Evaluating the correlation and prediction of trunk fat mass with five anthropometric indices in Chinese females aged 20–40years

Methods and results A sample of 850 China females aged 20–40 years were recruited and divided into four age groups with a 5-year range in each group. Five anthropometric indices were measured or calculated. FM trunk in kg was measured using a dual-energy X-ray absorptiometry scanner. Principal component analysis (PCA) and multiple regression analysis were performed to develop prediction equations. There was an increasing trend of FM trunk and five anthropometric indices in successively older age groups. Four formed principal components (PCs) interpreted over 99% of the total variation of five relative anthropometric indices in all age groups. Regression analyses showed that four PCs combined explained a greater variance ( R 2 = 45.2–81.6%) in FM trunk than did each of the five indices alone ( R 2 = 2.4–72.2%). Conclusions Our results suggested that there is an increasing trend of FM trunk and five anthropometric indices with aging; that age obviously influences the relationship of FM trunk and the anthropometric indices studied; and that the accuracy of predicting the FM trunk using five anthropometric indices combined is greater than using the five indices alone. Keywords Anthropometric index Obesity Trunk fat mass Percent trunk fat mass Principal component analysis Introduction Obesity has become a worldwide problem [1–3] . In China, over 40 million people suffer from obesity [4] . Reported studies show that excess trunk fat is associated with an increased risk of metabolic diseases [5–9] . In addition, there is specific distribution of body fat mass for the expenditure of energy and body building among various ethnic populations [10–13] . Therefore, the early measurement of trunk fat mass (FM trunk , or, so-called, central fat mass) has the potential importance of evaluating adipose distribution in the trunk area, and to predict future health risk in the Chinese population. There are three major techniques, computed tomography (CT); magnetic resonance imaging (MRI); and dual-energy X-ray absorptiometry (DEXA), which are used to measure FM trunk accurately. However, these techniques depend on complex techniques and are quite expensive. The indirect evaluation of FM trunk by anthropometric indices is simpler and cheaper than that by DEXA, CT, and MRI, and is especially suitable for routine clinical use in developing countries for large-scale detection of trunk fat mass. A number of studies in Caucasians have used several anthropometric indices, such as body mass index (BMI), waist circumference (WC), hip circumference (HC), waist-to-hip ratio (WHR), conicity index (CI), as predictors of total fat mass and FM trunk [14–18] and the results were inconsistent regarding the accuracy of prediction, due to the fact that the studies generally used only one, or not more than three, anthropometric indices to predict fat mass. To our knowledge, there are no data simultaneously using the above five anthropometric indices plus a DEXA scan, which is quick and safe and provides measurements of body composition (including fat mass, lean mass, and bone mass) with excellent precision compared to other methods [19,20] . Therefore, the objectives of the present study were: 1) to investigate the correlation between FM trunk and anthropometric indices; 2) to evaluate the accuracy of FM trunk predicted by the above five anthropometric indices; and 3) to develop four equations of predicting FM trunk . Methods Subjects The project was approved by the Research Administration Departments of Hunan Normal University. The female subjects were from a large population that was recruited for genetic studies aimed at searching for the gene underlying peak bone mass variation in the Chinese population. Our sample size was initially determined by comparing similar studies. Three patterns of recruitment (internet, poster and interview) were used. Firstly, we placed an advertisement about nutrition and bone health on the homepage of Hunan Normal University. Secondly, we put posters around Hunan Normal University inviting volunteers to our laboratory for a free measurement. Thirdly, we randomly interviewed subjects in their homes. After the subject had signed the informed consent document, a questionnaire was administered to obtain information about her age, medical history, family history, physical activity, alcohol use, dietary habits and smoking history under the direction of a clinician. We also adopted the exclusion criteria detailed by Deng et al. [21] to screen and recruit “healthy” subjects. Briefly, subjects with chronic diseases and conditions that may affect bone mass or bone metabolism, were excluded from the study. We did not exclude subjects with an extremely low or high BMI. Finally, a total of 850 healthy Han Chinese females aged 20–40 years were recruited from Changsha city, the PR of China, and its surrounding region. In this sample, the number of subjects classified by BMI was 185 (BMI < 18.5), 644 (BMI = 18.5–24.9), 19 (BMI = 25–29.9), 2 (BMI = 30–35) respectively. Measurements Height and weight were measured using standard altimeters and scales (made in China). BMI = (kg/m 2 ) was calculated as weight (kg) divided by square height (m 2 ). WC and HC were measured with an anthropometric tape over light clothing, with measurement of waist and hip circumference at the minimum circumference between the iliac crest and the rib cage, and at the maximum protuberance of the buttocks. All the anthropometric measurements were taken in our laboratory. CI was calculated using the following formula: CI = WC/[0.109 × square root of (weight/height)] [16] , where WC and height were measured in meters and weight was measured in kg. FM trunk and trunk lean mass in kg was measured in the conventional DEXA trunk region using a Hologic QDR 2000 DEXA scanner (Hologic Corp., Waltham, MA). The trunk fat mass percentage (%FM trunk ) was calculated as [FM trunk /(FM trunk + trunk lean mass)]. The coefficient of variation of FM trunk , obtained from 30 individuals who were measured twice, of the DEXA measurements was 0.99%. Statistical analysis All statistical analyses were performed with the SAS package (SAS Institute Inc., Cary, NC, USA). To investigate the effects of age, we divided the samples into four age groups with a 5-year (5-yr) range in each group (20–24-yr 25–29-yr, 30–34-yr and 35–39-yr). Pearson's correlation coefficients were used to investigate the linear correlation of FM trunk , %FM trunk with five anthropometric indices. Because BMI, WC, HC, WHR and CI were five highly related anthropometric indices, to avoid the disturbance of colinearity when simultaneously modeling the five indices in multiple regression for predicting FM trunk , a principal component (PC) analysis (PCA) was performed to form 4 PCs accounting for most variations in the five anthropometric indices [22] . Then the PC values were calculated by Eigenvalues of matrix and Eigenvectors, and were then used to estimate the regression coefficients and the proportions of the variance ( R 2 ) of FM trunk predicted by these PCs by multiple regression analysis. Regression analyses were also conducted to determine whether four PCs combined explained a greater variance ( R 2 ) in FM trunk than did BMI, WC, HC, WHR and CI alone and to investigate the correlation efficient of the measured FM trunk and the predicted FM trunk by four PCs or five anthropometric indices alone. According to the relationship of variables to each other in our analysis, we developed four simple equations to predict FM trunk using measured values of five anthropometric indices. Results Age was significantly correlated with FM trunk , %FM trunk , and five anthropometric indices in all the females, but was not significantly associated with them in each 5-yr age group (data not shown). As shown in Table 1 , there was an increasing trend of FM trunk , %FM trunk and five anthropometric indices in the successively older age groups. FM trunk and %FM trunk were significantly correlated with five anthropometric indices, with the correlation coefficients ranging from 0.11 to 0.85 ( P = 0.0001). However, the correlations between FM trunk and five anthropometric indices were generally higher than those between %FM trunk and the indices in each age group ( Table 2 ), therefore our analyses focused on the prediction of FM trunk with five anthropometric indices. Generally, the correlation coefficients between FM trunk and BMI are higher than those between WC, HC, WHR, CI and FM trunk in the same age group (e.g., the correlation coefficients between FM trunk and BMI, WC, HC, WHR and CI in 20–24-yr group were 0.79, 0.74, 0.64, 0.42 and 0.29 respectively). Additionally, there were obvious differences of correlation coefficients among different age groups (e.g., in 25–29-yr group the correlation coefficients were smaller than those in other age groups). As shown in Table 3 , the four PCs interpreted over 99.9% of the total variation of five relative anthropometric indices in the four age groups by PCA, with over 53.7% of the total variation accounted by PC1. Regression analyses ( Table 4 ) showed that four PCs combined explained a greater variance of FM trunk ( R 2 = 45.2–81.6%) than did five indices alone ( R 2 = 2.4–72.2%) (e.g., in the 20–24-yr group, the proportions are 71.4%, 63.1%, 54.8%, 40.7%, 17.9% and 8.4% respectively by the four PCs, BMI, WC, HC, WHR, and CI). Although four PCs combined, or five indices alone, correlated strongly with FM trunk , about 20–50% of the variance of FM trunk remained to be explained. Fig. 1 demonstrates that the correlations of FM trunk and the predicted FM trunk by four PCs combined were higher than those by five indices alone, indicating that using the four PCs predicts more precisely FM trunk than using five indices alone. Finally, we developed four prediction equations of FM trunk according to age groups ( Table 5 ). Discussion Our primary results in the present study show that there is an increasing trend of FM trunk and five anthropometric indices in successively older age groups; that age has remarkable effects on the relationship of FM trunk and the anthropometric indices studied; and that the accuracy of predicting the FM trunk using five anthropometric indices is higher than using BMI, WC, HC, WHR, and CI alone. Because FM trunk is a risk factor of metabolic disease, such as hypertension, hypertensive heart disease, coronary heart disease, diabetes and cardiovascular disease, etc. [23,24] , our predicting equations may be applied to practical evaluation to health risk, especially in undeveloped regions and developing countries. Our results suggested that the predictions of FM trunk using the five indices combined were more accurate than those using them separately. Previous evidence has shown that using the anthropometric indices independently has limits to predict fat mass, which may produce inconsistent results [15,16,25,26] . For example, BMI and WC were usually taken as proxies of overall fat mass and central fat mass, respectively [16,26] . However, in the present study, BMI is a better predictor of central fat mass than WC. The adipose distribution may partially affect the accuracy. Janssen et al. [14] and Ferland et al. [27] reported that in measurements such as BMI, WC and WHR, the relationship between different anthropometric indices and regional adipose tissue in the body was apparent. Therefore, the combination of five anthropometric indices (including HC and CI) can extract useful information. and thus improve the ability of predicting FM trunk . Janssen et al. [14] has found that BMI and WC combined was a better predictor of abdominal fat than either variable alone. Additionally, those indices are highly related and thus cannot simply be molded in multiple regression analysis to predict FM trunk . Hence, a principal component analysis was employed to extract the useful information of these related indices. It is obvious from Fig. 1 that the simple correlation coefficients are higher between the measured FM trunk and the predicted FM trunk by the four combined PCs than those by five anthropometric indices alone. Compared with Caucasians, relatively few data on the relationship between FM trunk and anthropometric indices have been reported in the Chinese population. Until now, only few studies have predicted visceral adipose tissue, abdominal fat mass and body fat mass using anthropometric indices and DEXA data in Caucasians and Latinos [28–32] , but there is no equation to evaluate FM trunk in the Chinese population by simple anthropometric methods. Considering the effects of ethnic-population differences and anthropometric indices on FM trunk , we performed the present study in a large Chinese female population to obtain corresponding predicting equations. Compared with established prediction equations in other populations [30–32] , we involved more indices in the prediction equations in our study. For example, an equation to predict body fat (BF) in white women [32] was expressed by W/H (weight/height) alone (BF = 1.181 × W/H − 24.18). To decrease the effects of age, it was appropriate to divide our sample into four 5-yr age range groups, because age was significantly correlated with FM trunk and five anthropometric indices in the whole sample, but was not associated with them in each age group. Consistent with previous studies [33,34] , our results showed that there was an increasing trend of FM trunk and five anthropometric indices in successively older age groups. Similar results were also reported in middle-aged Japanese women [35] . This tendency with aging may be partially explained by the decreased secretion of growth hormone (GH) [36–38] . For instance, Hoffman et al. [38] pointed out that GH diminished trunk adipose tissue in adults with GH deficiency syndrome. The variety of estrogen and declining physical activities with aging may also accelerate a more central body shape resulting from the higher proportion of FM trunk [39–41] . Our results showed that age had potential effects on the correlations between FM trunk and the anthropometric indices in four age groups studied. The difference in fat distribution in the different age groups may partially account for this observation. For example, in the 25–30 yr group, the correlation coefficients between FM trunk and anthropometric indices were smallest among all age groups. We presume that most Chinese females have the experience of procreating in this age phase, when the increasing level of estrogen may influence their fat distribution by controlling lipolysis [42] . Another study has shown that the estrogen-alpha gene polymorphisms were also associated with the distributions of regional fat in middle-aged women [43] . So, estrogen may largely affect the correlations between FM trunk and the five anthropometric indices in 25–29 yr group through the related route of adipose tissue metabolism. Other differences in body shape and lifestyle in different age groups may also account for the differential correlations between FM trunk and anthropometry [10] . Inevitably, the study has some questions that need to be explored further. Firstly, although five indices combined were strongly correlated with FM trunk , about 20–50% of the variance of FM trunk remained to be explained. It is necessary to find more underlying anthropometric indices for improving the accuracy of predicting FM trunk . Secondly, the present results from “healthy” samples aged 20–40 years, according to our exclusion criterion, might be inappropriate for other individuals beyond our “healthy” criterion and be incapable of generalising to older or younger populations, including the individuals with extreme BMI because BMI in our samples were mostly normal. In summary, the current study has found a complex relationship between the five anthropometric indices and FM trunk in a large cohort of Chinese adult females, with quite high validation of predicting FM trunk , and is the first to predict FM trunk using the five anthropometric indices combined in Chinese females. These results add to our understanding of the relationship between these indices and trunk adiposity, as well as the effects that age may have on the relationship between the two. Acknowledgments The study was partially supported by a key project grant (30230210), a general grant (30470534) from the National Science Foundation of China; three projects from Scientific Research Fund of Hunan Provincial Education Department (02A027, 03C226, 04B039); and a grant from the Natural Science Foundation of Hunan Province (04JJ1004). Investigator (HWD) was partially supported by grants from the Health Future Foundation of USA and grants from the National Health Institute (K01 AR02170-01A2, R01 GM60402 and 5R01 AR050496-02). References [1] D.N. Kaluski E.M. Berry Prevalence of obesity in Israel Obes Rev 6 2005 115 116 [2] D.M. Kim C.W. Ahn S.Y. Nam Prevalence of obesity in Korea Obes Rev 6 2005 117 121 [3] A. Milewicz D. Jedrzejuk F. Lwow A.S. Bialynicka J. Lopatynski G. Mardarowicz Prevalence of obesity in Poland Obes Rev 6 2005 113 114 [4] G.L. Liu An expert symposium Obesity. Zhong guo shi yong nei ke za zhi (in Chinese) 23 2003 513 524 [5] I.S. Okosun K.M. Chandra S. Choi J. Christman G.E. Dever T.E. Prewitt Hypertension and Type 2 diabetes comorbidity in adults in the United States: risk of overall and regional adiposity Obes Res 9 2001 1 9 [6] S. Lemieux Contribution of visceral obesity to the insulin resistance syndrome Can J Appl Physiol 26 2001 273 290 [7] A.H. Kissebah G.R. Krakower Regional adiposity and morbidity Physiol Rev 74 1994 761 811 [8] M. Krotkiewski P. Bjo¨rntorp L. Sjo¨stro¨m U. Smith Impact of obesity on metabolism in men and women. Importance of regional adipose tissue distribution J Clin Invest 72 1983 1150 1162 [9] R.E. Van Pelt E.M. Evans K.B. Schechtman A.A. Ehsani W.M. Kohrt Contributions of total and regional fat mass to risk for cardiovascular disease in older women Am J Physiol Endocrinol Metab 282 2002 E1023 E1028 [10] P. Deurenberg M. Yap W.A. van Staveren Body mass index and percent body fat: a meta analysis among different ethnic groups Int J Obes Relat Metab Disord 22 1998 1164 1171 [11] E. Duncan G. Schofield S. Duncan G. Kolt E. Rush Ethnicity and body fatness in New Zealanders N Z Med J 117 2004 U913 [12] J.M. Conway S.Z. Yanovski N.A. Avila V.S. Hubbard Visceral adipose tissue differences in black and white women Am J Clin Nutr 61 1995 765 771 [13] J.C. Lovejoy J.A. de la Bretonne M. Klemperer R. Tulley Abdominal fat distribution and metabolic risk factors: effects of race Metabolism 45 1996 1119 1124 [14] I. Janssen S.B. Heymsfield D.B. Allison D.P. Kotler R. Ross Body mass index and waist circumference independently contribute to the prediction of non-abdominal, abdominal subcutaneous, and visceral fat Am J Clin Nutr 75 2002 683 688 [15] R.W. Taylor D. Keil E.J. Gold S.M. Williams A. Goulding Body mass index, waist girth, and waist-to-hip ratio as indices of total and regional adiposity in women: evaluation using receiver operating characteristic curves Am J Clin Nutr 67 1998 44 49 [16] R.W. Taylor I.E. Jones S.M. Williams A. Goulding Evaluation of waist circumference, waist-to-hip ratio, and the conicity index as screening tools for high trunk fat mass, as measured by dual-energy X-ray absorptiometry, in children aged 3–19 yr Am J Clin Nutr 72 2000 490 495 [17] W.H. Xu C.E. Matthews Y.B. Xiang W. Zheng Z.X. Ruan J.R. Cheng Effect of adiposity and fat distribution on endometrial cancer risk in Shanghai women Am J Epidemiol 161 2005 939 947 [18] S. Czernichow S. Bertrais J.M. Oppert P. Galan J. Blacher P. Ducimetière Body composition and fat repartition in relation to structure and function of large arteries in middle-aged adults (the SU.VI.MAX study) Int J Obesity 29 2005 826 832 [19] J.E. Pritchard C.A. Nowson B.J. Strauss J.S. Carlson B. Kaymakci J.D. Wark Evaluation of dual energy X-ray absorptiometry as a method of measurement of body fat Eur J Clin Nutr 47 1993 216 228 [20] P.G. Kopelman Obesity as a medical problem Nature 404 2000 635 643 [21] F.Y. Deng M.Y. Liu M.X. Li S.F. Lei Y.J. Qin Q. Zhou Tests of linkage and association of the COL1A2 gene with bone phenotypes’ variation in Chinese nuclear families Bone 33 2003 614 619 [22] K.W. Chapman H.T. Lawless K.J. Boor Quantitative descriptive analysis and principal component analysis for sensory characterization of ultrapasteurized milk J Dairy Sci 84 2001 12 20 [23] C.J. Chang C.H. Wu W.J. Yao Y.C. Yang J.S. Wu F.H. Lu Relationships of age, menopause and central obesity on cardiovascular disease risk factors in Chinese women Int J Obes Relat Metab Disord 24 2000 1699 1704 [24] R.F. Gillum The association of body fat distribution with hypertension, hypertensive heart disease, coronary heart disease, diabetes and cardiovascular risk factors in men and women aged 18–79 years J Chron Dis 40 1987 421 428 [25] S. Lemieux D. Prud'homme C. Bouchard A. Tremblay J.P. Despres A single threshold value of waist girth identifies normal-weight and overweight subjects with excess visceral adipose tissue Am J Clin Nutr 64 1996 685 693 [26] R.S. Lindsay R.L. Hanson J. Roumain E. Ravussin W.C. Knowler P.A. Tataranni Body mass index as a measure of adiposity in children and adolescents: relationship to adiposity by dual energy X-ray absorptiometry and to cardiovascular risk factors J Clin Endocrinol Metab 86 2001 4061 4067 [27] M. Ferland J.P. Despres A. Tremblay S. Pinault A. Nadeau S. Moorjani Assessment of tissue distribution by computed tomography in obese women in association with body density and anthropometric measurement Br J Nutr 61 1989 139 148 [28] E. Bertin C. Marcus J.C. Ruiz J.P. Eschard M. Leutenegger Measurement of visceral adipose tissue by DXA combined with anthropometry in obese humans Int J Obes Relat Metab Disord 24 2000 263 270 [29] M.D. Jensen J.A. Kanaley J.E. Reed P.F. Sheedy Measurement of abdominal and visceral fat with computed tomography and dual-energy X-ray absorptiometry Am J Clin Nutr 61 1995 274 278 [30] T.T. Huang M.P. Watkins M.I. Goran Predicting total body fat from anthropometry in Latino children Obes Res 11 2003 1192 1199 [31] C. Pongchaiyakul V. Kosulwat N. Rojroongwasinkul S. Charoenkiatkul K. Thepsuthammarat M. Laopaiboon Prediction of percentage body fat in rural Thai population using simple anthropometric measurements Obes Res 13 2005 729 738 [32] I. Larsson B. Henning A.K. Lindroos I. Naslund C.D. Sjostrom L. Sjostrom Optimized predictions of absolute and relative amounts of body fat from weight, height, other anthropometric predictors, and age Am J Clin Nutr 83 2006 252 259 [33] K. Kotani K. Tokunaga S. Fujioka T. Kobatake Y. Keno S. Yoshida Sexual dimorphism of age related changes in whole-body fat distribution in the obese Int J Obes 18 1994 207 212 [34] M. Zamboni F. Armellini T. Harris E. Turcato R. Micciolo I.A. Bergamo-Andreis Effects of age on body fat distribution and cardiovascular risk factors in women Am J Clin Nutr 66 1997 111 115 [35] H. Ito A. Ohshima N. Ohto M. Ogasawara M. Tsuzuki K. Takao Relation between body composition and age in healthy Japanese subjects Eur J Clin Nutr 55 2001 462 470 [36] P. Bjorntorp The regulation of adipose tissue distribution in humans Int J Obes Relat Metab Disord 20 1996 291 302 [37] E. Corpas S.M. Harman M.R. Blackman Human growth hormone and human aging Endocr Rev 14 1993 20 39 [38] A.R. Hoffman B.M. Biller D. Cook J. Baptista B.L. Silverman L. Dao Genentech Adult Growth Hormone Deficiency Study Group. Efficacy of a long-acting growth hormone (GH) preparation in patients with adult GH deficiency J Clin Endocrinol Metab 90 2005 6431 6440 [39] S.S. Guo C. Zeller W.C. Chumlea R.M. Siervogel Aging, body composition, and lifestyle: the Fels Longitudinal Study Am J Clin Nutr 70 1999 405 411 [40] D. Mitchell M.N. Haan F.M. Steinberg M. Visser Body composition in the elderly: the influence of nutritional factors and physical activity J Nutr Health Aging 7 2003 130 139 [41] T.J. Parsons C. Power O. Manor Physical activity, television viewing and body mass index: a cross-sectional analysis from childhood to adulthood in the 1958 British cohort Int J Obes (Lond) 29 2005 1212 1221 [42] S.B. Pedersen K. Kristensen P.A. Hermann J.A. Katzenellenbogen B. Richelsen Estrogen controls lipolysis by up-regulating alpha2A-adrenergic receptors directly in human adipose tissue through the estrogen receptor alpha. Implications for the female fat distribution J Clin Endocrinol Metab 89 2004 1869 1878 [43] T. Okura M. Koda F. Ando N. Niino S. Ohta H. Shimokata Association of polymorphisms in the estrogen receptor alpha gene with body fat distribution Int J Obes Relat Metab Disord 27 2003 1020 1027