Associations between accelerometer-measured physical activity, sedentary behaviour and pericardial adiposity: a prospective cohort study of UK Biobank participants Authors: Wanru Yang, Zhang Chen, Simin Li, Bingshu E Chen, Xinzhi Li Published: 2026 ABSTRACT AbstractIntroductionPhysical activity (PA) reduces the risk of atherosclerotic cardiovascular disease by limiting the overgrowth of fat tissue, particularly visceral adiposity. However, no large-scale studies have investigated the relationship between PA and pericardial adiposity.MethodsThis study included 16 685 participants from the UK Biobank prospective cohort who underwent cardiac magnetic resonance assessment and had available PA data. PA was objectively measured using a wrist-worn triaxial accelerometer and categorised as light-intensity PA (LPA), moderate-intensity PA (MPA) and vigorous-intensity PA (VPA). Multivariable linear regression models were applied to estimate adjusted mean epicardial and pericardial adipose tissue (EPAT) areas across quartiles of total and intensity-specific PA, with linear trend analyses performed. Restricted cubic spline models were used to examine potential non-linear associations. The association between sedentary behaviour (SB) and EPAT area was also examined as a negative control.ResultsAccelerometer-derived PA duration across all intensities was negatively correlated with EPAT area. Participants in the highest quartile (Q4) demonstrated smaller EPAT areas relative to those in the lowest quartile (Q1), with relative mean reductions of 21.29% for total PA, 11.24% for LPA, 18.28% for MPA and 19.10% for VPA (all p trend ≤0.001). Conversely, prolonged SB showed a positive correlation with EPAT area. Compared with those in Q1, individuals in Q4 in SB presented 15.90% greater relative mean EPAT area (p for trend <0.001). After adjustment for visceral adipose tissue volume, these associations were attenuated but remained significant. Stratified analyses confirmed that PA-EPAT association persisted among individuals with atherosclerotic cardiovascular disease, diabetes, hypertension or hyperlipidaemia. Mediation analysis showed VAT largely mediated the relationship between PA and EPAT area.ConclusionsPA reduces epicardial adiposity by limiting the overgrowth of EPAT area. These findings highlight the importance of PA as a preventive strategy for improving obesity-related health management worldwide.Keywords: Public Health, Obesity, Cardiovascular Diseases WHAT IS ALREADY KNOWN ON THIS TOPIC Physical activity reduces the risk of atherosclerotic cardiovascular disease by limiting the overgrowth of fat tissue, particularly visceral adiposity. Physical activity reduces the risk of atherosclerotic cardiovascular disease by limiting the overgrowth of fat tissue, particularly visceral adiposity. WHAT THIS STUDY ADDS Using large-scale cohort data, this study demonstrates that physical activity can also reduce pericardial adiposity, as indicated by decreases in epicardial and pericardial adipose tissue (EPAT) area. Using large-scale cohort data, this study demonstrates that physical activity can also reduce pericardial adiposity, as indicated by decreases in epicardial and pericardial adipose tissue (EPAT) area. Physical inactivity was associated with increased pericardial adiposity. Physical inactivity was associated with increased pericardial adiposity. Visceral adipose tissue plays a significant role in modulating the associations between physical activity and EPAT area. Visceral adipose tissue plays a significant role in modulating the associations between physical activity and EPAT area. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Our findings underscore the importance of physical activity as a preventive strategy for reducing adiposity around the heart and improving the management of obesity-related health issues worldwide. Our findings underscore the importance of physical activity as a preventive strategy for reducing adiposity around the heart and improving the management of obesity-related health issues worldwide. Introduction Atherosclerotic cardiovascular disease (ASCVD) remains one of the major global health challenges, imposing a substantial burden on healthcare systems due to its high prevalence and mortality rates.1 Physical inactivity elevates the risk of ASCVD, whereas increased physical activity (PA) can mitigate disease incidence.2,3 For example, PA—particularly at higher intensities—has been shown to reduce cardiovascular risk4,5; whereas prolonged sedentary behaviour (SB) is associated with increased risk.6 Pericardial adiposity, as indicated by increased volume of epicardial and pericardial adipose tissue (EPAT), has been recognised as another risk factor of cardiovascular disorders in addition to central/visceral adiposity.7–9 Previous studies have demonstrated that increased EPAT is associated with a higher incidence of cardiovascular diseases,10–13 including heart attack14 and atrial fibrillation.15,16 Clinical and population-based studies have consistently shown that greater EPAT volume is independently linked to adverse cardiovascular outcomes, even after adjusting for multiple confounders.9 Similar to abdominal visceral adipose tissue, EPAT is believed to contribute to cardiovascular risk through mechanisms such as inflammation, endothelial dysfunction and structural remodelling of the heart.17 However, whether exercise can specifically reduce EPAT and thereby lower cardiovascular disease risk remains to be fully elucidated. While previous studies have consistently demonstrated associations between PA, SB and pericardial adiposity, several limitations persist. Most of these investigations have relied on small or highly specific populations, which may limit generalisability.18–20 The Multi-Ethnic Study of Atherosclerosis indeed tried to recruit a larger number of participants (over 6000) and found that lower SB and higher light or moderate-to-vigorous PA were associated with reduced pericardial adiposity.21 However, this study still relied on self-reported PA data and did not specifically examine epicardial adipose tissue. Furthermore, no previous research has identified the threshold of PA necessary for reducing EPAT risk. Based on data from a large-scale prospective cohort with device-measured PA, the primary objective of this study is to further investigate how time spent at different intensities of PA and SB influences EPAT, and to identify potential thresholds of activity duration needed for reductions in EPAT. By focusing on modifiable lifestyle factors, this research aims to deepen our understanding of preventable ASCVD risk and to provide evidence to support the development of effective public health recommendations and clinical guidelines. Methods Study design and participants The study population was sourced from the UK Biobank, a prospective cohort study that recruited over 500 000 participants aged 40–69 between 2006 and 2010. This database collected comprehensive health status and lifestyle information through questionnaires. Baseline physical measurements, including blood biochemical profiles, were recorded for all participants. Additionally, it houses the world’s largest repository of imaging data, including cardiac magnetic resonance (CMR) imaging for pericardial adipose tissue area measurements from 45 489 subjects available for our study.8 This work was conducted under the UK Biobank application No. 102278. Outcome, exposure and potential confounders EPAT quantification A subset of 45 489 UK Biobank participants who had completed CMR imaging were accrued in this study. The detailed inclusion and exclusion procedure are summarised in online supplemental figures 1 and 2. The EPAT area measurements used in this study were derived from the data returned to UK Biobank (Return dataset #2103) by other users (application ID: 2964). They extracted epicardial and pericardial fat depots from end-diastolic CMR four-chamber cine images using an automated tool previously developed and validated in UK Biobank and external cohorts.8,22 Images with Dice scores below 0.6 were considered to have poor segmentation quality and were excluded from further analysis. The frequency distribution histogram of EPAT area is shown in online supplemental figure 3. Physical activity and SB assessment Objective PA measures were collected by UK Biobank using accelerometers between 2013 and 2015 through participants wearing an Axivity AX3 triaxial accelerometer for 7 days. Participants who wore the accelerometer for less than 72 hours and those with missing data or poor device calibration were excluded.23 The total volume of PA was reflected by the average acceleration.4 Study determined weekly light PA (LPA), moderate PA (MPA) and vigorous PA (VPA) as 30 to 125 mg, >125 to 400 mg and >400 mg.5,23 The average daily SB (hours/day) was calculated by adding the average SB per hour in the database throughout the day. The frequency histogram of these exposures is shown in online supplemental figure 3. Subjectively reported (self-reported) PA data were obtained from the International Physical Activity Questionnaire (IPAQ), which assessed total PA, walking, moderate and vigorous PA over the last 7 days. Data were analysed in accordance with the IPAQ scoring standard metabolic equivalent of task (MET) levels included walking (MET values of walking is 3.3), moderate-intensity (MET values of MPA is 4.0) and vigorous-intensity (VPA) (MET values of VPA is 8.0).24 In order to analyse the relationship between the time spent on PA at different intensities and the outcomes, we divided the unit of the original data, MET-min/week, by the MET value corresponding to different intensities of PA to obtain the average time spent per week at different intensities of PA, and then converted the unit to 30 min/day for easy interpretation. Total PA was expressed in units of MET-30 min/week. Covariates Demographics and lifestyle: Sex and ethnicity were self-reported at the baseline assessment. Townsend Deprivation Index was calculated immediately prior to participant joining UK Biobank. The smoking, drinking status6,25,26 and dietary intake24,27 were based on the entry questionnaires. Obesity measures:28 total body fat percentage, waist circumference and abdominal visceral adipose tissue (VAT) volumes were calculated based on anthropometry data from UK Biobank and measured using abdominal MRI. Pre-existing and incidental cardiometabolic complications: including ASCVD, diabetes,29,30 hypertension and hyperlipidaemia. The included covariates and their corresponding field IDs are listed in online supplemental table 1. Statistical analysis We removed participants with missing covariates. We described the baseline characteristics of participants by quartiles of either objectively measured or subjectively reported PA and SB, using model-based standardised means or percentages of each category. Since EPAT area exhibits a slightly right-skewed distribution, log-transformed was used to maintain a normal distribution in all analyses. To assess the association between PA and EPAT area, we used a multivariable linear regression model to estimate adjusted means of EPAT area. All PA variables were categorised into quartiles based on their distributions, allowing for ranking individuals’ levels of PA. The results can be interpreted as the relative mean differences (Q4 vs Q1) in EPAT area,31 with all other variables held constant, for total PA (mg) or SB (hour/day),6 LPA or MPA (30 min/day)32 or VPA (10 min/day). P for trend was calculated using the median value of each quartile. Potential confounders were adjusted for multiple variables, including age, sex, ethnicity, Townsend Deprivation Index, smoking, alcohol consumption, red meat, processed meat, vegetable, fruit, salt added to food and oily fish. Multicollinearity in adjusted models was assessed using variance inflation factors (online supplemental table 3). Further linear regression analyses were adjusted using total body fat percentage and VAT. Restricted cubic splines were fitted to potentially capture the non-linear associations of activity duration needed for reductions in EPAT area. Four knots were placed at prespecified percentiles (5th, 35th, 65th, 95th) of each exposure distribution. For visualisation, adjusted dose-response curves were generated using marginal standardisation. Specifically, for each value of the exposure, predictions were obtained for all participants while retaining their observed covariate values, and the mean predicted EPAT area was then calculated after back-transformation to the original scale. 95% CIs were estimated using 500 bootstrap resamples. We stratified body fat percentage and VAT and explored their interactions with PA/SB. We investigated the mediating role of VAT in the effect of PA duration at different PA intensities on the area of EPAT. Given that increased PA has been shown to significantly reduce VAT,33 which proportionately associates with EPAT,9 we conducted a mediation analysis to further explore these relationships. Using the bruceR package, we adjusted for relevant covariates as previously described. CIs were estimated by bootstrapping to ensure robust statistical inference. To account for the compositional nature of accelerometer-measured PA, we performed isotemporal substitution analyses, as previously reported.32 In these analyses, all activity components except the behaviour being ‘substituted’ are included as covariates. The model coefficient estimates the association with EPAT when a fixed amount of time is reallocated from the reference behaviour to another activity, while keeping total activity time constant. For interpretability, effect estimates were scaled to 30 min/day for LPA, MPA and SB, and 10 min/day for VPA. EPAT was log-transformed for analysis, and results were back-transformed and expressed as percentage differences with 95% CIs. In subgroup analyses, we performed stratified analyses to examine whether the association between PA and EPAT area differed after stratification. We stratified participants by sex, body mass index (BMI), waist circumference, pre-existing and incidental ASCVD, diabetes, hyperlipidaemia, hypertension and ethnicity. We also performed separate analyses for Caucasians and other ethnicities. To examine the robustness of our findings, we conducted several sensitivity analyses. First, we repeated the analysis after excluding participants in the top 10% of EPAT area scores. Second, we also added other PA variables to the adjusted model. For these analyses, restricted cubic spline curves were fitted across the 1st to 99th percentiles of exposure within each subgroup to lessen instability at the extreme ends. All data processing and statistical analyses were performed using R version 4.4.1.34 All p values are based on two-sided tests with a significance level of 0.05. Results Characteristic of study cohort The first cohort consisted of 45 489 participants with available return data of CMR-assessed EPAT area. Among these, 2587 participants were excluded because of poor segmentation quality (Dice score <0.6). An additional 24 643 individuals were excluded because they lacked measured PA and SB. We further filtered out 422 participants whose data could not be calibrated or who did not wear the device long enough to obtain a stable measure of physical status. Individuals with extreme values of SB (>18 hours/day) were also excluded from the primary analyses.26 Finally, 1151 participants were removed due to missing covariates, resulting in an effective sample size of 16 685 (online supplemental figure 1). The second cohort contained 33 659 participants with available return data and subjective PA data for EPAT area with CMR assessment, as detailed in the online supplemental figure 2. Of all participants, 53.40% were women, and the mean age was 61.22±7.59 (SD) years. Compared with participants in the lower quartiles categorised by total PA acceleration, those in the higher quartiles had a smaller EPAT area, tended to be younger, had a higher proportion of women, consumed a healthier diet and had a lower BMI, percentage of total body fat and abdominal VAT. As the total PA volume increased, both the EPAT area and the prevalence of cardiovascular disease, hypertension and diabetes decreased (table 1). Table 1Baseline characteristics of participants categorised by total PA accelerationOverall(N=16 685)Quartile 1(<22.95 mg)N=4157Quartile 2 (22.95~27.54 mg)N=4176Quartile 3 (27.54~32.82 mg)N=4174Quartile 4(>32.82 mg)N=4178P valueEPAT area, mean (SD), cm223.74 (12.29)27.74 (13.40)24.26 (12.39)22.62 (11.67)20.34 (10.32)<0.001 (ANOVA)Age at accelerometry, mean (SD), year61.22 (7.59)63.45 (7.39)61.96 (7.49)60.62 (7.44)58.84 (7.28)<0.001 (ANOVA)Female, n (%)8908 (53.4)1907 (45.9)2273 (54.4)2348 (56.3)2380 (57.0)<0.001 (χ2)Ethnicity, n (% white)16 286 (97.6)4056 (97.6)4082 (97.7)4085 (97.9)4063 (97.3)0.075Townsend score, mean (SD)−1.92 (2.71)−1.83 (2.76)−2.07 (2.67)−1.92 (2.71)−1.89 (2.67)0.001 (ANOVA)Tobacco use, n (%)<0.001 (χ2) Current931 (5.6)276 (6.6)231 (5.5)206 (4.9)218 (5.2) Previous5591 (33.5)1471 (35.4)1411 (33.8)1331 (31.9)1378 (33.0) Never10 163 (60.9)2410 (58.0)2534 (60.7)2637 (63.2)2582 (61.8)Alcohol, mean (SD), g/week137.98 (143.45)134.85 (149.52)136.98 (142.60)138.09 (141.42)141.97 (140.07)0.143 (ANOVA)Body mass index, mean (SD), kg/m226.49 (4.15)28.02 (4.60)26.76 (4.08)26.08 (3.82)25.08 (3.45)<0.001 (ANOVA)VAT, mean (SD), litre3.64 (2.21)4.60 (2.40)3.79 (2.14)3.38 (2.01)2.82 (1.86)<0.001 (ANOVA)Total body fat, mean (SD), %30.08 (8.16)31.78 (8.45)30.92 (8.19)29.83 (7.79)27.78 (7.64)<0.001 (ANOVA)Red meat, mean (SD), frequency/week2.06 (1.36)2.18 (1.41)2.08 (1.36)2.01 (1.32)1.97 (1.35)<0.001 (ANOVA)Vegetable, mean (SD), spoons/day4.79 (3.04)4.58 (3.07)4.77 (2.94)4.82 (2.98)4.99 (3.14)<0.001 (ANOVA)Fruit, mean (SD), spoons/day3.22 (2.44)2.96 (2.21)3.13 (2.35)3.31 (2.63)3.48 (2.53)<0.001 (ANOVA)Processed meat, mean (SD), frequency/week1.42 (1.36)1.57 (1.43)1.37 (1.32)1.37 (1.32)1.39 (1.37)<0.001 (ANOVA)Use of table salt, n (%)0.015 (χ2) Never/rarely10 256 (61.5)2527 (60.8)2545 (60.9)2577 (61.7)2607 (62.4) Sometimes4304 (25.8)1048 (25.2)1087 (26.0)1103 (26.4)1066 (25.5) Usually1654 (9.9)441 (10.6)440 (10.5)369 (8.8)404 (9.7) Always471 (2.8)141 (3.4)104 (2.5)124 (3.0)101 (2.4)Oil fish, mean (SD), frequency/week1.10 (0.99)1.10 (1.00)1.10 (0.99)1.10 (0.98)1.08 (0.99)0.781 (ANOVA)ASCVD, n (%)1698 (10.2)630 (15.2)457 (10.9)334 (8.0)277 (6.6)<0.001 (χ2)Diabetes, n (%)991 (5.9)454 (10.9)245 (5.9)181 (4.3)111 (2.7)<0.001 (χ2)Hyperlipidaemia, n (%)5446 (32.6)1701 (40.9)1494 (35.8)1248 (29.9)1003 (24.0)<0.001 (χ2)Hypertension, n (%)5507 (33.0)1823 (43.9)1472 (35.2)1221 (29.3)991 (23.7)<0.001 (χ2)Central obesity, n (%)8611 (51.6)2739 (65.9)2383 (57.1)2001 (47.9)1488 (35.6)<0.001 (χ2)ANOVA, analysis of variance; ASCVD, atherosclerotic cardiovascular disease; EPAT, epicardial and pericardial adipose tissue; PA, physical activity; SD, standard deviation; VAT, visceral adipose tissue. Baseline characteristics of participants categorised by total PA acceleration ANOVA, analysis of variance; ASCVD, atherosclerotic cardiovascular disease; EPAT, epicardial and pericardial adipose tissue; PA, physical activity; SD, standard deviation; VAT, visceral adipose tissue. In the analyses of baseline characteristics stratified by questionnaire-based PA, participants in the higher quartiles also had a smaller EPAT area, lower percentage of total body fat and abdominal VAT, consistent with the findings from the accelerometer-measured PA analyses. However, other trends were less pronounced or even reversed in the questionnaire-based cohort; for example, age and total body fat and abdominal VAT did not show the same trends (online supplemental table 2). Associations between physical activity and EPAT area Collinearity diagnostics indicated no evidence of collinearity among the variables included in the models (online supplemental table 3). Restricted cubic spline analyses showed that higher levels of accelerometer-measured total PA, LPA, MPA and VPA were associated with smaller EPAT areas, whereas longer SB correlated with greater EPAT areas (figure 1). Collectively, most exposure and EPAT relationships displayed non-linear features in the base adjusted models (Model 1), additional adjustment for total body fat percentage and VAT largely attenuated the non-linear trend (figure 1). Restricted cubic splines for the associations between total PA/accelerometer-measured intensity-specific PA/SB time and EPAT area. Model 1: Adjusted for age, sex, ethnicity, Townsend Deprivation Index, smoking, alcohol consumption, dietary intake of red meat, processed meat, vegetable, fruit, salt and oily fish. Model 2: Additionally adjusted for total body fat percentage based on Model 1. Model 3: Additionally adjusted for VAT based on Model 1. Black vertical dashed lines represent current WHO recommendations (at least 150–300 min/week of MPA or 75–150 min/week of VPA). The red vertical dashed line indicates a previously reported threshold for SB (10.6 hours/day) associated with increased cardiovascular risk. EPAT, epicardial and pericardial adipose tissue; LPA, light physical activity; MPA, moderate physical activity; PA, physical activity; SB, sedentary behaviour; VPA, vigorous physical activity. We further evaluated the association between PA and EPAT area using the current WHO recommendations as reference thresholds.35 Previous studies have reported that SB exceeding 10.6 hours/day is associated with a significantly increased risk of cardiovascular events.6 This threshold was also considered in the present analysis. As shown in figure 1, EPAT area decreased more rapidly as MPA increased within the WHO-recommended range of 150–300 min/week. For VPA, a marked reduction in EPAT area was observed once activity reached approximately 75 min/week, corresponding to the lower limit of the WHO recommendation. In contrast, SB showed the opposite pattern, where EPAT area increased substantially when SB exceeded 10.6 hours/day (figure 1). Table 2 presents the associations of accelerometer-measured PA and SB with EPAT area. In Model 1, higher levels of PA were consistently associated with smaller EPAT area, whereas greater SB was associated with larger EPAT area. Compared with participants in the lowest quartile (Q1), those in the highest quartile (Q4) had a relative 21.29% smaller EPAT area for total PA, 11.24% smaller for LPA, 18.28% smaller for MPA and 19.10% smaller for VPA (all p for trend <0.001). Conversely, participants in the highest quartile of SB had a 15.90% larger EPAT area than those in the lowest quartile (p for trend <0.001). Further adjustment for total body fat percentage (Model 2) attenuated these associations, although all overall trends remained statistically significant. With additional adjustment for VAT in Model 3, the effect estimates were further attenuated to a greater extent. These results suggest that both general adiposity and visceral adiposity may partially explain the observed associations between PA, SB and EPAT area, with VAT appearing to account for a larger proportion of these relationships (table 2). Table 2Association between accelerometer-measured LPA (30 min/day), MPA (30 min/day), VPA (10 min/day) and SB (hour/day) with EPAT areaEPAT area, cm2Q1Q2Q3Q4Relative mean difference (Q4 vs Q1) (%)P for trendTotal volume Crude24.72 (24.35 to 25.09)21.47 (21.15 to 21.79)20.05 (19.76 to 20.35)18.10 (17.83 to 18.37)−26.77<0.001 Model 121.53 (20.65 to 22.44)19.56 (18.77 to 20.39)18.57 (17.81 to 19.36)16.94 (16.25 to 17.66)−21.29<0.001 Model 220.26 (19.49 to 21.06)19.30 (18.57 to 20.06)18.95 (18.23 to 19.70)18.32 (17.63 to 19.05)−9.56<0.001 Model 320.20 (19.51 to 20.92)19.85 (19.16 to 20.56)19.72 (19.04 to 20.43)19.31 (18.64 to 20.00)−4.43<0.001LPA (30 min/day) Crude23.56 (23.20 to 23.91)21.32 (21.00 to 21.65)20.23 (19.93 to 20.54)18.94 (18.66 to 19.23)−19.59<0.001 Model 120.45 (19.60 to 21.33)19.26 (18.47 to 20.09)18.87 (18.10 to 19.69)18.15 (17.40 to 18.93)−11.24<0.001 Model 219.83 (19.07 to 20.61)19.19 (18.47 to 19.95)19.08 (18.36 to 19.84)18.88 (18.17 to 19.63)−4.76<0.001 Model 320.10 (19.41 to 20.82)19.72 (19.04 to 20.43)19.77 (19.09 to 20.48)19.56 (18.89 to 20.27)−2.670.002MPA (30 min/day) Crude24.10 (23.74 to 24.47)21.41 (21.09 to 21.73)20.19 (19.88 to 20.49)18.50 (18.22 to 18.78)−23.25<0.001 Model 121.10 (20.24 to 22.00)19.32 (18.53 to 20.14)18.42 (17.67 to 19.22)17.25 (16.53 to 17.99)−18.28<0.001 Model 220.00 (19.24 to 20.79)19.24 (18.51 to 20.00)18.94 (18.21 to 19.69)18.53 (17.82 to 19.27)−7.32<0.001 Model 320.02 (19.33 to 20.73)19.82 (19.14 to 20.53)19.76 (19.07 to 20.47)19.45 (18.77 to 20.15)−2.86<0.001VPA (10 min/day) Crude22.33 (22.00 to 22.67)22.00 (21.66 to 22.33)20.85 (20.53 to 21.17)18.81 (18.53 to 19.10)−15.77<0.001 Model 121.01 (20.16 to 21.91)20.12 (19.29 to 20.98)18.86 (18.08 to 19.66)17.00 (16.30 to 17.72)−19.10<0.001 Model 219.87 (19.12 to 20.65)19.71 (18.96 to 20.49)19.24 (18.51 to 20.00)18.25 (17.56 to 18.98)−8.14<0.001 Model 320.02 (19.33 to 20.73)20.07 (19.38 to 20.79)19.75 (19.07 to 20.46)19.35 (18.68 to 20.05)−3.32<0.001SB (hour/day) Crude18.81 (18.53 to 19.10)20.05 (19.75 to 20.36)21.18 (20.86 to 21.50)24.06 (23.70 to 24.43)27.92<0.001 Model 117.93 (17.19 to 18.70)18.61 (17.84 to 19.41)19.14 (18.35 to 19.96)20.78 (19.92 to 21.67)15.90<0.001 Model 218.80 (18.08 to 19.54)18.91 (18.18 to 19.65)19.10 (18.38 to 19.86)20.02 (19.26 to 20.81)6.52<0.001 Model 319.54 (18.87 to 20.24)19.73 (19.04 to 20.44)19.75 (19.07 to 20.45)20.08 (19.39 to 20.80)2.760.002Note: Values are adjusted geometric means (95% CIs) of EPAT area estimated from linear regression models with log-transformed EPAT area and back-transformed to the original scale. Relative mean difference represents the percentage difference comparing Q4 with Q1. P for trend was calculated using the median value of each quartile as a continuous variable.Model 1: Adjusted for age, sex, ethnicity, Townsend Deprivation Index, smoking, alcohol consumption, dietary intake of red meat, processed meat, vegetable, fruit, salt and oily fish. Model 2: Additionally adjusted for total body fat percentage based on Model 1. Model 3: Additionally adjusted for VAT based on Model 1.EPAT, epicardial and pericardial adipose tissue; LPA, light physical activity; MPA, moderate physical activity; SB, sedentary behaviour; VPA, vigorous physical activity. Association between accelerometer-measured LPA (30 min/day), MPA (30 min/day), VPA (10 min/day) and SB (hour/day) with EPAT area Note: Values are adjusted geometric means (95% CIs) of EPAT area estimated from linear regression models with log-transformed EPAT area and back-transformed to the original scale. Relative mean difference represents the percentage difference comparing Q4 with Q1. P for trend was calculated using the median value of each quartile as a continuous variable. Model 1: Adjusted for age, sex, ethnicity, Townsend Deprivation Index, smoking, alcohol consumption, dietary intake of red meat, processed meat, vegetable, fruit, salt and oily fish. Model 2: Additionally adjusted for total body fat percentage based on Model 1. Model 3: Additionally adjusted for VAT based on Model 1. EPAT, epicardial and pericardial adipose tissue; LPA, light physical activity; MPA, moderate physical activity; SB, sedentary behaviour; VPA, vigorous physical activity. Table 3 summarised the associations between questionnaire-assessed PA and EPAT area. Overall, higher self-reported total PA, walking, MPA and VPA were inversely associated with smaller EPAT area in both crude and Model 1 analyses. In Model 1, participants in the top quartile of total PA exhibited 8.19% smaller EPAT area relative to those in the bottom quartile (p for trend <0.001). Comparable inverse relationships were detected for walking, MPA and VPA: individuals in their respective highest quartiles had 3.07%, 6.78% and 9.64% smaller EPAT areas than counterparts in the lowest quartiles. Subsequent additional adjustment for body fat percentage and VAT markedly attenuated all these associations; only the relationship for VPA remained significant (table 3). Table 3Association between questionnaire-based walking (30 min/day), MPA (30 min/day), VPA (30 min/day) and total PA (MET30 min/day) with EPAT areaEPAT area, cm2Q1Q2Q3Q4Relative mean difference (Q4/Q3 vs Q1) (%)P for trendTotal PA (MET30 min/day) Crude22.87 (22.62 to 23.11)21.22 (20.99 to 21.45)20.93 (20.71 to 21.16)20.96 (20.73 to 21.18)−8.35<0.001 Model 120.26 (19.70to 20.84)19.05 (18.52 to 19.60)18.70 (18.18 to 19.24)18.60 (18.08 to 19.13)−8.19<0.001 Model 219.30 (18.80 to 19.81)18.91 (18.42 to 19.41)18.92 (18.43 to 19.42)19.14 (18.64 to 19.64)−0.860.844 Model 320.24 (19.77 to 20.72)19.98 (19.51 to 20.45)19.95 (19.48 to 20.42)19.96 (19.49 to 20.43)−1.380.071Walking (30 min/day) Crude22.30 (22.06 to 22.54)21.40 (21.18 to 21.63)21.09 (20.86 to 21.33)21.13 (20.90 to 21.35)−5.25<0.001 Model 119.66 (19.11 to 20.23)19.15 (18.62 to 19.70)18.90 (18.37 to 19.45)19.06 (18.52 to 19.61)−3.070.001 Model 219.07 (18.58 to 19.58)19.01 (18.53 to 19.51)18.96 (18.47 to 19.47)19.30 (18.80 to 19.81)1.170.016 Model 320.13 (19.67 to 20.61)20.06 (19.59 to 20.53)19.93 (19.46 to 20.40)20.04 (19.57 to 20.51)−0.480.57MPA (30 min/day) Crude22.84 (22.57 to 23.11)21.44 (21.23 to 21.65)20.75 (20.51 to 20.99)21.08 (20.87 to 21.30)−7.70<0.001 Model 120.10 (19.54 to 20.69)19.26 (18.73 to 19.81)18.74 (18.21 to 19.29)18.74 (18.22 to 19.28)−6.78<0.001 Model 219.19 (18.69 to 19.70)19.03 (18.54 to 19.53)18.95 (18.45 to 19.45)19.14 (18.65 to 19.65)−0.250.7 Model 320.16 (19.68 to 20.64)20.12 (19.66 to 20.59)19.92 (19.45 to 20.40)19.94 (19.48 to 20.42)−1.060.059VPA (10 min/day) Crude22.52 (22.32 to 22.73)21.27 (21.06 to 21.47)20.65 (20.46 to 20.85)−8.29<0.001 Model 120.19 (19.63 to 20.75)19.05 (18.52 to 19.59)18.24 (17.74 to 18.76)−9.64<0.001 Model 219.38 (18.89 to 19.89)19.02 (18.53 to 19.52)18.77 (18.29 to 19.26)−3.16<0.001 Model 320.21 (19.75 to 20.68)20.02 (19.56 to 20.49)19.86 (19.40 to 20.33)−1.740.001Note: Values are adjusted geometric means (95% CIs) of EPAT area estimated from linear regression models with log-transformed EPAT area and back-transformed to the original scale. Relative mean difference represents the percentage difference comparing Q4 with Q1. P for trend was calculated using the median value of each quartile as a continuous variable.Model 1: Adjusted for age, sex, ethnicity, Townsend Deprivation Index, smoking, alcohol consumption, dietary intake of red meat, processed meat, vegetable, fruit, salt and oily fish. Model 2: Additionally adjusted for total body fat percentage based on Model 1. Model 3: Additionally adjusted for VAT based on Model 1.EPAT, epicardial and pericardial adipose tissue; LPA, light physical activity; MET, metabolic equivalent; MPA, moderate physical activity; PA, physical activity; VPA, vigorous physical activity. Association between questionnaire-based walking (30 min/day), MPA (30 min/day), VPA (30 min/day) and total PA (MET30 min/day) with EPAT area Note: Values are adjusted geometric means (95% CIs) of EPAT area estimated from linear regression models with log-transformed EPAT area and back-transformed to the original scale. Relative mean difference represents the percentage difference comparing Q4 with Q1. P for trend was calculated using the median value of each quartile as a continuous variable. Model 1: Adjusted for age, sex, ethnicity, Townsend Deprivation Index, smoking, alcohol consumption, dietary intake of red meat, processed meat, vegetable, fruit, salt and oily fish. Model 2: Additionally adjusted for total body fat percentage based on Model 1. Model 3: Additionally adjusted for VAT based on Model 1. EPAT, epicardial and pericardial adipose tissue; LPA, light physical activity; MET, metabolic equivalent; MPA, moderate physical activity; PA, physical activity; VPA, vigorous physical activity. Interaction and mediation of PA and SB with body fat for EPAT area When stratified by total body fat percentage, inverse associations of LPA, MPA and VPA with EPAT area, as well as a positive association between SB and EPAT area, were observed across all tertiles. There was no significant evidence of interaction by total body fat percentage (all p for interaction >0.05). A similar pattern was observed across VAT strata for LPA, MPA and SB, with broadly consistent associations and no statistically significant interactions. In contrast, the association between VPA and EPAT area differed significantly across VAT strata (p for interaction=0.003). Specifically, the inverse association was strongest among participants in the highest VAT tertile compared with those in the lowest and middle tertiles. These findings suggest that the association between VPA and lower EPAT area may be more pronounced among individuals with higher visceral adiposity (table 4). Table 4Linear associations between accelerometer-measured LPA (30 min/day), MPA (30 min/day), VPA (10 min/day) and SB (hour/day) with EPAT area stratified by different levels of total body fat percentage or VAT volumeEPAT areaEPAT areaEPAT areaP additive interaction(Total fat percentage Q1)(Total fat percentage Q2)(Total fat percentage Q3)β (95% CI)β (95% CI)β (95% CI)LPA (30 min/day)−0.01 (−0.02 to −0.01)−0.01 (−0.02 to −0.01)−0.02 (−0.02 to −0.01)0.206MPA (30 min/day)−0.05 (−0.06 to −0.04)−0.03 (−0.04 to −0.02)−0.05 (−0.06 to −0.04)0.213VPA (10 min/day)−0.07 (−0.09 to −0.06)−0.08 (−0.10 to −0.06)−0.10 (−0.13 to −0.07)0.247SB (hour/day)0.03 (0.02 to 0.03)0.01 (0.01 to 0.02)0.02 (0.02 to 0.03)0.632(VAT Q1)(VAT Q2)(VAT Q3)P additive interactionβ (95% CI)β (95% CI)β (95% CI)LPA (30 min/day)−0.01 (−0.01 to −0.00)−0.01 (−0.01 to −0.00)−0.01 (−0.02 to −0.01)0.235MPA (30 min/day)−0.02 (−0.03 to −0.01)−0.02 (−0.03 to −0.01)−0.04 (−0.05 to −0.02)0.181VPA (10 min/day)−0.03 (−0.04 to −0.01)−0.03 (−0.05 to −0.01)−0.09 (−0.11 to −0.06)0.003SB (hour/day)0.01 (0.00 to 0.02)0.01 (0.00 to 0.02)0.01 (0.01 to 0.02)0.403Note: Adjusted for age, sex, ethnicity, Townsend Deprivation Index, smoking, alcohol consumption, dietary intake of red meat, processed meat, vegetable, fruit, salt and oily fish.For total fat percentage, Q1: <26%; Q2: 26%~33.6%; Q3 >33.6%. For VAT (litres), Q1: <2.32; Q2: 2.32~4.25; Q3: >4.25.EPAT, epicardial and pericardial adipose tissue; LPA, light physical activity; MPA, moderate physical activity; SB, sedentary behaviour; VAT, visceral adipose tissue; VPA, vigorous physical activity. Linear associations between accelerometer-measured LPA (30 min/day), MPA (30 min/day), VPA (10 min/day) and SB (hour/day) with EPAT area stratified by different levels of total body fat percentage or VAT volume Note: Adjusted for age, sex, ethnicity, Townsend Deprivation Index, smoking, alcohol consumption, dietary intake of red meat, processed meat, vegetable, fruit, salt and oily fish. For total fat percentage, Q1: <26%; Q2: 26%~33.6%; Q3 >33.6%. For VAT (litres), Q1: <2.32; Q2: 2.32~4.25; Q3: >4.25. EPAT, epicardial and pericardial adipose tissue; LPA, light physical activity; MPA, moderate physical activity; SB, sedentary behaviour; VAT, visceral adipose tissue; VPA, vigorous physical activity. In mediation analyses, VAT appeared to account for a substantial proportion of the observed associations between PA/SB behaviours and EPAT (online supplemental table 4). The proportion attributable to the pathway through VAT was estimated to be 76% for LPA, 83% for MPA and 80% for VPA, compared with corresponding direct proportions of 24%, 17% and 20%, respectively. For SB, the pathway through VAT accounted for 83% of the observed association with EPAT, whereas the direct proportion was 17%. These findings suggest that visceral adiposity constitutes a key mechanistic pathway linking PA to EPAT deposition. Isotemporal substitution analyses of accelerometer-measured PA and EPAT In isotemporal substitution analyses, reallocating time from SB to more active behaviours was associated with lower EPAT. Replacing 30 min/day of SB with 30 min/day of LPA or MPA was associated with 1.20% (95% CI 0.83% to 1.57%) and 4.18% (95% CI 3.41% to 4.94%) smaller EPAT, respectively. Whereas replacing 10 min/day of SB with 10 min/day of VPA was associated with 8.89% (95% CI 7.73% to 10.02%) smaller EPAT (table 5). Similar inverse associations were observed when LPA or MPA was replaced by higher-intensity activity. Conversely, substituting VPA with SB, LPA or MPA was associated with substantially larger EPAT. Overall, these findings were consistent with the main analyses and suggest that reallocating time toward higher-intensity PA may be associated with reduced epicardial adiposity. Table 5Isotemporal substitution analyses of accelerometer-measured PA and EPATSubstitution Model 1EPAT areaSubstitution Model 2EPAT areaSubstitution Model 3EPAT areaSubstitution Model 4EPAT areaSB (30 min/day)Substituted1.22% (0.84 to 1.60)4.36% (3.53 to 5.19)32.20% (27.31 to 37.28)LPA (30 min/day)−1.20% (−1.57 to −0.83)Substituted3.11% (2.07 to 4.16)30.61% (25.79 to 35.62)MPA (30 min/day)−4.18% (−4.94 to −3.41)−3.01% (−3.99 to −2.03)Substituted26.68% (21.53 to 32.04)VPA (10 min/day)−8.89% (−10.02 to −7.73)−8.52% (−9.66 to −7.36)−7.58% (−8.85 to −6.29)SubstitutedNote: Values are percent differences (95% CIs) in EPAT area estimated from linear regression models with log-transformed EPAT and back-transformed to the original scale. Substitution Model 1 omitted SB; substitution Model 2 omitted LPA; substitution Model 3 omitted MPA; substitution Model 4 omitted VPA. Models were adjusted for age, sex, ethnicity, Townsend Deprivation Index, smoking status, alcohol consumption, dietary intake of red meat, processed meat, vegetables, fruit, salt added to food and oily fish.EPAT, epicardial adipose tissue area; LPA, light-intensity physical activity; MPA, moderate-intensity physical activity; PA, physical activity; SB, sedentary behaviour; VPA, vigorous-intensity physical activity. Isotemporal substitution analyses of accelerometer-measured PA and EPAT Note: Values are percent differences (95% CIs) in EPAT area estimated from linear regression models with log-transformed EPAT and back-transformed to the original scale. Substitution Model 1 omitted SB; substitution Model 2 omitted LPA; substitution Model 3 omitted MPA; substitution Model 4 omitted VPA. Models were adjusted for age, sex, ethnicity, Townsend Deprivation Index, smoking status, alcohol consumption, dietary intake of red meat, processed meat, vegetables, fruit, salt added to food and oily fish. EPAT, epicardial adipose tissue area; LPA, light-intensity physical activity; MPA, moderate-intensity physical activity; PA, physical activity; SB, sedentary behaviour; VPA, vigorous-intensity physical activity. Subgroup and sensitivity analyses When stratified by sex, the inverse association between VPA and EPAT appeared stronger in men than in women (online supplemental figures 4 and 5). When stratified by BMI, the associations of PA and SB with EPAT were more pronounced in participants with BMI ≥30 kg/m² than in those with BMI <30 kg/m². After further adjustment for total body fat percentage and VAT, only VPA remained strongly associated with EPAT in the BMI ≥30 kg/m² group (online supplemental figures 6 and 7). In analyses stratified by central obesity, the decline in EPAT appeared to plateau once MPA and VPA exceeded the WHO-recommended durations among participants without central obesity (online supplemental figures 8 and 9). When stratified by ASCVD status, increasing MPA and VPA were associated with a greater reduction in EPAT among participants without ASCVD than among those with ASCVD. Among participants with ASCVD, short-term VPA was associated with lower EPAT, whereas higher levels of VPA were associated with little further reduction (online supplemental figures 10 and 11). For type 2 diabetes, no significant associations of LPA, MPA, VPA or SB with EPAT were observed among participants with diabetes, whereas the associations remained evident among those without diabetes (online supplemental figures 12 and 13). Similar patterns were observed in analyses stratified by blood lipid levels and blood pressure (online supplemental figures 14–17). In ethnicity-stratified analyses, the findings among white participants were largely consistent with the main results, whereas no statistically significant associations were observed for other ethnic groups despite similar overall trends (online supplemental figures 18 and 19). Sensitivity analyses showed that excluding participants in the top 10% of EPAT did not dramatically alter the observed associations, although the associations were attenuated after further adjustment for VAT (online supplemental figure 20). In addition, further adjustment for different intensities of PA and SB did not change the main findings, supporting the robustness of the results (online supplemental figure 21). Discussion In this population-based study using UK Biobank data, longer durations of PA at various intensities, especially VPA, were associated with reduced EPAT area, while prolonged SB was linked to increased epicardial adiposity. The association between SB and EPAT area became particularly pronounced when daily sitting time exceeded 10.6 hours. After adjusting for total body fat, the associations between PA of all intensities, SB and EPAT area remained significant. Further adjustment for VAT substantially attenuated these relationships but did not eliminate them. Our mediation analysis indicates that a significant portion of the association between PA and SB and EPAT area is mediated by VAT, particularly in MPA, VPA and SB. This suggests that changes in EPAT area are not a direct result of exercise behaviour, but rather due to broader changes in VAT.36 Physiologically, both VAT and EPAT area are metabolically active ectopic fat reservoirs, sharing common upstream drivers, including lipid metabolism disorders and low-grade systemic inflammation.17,28 Since PA is known to significantly reduce abdominal visceral adiposity and improve systemic metabolic regulation,31 the reduction in EPAT area may at least partially result from improvements in abdominal visceral adiposity. The isotemporal substitution analyses further strengthened our findings by showing that replacing SB or lower-intensity with higher-intensity activities, was associated with progressively lower EPAT, highlighting the potential relevance of activity intensity reallocation for ectopic adipose distribution. Moreover, compared with low-intensity PA, a shorter duration of VPA was associated with a greater reduction in EPAT area. These findings further suggest that VPA may represent a practical and potent strategy to alleviate the burden of cardiovascular disease and obesity. Previous studies,18–20 though limited by small sample sizes, have shown that increased PA—especially at higher intensities—can reduce epicardial and pericardial adipose tissue. Conversely, greater SB is associated with higher cardiac fat accumulation. However, these studies are limited by indirect measurement methods and restricted study populations, which reduce the accuracy and generalisability of their findings. Additionally, reliance on self-reported PA, rather than objective measurements, is more likely to introduce potential recall and reporting biases. In contrast, our present study leverages the largest population-based sample to date, using objective PA measurements and comprehensive assessment of EPAT area. This robust design enables a more precise evaluation of the relationship between PA, SB and epicardial adiposity, providing valuable validation and clarification of these associations. Although some studies suggest that EPAT predicts atherosclerosis independently of obesity,37 the inclusion of VAT and total body fat as covariates reveals their significant influence on EPAT. In our study, after adjusting for total adiposity, the associations between PA, SB and EPAT were attenuated but remained statistically significant. When further adjustment was made for VAT, both PA and SB continued to show strong correlations with EPAT. However, in models that included VAT adjustment, the proportion of variance in EPAT area explained by PA and SB was reduced compared with models without VAT adjustment. These findings highlight the additional influence of both total adiposity and abdominal VAT volume on EPAT accumulation, suggesting that these factors partially mediate the relationship between lifestyle behaviours and EPAT. Recent research confirms VAT mediating role in the link between cardiorespiratory fitness, PA and EPAT.37 Similarly, population studies have demonstrated direct associations between waist circumference and epicardial adipose tissue,38 which proposed that EPAT may affect cardiovascular risk through mechanisms similar to VAT metabolism. These similarities suggest that exercise-induced reductions in VAT may also decrease EPAT. Importantly, our results indicate that the associations between PA, SB and EPAT persist even after VAT adjustment, suggesting that EPAT reduction can occur independently of VAT and overall adiposity. A recent study demonstrated that EPAT contributes to ASCVD risk,9 which underscores the importance of targeting EPAT through lifestyle interventions. By clarifying the association between PA and EPAT, our study provides valuable guidance for adopting active lifestyles and reducing SB, which may help lower EPAT and, consequently, improve cardiovascular health. This study possesses several notable strengths. First, the inclusion of over 16 000 participants provides exceptional statistical power. The cohort itself is characterised by a wide range of ages, health statuses and demographic backgrounds, allowing for more comprehensive and representative conclusions. Another key strength is the use of objective measurement techniques: PA and SB were assessed using accelerometers, which offer precise, quantitative data and minimise recall bias, while EPAT area was measured using high-resolution CMR, ensuring greater accuracy than traditional imaging or anthropometric methods. Furthermore, the study rigorously adjusts for important confounders such as VAT, total body fat and other metabolic risk factors. This comprehensive covariate adjustment strengthens the validity of the observed associations and allows for more nuanced analyses, including subgroup evaluations by obesity status. Collectively, these strengths make this study a significant contribution to the field and provide valuable insights for developing effective lifestyle interventions to reduce cardiovascular risk. Despite the advantages, this study has several limitations. First, only one ethnicity was included, limiting the generalisation of the findings to other populations. Second, EPAT measurements are performed several years after baseline data collection, which can introduce bias and affect the accuracy of the observed associations. A previous study showed that individual PA levels did not change much over time, but the time spent on VPA decreased slightly.39 Third, the cross-sectional design further limits the ability to establish causal relationships between PA, EPAT and cardiovascular outcomes, as there were no longitudinal analyses or repeated measurements. Fourth, the judgement of exercise time is the data collected when wearing the measuring instrument within a week, which may cause selection bias, because the enrolled participants will intentionally or unintentionally increase the intensity of their exercise, resulting in overestimation of the improvement of exercise on EPAT. Addressing this issue also requires a deliberate controlled trial. Finally, the biological mechanisms between PA and EPAT are unclear and not well explored. Future research should address these limitations, including more diverse populations, employ longitudinal designs with repeated measures and examine the mechanisms within to better inform targeted interventions for ASCVD prevention and management. Conclusion This study demonstrates significant associations between PA levels and pericardial adiposity in a large cohort. These findings underscore the potential impact of public health initiatives aimed at increasing PA, specifically higher intensity activity, to reduce the burden of cardiovascular disease on healthcare systems. Future research should seek to validate these results in more diverse populations, incorporate longer follow-up periods and use repeated measurements of both PA and pericardial adiposity. Supplementary material Footnotes Funding: This work was financially supported by the Macau Science and Technology Development Fund (0035/2025/ITP1). Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: The UK Biobank study has been approved by the North West Multi-Center Research Ethics Committee (11/NW/0382), and all participants provided informed consent for follow-up at recruitment. Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. Data availability free text: Original data are downloaded from the UK Biobank under application No. 102278. The other data that support the findings of this study are available from the corresponding author upon reasonable request. Data availability statement Data are available upon reasonable request. References Associated Data Supplementary Materials Data Availability Statement Data are available upon reasonable request.