EFFECT OF TRUNCAL ADIPOSITY ON PLASMA LIPID AND LIPOPROTEIN CONCENTRATIONS

EFFECT OF TRUNCAL ADIPOSITY ON PLASMA LIPID AND LIPOPROTEIN CONCENTRATIONS C.M. NIEDERAUER, T.L. BINKLEY, B.L. SPECKER Ethel Austin Martin Program in Human Nutrition, South Dakota State University, Brookings, South Dakota. Address reprint requests and corresponding author: Bonny L. Specker, PhD, EA Martin Program in Human Nutrition, EAM Bldg, Box 2204, South Dakota State University, Brookings, SD, 57007, Telephone: 605-688-4645, Fax: 605-688-4220, Email: bonny.specker@sdstate.edu Abstract: Objective: The purpose of this study was to determine the effect of body fat distribution on plasma lipid and lipoprotein concentrations in healthy individuals. Methods: Subjects were 290 women and 223 men aged 20-66 years old. Lipid panels were obtained and truncal adiposity measurements were determined by total body dual energy x-ray absorptiometry (DXA). Physical activity was measured using a modification of the Seven Day Physical Activity Recall Questionnaire and dietary intake was collected by 24-hour recall. General linear models were used with covariates included. Results: Percent of total body fat in the trunk had the strongest correlation to the android/gynoid regional fat mass ratio determined by regional DXA analyses in both women (r2=0.97, p<0.001) and men (r2=0.93, p<0.001). Total cholesterol (r=0.134, p< 0.05), LDL (r=0.140, p<0.05), HDL (r= -0.181, p<0.001), total cholesterol to HDL ratio (r=0.240, p<0.001) and triglyceride (r=0.300, p<0.001) concentrations had the strongest correlation to percent of total body fat in the trunk in women. Total cholesterol (r= 0.151, p<0.05), total cholesterol to HDL ratio (r=0.216, p<0.001), and triglyceride (r=0.227, p<0.001) concentrations had the strongest correlation to percent of total body fat located in the trunk among men. Conclusion: Lipid profiles, consistent with increased cardiovascular risk, were associated with percent of total body fat in the trunk independent of lifestyle factors. Key words: Ttruncal adiposity, lipid, lipoprotein, cholesterol, DXA. Introduction It is well established that obesity plays a role in coronary artery disease (CAD) and it has been hypothesized that the relationship of CAD to adiposity and fat distribution may be partly mediated through the effects of adiposity and fat distribution on the serum lipid profile (1). Most cardiovascular diseases are caused by atherosclerosis, or plaque accumulation in the arteries. Atherosclerotic plaques contain large amounts of cholesterol, and this cholesterol is derived primarily from blood lipoproteins (2) Serum concentrations of lipids and lipoproteins may be influenced by the amount and distribution of body fat, but these relationships are poorly understood. Many of the previous studies on the relationship between lipid metabolism and body fat have relied on indirect, anthropometric methods for determination of adiposity and fat distribution, or have not controlled for the influence of activity or fat intake on lipid and lipoprotein concentrations. This research supports the influence of demographic, environmental factors, and obesity on dyslipidemias. Our hypothesis was that lipids and lipoprotein concentrations are influenced by truncal adiposity independent of lifestyle factors. This study has the unique advantage of utilizing data from a large, randomly selected population of healthy women and men, over a wide age range, 20-66 years, with diverse lifestyles. Also, it has the advantage of accessing adiposity using DXA, rather than other more indirect anthropometric indicators. Information on activity levels and dietary intakes are also included in this study. Knowledge of the strength of association between lipid and lipoprotein concentrations and body fat distribution independent of activity levels and dietary fat intake will contribute to public health efforts to decrease the risks of cardiovascular disease by reducing truncal adiposity through efforts to encourage lifestyle changes promoting the loss of adiposity in this region. Methods Subjects Subjects were 513 participants in an ongoing study on the effect of lifestyle factors on bone and body composition for whom we had fasting blood samples (3). Of the 513 subjects, 52 were classified as non-rural and 461 were classified as rural. Participants were classified as non-rural if they never lived on a working farm or ranch. Individuals were classified as rural if they spent the majority of their life (>75%) living on a working farm or ranch and not working off the farm for more than 20 h/wk. The majority of these individuals were recruited by calling every 10th telephone number in local phone books. Within the rural population, two populations were studied, Hutterites (N=398) and non-hutterites (N=115). The Hutterite Brethren are a communal group of people residing on selfsufficient colonies. Each colony relies on highly efficient production of grain and livestock using modern equipment and technology to remain self-sufficient and isolated. Colony members dine in a common hall with a wide variety of food Accepted for publication February 13, 2005 154

THE JOURNAL OF NUTRITION, HEALTH & AGING choices. Approximately half of the U.S. Hutterite colonies are located in eastern South Dakota. The rural non-hutterites also rely on agriculture for a living, but reside on individually owned farms or ranches. These individuals were recruited by calling telephone numbers of individuals owning land zoned agriculture. All rural and non-rural non-hutterite subjects lived in counties with participating Hutterite colonies. Individuals with Type 1 diabetes, parathyroid disease, current treatment for cancer, or chronic use (>6 months) of immunosuppressants, anticonvulsants, or oral steroids were excluded from the original study. We also excluded women who were lactating (breastfeeding at their baseline visit), pregnant (women who were pregnant at baseline or within the previous 6 months if the infant was not breast-fed), or weaning (discontinuation of breastfeeding within the previous 12 months); individuals who were taking cholesterol-lowering medicine; or individuals for whom we did not have a fasting blood sample. Since estrogen status is a potential covariate for women, we categorized them as either estrogen replete (postmenopausal and receiving hormone replacement therapy (HRT); pre-menopausal) (N=253), or deplete (post-menopausal and not receiving HRT) (N=37). There were 5 women who were having a menstrual cycle but considered themselves to be menopausal. They were included in the estrogen replete group. The study was approved by the South Dakota State University Human Subjects Review Board and all participants provided written consent. Procedures Plasma profiles including total cholesterol, low-density lipoprotein cholesterol (LDL), high-density lipoprotein cholesterol (HDL), triglyceride, and the ratio of total cholesterol to HDL (total cholesterol/hdl) were measured on fasting plasma samples by the Clinical Laboratories of the Midwest (Sioux Valley Hospital, Sioux Falls, SD). The Friedwald formula was used to calculate LDL. Coefficients of variation are 1.0 % for total cholesterol, HDL and triglycerides. Physical activity was measured using a modification of the Seven Day Physical Activity Recall Questionnaire (4). This questionnaire was completed by an interviewer and included the amount of time in vigorous, moderate, light, sitting and sleep activity per weekday and weekend during the previous week. Percent time in moderate plus vigorous activity and percent time in vigorous activity were used as two of our primary activity variables. Average number of miles walked per day and flights of stairs climbed per day also were included. Twenty-four hour dietary recall interviews, including information on vitamin and mineral supplement use, were obtained to estimate dietary intakes of macro- and micronutrients. Study personnel conducted the dietary interviews using food models. Hutterite colony cooks assisted in providing recipes for specific Hutterite foods. Nutrient intakes were determined using the Nutritionist Pro software (First Data Bank, San Bruno, CA). Truncal adiposity measurements were determined using a Hologic QDR4500A (Waltham, MA) DXA. Total body percent fat (fat mass/total mass), trunk percent fat (trunk fat mass/total trunk mass), and percent of total body fat in the trunk (trunk fat mass/total body fat mass) were calculated to determine their relative contribution to lipid and lipoprotein results. Coefficients of variation as determined by the manufacturer for total body percent fat is 1.3%. The trunk region consists of the tissue area bordered by a horizontal line below the chin, vertical borders lateral to the ribs, and oblique lines passing through the femoral neck. One male individual (149.2 kg) was excluded from this study due to extreme obesity and inability of DXA to properly determine body fat measurement. A manual analysis of regional fat distribution also was completed on a subset of 60 scans using the Hologic Discovery Software (QDR System Software Version 12.01). The method used was similar to that of Walton and coworkers, which was previously found to be analogous to the waist-to-hip ratio (1). Two regions of interest were defined: android region and gynoid region. Both regions were of identical height but the width was adjusted to include all the soft tissue for each subject. The height of the regions was defined as two-thirds of the distance between the top of the iliac crest and knee joint (Figure 1). The android region was placed on the iliac crest and the gynoid region placed mid pelvis at the top of the femoral head. The android/gynoid fat mass ratio was calculated as fat mass in the android region divided by fat mass in the gynoid region. In order to compare this ratio with other adiposity measures, men and women were divided separately into tertiles to determine low, medium and high total body percent fat of each subset. Ten men and 10 women were randomly selected from each tertile of total body fat percentage and their scans were analyzed to determine their android/gynoid fat mass ratio. Height without shoes was measured in duplicate to the nearest 0.5 cm using a portable SECA stadiometer (Hamburg, Germany). If measurements differed by more than 0.5 cm the measurement was repeated. Weight with light clothing was measured to the nearest 0.1 kg using a SECA scale (Model 770). Body mass index (BMI) was calculated as weight (kg)/height (m)2. Statistical Analysis Statistical analyses were carried out using the JMP software package (version 4.0, SAS Institute, Inc., Cary, NC). Triglyceride concentrations were log-transformed for normalization. Statistical analyses were stratified by sex. Forward stepwise regression analyses were performed to identify significant covariates to include in the models predicting plasma lipid and lipoprotein concentrations. Covariates initially considered for each lipid and lipoprotein concentration included age; height; weight; group (rural, nonrural, Hutterite); percentage of total kilocalories consumed from protein, fat, carbohydrates; total kilocalories, fat, saturated 155

fat, polyunsaturated fat, cholesterol, fiber, alcohol consumed; kilocalories, protein, carbohydrates, and fat grams consumed per kilogram of body weight; percentage of day spent in vigorous activity and vigorous plus moderate activity; average number of miles walked per day; flights of stairs climbed per day; parity; and estrogen status (replete or deplete). Only covariates significant at p< 0.05 were allowed to remain in the model. Measures of adiposity were then added to this basic model individually to determine their significance in predicting plasma lipid and lipoprotein concentrations. A separate basic model was determined for each lipoprotein measurement. Tukey s HSD was used to identify significant differences among group means. Figure 1 DXA image showing defined regions of interest in calculating the android/gynoid fat mass ratio. Results Study group characteristics are shown in Table 1. Fasting blood samples were obtained on 290 women and 223 men. Women were consuming less of all dietary factors than men except for grams of carbohydrates (both as g/kg body weight and percent of total kilocalories). Women also had less time in moderate plus vigorous activity, vigorous activity, and miles walked per day compared to men. Plasma LDL, triglyceride, and total cholesterol/hdl concentrations were lower, and HDL concentrations were higher, in women compared to men. No population differences were observed in the lipid and lipoprotein measurements. Table 1 Demographic and Descriptive Data (means + SD) Women Men P Value 156 Number (Rural/ Non-rural/ 23/23/244 40/29/154 Hutterite) Age (years) 40 + 13 40 + 13 0.79 Dietary Intake Kcal (g) 1782 + 538 2416 + 819 <0.001 % Protein 17 + 4 19 + 6 <0.001 % Fat 38 + 8 38 + 9 0.70 % CHO 44 + 9 41 + 10 <0.001 Total fat (g) 76 + 30 104 + 43 <0.001 Saturated FA (g) 29 + 11 38 + 17 <0.001 Monounsaturated FA (g) 25 + 13 34 + 17 <0.001 Polyunsaturated FA (g) 11 + 7 14 + 9 0.001 Cholesterol (mg) 390 + 267 509 + 358 <0.001 Fiber (g) 13 + 6 15 + 7 0.001 Alcohol (g) 1.5 + 5.9 7.7 + 17.6 <0.001 Kcal/kg 26 + 10 28 + 11 0.04 Pro/kg 1.1 + 0.4 1.3 + 0.5 <0.001 Fat/kg 1.1 + 0.5 1.2 + 0.5 0.06 CHO/kg 2.9 + 1.2 2.9 + 1.4 0.97 Physical Activity % moderate + vigorous 20 + 11 24 + 11 <0.001 % vigorous 3.1 + 5.6 4.6 + 6.1 0.005 Average miles walked/day 2.0 + 1.3 2.5 + 1.7 <0.001 Flights stairs/day 10 + 7 9 + 9 0.34 Anthropometrics & Body Composition Height (cm) 162.3 + 5.7 177.1 + 8.1 <0.001 Weight (kg) 73.4 + 15.0 90.5 + 14.8 <0.001 BMI (kg/m2) 28 + 5.9 29 + 5 0.05 Total body % fat 35 + 7 23 + 6 <0.001 Trunk % fat 32 + 8 24 + 8 <0.001 % of total in trunk 42 + 6 50 + 7 <0.001 Android/gynoid fat mass 0.9 + 0.4 1.1 + 0.4 0.009 Estrogen Status Replete/Deplete 253/37 Parity 3.0 + 2.9 Mean Lipid Profile (mg/dl) Total cholesterol 190 + 40 195 + 44 0.19 LDL cholesterol 111 + 35 118 + 38 0.03 HDL cholesterol 54 + 14 43 + 11 <0.001 Triglyceride* 109 (63, 187) 142 (72, 279) <0.001 Total/HDL Ratio 3.7 + 1.2 4.8 + 1.6 <0.001 * Geometric mean (-1 SD, + 1 SD); FA = fatty acid

THE JOURNAL OF NUTRITION, HEALTH & AGING Findings Among Women In a bivariate analyses, trunk percent fat was strongly correlated with total body percent fat, and percent of total body fat in the trunk was strongly correlated with the android/gynoid fat mass ratio (Table 2). All measures of adiposity correlated with BMI in women. Significant positive correlations were found among all lipid and lipoprotein concentrations and body fat measures except LDL and HDL concentrations and the android/gynoid fat mass ratio. HDL concentrations were inversely associated with all adiposity measures except total body percent fat. Findings Among Men Trunk percent fat was strongly correlated with total body percent fat, and percent of total body fat in the trunk was strongly correlated with the android/gynoid fat mass ratio (Table 2). All measures of adiposity correlated with BMI. All lipid and lipoprotein concentrations had a significant correlation with all adiposity measures except for LDL concentrations, which were correlated only with percent of total body fat in the trunk and the android/gynoid fat mass ratio. HDL concentrations were inversely correlated with all adiposity measures in men. Table 2 Simple correlation coefficients for fasting lipids and lipoprotein concentrations (mg/dl) and measures of adiposity for women (top number) and men (bottom number). Total Body % Fat Trunk % Fat % of Total Body Fat Android/ BMI (kg/m2) in Trunk Gynoid Fat Mass Total Body % Fat 1.00 - - - - Trunk % Fat 0.95 ** 0.98 ** 1.00 - - - % of Total Body Fat in Trunk 0.48 ** 0.71 ** 0.63 ** 0.78 ** 1.00 - - Android/Gynoid Fat Mass 0.47 * 0.72 ** 0.97 ** 0.72 ** 0.88 ** 0.93 ** 1.00 - BMI (kg/m2) 0.83 ** 0.84 ** 0.52 ** 0.70 ** 0.80 ** 0.84 ** 0.66 ** 0.70 ** 1.00 Total Cholesterol 0.34 ** 0.36 ** 0.31 ** 0.36 * 0.24 ** 0.18 * 0.21 * 0.28 ** 0.45 * 0.21 * LDL 0.23 ** 0.26 ** 0.25 ** 0.14 0.17 * 0.08 0.12 0.17 * 0.42 * 0.10 HDL -0.09-0.15 * -0.22 ** -0.35-0.19 ** -0.38 ** -0.40 ** -0.32 ** -0.40 * -0.39 ** Cholesterol/HDL 0.32 ** 0.39 ** 0.43 ** 0.61 ** 0.34 ** 0.38 ** 0.42 ** 0.42 ** 0.59 ** 0.42 ** Triglycerides 0.50 ** 0.58 ** 0.55 ** 0.77 ** 0.50 ** 0.50 ** 0.54 ** 0.50 ** 0.64 ** 0.49 ** *p<0.05 ** p<0.001 The basic models for the relationships between plasma lipid and lipoprotein concentrations and significant covariates are summarized in Table 3. Also provided are the partial correlation coefficients of each of the adiposity measures when added individually to the regression models containing the covariates listed. Total plasma cholesterol, LDL concentrations, and the total cholesterol/hdl ratio were associated with trunk percent fat and percent of total body fat in the trunk. There was an inverse relationship between HDL concentrations and percent of total body fat in the trunk. Triglyceride concentrations were correlated with all adiposity measures except BMI. A scatterplot of the relationship between the total cholesterol/hdl ratio and percent of total body fat in the trunk is shown in Figure 2. Tables 4 summarizes the basic regression models for the relationships between plasma lipid and lipoprotein concentrations and significant covariates, as well as the partial correlation coefficients of each of the adiposity measures when added individually to the basic regression models. Total plasma cholesterol was only associated with percent of total body fat in the trunk and BMI. LDL concentrations were not associated with any adiposity measures among men, while HDL concentrations were inversely associated with only trunk percent fat. The total cholesterol/hdl ratio was associated with trunk percent fat and percent of total body fat in the trunk. As in women, triglyceride concentrations were associated with all of the body fat measures except BMI. A scatterplot of the total cholesterol/hdl ratio and percent of total body fat in the trunk is shown in Figure 3. 157

Table 3 Relationships among plasma lipids and lipoprotein concentrations and measures of adiposity in women after controlling for covariates. Symbols are the direction of the relationship between the lipid or lipoprotein concentrations and the specific covariate. Women Cholesterol LDL HDL Cholesterol/ Triglycerides (mg/dl) (mg/dl) (mg/dl) HDL Ratio (mg/dl) Basic Model Age (+) Age (+) Age (+) Height (-) Age (+) Fat Intake/kg (-) Fat Intake/kg(-) Weight (-) Fat Intake (+) Weight (+) Sat Fat Intake (+) Sat Fat Intake (+) Height (+) Fat Intake/kg) (-) Height (-) PUFA Intake (+) Parity (-) PUFA Intake (+) Sat Fat Intake (+) Sat Fat Intake (+) Parity (-) Cholesterol Intake (-) CHO/kg (-) % Kcals Fat (-) % Kcals Protein (+) % Kcals CHO (+) % Time Mod+Vig (-) % Time Vigorous (-) Estrogen Deplete (+) Stair flights/day (+) Basic Model r 2 0.302 0.171 0.189 0.213 0.348 Body composition variables added individually to the basic model described above. Values below are the partial correlation coefficient (r) and (r 2 ) when each adiposity measure was individually added to basic model: Total Body % Fat 0.059 (0.003) 0.073 (0.005) 0.014 (0.0002) 0.043 (0.002) 0.133 (0.018) * Trunk % Fat 0.108 (0.012) * 0.116 (0.013) * - 0.090 (0.008) 0.150 (0.023) * 0.266 (0.071) ** % of Total Body Fat 0.134 (0.018) * 0.140 (0.020) * - 0.181 (0.033) ** 0.240 (0.058) ** 0.300 (0.090) ** in Trunk BMI (kg/m 2 ) - 0.039 (0.002) 0.010 (0.000) 0.046 (0.002) 0.002 (0.000) 0.008 (0.000) PUFA = polyunsatured fatty acids; Sat Fat = saturated fat intake; CHO = carbohydrate; * p<0.05 ** p<0.001 Table 4 Relationships among plasma lipids and lipoprotein concentrations and measures of adiposity in men after controlling for covariates. Symbols are the direction of the relationship between the lipid or lipoprotein concentrations and the specific covariate. Women Cholesterol LDL HDL Cholesterol/ Triglycerides (mg/dl) (mg/dl) (mg/dl) HDL Ratio (mg/dl) Basic Model Age (+) Age (+) Weight (-) Weight (+) Weight (+) PUFA Intake (+) Height (+) Height (-) Height (-) Miles Walked/d (+) MUFA Intake (+) % Kcals Fat (-) % Kcals Fat (+) % Kcals CHO (-) % Kcals CHO (-) % Kcals Protein (-) Alcohol Intake (-) Basic Model r 2 0.063 0.078 0.186 0.251 0.333 Body composition variables added individually to the basic model described above. Values below are the partial correlation coefficient (r) and (r 2 ) when each adiposity measure was individually added to basic model: Total Body % Fat 0.080 (0.006) - 0.010 (0.000) - 0.105 (0.011) 0.057 (0.003) 0.143 (0.021) * Trunk % Fat 0.010 (0.102) 0.010 (0.000) - 0.133 * (0.018) 0.122 (0.015) * 0.193 (0.037) ** % of Total Body Fat in Trunk 0.151 (0.023) * 0.048 (0.002) - 0.108 (0.012) 0.216 (0.046) ** 0.227 (0.052) ** BMI (kg/m 2 ) 0.139 (0.019) * 0.049 (0.002) - 0.068 (0.068) - 0.040 (0.002) - 0.101 (0.010) PUFA = polyunsatured fatty acids; MUFA = monounsaturated fatty acids; Sat Fat = saturated fat intake; CHO = carbohydrate; * p<0.05; ** p<0.001 158

THE JOURNAL OF NUTRITION, HEALTH & AGING Figure 2 A scatterplot of the correlation between the total cholesterol/hdl ratio and percent of total body fat in the trunk in women (r= 0.240, p<0.001). Figure 3 A scatterplot of the correlation between the ratio of the total cholesterol/hdl ratio and percent of total body fat in the trunk in men (r= 0.216, p<0.001). Discussion Our results are similar to previous reports of an association between lipid concentrations and centrally located body fat (1, 5-7). The positive correlations between measures of central body fat and cholesterol, LDL, and triglyceride concentrations, and a negative association with HDL concentrations, among our study population of primarily pre-menopausal women are consistent with those previously reported in postmenopausal women (5-7). Walton and coworkers did not find a significant correlation between HDL or LDL concentrations and central body fat among men (1). However, they did find a significant inverse association between the HDL2 subfraction and central body fat. We found an inverse relationship between HDL and trunk percent fat among men. Additionally, we found an inverse correlation between the total cholesterol/hdl ratio and both trunk percent fat and percent of total body fat in the trunk. In both women and men, lipids and lipoproteins were more correlated to percent trunk fat and percent of total fat in the trunk than to the total body percent fat or BMI. It has been suggested that visceral fat has a higher turnover rate than subcutaneous fat and therefore may have a greater influence on the plasma lipid profile (6). Although DXA cannot differentiate between visceral and subcutaneous fat (8), it has been shown that specific regions on the DXA can indirectly estimate visceral fat content in adult women (9). Our study has the unique advantage of utilizing manual analysis of specific android and gynoid regions on total body DXA scans. We found that the android/gynoid fat mass ratio was highly correlated to the percent of total body fat located in the trunk in both women and men. This supports the use of percent of total body fat located in the trunk as a more cost-effective and timeeffective means to determine central body fat distribution than manually defining android and gynoid regions. Determination of the android/gynoid fat mass ratio is not routinely calculated from total body DXA scans and requires the DXA operator to re-analyze DXA scans. BMI, a routinely calculated measurement, may be inadequate in estimating risk of altered lipid and lipoprotein concentrations. Although BMI was statistically correlated with all measures of adiposity, the correlation was less between BMI and percent of total body fat in the trunk region than for the other adiposity measures. In addition, BMI was not significantly associated with the majority of the lipid measurements that were obtained in this study. Although several activity and dietary factors were included as significant covariates in models the contribution of those factors on lipid concentrations is less than that of body fat distribution as shown in Tables 3 and 4. In women more activity and dietary factors were found to be significant covariates while the lipid concentrations were also more strongly associated with central body fat distribution than in men. There are several limitations to this study. This analysis used cross-sectional data. Whether changes in lipid profiles indicative of cardiovascular risk are associated with changes in truncal adiposity over time would need to be determined. Information on cigarette smoking was not originally included in our questionnaires due to absence of smoking among members of the Hutterite colonies, and therefore was not included as a covariate. The strengths of this study include utilization of data from a large, randomly ascertained population of healthy men and women, over a wide age range, 20-66 years, with diverse lifestyles. We also had the advantage of accessing adiposity using DXA rather than more indirect anthropometric indicators. In addition, we were able to control for activity levels and 159

dietary intakes, both of which are important confounders. Our results provide rationale to further assess adiposity beyond BMI in order to determine risk for altered lipids and lipoprotein concentrations. Decreases in truncal adiposity may lead to decreased atherosclerotic risk mediated by lipids and lipoproteins in relatively healthy men and women. References 1. Walton C., Lees B., Crook D., Worthington M., Godsland I.F., Stevenson J.C., Body fat distribution, rather than overall adiposity influences serum lipids and lipoproteins in healthy men independently of age. Am. J. Med.,1995, 99:459-465. 2. Steinberg D., Parthasarathy S., Carey T.E., Beyond Cholesterol: modifications for low-density lipoprotein that increase its atherogenicity. N. Engl. J. Med.,1989, 320:915-924. 3. Specker B.L., Binkley T.L., Fahrenwald N. Rural vs. non-rural differences in BMC, volumetric BMD, and bone size: a population-based cross-sectional study. Bone, 2004, 35:1389-1398. 4. Paffenbarger R.S. Jr., Wing A.L., Hyde R.T., Physical activity as an index of heart attack risk in college alumni, Am. J. Epidemiol.,1978, 108:161-175. 5. Haarbo J., Hassager C., Schlemmer A., Christiansen C., Influence of smoking, body fat distribution, and alcohol consumption on serum lipids, lipoproteins, and apolipoproteins in early postmenopausal women, Atherosclerosis, 1990, 84:239-244. 6. Despres J.P., Moorjani S., Lupien P.J., Tremblay A., Nadeau A., Bouchard C., Regional distribution of body fat, plasma lipoproteins, and cardiovascular disease, Arteriosclerosis, 1990,10: 497-511. 7. Haarbo J., Hassager C., Riss B.J., Christiansen C., Relation of body fat distribution to serum lipids and lipoproteins in elderly postmenopausal women, Atherosclerosis 1989, 80: 57-62. 8. Taylor R., Cannan R., Gold E., Lewis-Berned N., Goulding A., Regional body fat distribution in New Zealand girls aged 4-16 years: a cross-sectional study by dualenergy x-ray absorptiometry, Int. J. Obes., 1996, 20:763-67. 9. Treuth M.S., Hunter G.R., Kekes-Szabo T., Estimating intraabdominal adipose tissue in women by dual energy X-ray absorptiometry, Am. J. Clin. Nutr., 1995, 62: 527-534. 160