Study design

The study visits will be conducted at the Department of Biomedical Sciences, University of Copenhagen. 240 healthy community-dwelling elderly aged 65 and older in good general health will be recruited and block randomized to either of the interventions for 52 weeks. Assessments are carried out at baseline, 3 months, 6 months, and 12 months and followed up after 2 years, 5 years and 10 years (Fig. 2).

Fig. 2

Flowchart of the FAXAge experimental set-up

Participants, recruitment strategy, and randomization

Participants are excluded if they have received another investigational drug or intervention within 1 year, as prior treatments could influence the responses observed in this study, making it difficult to isolate effects to the current intervention. Individuals engaging in more than one hour of systematic strenuous exercise or strength training a week are excluded to maintain similar baseline activity levels across participants and reduce variability in fitness, making the effects of the intervention clearer. Individuals already practicing time-restricted eating are excluded, as they will not show clear pre- and post-intervention responses. We wish to include a representative part of the healthy elderly population; however, we expect significant bias in our data if we only include individuals without chronic diseases. For this reason individuals with severe or dysregulated diseases are excluded, as the disease state may independently impact the intervention outcomes, while allowing one well-managed chronic disease helps limit heterogeneity and risks for the participants. Smoking is an exclusion criterion due to its effects on cardiometabolic health, inflammation, and exercise capacity, all of which could confound study results. Finally, individuals using systemic glucocorticoids, androgens, or antiandrogens are excluded because these medications have strong effects on metabolism, muscle adaptation, appetite, and glucose tolerance, potentially masking or exaggerating the effects of time-restricted feeding and exercise interventions. Participants are recruited via online advertising, on social media, and in the DaneAge Association membership magazine. Participants will be randomized according to a permuted block randomization scheme with a block size of 4, stratified according to biological sex. The randomization list will be generated prior to commencing the study.

Interventions

Participants randomized to the EXE and FAX group will perform supervised resistance training two times per week as well as cardio training at home two times per week monitored by fitness watches to track heart rate. This combined approach is chosen since both high muscle strength and cardiorespiratory fitness are independently associated with lower all-cause mortality, whereas low levels correlate with increased risk (Schnohr et al. 2025). Accordingly, this intervention is designed to enhance both muscle strength and cardiorespiratory fitness, enabling the investigation of their association with changes in markers of biological age.

The strength training intensity will be individualized and increased over time according to changes in the participant’s training status, aiming to maintain a similar relative load across participants by adjusting loads to a target range of repetitions in reserve. The training program will consist of the same exercises over the 52 weeks and is designed in a progressive manner as this approach has been shown to enable long-term adaptations in muscular strength through progressive increases in mechanical loading and total training stimulus, thereby limiting training plateaus over time (American College of Sports Medicine 2009). It will consist of approximately 1 h of strength training per session with at least 48 h between each session. Participants will perform a standardized, machine-based (Technogym S.p.A, Cesena, Italy) strength training program involving exercises for the lower and upper body. The program includes both multi-joint (leg press, chest press, lateral pulldown and low row) and single-joint (leg extension and lying leg curl) exercises, with an emphasis on multi-joint movements, which have been shown to elicit greater improvements in muscle strength and contribute to improvements in cardiorespiratory fitness compared with single-joint exercises when total training load is matched (Paoli 2017). This effect has been attributed to a higher oxygen demand associated with the involvement of a larger active muscle mass (Paoli 2017).

The training protocol is based on the LISA study and guidelines to resistance training of older adults from the American College of Sports Medicine (American College of Sports Medicine 2009; Eriksen et al. 2016). The program is initiated with a 4-week familiarization phase characterized by high-repetition, low-load training, to ensure proper technique, allow musculoskeletal tissue adaptation and to reduce the risk of injury (Signorile 2013). The training period consists of 4 blocks of 13 weeks (the first block includes the 4-week familiarization phase). Each block begins with three sets of 12 repetitions per exercise, with the number of repetitions decreasing in a stepwise manner every three weeks, ultimately concluding with four sets of six repetitions in the last three weeks. Within each block, training will be organized into consecutive 3-week phases during which repetition range, target repetitions in reserve, and number of sets will be maintained constant, allowing sufficient time for adaptation before progressing to the next phase. Load is continuously adapted for each training session based on participant-reported repetitions in reserve in the previous session. When the upper target is reached and maintained for ≥ 2 sessions, load is increased. Target repetitions in reserve vary for each 3-week phase in accordance with the prescribed number of repetitions – the fewer repetitions, the fewer target repetitions in reserve. Every 13th week a deload phase is implemented to facilitate recovery and mitigate fatigue development (Bell et al. 2023). During this week, training volume is reduced ~ 30 to 50% by reducing the number of sets for each exercise and increasing target repetitions in reserve while maintaining the number of repetitions. At the end of each block, an indirect one repetition maximum load test is performed using the Brzycki formula (Brzycki 1993) at a target of six repetitions, to test alignment between participant-reported repetitions in reserve and one repetition maximum based loading to ensure load is progressive (American College of Sports Medicine 2009). After completion of each block, the same progression pattern is repeated in the subsequent block with absolute loads adjusted from the one maximum load test, starting at ~ 65% of calculated one repetition maximum load. Training sessions will be supervised and will take place at the Department of Biomedical Sciences, University of Copenhagen.

Cardio training will be of self-chosen modality and at moderate to vigorous intensity (> 70% HRmax) since vigorous-intensity exercise may result in greater increases in aerobic capacity than moderate-intensity (Gormley et al. 2008) and that participation in vigorous activities is associated with lower mortality rates (Lee and Paffenbarger 2000). The desired duration will increase from 30 min per session in the first month, 45 min per session in the second month to 60 min per session in the following 3–12 months. Adherence, workout duration, and average and maximal heart rate will be monitored by fitness watches provided to each participant.

Participants randomized to the TRF group will be instructed to abstain from any caloric intake during the targeted fasting window of 16 continuous hours and consume ad libitum during the 8-hour eating window. Participants can choose their preferred eating window and are encouraged to drink plenty of water during fasting. Participants will receive an adherence diary including an eating time log for noting the time of first and last calorie consumption each day.

Participants randomized to the FAX group will be instructed to both exercise and fast during the 52-week intervention.

Measurements

Participants go through a battery of tests before the intervention (Fig. 3), 3 months and 6 months into the intervention and after the intervention. Additionally, follow-up testing is performed at 2, 5 and 10 years after the intervention. Follow-up testing is deemed critical because it may allow us to discover long-term beneficial or detrimental effects of these interventions. In the section below we will go through the different data modalities that will be gathered as part of the intervention and the rationale for why these measures were chosen. Tests are performed according to defined standard operating procedures (SOPs) in the same order for all participants at all visits, and to ensure a minimum of bias, test leaders are instructed to not look at the participants’ previous test results before a new test. In addition, biological samples will be processed using harmonized workflows with batch tracking, internal controls, and standardized normalization procedures where applicable. For molecular analyses, raw data and metadata will be retained to allow reprocessing and harmonized downstream analyses as analytical methods evolve. For phenotypic and physiological measurements, personnel will be trained in predefined acquisition procedures to minimize interoperator bias.

Fig. 3

Visualization of the test-battery. A. Participants are tested at baseline, after 3 months, 6 months and 1 year, and follow-up will be performed 2, 5 and 10 years after baseline. B. A stool sample is collected at home for microbiome analysis within 18 hours of the test day and brough to the lab. A blood sample is collected to assess multiple blood markers after which participants fill out questionnaires, have their blood pressure measured and a digital photo is taken and a voice recording ise performed. Body composition is measured with a DXA scan and VO2max is estimated with Seismofit. Lastly participants perform a physical test battery consisting of handgrip strength, gait speed and sit-to-stand tests as well as a graded cardiopulmonary VO2peak test

Primary outcome measure

As the primary outcome measure of the trial, Dunedin Pace of Aging DNA methylation analyses (Belsky et al. 2020) will be performed using the Illumina 450 K array on peripheral blood mononucleated cells. It was chosen to isolate these cells instead of performing investigations on whole blood to reduce the potential effect of blood composition changes that may be a side effect of interventions. Raw DNA methylation data will be stored to enable re-analysis using both current and future generations of epigenetic clocks and computational pipelines. Long-term follow-up samples will initially be analyzed using the same methylation platform and primary clock algorithms to ensure longitudinal consistency, while harmonized re-analysis with updated methods may subsequently be performed as the field evolves. In addition, specific DNA methylation changes in the epigenome will be investigated in secondary explorative outcomes in connection with RNA-sequencing results with appropriate false discovery rate corrections made.

Exploratory outcome measures

Blood work

Venous blood is collected following a 12-h overnight fast at the start of the test day and used for analysis of multiple parameters (Table 1).

Table 1 Measured blood markers

Measurement of inflammatory markers is important, as chronic, low-grade inflammation is a hallmark of aging and has been linked to an increased risk of frailty and mortality (Flanagan et al. 2020; Arosio et al. 2023; Ferrucci and Fabbri 2018). Glycemic control markers are measured because impaired glucose regulation is common among older adults (Dubowitz et al. 2014) and is associated with elevated risk of most chronic age-associated diseases (Schnell and Standl 2006). Complete blood cell count parameters are assessed since aging influences hematopoiesis, and white blood cell counts serve as predictors of all-cause mortality (Groarke and Young 2019).

RNA-sequencing and DNA methylation changes in blood mononucleated cells will provide insights into transcriptional and signaling processes at the cellular level. Metabolomic profiling will be performed on plasma to assess the changes, exercise and fasting induce at the molecular level and what adaptations and cellular pathways are involved. Lastly, biochemical and metabolic markers are measured to detect organ system decline and age-related metabolic disturbances. Combined with physiological measures this will provide a comprehensive understanding of how fasting and exercise influence aging.

A finger prick is also performed on the tip of the middle finger for thin blood smear senescence prediction as recently described (Heckenbach et al. 2022).

Stool sample

Changes in the gut microbiome have been seen with age and may predict health outcomes in the elderly (Bradley and Haran 2024; N V 2025). Notably, gut microbiome is strongly affected by feeding habits and we expect TRF will have a considerable effect on this (Paukkonen et al. 2024). For that reason, all participants will be given a feces kit to collect a stool sample prior to each test day. The stool sample will be collected on the test day and stored at −80 degrees Celsius until analysis. For analysis, we will perform 16S rRNA sequencing.

Questionnaires

To gain understanding of overall health, sleep quality, and cognitive decline participants answer the following questionnaires:

Health examination

Participants will answer a modified version of the Danish Health Examination Survey questionnaire, including questions about chronic diseases, dietary habits, alcohol, stress, physical activity and for female participants their menstrual cycle (see appendix).

Sleep quality

Pittsburgh Sleep Quality Index (PSQI) questionnaire will be used to assess subjective sleep quality and disturbances over the past month. Questions include assessment of duration, latency, efficiency, disturbances, and daytime dysfunction. Sleep will be further evaluated through wearables.

Montreal Cognitive Assessment (MoCA)

The MoCA Full test is performed to assess a wide range of cognitive domains, including memory, attention, executive function, language, visuospatial abilities and orientation, to possibly identify mild cognitive impairment (MCI). This allows for identification of subtle cognitive deficits and is better suited in healthy elderly than the mini-mental state examination (Nasreddine et al. 2005). Nevertheless, several considerations should be made concerning the MoCA test as a secondary outcome. First, the MoCA test may be subject to ceiling effects due to healthy individuals scoring close to max (Bernier et al. 2023), second it is possible that a one-year trial will not yield sufficient changes in MoCA scoring during the trial period (Bernier et al. 2023), third repeat testing may impact the outcome of the test. Nevertheless, we chose this test due to the feasibility in terms of time-management during trial days, the clinical application of test and because we have a significant long-term follow-up time where cognitive decline is expected and where the interventions may impact these outcomes.

Vitals

Participants’ resting pulse and blood pressure are measured three times on the left arm using an electric blood pressure monitor in a seated position with both feet flat on the floor and the arm resting on the table. Blood pressure is measured after filling out questionnaires to ensure that the participant is relaxed. Three consecutive readings are obtained automatically, and the mean of the values is used. If the variability between the first three consecutive readings exceeds threshold, the blood pressure measurement is repeated and the outlier is discarded (University of Maryland 2009).

Portrait photo and voice recording

A portrait photo as well as a voice recording, answering a standardized question, will be obtained of each participant. For facial photographs, standardized image acquisition procedures will be implemented, including controlled lighting, fixed camera positioning and distance, neutral facial expression, and standardized background conditions. Participants will be instructed to avoid excessive makeup, major facial accessories, and recent cosmetic procedures when possible prior to imaging sessions. Audio clips are expected to be 1 min. Both will be used for determination of the participants’ biological age by using facial image-based and voice-based age prediction algorithms using deep neural networks trained to predict age (Teklu et al. 2025; Kwasny and Hemmerling 2021).

Body composition

Body composition and bone mineral density are evaluated with dual energy X-ray absorptiometry (DEXA)-scan (Lunar iDEXA, Madison, Wisconsin, USA) using enCORE software, V.18 (Krugh 2025). A whole-body scan is performed to determine fat-percentage, visceral fat mass, lean body mass and bone mineral density.

Handgrip strength

Grip strength is measured to assess the maximal isometric handgrip strength in kilograms with a Jamar Smart Hand Dynamometer (Jamar, Nottinghamshire, UK). Participants are seated in a chair with a straight back with the elbow bent in a 90-degree angle. The dynamometer is pointing vertically upward. The measurement is a 5 s maximum-effort measure and is repeated three times on each hand with one minute rest in between trials on the same hand. The participants are verbally encouraged during each maximum-effort measure. The highest value obtained for each arm is reported and used for data analysis.

Gait speed

Participants’ gait speed and function are assessed by a standardized 4-m test, where participants are instructed to walk two times—once at their normal pace and once as fast as they can. Participants are also video recorded while walking and gait speed, acceleration, balance, cadence, step length, posture and joint angles will be estimated using Tracked Biotechnologies’ TrackedGait system (Tracked Biotechnologies, Virginia, USA).

Sit-to-stand

A 30 second sit-to-stand test is performed to assess strength in the lower extremities. The test is performed using a chair (45 cm) without armrest. Participants are shown how to perform the test and instructed to place their feet flat on the ground, cross the arms and hold them against the chest. Participants must stand fully and contact the seat fully. The test is completed two times and the score is the highest number of full stands completed within 30 s.

VO 2 measurements

Estimated VO2max

VO2max will be estimated at rest using seismocardiographic measurements (Ventriject, Copenhagen, Denmark) (Hansen et al. 2023). This will be done immediately after the DXA scan where participants have been lying down relaxed for ~8-10 minutes.

Direct measure

A cardiopulmonary exercise test (CPET) is performed on a Monark LC7TT bike (Monark, Varberg, Sweden) using the Quark CPET metabolic cart (Cosmed, Rome, Italy) to determine the peak oxygen consumption rate using indirect calorimetry. Participants perform a graded exercise test with warm-up and increment intensity adjusted to their Seismofit-estimated VO2max. Estimated peak power output is calculated from the estimated VO2max, a sex-specific coefficient, body weight and a regression-based constant interpreted from Eriksen et al (2014). From estimated peak power output, warm-up intensity and increment size are calculated to reach a total test duration of 10–12 min including a 5-min warm up (Buchfuhrer et al. 1983).

Activity monitors

Activity monitors (Huawei Band 10, Huawei, Shenzhen, China) will be provided to all participants to monitor daily activity. The activity monitors will be used by EXE and FAX participants to monitor home workouts of self-chosen modality and are used to ensure that average heart rate is > 70% HRmax for each session. For CON and TRF the activity monitors are used to assess average activity level. Furthermore, the monitors are used to assess sleep quality, which will be compared to the PSQI scores.