Associations of Objectively Measured Physical Activity and Abdominal Fat Distribution ANNELOTTE PHILIPSEN1, ANNE-LOUISE SM1DT HANSEN2, MARIT EIKA J0RGENSEN1, S0REN BRAGE3 BENDIX CARSTENSEN1, ANNELLI SANDBAEK2, THOMAS PETER ALMDAL4, JEPPE GRAM5 ERLING BJERREGAARD PEDERSEN6, TORSTEN LAURITZEN2, and DANIEL RINSE WITTE7 1Stem Diabetes Center A/S, Gentofte, DENMARK; 2Section o f General Medical Practice, Department o f Public Health, Aarhus University, Aarhus, DENMARK; 3MRC Epidemiology Unit, University o f Cambridge, Cambridge, UNITED KINGDOM; Department o f Internal Medicine F, Gentofte Hospital, Gentofte, DENMARK; 5Department o f Endocrinology, Hospital o f Southwest Denmark, Esbjerg, DENMARK; 6University Clinic in Nephrology and Hypertension, Department o f Medical Research, Holstebro Hospital, Holstebro, DENMARK; department o f Public Health, Public Research Center for Health Strassen, LUXEMBOURG
ABSTRACT PHILIPSEN, A., A.-L. S. HANSEN, M. E. J0RGENSEN, S. BRAGE, B. CARSTENSEN, A. SANDBAEK, T. P. ALMDAL, J. GRAM, E. B. PEDERSEN, T. LAURITZEN, and D. R. WITTE. Associations of Objectively Measured Physical Activity and Abdominal Fat Distribution. Med. Sci. Sports Exerc., Vol. 47, No. 5, pp. 983—989, 2015. Introduction/Purpose: Visceral adipose tissue (VAT) and physical activity are both independent predictors of Type 2 diabetes. Physical activity and overall obesity are inversely associated with each other. Yet the nature of the association between objectively measured dimensions of physical activity and abdominal fat distribution has not been well characterized. We aimed to do so in a middle-age to elderly population at high risk of diabetes. Methods: A crosssectional analysis of 1134 participants of the ADDITION-PRO study. VAT and subcutaneous adipose tissue (SAT) were assessed onedimensionally by ultrasonography and physical activity with combined accelerometry and HR monitoring. Linear regression of physical activity energy expenditure (PAEE) and time spent in different physical activity intensity levels on VAT and SAT was performed.
Cl = -0 .0 to -0.7). Conclusions: In this population with low physical activity levels, cross-sectional findings indicate that increasing overall physical activity and decreasing time spent sedentary is important to avoid the accumulation of metabolically deleterious VAT. Key Words: VISCERAL FAT, ACCELEROMETRY, TYPE 2 DIABETES, ULTRASONOGRAPHY
isceral adipose tissue (VAT) is an independent predictor of Type 2 diabetes and all-cause mortality above and beyond general obesity and overall ab dominal obesity (14,30). The pathogenic nature of VAT is thought to be related to the higher degree of adverse meta bolic activity seen in VAT (12). Conversely, increased physical activity (PA) is independently associated with decreased risk of Type 2 diabetes and mortality (6,27,31).
Previous studies have firmly established an inverse rela tionship between PA and obesity, and there is also evidence that higher levels of PA reduce overall abdominal obesity (5). However, the nature of the association between PA and VAT has not been well characterized. Studies in this field typically have small samples sizes or use proxy measures for PA and abdominal fat distribution. Yet, in the context of the current obesity pandemic and concurrent rising levels of Type 2 diabetes disease worldwide, PA recommendations are a key component of obesity management. A detailed in vestigation of the association between these two risk factors is therefore warranted. PA can be assessed in several dimensions. Current official PA recommendations focus both on total volume, often ex pressed as PA energy expenditure (PAEE), and on the sep arate entities duration, frequency, and intensity— including vigorous, moderate, and light PA (LPA), and typically advocate that adults should accumulate 30 min or more of
V
Address for correspondence: Dr. Annelotte Philipsen, Steno Diabetes Center A/S, Niels Steensensvej 2, NLD 2.06, Gentofte 2820, Denmark; E-mail: [email protected]. Submitted for publication July 2014. Accepted for publication August 2014. 0195-9131/15/4705-0983/0 MEDICINE & SCIENCE IN SPORTS & EXERCISE® Copyright © 2014 by the American College of Sports Medicine DOI: 10.1249/MSS.0000000000000504
983
EPIDEMIOLOGY
Results: Median body mass index (BMI) was 26.6 kg m -2 and PAEE was 28.1 kJ-kg“ '-d“ \ with 18.9 h-cT1 spent sedentary, 4.5 h-cT1 in light-intensity physical activity, and 0.4 h-d in moderate-intensity physical activity. PAEE was significantly negatively associ ated with VAT, and in women, PAEE was also significantly negatively associated with SAT. The difference in VAT was -1 .1 mm (95/o confidence interval [Cl] = —1.8 to —0.3) per 10-kJ-kg 1d 1 increment, and the corresponding difference in SAT for women was -0 .6 mm (95% Cl = -1 .2 to -0.04) in models adjusted for age, sex, and waist circumference. Exchanging 1 h of light physical ac tivity with moderate physical activity was significantly associated with VAT (-4 .5 mm, 95% Cl = -7 .6 to -1.5). Exchanging one sedentary hour with light physical activity was significantly associated with both VAT (-0 .9 mm, 95% Cl = -0 .1 to -1 .8 ) and SAT ( -0 .4 mm, 95%
moderate-to-vigorous PA (MVPA) daily in bouts of at least 10 min (9,33). During the last few years, studies have also shown that sedentary behavior (SED), defined as any waking behavior characterized by an energy expenditure 1.5 to 3.0 METs) (h-d '), and iii) hours spent in MVPA (>3.0 METs) (h d-1). For the analysis in the present article, only measures from participants with a minimum of 24 h of monitor wear time were considered valid. Obesity measures. Height was measured without shoes to the nearest 0.1 cm with a stadiometer (Seca; Med ical Scales and Measuring Systems, Hamburg, Germany). Weight was measured with the participants barefoot wearing light indoor clothing using a Tanita Body Composition Analyser (Tokyo, Japan). BMI was defined as weight (kg) divided by height (m) squared. Waist circumference was assessed as the average of two measurements to the nearest 0.1 cm at the midpoint between the iliac crest and the lower rib with the participant standing. A D-loop tape was used without applying pressure to the skin. Abdominal fat distri bution was assessed by ultrasonography (Logiq 9 machine; GE Healthcare, Waukesha, WI) by trained sonographers according to a strict validated protocol with adequate re producibility (4,17,18,24). Visceral fat was measured using a 4C abdominal convex transducer and subcutaneous fat with a 9-L small parts linear transducer. With the participant lying down, VAT thickness was defined as the depth (cm) from the peritoneum to the lumbar spine and SAT thickness
http://www.acsm-msse.org
was defined as the depth (cm) from the skin to the linea alba. Measurements were made at the point where the waist circumference crosses the midline. Coefficients of variation for intra- and interobserver variation were in the range 3.4%—6.1% except for interobserver variation for subcutane ous fat (9.5%) (17). S ta tis tic a l M e th o d s
Linear regression analysis of PAEE and time spent at different PA intensity levels on VAT and SAT was per formed with adjustment for age, sex, and smoking. We also present results with further adjustment for waist circumfer ence. Interaction terms were included in all PA models to test for the presence of effect modification by sex. In the PAEE models, the presence of a nonlinear association was tested by inclusion of a quadratic. Given that PAEE and PA intensity variables are composite exposures and therefore not independent of each other, they were analyzed in sepa rate models. As the three intensity variables are mutually exclusive and sum up to 24 h-d-1, they were analyzed by including SED and MVPA in the same model, leaving out LPA. In this manner, the regression coefficient for SED is the difference in VAT (or SAT) between individuals who differ 1 h in SED (and hence LPA) and do the same amount of MVPA. This can be thought of as the effect size on VAT (or SAT) of exchanging 1 h of LPA with 1 h of SED, as suming the association is causal. Similarly, the coefficient for MVPA corresponds to the effect of exchanging 1 h of LPA with 1 h of MVPA. All analyses were performed using SAS 9.2.
Data on 1134 ADDITION-PRO participants were avail able for complete case analysis. Figure 1 shows the de tails on study sample selection, and Table 1 summarizes the characteristics of the study sample. This is a middle-age
M en
n(%)
606 (53) Age 66.3 (62.2-71.2) Ethnicity Caucasian, n (%) 592 (98) Current smokers, n (%) 437 (72) BMI (kg-m 2) 26.9 (24.8-29.7) WC (cm) 99.8 (92.5-107.3) VAT (cm) 8.6 (6.9-10.4) SAT (cm) 2.2 (1.6-2.8) Average time Actiheart was worn (h) 168 (161-173) PAEE (kJ-kg 1-d 1) 30.0 (21.3-41.4) SED, including sleeping, 1.5 to 3.0 METs (h-d-1 ) 4.6 (3.3—5.8) Moderate-intensity PA, >3.0 to 0.4 (0.2-0.9) 6.0 METs (h-d-1) Vigorous-intensity PA, >6.0 METs (h-d-1) 0 (0.0-0.0)
W om en
528 (47) 66.3 (60.7-71.1) 510 (97) 280 (53) 26.0 (22.7-28.9) 87.7 (80.0-96.7) 6.5 (5.3-8.0) 2,8 (2.1-3.5) 165 (148-170) 26.5 (19.4-36.9) 19.2 (17.4-20.6) 4.4 (3.1-6.0) 0.3 (0.1-0.6) 0 (0.0-0.0)
Data are median values (interquartile range).
to elderly Caucasian population with a mean BM1 of 26.9 kg-m ", Of note, very little PA of vigorous intensity was undertaken (mean = 0.03 h). Therefore, the MVPA variable represents almost only moderate-intensity activity (MPA) and will be referred to as such in the rest of this article. Table 2 shows regression coefficients for the associations between abdominal fat depots and PA measures. P A E E . We found a statistically significant negative asso ciation between PAEE and VAT, even after adjustment for waist circumference. There was no statistically significant sex interaction (P = 0.14). A 10-kJ kg_ 1d _ 1 higher PAEE was associated with 3.3-mm lower VAT. A sex interaction term was significant for SAT (P = 0.002). There was a negative association for both sexes, but it was only statistically signifi cant for women. This association remained statistically signif icant after adjustment for waist circumference. Standardizing the coefficients by the population-specific SD for VAT and sex-specific for SAT showed that the strength of the associa tions was similar for VAT and SAT for women, whereas the effect in men for SAT was more than three times smaller. PA in te n s itie s . Moderate PA was inversely statistically significantly associated with VAT. Exchanging 1 h-d-1 of LPA with moderate PA was associated with a 4.5-mm lower
Attending Addition-Pro health examination (N=2082) in 2009-2011
I Excluded by design: Persons at 2 study centres where ultrasound was not done, n=360 Persons with incident diabetes*, n=336
Excluded due to missing values: Missing or very incomplete clinical record form, n=7 Actiheart data, n=358 VAT and/or SAT, n=20
I Study population (N=1134)
* Persons with incident diabetes since baseline screenings were not fasting for ultrasound measurements.
FIGURE 1—Study sample selection.
FAT DISTRIBUTION AND PHYSICAL ACTIVITY
Medicine & Science in Sports & Exercise®
985
EPIDEMIOLOGY
RESULTS
TABLE 1. Participant characteristics.
TA B LE 2. R egression c o e ffic ie n ts and 9 5 % co n fid e n ce in te rva ls fo r the a sso cia tio n betw een o b je c tiv e ly m e asure d ph y s ic a l a c tiv ity and vis c e ra l and s u b c u ta n e o u s a d ip o s e tis s u e .
Outcome V A T (cm ) M u ltiv a ria b le * P, sex in te ra ctio n M en W om en M u ltiv a ria b le * + WC M en W om en SAT (cm ) M u ltiv a ria b le * P, sex in te ra ctio n M en W om en M u ltiv a ria b le * + WC M en W om e n
ABSTRACT PHILIPSEN, A., A.-L. S. HANSEN, M. E. J0RGENSEN, S. BRAGE, B. CARSTENSEN, A. SANDBAEK, T. P. ALMDAL, J. GRAM, E. B. PEDERSEN, T. LAURITZEN, and D. R. WITTE. Associations of Objectively Measured Physical Activity and Abdominal Fat Distribution. Med. Sci. Sports Exerc., Vol. 47, No. 5, pp. 983—989, 2015. Introduction/Purpose: Visceral adipose tissue (VAT) and physical activity are both independent predictors of Type 2 diabetes. Physical activity and overall obesity are inversely associated with each other. Yet the nature of the association between objectively measured dimensions of physical activity and abdominal fat distribution has not been well characterized. We aimed to do so in a middle-age to elderly population at high risk of diabetes. Methods: A crosssectional analysis of 1134 participants of the ADDITION-PRO study. VAT and subcutaneous adipose tissue (SAT) were assessed onedimensionally by ultrasonography and physical activity with combined accelerometry and HR monitoring. Linear regression of physical activity energy expenditure (PAEE) and time spent in different physical activity intensity levels on VAT and SAT was performed.
Cl = -0 .0 to -0.7). Conclusions: In this population with low physical activity levels, cross-sectional findings indicate that increasing overall physical activity and decreasing time spent sedentary is important to avoid the accumulation of metabolically deleterious VAT. Key Words: VISCERAL FAT, ACCELEROMETRY, TYPE 2 DIABETES, ULTRASONOGRAPHY
isceral adipose tissue (VAT) is an independent predictor of Type 2 diabetes and all-cause mortality above and beyond general obesity and overall ab dominal obesity (14,30). The pathogenic nature of VAT is thought to be related to the higher degree of adverse meta bolic activity seen in VAT (12). Conversely, increased physical activity (PA) is independently associated with decreased risk of Type 2 diabetes and mortality (6,27,31).
Previous studies have firmly established an inverse rela tionship between PA and obesity, and there is also evidence that higher levels of PA reduce overall abdominal obesity (5). However, the nature of the association between PA and VAT has not been well characterized. Studies in this field typically have small samples sizes or use proxy measures for PA and abdominal fat distribution. Yet, in the context of the current obesity pandemic and concurrent rising levels of Type 2 diabetes disease worldwide, PA recommendations are a key component of obesity management. A detailed in vestigation of the association between these two risk factors is therefore warranted. PA can be assessed in several dimensions. Current official PA recommendations focus both on total volume, often ex pressed as PA energy expenditure (PAEE), and on the sep arate entities duration, frequency, and intensity— including vigorous, moderate, and light PA (LPA), and typically advocate that adults should accumulate 30 min or more of
V
Address for correspondence: Dr. Annelotte Philipsen, Steno Diabetes Center A/S, Niels Steensensvej 2, NLD 2.06, Gentofte 2820, Denmark; E-mail: [email protected]. Submitted for publication July 2014. Accepted for publication August 2014. 0195-9131/15/4705-0983/0 MEDICINE & SCIENCE IN SPORTS & EXERCISE® Copyright © 2014 by the American College of Sports Medicine DOI: 10.1249/MSS.0000000000000504
983
EPIDEMIOLOGY
Results: Median body mass index (BMI) was 26.6 kg m -2 and PAEE was 28.1 kJ-kg“ '-d“ \ with 18.9 h-cT1 spent sedentary, 4.5 h-cT1 in light-intensity physical activity, and 0.4 h-d in moderate-intensity physical activity. PAEE was significantly negatively associ ated with VAT, and in women, PAEE was also significantly negatively associated with SAT. The difference in VAT was -1 .1 mm (95/o confidence interval [Cl] = —1.8 to —0.3) per 10-kJ-kg 1d 1 increment, and the corresponding difference in SAT for women was -0 .6 mm (95% Cl = -1 .2 to -0.04) in models adjusted for age, sex, and waist circumference. Exchanging 1 h of light physical ac tivity with moderate physical activity was significantly associated with VAT (-4 .5 mm, 95% Cl = -7 .6 to -1.5). Exchanging one sedentary hour with light physical activity was significantly associated with both VAT (-0 .9 mm, 95% Cl = -0 .1 to -1 .8 ) and SAT ( -0 .4 mm, 95%
moderate-to-vigorous PA (MVPA) daily in bouts of at least 10 min (9,33). During the last few years, studies have also shown that sedentary behavior (SED), defined as any waking behavior characterized by an energy expenditure 1.5 to 3.0 METs) (h-d '), and iii) hours spent in MVPA (>3.0 METs) (h d-1). For the analysis in the present article, only measures from participants with a minimum of 24 h of monitor wear time were considered valid. Obesity measures. Height was measured without shoes to the nearest 0.1 cm with a stadiometer (Seca; Med ical Scales and Measuring Systems, Hamburg, Germany). Weight was measured with the participants barefoot wearing light indoor clothing using a Tanita Body Composition Analyser (Tokyo, Japan). BMI was defined as weight (kg) divided by height (m) squared. Waist circumference was assessed as the average of two measurements to the nearest 0.1 cm at the midpoint between the iliac crest and the lower rib with the participant standing. A D-loop tape was used without applying pressure to the skin. Abdominal fat distri bution was assessed by ultrasonography (Logiq 9 machine; GE Healthcare, Waukesha, WI) by trained sonographers according to a strict validated protocol with adequate re producibility (4,17,18,24). Visceral fat was measured using a 4C abdominal convex transducer and subcutaneous fat with a 9-L small parts linear transducer. With the participant lying down, VAT thickness was defined as the depth (cm) from the peritoneum to the lumbar spine and SAT thickness
http://www.acsm-msse.org
was defined as the depth (cm) from the skin to the linea alba. Measurements were made at the point where the waist circumference crosses the midline. Coefficients of variation for intra- and interobserver variation were in the range 3.4%—6.1% except for interobserver variation for subcutane ous fat (9.5%) (17). S ta tis tic a l M e th o d s
Linear regression analysis of PAEE and time spent at different PA intensity levels on VAT and SAT was per formed with adjustment for age, sex, and smoking. We also present results with further adjustment for waist circumfer ence. Interaction terms were included in all PA models to test for the presence of effect modification by sex. In the PAEE models, the presence of a nonlinear association was tested by inclusion of a quadratic. Given that PAEE and PA intensity variables are composite exposures and therefore not independent of each other, they were analyzed in sepa rate models. As the three intensity variables are mutually exclusive and sum up to 24 h-d-1, they were analyzed by including SED and MVPA in the same model, leaving out LPA. In this manner, the regression coefficient for SED is the difference in VAT (or SAT) between individuals who differ 1 h in SED (and hence LPA) and do the same amount of MVPA. This can be thought of as the effect size on VAT (or SAT) of exchanging 1 h of LPA with 1 h of SED, as suming the association is causal. Similarly, the coefficient for MVPA corresponds to the effect of exchanging 1 h of LPA with 1 h of MVPA. All analyses were performed using SAS 9.2.
Data on 1134 ADDITION-PRO participants were avail able for complete case analysis. Figure 1 shows the de tails on study sample selection, and Table 1 summarizes the characteristics of the study sample. This is a middle-age
M en
n(%)
606 (53) Age 66.3 (62.2-71.2) Ethnicity Caucasian, n (%) 592 (98) Current smokers, n (%) 437 (72) BMI (kg-m 2) 26.9 (24.8-29.7) WC (cm) 99.8 (92.5-107.3) VAT (cm) 8.6 (6.9-10.4) SAT (cm) 2.2 (1.6-2.8) Average time Actiheart was worn (h) 168 (161-173) PAEE (kJ-kg 1-d 1) 30.0 (21.3-41.4) SED, including sleeping, 1.5 to 3.0 METs (h-d-1 ) 4.6 (3.3—5.8) Moderate-intensity PA, >3.0 to 0.4 (0.2-0.9) 6.0 METs (h-d-1) Vigorous-intensity PA, >6.0 METs (h-d-1) 0 (0.0-0.0)
W om en
528 (47) 66.3 (60.7-71.1) 510 (97) 280 (53) 26.0 (22.7-28.9) 87.7 (80.0-96.7) 6.5 (5.3-8.0) 2,8 (2.1-3.5) 165 (148-170) 26.5 (19.4-36.9) 19.2 (17.4-20.6) 4.4 (3.1-6.0) 0.3 (0.1-0.6) 0 (0.0-0.0)
Data are median values (interquartile range).
to elderly Caucasian population with a mean BM1 of 26.9 kg-m ", Of note, very little PA of vigorous intensity was undertaken (mean = 0.03 h). Therefore, the MVPA variable represents almost only moderate-intensity activity (MPA) and will be referred to as such in the rest of this article. Table 2 shows regression coefficients for the associations between abdominal fat depots and PA measures. P A E E . We found a statistically significant negative asso ciation between PAEE and VAT, even after adjustment for waist circumference. There was no statistically significant sex interaction (P = 0.14). A 10-kJ kg_ 1d _ 1 higher PAEE was associated with 3.3-mm lower VAT. A sex interaction term was significant for SAT (P = 0.002). There was a negative association for both sexes, but it was only statistically signifi cant for women. This association remained statistically signif icant after adjustment for waist circumference. Standardizing the coefficients by the population-specific SD for VAT and sex-specific for SAT showed that the strength of the associa tions was similar for VAT and SAT for women, whereas the effect in men for SAT was more than three times smaller. PA in te n s itie s . Moderate PA was inversely statistically significantly associated with VAT. Exchanging 1 h-d-1 of LPA with moderate PA was associated with a 4.5-mm lower
Attending Addition-Pro health examination (N=2082) in 2009-2011
I Excluded by design: Persons at 2 study centres where ultrasound was not done, n=360 Persons with incident diabetes*, n=336
Excluded due to missing values: Missing or very incomplete clinical record form, n=7 Actiheart data, n=358 VAT and/or SAT, n=20
I Study population (N=1134)
* Persons with incident diabetes since baseline screenings were not fasting for ultrasound measurements.
FIGURE 1—Study sample selection.
FAT DISTRIBUTION AND PHYSICAL ACTIVITY
Medicine & Science in Sports & Exercise®
985
EPIDEMIOLOGY
RESULTS
TABLE 1. Participant characteristics.
TA B LE 2. R egression c o e ffic ie n ts and 9 5 % co n fid e n ce in te rva ls fo r the a sso cia tio n betw een o b je c tiv e ly m e asure d ph y s ic a l a c tiv ity and vis c e ra l and s u b c u ta n e o u s a d ip o s e tis s u e .
Outcome V A T (cm ) M u ltiv a ria b le * P, sex in te ra ctio n M en W om en M u ltiv a ria b le * + WC M en W om en SAT (cm ) M u ltiv a ria b le * P, sex in te ra ctio n M en W om en M u ltiv a ria b le * + WC M en W om e n
