Association of general and abdominal obesity with age, endocrine and metabolic factors in Asian men

Abstract

Objective: This study made use of the percent abdominal fat to define abdominal obesity (AbO) and examined the differential associations of general obesity (GOb) and AbO with age, metabolic and endocrine factors.

Methods: Metabolic, endocrine and anthropometric factors and body composition were measured in 481 Asian men.

Results: A DEXA-derived ≥25% abdominal fat (PAbdF) was used to define men with AbO. Age was directly associated with PAbdF and percent total body fat (PBF). Exercise intensity was negatively associated with PBF. Both PBF and PAbdF were associated with HDL and LDL, but have opposite correlation with triglyceride. Furthermore, both PBF and PAbdF were associated with the number of metabolic syndrome (MetS) risk factors. Men with GOb had lower levels of percent lean mass (PLM), testosterone and bioavailable testosterone, and higher insulin and glucose levels. Men with AbO had lower arm and leg fat, higher insulin levels and triglycerides.

Conclusions: Men with GOb and AbO had different pattern of body composition. Age may be a contributory factor in AbO and a sedentary lifestyle may contribute to GOb. Both GOb and AbO are associated with an increased risk of MetS, with GOb more predispose to risk of diabetes, while AbO more at risk for cardiovascular diseases.

• Abdominal obesity
• hormones
• metabolic syndrome
• obesity

Introduction

Subcutaneous adipose tissue (SCAT) and visceral adipose tissue (VAT) are associated with general obesity (GOb) and abdominal obesity (AbO), respectively. They are thought to predispose to variable degrees of major disorders, including diabetes, cardiovascular disease and hypertension [1–8]. The variability noted in the association of GOb and AbO with various diseases may, in part, be due to the criteria used to classify GOb and AbO in earlier studies. A Body Mass Index (BMI) of ≥30 kg/m² is most commonly used definition of general obesity [9]. On the other hand, abdominal obesity for men has been defined by waist circumference (W), waist-hip ratio (W/H) or waist-height ratio [10,11]. However, inherent in all these indices is the fact that these cut-off values were arbitrarily set and that they may not be representative of all population groups. In our earlier study, we noted that instead of a BMI of 30 kg/m² as a cut-off value for obesity, a BMI of between 25 to 27 kg/m² was more appropriate for the Asian populations [12]. Our study further showed that more than 50% of non-obese men, based on having a DEXA-derived percent total body fat (PBF) of less than 25%, were misclassified as obese [12]. The high degree of misclassification, as was also shown by other investigators [13], is troublesome. The continued use of BMI as a surrogate of PBF has been encouraged by its simplicity and reproducibility but BMI does not account for all the adiposity-related risk in obese individuals [14,15].

Although W, W/H and W/Ht have been used as indices of central or abdominal obesity, there is still a lack of a more accurate measure to define AbO. The establishment of more appropriate indices of obesity is needed for future studies of obesity and its associated health problems. The present cross-sectional study evaluated the usefulness of the DEXA-derived percent abdominal fat (PAbdF) to define abdominal obesity (AbO) and examined whether there are differences between the GOb and AbO in the overall body composition, metabolic, cardiovascular and endocrine profiles in a group of Asian men.

Materials and methods

Subjects

This study was approved by the Institutional Review Board of the National University Hospital of Singapore and each volunteer gave his written informed consent. Five-hundred and twenty-nine Singaporean Chinese men, aged between 29y and 72y, living in the community were recruited through general invitation through the media and word of mouth. As the primary objective of the study was to evaluate the determinants of the natural aging process, only men without a history of medical illnesses such as cancer, hypertension, thyroid dysfunction, diabetes, osteoporotic fracture, cardiovascular events, major sleep disorders, major joint surgery, or bone fracture were included in the study. Subjects were not paid for their participation. The cohort of men represented the diverse spectrum of Chinese in Singapore, ranging from those with low to high levels of education, working and non-working men (retirees), and those in various types of vocation [16]. Their profiles were typical of Singapore, which is a highly urbanized city-state with no rural population. Each subject answered a self-administered and investigator-guided questionnaire. The method was previously reported [17]. Out of the 529 subjects recruited data from 482 Singaporean Chinese men were included in the analyses.

Methods

General questionnaire

Each subject answered a self-administered and investigator-guided questionnaire. Questions asked included their medical, social, sex, exercise regime and family history.

Biochemical and hormone measurements

An overnight 12 h fasting blood sample was collected in the morning between 9.00am and 11.00am and the sera were stored at −80 °C until analyses. Serum levels of total cholesterol (TC) and triglycerides (TG), high-density lipoprotein-cholesterol (HDL), low-density lipoprotein cholesterol (LDL) and fasting glucose level (GLU) were measured by methods reported earlier [12]. Serum testosterone (T) and estradiol (E2), dehydroepiandrosterone sulphate (DHEA/S), sex hormone binding globulin (SHBG) and cortisol (Cor) were measured by established radioimmunoassay methods reported earlier [12]. Serum concentrations of insulin-like growth factor-1 (IGF1) and insulin like growth factor binding protein-3 (BP3) were measured using immunoradiometric assay kits (Diagnostic Systems Laboratories, Inc., Webster, TX) as reported earlier [18,19]. Serum concentrations of insulin (INS) were measured in-house using the Axsym platform from Abbott Laboratories (Irving, TX). Bioavailable testosterone (BioT) was calculated using the computer formula of Vermeulen, which is available on the ISSAM website (www.issam.ch).

Whole body DEXA scan

Every man had a whole body scan using the DEXA Hologic (Bedford, MA). The percent total body fat (PBF) was calculated by the DEXA machine based on the Siri formula. Percent body lean mass (L), and fat mass (F) in the different regions were computed from the total body scan on the DEXA. Hence, for the present study, the percent trunk lean and fat mass (PTkL and PTkF), percent abdominal lean and fat mass (PAbdL and PAbdF), percent arm lean and fat mass (PArmL and PArmF), and percent leg lean and fat mass (PLegL and PLegF) together with percent total lean mass (PLM) were used in the analyses.

Anthropometric measurements

The three most common anthropometric measurements for abdominal obesity used are waist circumference (W), waist/hip ratio (W/H) and waist/height ratio (W/Ht). Waist circumference in centimeter was measured midway between the lower costal margin and iliac crest during the end-expiratory phase, while hip circumference (H) was measured in centimeters. Height (Ht) in centimeter was measured without shoes.

Blood Pressures

Brachial systolic (Sys) and diastolic (Dia) blood pressures were measured by trained clinical researchers using a standardized manual sphygmomanometric method after subjects had five minutes of rest. Both blood pressures were recorded as millimeter of mercury (mmHg).

Intensity of Exercise (METmin)

The intensity of exercise, expressed as metabolic equivalent of task-minutes (METmin), was shown in our earlier studies to affect body composition, and various metabolic and endocrine factors. In order to adjust for the effect of exercise, the METmin was used as a covariate in all the analyses. Computation of the total exercise score (METmin) was reported earlier [20]. This takes into account the type, duration and frequency/week of exercise for each participant. These data were gathered using a self-administered and investigator-guided questionnaire.

Risk factors of the metabolic syndrome

The present study made used of NCEP ATPIII definition of the metabolic syndrome (MetS) [21] except that a waist circumference (W) of 92.5 cm appropriate for the local population [12] was used instead of 102 cm as recommended by the NCEP ATPIII [21]. The number of possible metabolic syndrome risk factors (NoMetS) for each man, therefore, ranged from 0 to 5.

Definitions of general and abdominal obesity)

Percent total body fat (PBF) computed from the DEXA-whole body scan was used to define general obesity. Obesity is defined when the PBF is ≥25% for men [22]. Hence, all the men were categorized as having general obesity (GOb) when the PBF was ≥25%.

For the definition of abdominal obesity (AbO), we constructed a frequency distribution of all DEXA-derived PAbdF of the 482 men. We then arbitrarily used the PAbdF at the 95 percentile as the cut-off value for AbO. This cut-off value was found to be equal to a PAbdF of 25%. Therefore, for the present study, men with PAbdF of ≥25% were considered to have abdominal obesity (AbO). However, it is noted that the DEXA-derived PAbdF does not distinguish between abdominal subcutaneous from visceral fat. It is the sum total of fat in the abdominal area.

Table 1 shows the distribution of the 482 men with and without AbO against men with or without GOb. Of the 25 men with AbO (PAbdF ≥25%), only one had concurrently a PBF of 27% thus was with GOb. For a clean classification of AbO and GOb, we therefore removed this case from all subsequent analyses, leaving 481 men in the analyses. The Receiver Operating Characteristic (ROC) curves were constructed with PAbdF set at 25% against W, W/H and W/Ht for abdominal obesity in order to determine their respective cut-off levels at the point of the 100% sensitivity level where there is no overlap between the obese and non-obese groups.

Table 1. Distribution of men with general obesity (PBF ≥ 25%) and with abdominal obesity (PAbdF ≥ 25%).

	PAbdF < 25%, n = 457	PAbdF ≥ 25%, n = 25
PBF < 25%, n = 451	427	24
PBF ≥ 25%, n = 31	30	1

Statistical analysis

Statistical analyses were performed using SPSS for windows version 21.0 (Armond, NY). Basic descriptive statistics and multivariate linear analyses coupled with the Bonferroni correction for multiple comparisons were used on continuous measurements. To adjust for their effects, age and METmin were used as co-variates in all the multivariate analyses. Multivariate linear analyses were carried out on the three groups: non-obese men (NOB), general obese men (OB) and abdominal obese men (AO). To evaluate the parameters which were independently correlated to the PBF and PAbdF, the “stepwise” method of linear regression analyses was adopted. Significant correlation was set at a p value of <0.05. ROC curves for PAbdF versus W, waist/hip ratio (W/H) and waist/height ratio (W/Ht) were constructed using this SPSS package. All p values were based on the two-sided analysis. Comparisons were only considered as statistically significant when, on a 2-tail analysis, the p values is ≤0.05.

Results

Table 2 shows the 100% sensitivity cut-off values for W, W/H and W/Ht from the ROC curves when AbO was set at a PAbdF of 25%. However, at the respective 100% sensitivity cut-off values, the rates of false positive for AbO using the three anthropometric indices were high, ranging from about 53% to 66% (Table 2).

Table 2. Sensitivity and 1-Specificity for W, W/H and W/Ht when the cut-off value for abdominal obesity was set at a percent abdominal fat (PAbdF) of 25%.

	Cut-off value when Sensitivity = 100%	1-Specificity
W (cm)	86.5	0.638
W/H	0.887	0.534
W/Ht	0.511	0.655

Age was directly correlated to PBF and PAbdF, with an apparently greater association with PAbdF than with PBF (Table 3). Increasing intensity of physical exercise (METmin) was significantly correlated with lower PBF but not with PAbdF (Table 3). Both PBF and PAbdF were significantly associated with HDL and LDL, but have opposite correlations with TG. Percent total body fat (PBF) was negatively correlated to TG, whereas PAbdF was positively correlated to TG (Table 3). As PBF increases, the number of metabolic risk factors increases (Table 3). Percent abdominal fat (PAbdF), on the other hand, was positively correlated to systolic blood pressure (Table 3). Both PBF and PAbdF were highly significantly and positively correlated to insulin levels (Table 3). Only PBF was negatively correlated to testosterone (T) (Table 3). The opposite is true for SHBG: only PAbdF was significantly and negatively correlated to SHBG (Table 3). All other endocrine factors including E2, DHEA/S, BioT, Cor, IGF1, BP3, IGF/BP and other metabolic factors including TC, TC/HDL and diastolic blood pressure were not significantly correlated with either PBF or PAbdF.

Table 3. Multiple linear regression analyses of percent body fat (PBF) and percent abdominal fat (PAbdF) separately with metabolic and endocrine factors using the stepwise method.

PBF vs	Beta	p value	PAbdF vs	Beta	p Value
Age (y)	0.085	0.044	Age	0.184	<0.001
METmin	−0.150	<0.001	MET-min	−0.007	NS
TG (mmol/l)	−0.121	0.031	TG	0.244	<0.001
HDL (mmol/l)	−0.143	0.004	HDL	−0.194	<0.001
LDL (mmol/l)	0.101	0.020	LDL	0.122	0.004
NoMetS	0.335	<0.001	Sys BP	0.114	0.011
T (ng/ml)	−0.091	0.030	T	−0.024	NS
SHBG (nmol/l)	−0.028	NS	SHBG	−0.220	<0.001
INS (mIU/l)	0.435	<0.001	INS	0.304	<0.001

Significant differences in some metabolic, endocrine factors and body composition were noted between general obese (OB) and abdominal obese (AO) men when compared to their corresponding non-obese counterparts (NOB). These differences occurred in the presence of no differences in age and METmin in both the OB and AO groups when compared to the corresponding NOB group (Table 4).

Table 4. Comparisons among non-obese men (NOB), general obese men (OB) and abdominal obese men (AO) using the multivariate linear analyses. All values state the mean and ±SE.

	NOB (n = 427)	OB (n = 30)	AO (n = 24)	p Values
Age (y)	50.9 ± 0.41	53.1 ± 1.54	53.5 ± 1.73	NS
MET-min	677 ± 40	534 ± 151	907 ± 169	NS
PLM (%)	74.5 ± 0.20	64.9 ± 0.75	74.7 ± 0.83	OB versus NOB, AO (<0.001, <0.001)
PBF (%)	17.1 ± 0.19	26.6 ± 0.72	18.1 ± 0.81	OB versus NOB, AO (<0.001, <0.001)
PAbdL (%)	17.3 ± 0.056	17.6 ± 0.21	18.8 ± 0.24	NOB, OB versus AO (<0.001, <0.001)
PLegL (%)	28.4 ± 0.067	28.7 ± 0.25	27.4 ± 0.28	NOB, OB versus AO (0.003, 0.002)
PTkF (%)	41.4 ± 0.13	42.0 ± 0.51	44.0 ± 0.58	NOB, OB versus AO (<0.001, 0.030)
PAbdF (%)	20.9 ± 0.11	21.2 ± 0.42	25.9 ± 0.48	NOB, OB versus AO (<0.001, 0.001)
PArmF (%)	9.22 ± 0.11	9.84 ± 0.41	7.39 ± 0.46	NOB, OB versus AO (<0.001, 0.001)
PLegF (%)	23.8 ± 0.19	23.6 ± 0.71	18.4 ± 0.81	NOB, OB versus AO (<0.001, 0.001)
T (ng/ml)	5.18 ± 0.09	4.33 ± 0.35	4.74 ± 0.39	NOB versus OB (0.049)
BioT (ng/dl)	278 ± 5.5	221 ± 21	254 ± 23	NOB versus OB (0.012)
INS (mIU/l)	6.93 ± 0.20	12.2 ± 0.74	9.45 ± 0.83	NOB versus OB, AO (<0.001, 0.010) OB versus AO (0.039)
TG (mmol/l)	1.44 ± 0.034	1.48 ± 0.13	1.94 ± 0.15	NOB versus AO (0.003)
GLU (mmol/l)	4.91 ± 0.035	5.32 ± 0.13	5.08 ± 0.15	NOB versus OB (0.008)
HDL(mmol/l)	1.35 ± 0.014	1.23 ± 0.054	1.19 ± 0.059	NOB versus AO (0.035)
TC/HDL	4.49 ± 0.057	4.53 ± 0.21	5.25 ± 0.24	NOB versus AO (0.005)
Systolic B/P	127.2 ± 0.74	131.5 ± 3.0	135.0 ± 2.6	NOB versus AO (0.036)
Diastolic B/P	79.7 ± 0.45	79.2 ± 1.71	84.4 ± 1.87	NOB versus AO (0.041)
NoMetS	0.74 ± 0.043	1.89 ± 0.16	1.44 ± 0.18	NOB versus OB, AO (<0.001, <0.001)

Percent total lean mass (PLM) in OB group was significantly lower, by about 10%, as compared to non-obese men (NOB, Table 4). No significant differences were noted in the regional distribution of lean and fat mass when compared to non-obese men (Table 4). No significant differences in either percent total lean mass (PLM) and PBF were noted between AO and NOB groups (Table 4). However, there were significant differences in the regional distribution of body composition between AO and NOB groups. Abdominal obese-men had significantly higher levels of abdominal and trunk fat in the presence of lower fat mass in both the arms and legs (Table 4). There was a 1% lower lean mass in the legs and correspondingly, a 1% higher lean mass in the abdomen (Table 4).

Men in OB and AO groups had significantly higher insulin levels (INS) than in NOB group (Table 4). In addition, raised INS levels in OB group were significantly higher than those in AO group (Table 4). While no other significant differences in other endocrine factors were noted between AO and NOB groups, T and BioT in OB group were significantly lower than corresponding levels in NOB group (Table 4). There were no significant differences for E2, DHEA/S, IGF1, BP3, IGF/BP and Cor among the OB, AO and NOB groups.

Among the metabolic factors, GLU was significantly higher in OB group when compared to NOB (Table 4). On the other hand, in the AO group, TG was significantly higher when compared to non-obese men (NOB) (Table 4). Men in the AO group had significantly higher TC/HDL, and systolic and diastolic blood pressure and lower HDL when compared to corresponding levels in the NOB group (Table 4). In both OB and AO groups, there was significantly greater number of metabolic syndrome risk factors when compared to non-obese men in NOB group (Table 4).

Discussion

In the present study, we have used, arbitrarily, the DEXA-derived ≥25% PAbdF to define abdominal obesity (AbO). Using the ROC curves with the 100% sensitivity set at 25% PAbdF, the cut-off values for defining AbO using the anthropometric parameters of W, W/H and W/Ht were 86.5 cm, 0.887 and 0.511, respectively. Interestingly, except for W, the newly derived W/H and W/Ht cut-off values for AbO were fairly close to the >0.90 and >0.5 recommended earlier [10,11] and the 86.5 cm shown in the present study is in agreement with the suggested W cut-off values of 90 cm for the Asian population [23]. However, the high rates of false positives of these anthropometric indices for defining AbO set at >+=25% PAbdF, confirm the limitations of using these indices for studies of abdominal obesity.

We have shown that general obesity and abdominal obesity, whilst usually occurring in different individuals, are not entirely mutually exclusive. One out of 25 cases classified as having abdominal obesity had a PBF of 27% and therefore was concurrently classified as having general obesity. The other 24 cases of AbO had PBF ranging from 15.7% to 24.5%. But generally, AbO is not a subset of GOb and apparently lean individuals with normal PBF can have a significant accumulation of abdominal fat and have been referred to as metabolically obese normal weight (MONW) [24]. The results suggest that we can identify individuals with abdominal obesity using 25% of abdominal fat as the cut-off value. Furthermore, the existence of 1 out of 25 who was both abdominally and generally obese, suggest that individuals with both types obesities might exist. It is unclear at this juncture whether the very-low rate (1/25) of men having concurrently general and abdominal obesity is peculiar to this Asian population or whether it is true also for Caucasian men. Future studies on this group of individuals to evaluate whether their adverse associations to metabolic, cardiovascular and other diseases might be different to individuals who are MONW or are generally obese alone.

As reflected in the differences noted in the body composition, general and abdominal obesity may have different etiologies. In men with general obesity, the increase in PBF of about 10% was associated with no changes in regional distribution of lean and fat mass. In AbO, on the other hand, the opposite is true. There were no significant differences in PLM and PBF. However, significant differences were noted in the regional distribution of fat mass. Increased fat accumulation in the abdomen was paralleled by decreased fat accumulation in the limbs.

While it is not clear what the mechanisms for the specific patterns of fat accumulation in both GOb and AbO are, the association studies suggest that some factors may play a role in their development. Percent abdominal fat was more highly than PBF associated with age, implying that AbO is more age-dependent which is in agreement with earlier observations [1]. The PBF but not PAbdF was negatively associated with exercise intensity implies that a sedentary lifestyle may increase the risk of general obesity. In addition, while the promotion of physical exercise may help to reduce the risk of general obesity, it will not help reduce the risk of abdominal obesity as much [20]. This is in contrast to some reports which have suggested that exercise can reduce waist size and VAT independent of changes in BMI [25].

Earlier studies have shown that although general obesity is associated with insulin resistance, the relative contribution of VAT and SCAT to insulin resistance has been conflicting [16,26]. It has been shown that both SCAT and VAT were strongly and independently correlated with insulin resistance but that VAT is a more potent predictor of insulin resistance than SCAT [27,28]. Other studies, however, have found that abdominal SCAT contributes to insulin resistance independent of VAT [29,30]. The present study showed that both GOb and AbO were independently associated with higher INS, implying increased insulin resistance in both GOb and AbO; and the difference may be in the degree of insulin resistance. It has been suggested that high levels of SCAT are associated with insulin resistance and increasing INS secretion [27]. We have shown that in GOb men, the INS level was raised to levels higher than those in men with AbO. In addition, this raised INS in GOb men was accompanied by significantly-raised fasting-glucose level, suggesting that the increased in INS secretion may not have adequately compensated the SCAT-related insulin resistance. Conflicting results of earlier studies may be due to the misclassification of general obesity and abdominal obesity when anthropometric indices are used to classify GOb and AbO, as was discussed earlier.

Our results showed that both GOb and AbO carry an increased risk of the metabolic syndrome (MetS) in agreement with earlier studies [31,32]. But the profile of the increased risk of MetS may be slightly different. Both GOb and AbO may be associated with increased insulin resistance, but with GOb the risk for diabetes may be higher than for AbO in agreement with earlier studies [31]. On the other hand, men with AbO may be predisposed to have higher TG, systolic and diastolic blood pressures, and lower HDL suggesting that AbO may carry a higher risk for cardiovascular diseases [32–34]. The relationship between high TG levels and glucose control in subjects with abdominal obesity is unclear.

Obesity results from the hyperplasia and hypertrophy of fat cells. Many endocrine factors such as adipsin, complement D and leptin are secreted by fat cells. Fat depots have been viewed as endocrine organs [35]. Androgens, estrogens, DHEA, cortisol, and growth hormone (GH) are known to affect body fat distribution [4]. The present study, indeed, indicated that hormones are associated with obesity. Of all the hormones including, IGF1, BP3, IGF/BP, Cor, E2 and DHEA/S, only T, SHBG and BioT were noted to have significant associations with total body and abdominal fat. However, the associations of these hormones with GOb and AbO are different. Percent total body fat (PBF) was significantly and negatively correlated with T but not with SHBG. On the other hand, PAbdF was significantly and negatively correlated to SHBG but not with T, in agreement with earlier studies which showed that VAT was associated with T, free-T or SHBG [36]. Our findings of significantly lower T and BioT in GOb men but not in AbO men suggest that general obesity, rather than abdominal obesity is more related to androgen metabolism than is AbO. Other studies have shown that negative correlation of T with various anthropometric indices for general and central obesities [37,38]. However, it is unclear whether the low androgen level is the cause or the effect of general obesity. In addition, contrary to an earlier report, estradiol showed no correlation with either general or abdominal obesity [39]. The conflicting observations may be due, again, to the misclassifications of obesity in studies using anthropometric indices.

We have use the DEXA-derived ≤25% of PBF and ≥25% of PAbdF to define general obesity and abdominal obesity for this study. We have shown that while GOb and AbO men have some similar characteristics, there were distinctive differences in body composition, metabolic and endocrine features between GOb and AbO. Men with general obesity had raised PBF in the presence of unchanged regional distribution of lean and fat mass. Men with abdominal obesity, on the other hand, had unchanged PBF and percent total lean mass, but raised PAbdF and PTkF accompanied by reduced PArmF and PLegF. While age may be a confounding factor in the accumulation of abdominal fat, a sedentary lifestyle appears to be related to the increase in general adiposity in the current subjects. Both GOb and AbO are associated with an increased risk of MetS, with GOb men more at risk of diabetes and AbO men more at risk of cardiovascular diseases which is supported by earlier studies [5,6,40,41] but contrary to others [8]. General and not abdominal obesity may also be associated with a dysfunctional androgen metabolism.

The etiology of the selective accumulation of body fat in GOb and AbO is unclear from the present study. Further, it is unclear whether the lower androgen level in GOb is a cause or an effect of the obesity. It would appear that the promotion of regular physical exercise would probably have beneficial effects in men with GOb and AbO, and would likely be more beneficial to men with general obesity.

The major disadvantage of the present study is that as a cross-sectional study, no definitive causal relationships could be attributed, but only suggested. However, the advantages of the present study include the use of DEXA-derived total and regional distribution of body composition, avoiding the use of anthropometric indices, with their inherent risks of misclassification of obesity, to categorize GOb and AbO. In addition, the use of age, and exercise intensity (MET-min) as co-variates in the multivariate analyses ensured that the potential confounding influences of these factors could be adjusted for.

Much has been published on the different association of general and abdominal obesity with metabolic, cardiovascular diseases and hormones. However, many of the studies based their classification of general and abdominal obesity on anthropometrics indices such as BMI, waist circumference (W) and waist/hip (W/H) and waist/height (W/Ht) ratios. These indices are known to have misclassification rates as high as 50% therefore making interpretations of observed results very conflicting and confusing.

The current study made use of the dual-energy x-ray absorptiometry (DEXA)-derived percent total body fat and percent abdominal fat to define general and abdominal obesity respectively. The criteria used to define general obesity avoided the misclassification present in many studies using the anthropometric indices. The used of 25% of abdominal fat as a cut off to define abdominal obesity resulted in a group with abdominal obesity without concurrently having general obesity. With this index for classifying AbO, we were able to show similar and yet distinct differences in the association of GOb and AbO with metabolic, body composition and endocrine factors.

Acknowledgements

We would like to acknowledge the technical assistance from staff of the Endocrine Research and Service Laboratory of the Department of Obstetrics and Gynecology, National University of Singapore.

Declaration of interest

This study was designed, conducted and data collected while Prof. Victor H. H. Goh was at the Department of Obstetrics and Gynecology, National University of Singapore. Prof. Hart was intimately involved in the interpretation, drafting and critical revision of the article for submission.

The authors report no declaration of interest. This study was supported, in part, by funds from the Academic Research Fund of the National University of Singapore, Singapore.