changes.

Supplements Tested on Biological Age Clocks

Omega-3 fatty acid supplementation slowed biological aging by several months over three years, whereas vitamin D showed no effect across tested epigenetic clocks. In the DO-HEALTH trial, Bischoff-Ferrari and colleagues administered one gram of omega-3 fatty acids daily to older adults for three years. Omega-3 alone slowed PhenoAge, GrimAge2, and DunedinPACE trajectories. Standardized effect sizes ranged from 0.16 to 0.32 units, corresponding to 2.9 to 3.8 months of slowed aging. The study authors described this as a small protective effect.

Daily supplementation with 2,000 International Units (IU) of vitamin D was not associated with changes in any clock. No other single dietary supplement on this list was tested alone on a biological age measure. The phytonutrient and probiotic supplements in the Fitzgerald trial were delivered as part of a multi-component lifestyle intervention. Claims regarding other anti-aging compounds remain untested in the clinical trials reviewed here. Evidence for individual compounds is analyzed in the guide on longevity supplements.

Retest Intervals in Published Studies

Clinical trials retested biological age markers at intervals between eight weeks and three years, with no trial defining a standard minimum retest window. In the CALERIE trial, the reduction in DunedinPACE was present at 12 months and maintained at 24 months. In the bariatric cohort, reductions in phenotypic age were established at 12 months and remained stable at 24 months. No trial reviewed here established an optimal monitoring frequency for healthy individuals.

Retest Reliability Across Different Biological Age Clocks

Different biological age algorithms gave conflicting results in the same people. In the CALERIE trial, calorie restriction slowed DunedinPACE while leaving PhenoAge and GrimAge without statistically detectable shifts. The CALERIE authors stated that there is no gold standard measure of biological aging. Technical variation can also distort repeated testing results.

An analysis by Higgins-Chen and colleagues (Nature Aging, 2022) found technical noise produced discrepancies up to nine years between replicates for six prominent epigenetic clocks. Their computational adjustment based on principal components brought most replicates within 1.5 years. In a validation study by Belsky and colleagues (eLife, 2022), DunedinPACE showed high test-retest reliability with an Intraclass Correlation Coefficient (ICC) of 0.96.

Studies on this list used divergent testing methods. These methods included saliva Horvath DNAmAge, blood DunedinPACE, blood KDM biological age, routine blood phenotypic age, and leucocyte telomere length. Algorithms and their underlying mathematical designs are detailed in the guide to epigenetic clocks. Biological age concepts and calculations are explained in the guide on biological age. To track a change, repeat the same test with the same algorithm. No study on this page sets a minimum interval between tests.