Body Composition, Cardiorespiratory Fitness, and Low-Grade Inflammation in Middle-Aged Men and Women
Article Outline
The objective of the present study was to determine the respective contributions of visceral adipose tissue (AT) accumulation and cardiorespiratory fitness to variation of inflammatory markers in men and women. Circulating levels of C-reactive protein, tumor necrosis factor-α, interleukin-6, and adiponectin were obtained with visceral AT (computed tomography) and fitness (physical working capacity test) levels in a sample of healthy men (n = 120) and women (n = 152) covering a wide range of adiposity. An inflammation score was developed based on gender-specific percentile values of each inflammatory marker (0 or 1), which yielded a score ranging from 0 (low) to 4 (high). Visceral AT was positively associated with C-reactive protein and interleukin-6 levels (r ≥0.35, p <0.0001), but negatively associated with adiponectin (r = −0.29, p ≤0.0003) after adjustment for fitness. After adjusting for visceral AT, fitness was not associated with variation in inflammatory markers in women and only with adiponectin in men (r = −0.20, p = 0.03). In participants with low visceral AT (<130 cm2 for men and <100 cm2 for women), prevalences of participants with an increased inflammation score were 23.9% and 28.0%, respectively, for participants with high and low fitness, whereas in subjects with increased visceral AT, prevalences of a high inflammation score were 60.0% and 61.7%, respectively, for participants with high and low fitness. In conclusion, these results suggest that the previously reported association between poor fitness and low-grade inflammation may be largely attributable to increased visceral AT accumulation and its associated state of insulin resistance, conditions frequently observed in subjects with poor cardiorespiratory fitness.
Low cardiorespiratory fitness (CRF) levels have been associated with cardiovascular disease (CVD)1 and type 2 diabetes2 incidence and studies have reported an important role for low CRF levels in predicting a deteriorated atherogenic and diabetogenic risk profile.3 There is also evidence that the relation between low CRF levels and increased low-grade inflammation might be independent of obesity. However, no studies have examined whether body fat distribution is a stronger determinant of the relation between poor CRF and low-grade inflammation compared with body mass index (BMI). The first objective of the present study was therefore to investigate whether increased visceral adipose tissue (AT) accumulation or poor CRF predicted altered plasma concentrations of inflammatory cytokines C-reactive protein (CRP), interleukin-6, tumor necrosis factor-α, and adiponectin. A second objective was to verify to what extent the relation between low CRF and inflammation in middle-aged men and women could be attributed to concomitant variation in BMI and visceral adiposity.
Methods
Subjects of this study were asymptomatic men and women without diabetes who participated in the Québec Family Study (QFS). The study sample is composed of healthy men and women on whom we had data on body composition, CRF, and plasma levels of cardiometabolic risk markers. Briefly, the QFS is a population-based study of French-Canadian families living in and around the Québec City area. The QFS was approved by the medical ethics committee of Université Laval (Québec, Canada). Subjects were recruited through the media and gave their written informed consent to participate in the study. Only healthy men and women, 18 to 65 years of age, who were not under treatment for CVD, diabetes, dyslipidemias, or endocrine disorders were considered for the present analyses.
Height, body weight, and waist circumference were measured according to standardized procedures. Body density was measured by the hydrostatic weighing technique.4 The mean of 6 measurements was used to calculate percent body fat from body density using the equation of Siri.5 Fat mass was obtained by multiplying body weight by percent body fat. Measurement of abdominal AT areas was performed by computed tomography with a Siemens Somatom DHR scanner (Siemens, Erlanger, Germany). Briefly, participants were examined in the supine position with the 2 arms stretched above the head. The scan was performed at the abdominal level (L4 and L5 vertebrae) using an abdominal scout x-ray to standardize the position of the scan to the nearest millimeter. Total AT area was calculated by delineating the abdominal scan with a graph pen and then by computing total abdominal AT area with an attenuation range of −190 to −30 Hounsfield Units. Abdominal visceral AT area was measured by drawing a line within the muscle wall surrounding the abdominal cavity. Abdominal subcutaneous AT area was calculated by subtracting visceral AT area from total abdominal AT area.
After a 12-hour overnight fast, blood samples were collected from an antecubital vein into Vacutainer tubes containing ethylenediaminetetra-acetic acid (Miles Pharmaceuticals, Rexdale, Ontario, Canada) for measurement of plasma lipid and lipoprotein levels. Plasma cholesterol and triglyceride concentrations were determined in plasma and lipoprotein fractions using a Technicon RA-500 analyzer (Bayer Corporation, Tarrytown, New York) and enzymatic reagents were obtained from Randox Laboratories Ltd. (Crumlin, United Kingdom). Plasma very low-density lipoproteins (d <1.006 g/ml) were isolated by ultracentrifugation,6 and the high-density lipoprotein fraction was obtained after precipitation of low-density lipoprotein in the infranatant (d >1.006 g/ml) with heparin and MnCl2.7 Cholesterol content of the infranatant fraction was measured before and after the precipitation step allowing the calculation of low-density lipoprotein cholesterol. Enzyme-linked immunosorbent assays were used to measure plasma adiponectin (B-Bridge International, Inc., San Jose, California), interleukin-6, and tumor necrosis factor-α (R&D Systems, Inc., Minneapolis, Minnesota). Plasma CRP levels were measured with a highly sensitive immunoassay that used a monoclonal antibody coated with polystyrene particles (high-sensitivity CRP); the assay was performed with a Behring BN-100 nephelometer (Dade Behring).
A 3-hour 75-g oral glucose tolerance test was performed in the morning after an overnight fast. Blood samples were collected in ethylenediaminetetra-acetic acid–containing tubes through a venous catheter placed in an antecubital vein for determination of plasma glucose and insulin levels. Plasma glucose was measured enzymatically, whereas plasma insulin was measured by radioimmunoassay with polyethylene glycol separation.8 Total glucose and insulin areas under the curve during the oral glucose tolerance test were determined with the trapezoid method.
CRF of each participant was assessed by a progressive submaximal physical working-capacity test performed on a modified Monark cycle ergometer. Heart rate was measured through 1 electrocardiographic derivation and recorded during 3 consecutive 6-minute workloads, each separated by a 1-minute rest. The test was designed to exceed a heart rate of 150 beats/min at the end of the last workload. Physical working capacity with power output at 150 beats/min was then calculated from the linear relation between heart rate and power output. To take into account individual differences in body weight, physical working capacity with power output at 150 beats/min was expressed by kilogram of body weight.
Data are presented as mean ± SD or median (interquartile range) in the tables and mean ± SEM in the figures. Baseline anthropometric and metabolic characteristics of participants (described earlier) are presented in men and women separately. The gender-specific relation between visceral AT accumulation, CRF and fat-free mass, and inflammatory markers was assessed by adjusted and unadjusted Spearman rank correlations. Inflammatory markers CRP, tumor necrosis factor-α, interleukin-6, and adiponectin were quantified in men and women with low or high visceral AT accumulation (<130 or ≥130 cm2 in men and <100 or ≥100 cm2 in women) and low or high CRF (<10.7 or ≥10.7 kpm/kg in men and <6.9 or ≥6.9 kpm/kg in women, which corresponded to gender-specific median values). To estimate low-grade inflammation, an inflammation score based on plasma levels of CRP, adiponectin, interleukin-6, and tumor necrosis factor-α was developed. One point was attributed to participants each time an inflammatory marker was >50th percentile value (Table 1) or <50th percentile value for adiponectin. Participants with a low inflammation score had 0 point, 1 point, or 2 points, whereas participants with a high inflammation scored score had 3 or 4 points. Inflammation score was compared in participants with high or low visceral AT accumulation, BMI, or CRF and differences among groups were evaluated with 1-way analysis of variance. A p value <0.05 was considered statistically significant. All statistical analyses were performed with the SAS package (SAS Institute, Cary, North Carolina).
Table 1. Baseline anthropometric and metabolic characteristics of the 120 men and 152 women of the study
| Variables | Men | Women |
|---|---|---|
| (n = 120) | (n = 152) | |
| Age (years) | 38.2 ±13.9 | 35.5±12.5 |
| Physical working capacity (kpm/kg) | 11.2±3.4 | 7.0±2.3 |
| BMI (kg/m2) | 26.3±4.4 | 27.3±6.7 |
| Waist circumference (cm) | 90.8±12.7 | 82.5±15.1 |
| AT accumulation (cm2) | ||
| Total | 333±183 | 424±216 |
| Visceral | 113±64 | 86±52 |
| Subcutaneous | 220±135 | 338±176 |
| Fat mass (kg) | 18.6±9.6 | 24.5±13.2 |
| Fat-free mass (kg) | 61.6±8.0 | 47.1±6.4 |
| Systolic blood pressure (mm Hg) | 114±13 | 111±12 |
| Diastolic blood pressure (mm Hg) | 71±9 | 69±9 |
| Total cholesterol (mmol/L) (mg/dl) | 4.96(191.5)±0.99(38.2) | 4.79(184.9)±1.25(48.3) |
| LDL cholesterol (mmol/L) (mg/dl) | 3.17(122.4)±0.89(34.4) | 2.86(110.4)±0.82(31.7) |
| HDL cholesterol(mmol/L) (mg/dl) | 1.11(42.9)±0.25(9.7) | 1.29(49.8)±0.34(13.1) |
| Total cholesterol/HDL cholesterol ratio | 4.69±1.33 | 4.09±3.28 |
| Triglycerides (mmol/L) (mg/dl) | 1.41(124.8)(0.90 [79.6]–1.91 [169.0]) | 1.13(100.0)(0.87 [77.0]–1.54 [136.3]) |
| Apolipoprotein B (g/L) | 1.01±0.24 | 0.91±0.21 |
| Glucose (mmol/L) | 5.25(94.6)±0.48(8.6) | 4.99(90.7)±0.43(7.7) |
| Insulin (pmol/L) | 52.0(7.49)(39.0 [5.62]–77.0 [11.09]) | 53.0(7.63)(38.5 [5.54]–77.0 [11.09]) |
| Glucose AUC (mmol/L/min × 10−3) | 1.20(21.6)±0.23(4.1) | 1.12(20.2)±0.21(3.8) |
| Insulin AUC (pmol/L/min × 10−3) | 56.5(8.14)(41.1 [5.92]–93.4 [13.45]) | 59.1(8.51)(45.5 [6.55]–86.7 [12.48]) |
| CRP (mg/L) | 0.84(0.46–2.23) | 1.39(0.73–3.61) |
| Adiponectin (μg/ml) | 4.78(3.55–6.90) | 6.83(4.57–9.19) |
| Tumor necrosis factor-α (pg/ml) | 1.69(1.35–1.99) | 1.65(1.27–1.95) |
| Interleukin-6 (pg/ml) | 1.29(0.94–1.94) | 1.42(0.99–2.09) |
Results
Table 1 presents anthropometric characteristics and mean plasma levels of markers of the lipoprotein–lipid profile, glucose–insulin homeostasis, and inflammation in men and women separately. Spearman correlation coefficients among visceral AT, insulin area under the curve, CRF, and circulating levels of inflammatory markers are presented in Table 2. In men and women, visceral AT was positively associated with CRP and interleukin-6 and negatively correlated with adiponectin before and after adjustment for CRF, whereas no association was found with tumor necrosis factor-α. Insulin area under the curve was also associated with variation in inflammatory markers (tumor necrosis factor-α in men only). In men, CRF was not associated with variation in inflammatory markers in unadjusted analyses but showed a negative relation with adiponectin after adjusting for visceral AT. In women, CRF was negatively associated with CRP and interleukin-6 levels, but these relations were no longer observed after adjusting for visceral AT. In men and women, Spearman correlation coefficients between fat mass and inflammatory markers were in the same range, but not as strong as correlations between visceral AT and inflammatory markers (not shown). Table 3 presents independent “contributors” to the variance in inflammation score. In men and women, visceral AT was found to be the best predictor of the score. In men, CRF and insulin area under the curve appeared to add a small but significant contribution to visceral AT in predicting variation in the inflammation score. In women, age, but not CRF or insulin area under the curve, added a significant contribution to visceral AT in predicting variance in the inflammation score. Mean levels of inflammatory markers in men and women classified by visceral AT accumulation and CRF are depicted in Figure 1. Figure 2 presents the proportion of participants characterized by an increased inflammation score classified by (1) BMI and CRF or (2) visceral AT accumulation and CRF. Compared with normal-weight participants (BMI <25 kg/m2), overweight/obese participants (BMI ≥25 kg/m2) were more likely to be characterized by an increased inflammation score. In overweight/obese participants, those with low CRF levels were more likely to be characterized by an increased inflammation score compared with overweight/obese participants with increased CRF levels. Similar results were obtained when men and women were separated by waist circumference and CRF (not shown). However, when visceral AT was considered as the obesity parameter, those with an increased visceral AT accumulation were more likely to be characterized by an increased inflammation score, irrespective of CRF status. Visceral AT accumulation was similar in participants with high or low CRF within each visceral AT accumulation category. Mean CRF levels were 8.6 ± 1.6 kpm/kg in men considered “unfit” and 13.7 ± 2.8 kpm/kg for men considered “fit.” In women, mean CRF levels were 5.2 ± 1.1 and 8.8 ± 1.7 kpm/kg, respectively, for unfit and fit women. Similarly, mean visceral AT accumulations were 63.4 ± 1.6 and 162.3 ± 52.2 cm2 for men classified as having low and high visceral AT. In women, mean visceral AT accumulations were 47.6 ± 14.2 and 124.8 ± 47.2 cm2, respectively, for those with low and high visceral AT.
Table 2. Spearman unadjusted and adjusted correlation coefficients between visceral adipose tissue accumulation, insulin area under the curve during the oral glucose tolerance test, cardiorespiratory fitness, and inflammatory markers
| CRP | Adiponectin | IL-6 | TNF-α | |
|---|---|---|---|---|
| Men | ||||
| Visceral AT | ||||
| Unadjusted | 0.43⁎ | −0.27⁎ | 0.39⁎ | 0.00 |
| Adjusted for CRF | 0.42⁎ | −0.34⁎ | 0.39⁎ | 0.03 |
| Insulin AUC | ||||
| Unadjusted | 0.22⁎ | −0.35⁎ | 0.27⁎ | 0.20⁎ |
| Adjusted for CRF | 0.18⁎ | −0.39⁎ | 0.28⁎ | 0.21⁎ |
| CRF | ||||
| Unadjusted | −0.11 | −0.04 | −0.08 | 0.02 |
| Adjusted for visceral AT | 0.08 | −0.19⁎ | 0.09 | 0.04 |
| Women | ||||
| Visceral AT | ||||
| Unadjusted | 0.45⁎ | −0.24⁎ | 0.45⁎ | 0.06 |
| Adjusted for CRF | 0.35⁎ | −0.29⁎ | 0.37⁎ | 0.01 |
| Insulin AUC | ||||
| Unadjusted | 0.18⁎ | −0.31⁎ | 0.30⁎ | 0.04 |
| Adjusted for CRF | 0.09 | −0.32⁎ | 0.22⁎ | 0.01 |
| CRF | ||||
| Unadjusted | −0.32⁎ | 0.00 | −0.35⁎ | −0.13 |
| Adjusted for visceral AT | −0.14 | −0.14 | −0.15 | −0.09 |
⁎p ≤0.03. |
Table 3. Multivariate regression analyses showing independent contribution of visceral adiposity to variance in inflammation score
| Dependent Variable | Independent Variables | Total R2 | p Value |
|---|---|---|---|
| Men | |||
| Inflammation score | VisceralAT | 11.8% | <0.0001 |
| +CRF | 16.8% | 0.01 | |
| +insulinAUC | 19.6% | 0.05 | |
| +age | NS | ||
| Women | |||
| Inflammation score | VisceralAT | 22.2% | <0.0001 |
| +age | 28.4% | 0.0006 | |
| +CRF | NS | ||
| +insulinAUC | NS |
Figure 1.
Plasma levels of CRP, adiponectin, interleukin-6, and tumor necrosis factor-α in (A) men and (B) women classified by visceral AT accumulation (<130 or ≥130 cm2 in men and <100 or ≥100 cm2 in women) and CRF levels (<1.7 or ≥10.7 kpm/kg in men and <6.9 or ≥6.9 kpm/kg in women). 1, 2, 3Significantly different from corresponding subgroup (p <0.05, on log-transformed values).
Figure 2.
Proportion of subjects with an increased inflammation score in participants classified by (A) BMI (<25 or ≥25 kg/m2) and CRF (<10.7 or ≥10.7 kpm/kg in men and <6.9 or ≥6.9 kpm/kg in women) or (B) visceral AT accumulation (<130 or ≥130 cm2 in men and <100 or ≥100 cm2 in women) and CRF. The inflammation score was defined as the sum of the number of inflammatory markers >50th percentile (CRP, tumor necrosis factor-α, and interleukin-6) or <50th percentile for adiponectin. 1, 2, 3Significantly different from corresponding subgroup (p <0.05).
Respective contributions of visceral AT accumulation and glucose or insulin areas under the curve during the oral glucose tolerance test and respective contributions of CRF and glucose or insulin areas under the curve to the variation in inflammation score were examined in a 2 × 2 analysis of variance. Figure 3 shows that in participants with visceral obesity, insulin area, but not glucose area, is associated with a higher increased inflammation score. Figure 3 also shows that independently of CRF, increased areas of glucose or insulin were associated with an increased inflammation score. Figure 4 depicts the relation between BMI and visceral AT in men and women separately. Although these measurements are highly correlated, Figure 4 shows that, at any given BMI value, there is a substantial individual variation in visceral adiposity, with men more prone to visceral obesity than women.
Figure 3.
Mean inflammation score in participants classified by plasma glucose area under the curve (AUC) measured during the oral glucose tolerance test and by visceral AT accumulation (A) or CRF levels (B) and by insulin AUC measured during the oral glucose tolerance test and by visceral AT accumulation (C) or CRF levels (D). The inflammation score was defined as the sum of the number of inflammatory markers >50th percentile (CRP, tumor necrosis factor-α, and interleukin-6) or <50th percentile for adiponectin. In men, cut-off values of 1.16 mmol/L/min × 10−3 and of 56.5 pmol/L/min × 10−3, respectively, for glucose and insulin areas under the curve during the oral glucose tolerance test were used to classify participants with low or high values corresponding to median values observed in the sample of men. In women, these respective cut-off values corresponded to 1.10 mmol/L/min × 10−3 and 59.1 pmol/L/min × 10−3. 1, 2, 3Significantly different from corresponding subgroups (p <0.05).
Figure 4.
Relation between BMI and visceral AT accumulation in men (straight line) and women (dashed line).
Discussion
Results of the present investigation confirm that overweight/obese participants with low CRF levels are characterized by an increased prevalence of low-grade inflammation compared with participants with high CRF. However, our results show for the first time that the relation between obesity and inflammation is attributable to a very large extent to visceral AT accumulation rather than to excess body weight per se. In addition, the state of insulin resistance associated with visceral adiposity appeared to represent a potential link between low CRF levels and low-grade inflammation. Hence, we propose that unfit subjects are characterized by an inflammatory state because of their increased visceral AT accumulation and associated insulin-resistance state.
Results from our laboratory have previously shown that, despite having the same BMI (∼25 kg/m2), men with low CRF had approximately 30% more visceral AT than men with high CRF.9 Moreover, men with low CRF had significantly less fat-free mass. Wong et al10 also reported a relation between visceral AT and CRF that was independent of BMI; because BMI captures fat and fat-free mass, these observations suggest that measurement of BMI is not the optimal clinical tool to evaluate the risk associated with high visceral AT accumulation or low CRF. Moreover, our results indicate that at any BMI level subjects might have very different levels of visceral AT, suggesting that BMI does not necessarily “capture” the amount of atherogenic fat accumulation.
In the Women's Health Study, the relation among physical inactivity, inflammation, and CVD risk was also studied.11 It was found that 59% of the CVD risk associated with physical inactivity was attributable to underlying risk factors including BMI. In that study, increased inflammatory marker concentrations such as CRP, fibrinogen, and intracellular adhesion molecule-1 accounted for 33% of the increased CVD risk associated with physical inactivity. These findings clearly reinforce the role of low-grade inflammation as an important consequence of physical inactivity and the influence of the latter on other CVD risks. However, considering the association between visceral adiposity and inflammation reported in the present study, it might be reasonable to suggest that the increased CVD/diabetes risk associated with physical inactivity could be attributable to increased levels of visceral AT associated with physical inactivity. In a sample of 84 healthy men and women, Fischer et al12 studied the same inflammatory markers as in the present study in addition to BMI and leisure-time physical activity. Similar to what we have observed in the present study, they found that plasma CRP and interleukin-6 levels were higher and adiponectin levels lower in obese and inactive participants. They also found that active and obese participants were characterized by the highest circulating tumor necrosis factor-α levels. In men of the Aerobics Center Longitudinal Study, Church et al3 reported that fibrinogen, white blood cell count, uric acid, and metabolic syndrome score were the highest in men characterized by low CRF and obesity defined by BMI. However, body fat distribution was not assessed in these studies. We have also recently shown in a prospective, population-based study that sedentary subjects with abdominal obesity have higher plasma levels of several inflammatory markers that increase CVD risk.13
In the present study, men and women were free from CVD and diabetes. Moreover, participants were considered middle-aged and very few had high plasma CRP levels, even in participants with an increased visceral AT accumulation and poor CRF. Because inflammatory markers are “emerging” CVD risk markers, studies on tracking of inflammatory markers are lacking. Therefore, we believe that the results of the present study should be validated in prospective studies with hard CVD endpoints and in older populations of various ethnicities. Moreover, there is evidence supporting a role for acute exercise on plasma levels of certain cytokines such as interleukin-6, which might be independent of adiposity.14 However, in the present study, blood samples were performed in subjects who did not exercise for ≥48 hours before blood collection. We assessed physical fitness by a submaximal test rather than by direct measurement of oxygen consumption during a maximal test. However, it has been shown that this submaximal test could well discriminate very fit from nonfit subjects.15 Submaximal test results are therefore well correlated with maximum oxygen consumption and have the advantage of not requiring maximal effort to classify subjects by fitness level at lower cost and greater safety than maximum oxygen consumption. Nevertheless, a possible limitation of this study is the use of an arbitrary inflammation score based on cytokines that are derived from or closely associated with visceral AT. However, we believe that a cumulative marker of inflammation based on several markers might better assess true individual differences in overall inflammation than inflammatory markers studied individually. We observed that fitness did not correlate with any of the individual inflammatory markers in men, whereas it was negatively associated with CRP and interleukin-6 in women. Further studies are warranted to explain the basis of this gender-specific association.Supplementary Fig. 1a and 1b
Acknowledgment
The authors express their gratitude to the subjects for their excellent collaboration and to the staff of CHUL. They especially thank Guy Fournier, BSc, and Lucie Allard, BSc, of the Hôpital Laval Research Centre; Germain Thériault, MD, of Université Laval; and Claude Leblanc, MSc, of the Physical Activity Sciences Laboratory, for their help in collection of data and for their contribution to the study.
Supplementary data
Supplementary Fig. 1a.
Supplementary Fig. 1b.
References
- . Influences of cardiorespiratory fitness and other precursors on cardiovascular disease and all-cause mortality in men and women. JAMA. 1996;276:205–210
- . The association between cardiorespiratory fitness and impaired fasting glucose and type 2 diabetes mellitus in men. Ann Intern Med. 1999;130:89–96
- . Relative associations of fitness and fatness to fibrinogen, white blood cell count, uric acid and metabolic syndrome. Int J Obes Relat Metab Disord. 2002;26:805–813
- . Evaluation and Regulation of Body Build and Composition. Englewood Cliffs, NJ: Prentice-Hall. 1974;20–37
- . The gross composition of the body. Adv Biol Med Phys. 1956;4:239–280
- . The distribution and chemical composition of ultracentrifugally separated lipoproteins in human serum. J Clin Invest. 1955;34:1345–1353
- . Quantitative estimation of proteins by electrophoresis in agarose gel containing antibodies. Anal Biochem. 1966;15:45–52
- . [Determination of plasma glucose by hexokinase-glucose-6-phosphate dehydrogenase method]. Schweiz Med Wochenschr. 1971;101:615–618
- . Visceral adipose tissue accumulation, cardiorespiratory fitness, and features of the metabolic syndrome. Arch Intern Med. 2007;167:1518–1525
- . Cardiorespiratory fitness is associated with lower abdominal fat independent of body mass index. Med Sci Sports Exerc. 2004;36:286–291
- . Physical activity and reduced risk of cardiovascular events: potential mediating mechanisms. Circulation. 2007;116:2110–2118
- . Plasma levels of interleukin-6 and C-reactive protein are associated with physical inactivity independent of obesity. Scand J Med Sci Sports. 2007;17:580–587
- . Inflammatory biomarkers, physical activity, waist circumference, and risk of future coronary heart disease in healthy men and women. Eur Heart J. 2009;in press
- . The biological roles of exercise-induced cytokines: IL-6, IL-8, and IL-15. Appl Physiol Nutr Metab. 2007;32:833–839
- . Principal components of fitness: relation to physical activity and lifestyle. Can J Appl Physiol. 1994;19:200–214
PII: S0002-9149(09)00737-1
doi:10.1016/j.amjcard.2009.03.027
© 2009 Elsevier Inc. All rights reserved.
