620 Original article

Association of new indices; visceral adiposity index and body adiposity index, with parameters in obese patients with or without type 2 mellitus Nearmeen M. Rashad, George Emad

Department of Internal Medicine, Faculty of Background Medicine, Zagazig University, Zagazig, Egypt Obesity is the cornerstone of metabolic syndrome (MetS); it is not possible to use Correspondence to Nearmeen M. Rashad, MD, BMI to differentiate between lean mass and fat mass. We aimed to investigate, for Department of Internal Medicine, Faculty of the first time, the possible association of new obesity indices; visceral adiposity Medicine, Zagazig University, 44519, Zagazig, index (VAI) and body adiposity index (BAI), with parameters of MetS in Egyptian Egypt. Tel: +20 122 424 8642; e-mails: [email protected], [email protected] obese patients. Patients and methods Received 3 January 2019 This was a case–control study that included unrelated 150 obese patients and 50 Accepted 6 February 2019 Published: 18 August 2020 healthy controls. Obese patients were then subdivided into two subgroups, nondiabetic patients (n=85) and 65 patients with mellitus. We The Egyptian Journal of Internal Medicine 2019, 31:620–628 measured the anthropometric measures; BMI, waist/hip ratio, waist/height ratio, BAI, and VAI. Results Among obese patients, we found significant positive correlations between parameters of MetS and obesity indices. Among obesity indiced, the highly significant positive correlations were found between VAI and parameters of MetS. After adjusting for the traditional risk factors, logistic regression analysis test found that the VAI value was the best predictor of type 2 diabetes mellitus in comparison with BMI and BAI. Receiver operating characteristic curve was used to assess the power of obesity indices; the sensitivity and the specificity of BMI were 94.7 and 99.9%, for VAI, they were 74.4 and 99.9%, and, for BAI, they were 83.3 and 58%, respectively. Conclusion BMI is still the most powerful diagnostic tool for obesity. Although, in certain conditions, where there are limitations of using BMI, we can use other obesity indices, VAI and BAI could be used to discriminate cardiovascular risk among obese patients.

Keywords: body adiposity index, BMI, metabolic syndrome, obesity, visceral adiposity index

Egypt J Intern Med 31:620–628 © 2020 The Egyptian Journal of Internal Medicine 1110-7782 apnea, and respiratory problems, as well as some Introduction types of cancers [9,10]. The metabolic syndrome (MetS) is a set of interrelated risk factors including , dyslipidemia, BMI is the widely used measure of obesity. However, obesity, and high blood glucose [1,2]. Obesity is a BMI is unable to differentiate between lean mass and key phenotype leading to atherogenic and diabetogenic fat mass, and hence it is limited by differences in body profiles [3]. Insulin resistance, together with central/ adiposity for a given BMI across age, sex, and ethnicity abdominal or visceral obesity, have been proposed as [11]. Waist circumference (WC), waist-to-hip ratio key risk factors in the development of the MetS [4–6]. (WHR), and waist-to-height ratio (WHtR) have been developed and studied. WC has been proposed to be Obesity is actually an epidemic problem in the world; it the best among these measures, with excellent has become truly a global problem affecting countries correlation with abdominal imaging and high rich and poor. An estimated 500 million adults association with risk factors, worldwide are obese, and 1.5 billion are especially diabetes [12,13]. However, WC does not or obese [7]. Particularly the prevalence of obesity in account for differences in height, thereby, potentially, Egypt has increased at an alarming rate during the last three decades, affecting 22% of adult male individuals and 48% of adult female individuals [8]. It is associated This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 with several comorbidities including hypertension, License, which allows others to remix, tweak, and build upon the work dyslipidemia, type 2 diabetes mellitus (T2DM), non-commercially, as long as appropriate credit is given and the new coronary heart disease, stroke, , sleep creations are licensed under the identical terms.

© 2020 The Egyptian Journal of Internal Medicine | Published by Wolters Kluwer - Medknow DOI: 10.4103/ejim.ejim_4_19 Association of new obesity indices Rashad and Emad 621 overevaluating and underevaluating risk for tall and from diabetes and endocrinology outpatient clinic of short individuals, respectively [14]. Consequently, Internal Medicine Department of Zagazig University several researchers independently proposed the Hospitals and 50 healthy lean controls, who were WHtR as an alternative to WC. matched to cases by age, sex, and ethnic origin.

Notably, the body adiposity index (BAI) was proposed Obese patients were stratified into two subgroups: in 2011 [15]. This is a composite index based on nondiabetic patients (n=85) and T2DM patients hip circumference and height. The authors suggested (n=65), The diagnosis of T2DM was according to that this index showed a high correlation with body the American Diabetes Association criteria reported in fat measured using dual-energy radiography 2017: fasting plasma glucose levels of more than or absorptiometry. They added that this correlation was equal to 126 mg/dl (7.0 mmol/l) or 2-h postprandial higher than the correlation between BMI and body fat, plasma glucose levels of more than or equal to 200 mg/ measured using dual-energy radiography dl (11.1 mmol/l). All patients were subjected to absorptiometry among both African–American and thorough history taking and full clinical assessment Mexican–American men and women [15]. Thus, including blood pressure, WC (a level midway between BAI overcomes the limitation of BMI in the lowest rib and the iliac crest), and hip differentiating between fat and lean mass. Thus, circumference (widest diameter over the greater they concluded that the BAI is a useful predictor of trochanters). Anthropometric variables including obesity and suggested that it involves more simple BMI were calculated as weight (kg)/height (m2), measurements because weight is not needed [15]. WHR as WC (cm)/hip circumference (cm), and WHtR was calculated as waist (cm)/height (cm). In addition, visceral adiposity index (VAI) is a Moreover, VAI in female individualsÀÁ wasÀÁ calculated WC × TG × 1:52 mathematical model that uses both anthropometric as follows: 36:58þðÞ1:89 × BMI! 0:81 HDLC .It (BMI and WC) and functional [triglycerides (TG) normally equals 1 in healthy nonobese patients with and high-density lipoprotein (HDL) cholesterol] normal adipose distribution and normal TG and HDL simple parameters [16]. This index, which could be levels [15]. Finally, BAI, which is approximately equal considered a simple surrogate marker of visceral to the percentage of body fat for adult men and women adipose dysfunction, showed a strong association with of differing ethnicities, was calculated as HipcircuferanceðcmÞ both the rate of peripheral glucose utilization during the : 18 [16]. heightðmÞ1 5 euglycemic–hyperinsulinemic clamp and with visceral measured with MRI. Interestingly, it The MetS was diagnosed according to International showed a strong independent association with both Diabetes Federation criteriaas WC more than or equal cardiovascular and cerebrovascular events [14] and to 80 cm in women and more than or equal to 94 cm in showed better predictive power for incident diabetes men and the presence of at least two of the following events than its individual components (WC, BMI, characteristics: (a) fasting blood glucose more than or TG, and HDL cholesterol) [17]. equal to 100 mg/dl or previously diagnosed impaired fasting glucose; (b) blood pressure more than or equal On the basis of the previous facts, obesity is the to 130/85 mmHg or treated for hypertension; (c) TG cornerstone of the MetS; BMI is unable to more than or equal to 150 mg/dl; (d) HDL cholesterol differentiate between lean mass and fat mass. Further less than 40 mg/dl in men or less than 50 mg/dl in research is warranted, and, to address this need, we have women or taking treatment for low HDL [17]. focused on various anthropometric measures including the recently proposed BAI and VAI, which have not Patients with cancer, stroke, or liver, kidney, thyroid, been extensively analyzed and compared with BMI. and cardiovascular or any active inflammatory diseases Therefore, the purpose of this current novel study is were excluded from this study. None of the participants to clarify the possible relationships of traditional obesity had history of abdominal surgery that could have an indices (WC, BMI, and WHtR) and new obesity indices impact on abdominal fat distribution, as well as (BAI and VAI) with parameters of MetS in Egyptian receiving medications that affect endocrine obese women. parameters, glucose metabolism, and/or for weight reduction or participating in a dietary or Patients and methods program during the preceding 6 months or the Patients immediately preceding month (anti-inflammatory This study included 200 unrelated patients. One drugs). There was no concurrent minor infection hundred fifty obese patients (BMI>30) recruited reported during the study or during the month 622 The Egyptian Journal of Internal Medicine, Vol. 31 No. 4, October-December 2019 preceding the study. The ethical committee of Faculty We considered P to be significant at less than 0.05 with of Medicine, Zagazig University, approved our study a 95% confidence interval (CI). protocol, and all participants signed the written informed consent. Results Anthropometric and biochemical characteristics of the Blood sampling studied groups Blood samples were drawn from all patients after an Anthropometric and biochemical characteristics of the overnight fast and divided into three portions: 1 ml of study patients are summarized in Table 1. Obese whole blood was collected into evacuated tubes patients had significantly higher values of systolic containing EDTA for glycated hemoglobin blood pressure, diastolic blood pressure, fasting (HbA1c), and 1 ml of whole blood was collected blood glucose, HbA1c values, FSI, HOMA-IR, into evacuated tubes containing potassium oxalate total cholesterol, LDL, and TG levels compared and sodium fluoride (2 : 1) for fasting blood plasma with lean controls. Moreover, all obesity indices and glucose (FPG). Serum were separated immediately parameters (WC, BMI, WHR, WHtR, BAI, and from the remaining part of the sample and stored at VAI) were significantly higher in obese women −20°C until analysis. compared with lean patients. On the contrary, obese patients had significantly lower levels of HDL Biochemical analysis cholesterol and HOMA-B than healthy lean We determined FPG levels using the glucose oxidase individuals (P<0.001). method (Spinreact, Girona, Spain). Total cholesterol, HDL cholesterol, and TG levels were measured by Table 1 Anthropometric and biochemical characteristics of routine enzymatic methods (Spinreac). The low- the studied groups density lipoprotein (LDL) cholesterol level was Lean healthy Obese P value calculated using the Friedewald formula [18]. control (N=50) patients (N=150) Immunochemical assays Age (years) 45.58±11.4 44.64±8.2 0.15 Fasting serum insulin (FSI) concentrations were Sex (male/female) 15/35 42/108 0.93 2 measured using high-sensitivity linked BMI (kg/m ) 22.2±3.182 35.01±7.21 <0.001* Waist 91.4±14.71 111.01±13.8 <0.001* immunosorbent assay kit provided by (Biosource circumference (cm) Europe S.A., Nivelles, Belgium). Homeostasis model Waist/hip ratio 0.94±0.013 1.25±0.176 <0.001* assessments of insulin resistance (HOMA-IR) were Waist/height ratio 0.54±0.09 0.64±0.089 <0.001* calculated as follows: [FSI (mU/ml)×FPG (mg/dl)/ Body adiposity 22.97±8.7 35.86±15.4 <0.001* 405], and β-cell function (HOMA-β) was calculated index < as {20×[FSI (lU/ml)]/[FPG (mmol/l)−3.5]} [19]. Visceral adiposity 0.89±0.14 5.92.±1.34 0.001* index Systolic blood 110.2±4.79 129.5±16.2 <0.001* Statistical analysis pressure (mmHg) Statistical analyses were performed using the statistical Diastolic blood 75.4±4.21 84.43±9.13 <0.001* pressure (mmHg) package for the social sciences for Windows (version Total cholesterol 176.1±7.85 221.2±75.2 <0.001* 17.0; SPSS Inc., Chicago, Illinois, USA). Data were (mg/dl) expressed using descriptive statistics (mean±SD) and Triglycerides (mg/ 129.5±12.3 189.9±34.6 <0.001* were analyzed using the t test. Pearson’s correlation dl) coefficient was used to assess the association between LDL cholesterol 94.8±0.98 139.3±73.9 <0.001* (mg/dl) obesity indices and other studied metabolic parameters HDL cholesterol 48.4±4.41 46.6±8.95 0.178 in obese women. Receiver operating characteristic (mg/dl) (ROC) analysis was performed to assess the Fasting blood 86.8±5.08 147.9±55.9 <0.001* potential accuracy of BMI, VAI, and BAI, the area glucose (mg/dl) Fasting serum 7.4±1.82 13.18±6.37 <0.001* under the curve (AUC), and the cutoff values for insulin (μU/ml) diagnosis of T2DM among obese patients. A HbA1c (%) 5.36±0.41 7.26±1.76 <0.001* stepwise multiple linear regression analysis was HOMA-IR 1.57±0.40 5.8±4.14 <0.001* performed to detect the main predictors of VAI and HOMA-β 118.6±37.1 84.8±27.9 <0.001* BAI values in the obese group. Logistic regression Data are presented as mean±SD. HbA1c, glycated hemoglobin; analysis was performed to assess the predictive HDL, high-density lipoprotein; HOMA-IR, homeostasis model assessments of insulin resistance; HOMA-β, an index of β-cell powers of VAI as well as BAI in the diagnosis function; LDL, low-density lipoprotein. *P value less than 0.05 of MetS in obese patients with and without T2DM. when compared with control group. Association of new obesity indices Rashad and Emad 623

General characteristics of obese patients stratified by Table 2 Laboratory and anthropometric parameters in fasting blood plasma glucose as diabetic and nondiabetic obese and type 2 diabetes mellitus obese nondiabetic obese patients patients’ groups We found statistically significant higher values of Obese Obese diabetic P value obesity indices BMI, WC, WHR, WHtR VAI, and nondiabetic patients (N=65) = BAI in obese T2DM patients than in nondiabetic patients (N 85) obese patients (P<0.001). Moreover, obese T2DM Age (years) 43.3±5.8 42.5±8.02 0.49 patients had statistically significant higher values of Sex (male/ 22/48 20/60 0.49 female) systolic blood pressure, diastolic blood pressure, TC, BMI (kg/m2) 35.3±8.01 33.78±6.25 0.009 LDL, FPG, FSI, HbA1c, and HOMA-IR. On the Waist 113.2±15.35 103.8±10.22 <0.001* contrary, obese T2DM patients had significantly lower circumference levels of HOMA-β than nondiabetic obese patients (cm) (P<0.001). Waist/hip ratio 1.02±0.15 1.079±0.19 0.050 Waist/height 0.67±0.09 0.6±0.065 <0.001* ratio Correlations between anthropometric measures Body adiposity 26.6±4.4 43.9±11.4 <0.001* and parameters of metabolic syndrome in obese index patients Visceral 6.98±1.28 8.74±0.96 <0.001* Our results showed significant positive correlations adiposity index between parameters of MetS, including WC, systolic Systolic blood 122.1±4.77 158.12±8.39 <0.001* pressure and diastolic blood pressure, low HDL as well as TG, (mmHg) and all anthropometric measures in obese cases (BMI, Diastolic blood 78.2±5.32 90.8±7.46 <0.001* WHR, WHtR, BAI, and VAI). Interestingly, among pressure obesity indices, the highest positive correlation was (mmHg) Total cholesterol 222.9±95.96 256.7±41.55 <0.001* found between VAI and parameters of MetS (mg/dl) (P<0.001) (Tables 2 and 3). Triglycerides 193.12±46.7 197.9±17.4 <0.359 (mg/dl) < Multiple stepwise linear regression analyses in obese LDL cholesterol 159.9±93.7 174.6±42.11 0.001* patients to assess the main independent parameters (mg/dl) associated with visceral adiposity index HDL cholesterol 37.4±8.94 35.9±9.02 0.324 (mg/dl) Stepwise linear regression analysis test revealed that Fasting blood 84.3±12.37 196.1±9.24 <0.001* VAI was independently correlated with TGs, HDL, glucose (mg/dl) fasting blood glucose, HOMA-IR, and HOMA-B Fasting serum 7.78±2.98 17.9±4.46 <0.001* (P<0.001) (Table 4). insulin (μIU/ml) HbA1c (%) 5.07±0.47 9.23±1.3 <0.001* HOMA-IR 2.9±0.45 5.9±0.67 <0.001* Multiple stepwise linear regression analyses in obese β < patients to assess the main independent parameters HOMA- 97.5±10.03 61.4±12.07 0.001* associated with body adiposity index Data are presented as mean±SD. HbA1c, glycated hemoglobin; Stepwise linear regression analysis test found that BAI HDL, high-density lipoprotein; HOMA-IR, homeostasis model assessments of insulin resistance; HOMA-β, an index of β-cell was independently correlated with fasting blood functions; LDL, low-density lipoprotein. *P value less than 0.05 glucose, BMI, total cholesterol TGs, and diastolic when compared with obese nondiabetic patients’ group. blood pressure (P<0.001) (Table 5). Accuracy of visceral adiposity index and BMI for discriminating type 2 diabetes risk among obese Logistic regression analysis evaluating the association patients by receiver operating characteristic analysis of BMI, visceral adiposity index, and body adiposity We further investigated the potential diagnostic value index with type 2 diabetes mellitus among obese patients of VAI and BMI by ROC curves, which are presented After adjusting for the traditional risk factors, logistic in Fig. 1. In obese patients, when we discriminate type regression analysis test was carried out to evaluate the 2 diabetes among nondiabetic patients, the cutoff predictor of T2DM among obese patients. VAI, values of BMI and VAI were 28.82 and 7.31, and = – BAI, and BMI were statistically significant the AUC values were 0.958 (95% CI 0.931 0.985) predictors of T2DM among obese patients and 0.952 (95% CI=0.923–0.980), respectively. In (P<0.001) (Table 6), whereas the odds ratio of addition, the sensitivity and the specificity of BMI VAI (odds ratio=3.498, 95% CI=1.987–6.158) was were 94.7 and 99.9% and those of VAI were 74.4 and higher than that of BAI and BMI (odds ratio=1.149, 99.9%, respectively. Thus VAI as well as BMI could be 95% CI=1.085–1.218; odds ratio=0.791, 95% useful diagnostic tests to discriminate T2DM from CI=0.728–0.860, respectively). nondiabetic obese patients. 624 The Egyptian Journal of Internal Medicine, Vol. 31 No. 4, October-December 2019

Table 3 Pearson’s correlation coefficient between anthropometric indices and parameters of metabolic syndrome among obese patients BMI (kg/m2) Waist/hip ratio Waist/height ratio BAI VAI Waist circumference (cm) r 0.747 0.121 0.983 0.125 0.228 P <0.001* 0.139 <0.001* 0.128 <0.001* Systolic blood pressure (mmHg) r 0.255 0.125 0.235 0.527 0.326 P <0.01* <0.062 0.004 <0.001* 0.062 Diastolic blood pressure (mmHg) r 0.140 0.122 0.103 0.139 0.471 P 0.089 0.138 0.212 <0.091 <0.001* Triglycerides (mg/dl) r 0.833 0.389 0.425 0.230 0.804 P <0.001* <0.001* <0.001* <0.01* <0.001* HDL cholesterol (mg/dl) r −0.383 −0.356 −0.020 −0.383 −0.592 P <0.001* <0.001* 0.811 <0.001* <0.001* Fasting blood glucose (mg/dl) r 0.355 0.362 0.144 0.539 0.243 P <0.001* <0.001* <0.080 <0.001* <0.01* BAI, body adiposity index; HDL, high-density lipoprotein; VAI, visceral adiposity index. *P value less than 0.05.

Table 4 Multiple stepwise linear regression analysis testing the influence of the main independent variables against visceral adiposity index (dependent variable) in obese women Model Unstandardized Standardized coefficients tPvalue 95% CI coefficients B SE B Lower bound Upper bound 1 Constant 0.985 0.459 2.148 <0.05* 0.079 1.892 Triglycerides (mg/dl) 0.039 0.002 0.804 16.417 <0.001* 0.034 0.044 2 Constant 5.067 0.669 7.578 <0.001* 3.745 6.388 Triglycerides (mg/dl) 0.033 0.002 0.678 15.052 <0.001* 0.028 0.037 HDL cholesterol (mg/dl) −0.062 0.008 0.339 −7.523 <0.001* −0.079 −0.046 3 Constant 3.335 0.617 5.402 <0.001* 2.115 4.555 Triglycerides (mg/dl) 0.034 0.002 0.708 18.298 <0.001* 0.031 0.038 HDL cholesterol (mg/dl) −0.056 0.007 0.306 −7.89 <0.001* −0.071 −0.042− Fasting blood glucose (mg/dl) 0.008 0.001 0.267 7.408 <0.001* 0.006 0.010 4 Constant 3.236 0.591 5.477 <0.001* 2.068 4.404 Triglycerides (mg/dl) 0.033 0.002 0.685 18.267 <0.001* 0.030 0.037 HDL cholesterol (mg/dl) −0.060 0.007 −0.326 −8.717 <0.001* −0.074 −0.046 Fasting blood glucose (mg/dl) 0.016 0.002 0.539 6.804 <0.001* 0.011 0.021 HOMA-IR 0.122− 0.032 0.306 3.818 <0.001* 0.185 −0.059 5 Constant 1.859 0.872 2.133 <0.05* 0.137 3.582 Triglycerides (mg/dl) 0.033 0.002 0.691 18.600 <0.001* 0.030 0.037 HDL cholesterol (mg/dl) −0.058 0.007 −0.315 −8.426 <0.001* −0.072 −0.044 Fasting blood glucose (mg/dl) 0.021 0.003 0.713 6.303 <0.001* 0.014 0.028 HOMA-IR 0.155 0.035 0.388 4.405 <0.001* 0.224 0.085 HOMA-B −0.008 0.004 −0.128 −2.127 <0.05* −0.001 0.015 CI, confidence interval; HDL, high-density lipoprotein; HOMA-IR, homeostasis model assessments of insulin resistance. *P value less than 0.05. Association of new obesity indices Rashad and Emad 625

Table 5 Multiple stepwise linear regression analysis testing the influence of the main independent variables against body adiposity index (dependent variable) in obese women Model Unstandardized Standardized coefficients tPvalue 95% CI coefficients β SE B Lower bound Upper bound 1 Constant 14.483 2.981 4.858 <0.001* 8.592 20.374 Fasting blood glucose (mg/dl) 0.146 0.019 0.539 7.749 <0.001* 0.109 0.183 2 Constant 3.573 5.272 0.678 0.499 13.993 6.847 Fasting blood glucose (mg/dl) 0.173 0.019 0.640 9.058 <0.001* 0.136 0.211 BMI (kg/m2) 0.376 0.093 0.287 4.062 <0.001* 0.193 0.559 3 Constant 4.224 5.205 0.812 0.418 14.511 6.063 Fasting blood glucose (mg/dl) 0.176 0.019 0.651 9.326 <0.001* 0.139 0.214 BMI (kg/m2) 0.837 0.221 0.639 3.793 <0.001* 0.401 1.273 Total cholesterol (mg/dl) 0.076 0.033 0.379 2.295 <0.05* 0142 0.011 4 Constant 25.692 7.310 3.514 <0.01* 40.141 11.242 Fasting blood glucose (mg/dl) 0.132 0.021 0.486 6.211 <0.001* 0.090 0.174 BMI (kg/m2) 0.749 0.211 0.572 3.545 <0.01* 0.331 1.166 Total cholesterol (mg/dl) 0.203 0.045 1.009 4.531 <0.001* 0.292 0.115 Triglycerides (mg/dl) 0.313 0.078 0.706 3.994 <0.001* 0.158 0.467 5 Constant 51.112 12.157 4.204 <0.001* 75.143 27.082 Fasting blood glucose (mg/dl) 0.086 0.027 0.317 3.135 <0.01* 0.032 0.140 BMI (kg/m2) 0.648 0.211 0.495 3.074 <0.01* 0.231 1.064 Total cholesterol (mg/dl) 0.209 0.044 1.037 4.741 <0.001* 0.296 0.122 Triglycerides (mg/dl) 0.351 0.078 0.793 4.491 <0.001* 0.197 0.506 Diastolic blood pressure (mmHg) 0.356 0.137 0.215 2.589 <0.05* 0.084 0.627 CI, confidence interval. *P value less than 0.05.

Table 6 Logistic regression analysis of body adiposity index, visceral adiposity index and BMI in type 2 diabetic versus nondiabetic obese women β SE tPvalue Odds ratio 95% CI Lower Upper Constant −6.363 1.812 12.331 <0.001* 0.002 VAI 1.252 0.289 18.835 <0.001* 3.498 1.987 6.158 BAI 0.139 0.029 22.282 <0.001* 1.149 1.085 1.218 BMI −0.234 0.042 30.566 <0.001* 0.791 0.728 0.860 BAI, body adiposity index; CI, confidence interval; VAI, visceral adiposity index. *P value less than 0.05.

Accuracy of body adiposity index and BMI for discriminating type 2 diabetes risk among obese Discussion patients by receiver operating characteristic analysis Obesity plays a key role in the pathophysiology of all We further investigated the potential diagnostic MetS parameters and can be considered as an value of BAI and BMI using ROC curves, which independent predictor of MetS [6]. Pradeep et al. are presented in Fig. 2. In obese patients, when we [20] suggested that central obesity as an indicator of discriminate type 2 diabetes among nondiabetic body fat can be easily and cost-effectively estimated by patients, the cutoff values of BMI and BAI were measuring BMI and WC, which might discriminate 28.82 and 27.58 and the AUC values were 0.958 MetS from non-MetS status. (95% CI=0.931–0.985) and 0.770 (95% CI=0.703–0.837), respectively. In addition, the BMI was not a good indicator of MetS risk, particularly sensitivity and the specificity of BMI were 94.7 and when it is the only indicator, because it is not able to 99.9% and those of BAI were 83.3 and 58%, differentiate between adipose and muscle tissue. Taken respectively. together, we investigated, in the present study, for the 626 The Egyptian Journal of Internal Medicine, Vol. 31 No. 4, October-December 2019

Figure 1 Figure 2

ROC curve for both BMI and VAI in discriminating obese women with ROC curve for both BMI and BAI in discriminating obese women with T2DM from those without diabetes. ROC, receiver operating char- T2DM from those without diabetes. BAI, body adiposity index; ROC, acteristic; T2DM, type 2 diabetes mellitus; VAI, visceral adiposity receiver operating characteristic; T2DM, type 2 diabetes mellitus. index. In the current study, we found a significant positive first time, the possible association of new obesity correlation between BAI and all parameters of MetS, indices, BAI and VAI, as well as traditional obesity except WC and diastolic blood pressure among obese indices (WC, BMI, WHtR) and parameters of MetS cases. in Egyptian obese patients. Similar to our results, Bergman et al. [16] reported that The results presented herein are innovative, as this BAI is a good tool to estimate adiposity in the study performs a robust evaluation of new obesity Caucasian population. Furthermore, they suggested indices VAI and BAI as diagnostic tests of MetS. that it is more practical and easier than other Noteworthy, our results confirmed that VAI and complex mechanical methods. BAI values were significantly higher in obese patients compared with the lean group. Moreover, in As regards traditional obesity indices (WC, BMI, and the T2DM group, the values of VAI and BAI were WHR), our results revealed that, among them, BMI higher compared with the nondiabetic group. had the best positive correlation with MetS parameters. Nonetheless, the new obesity indices had the best As regards correlation of anthropometric measures positive correlation with MetS parameters. with the parameters of MetS, we found significant positive correlations between parameters of MetS, Gharipour et al. [21] suggested that WC could be including WC, systolic and diastolic blood pressure, considered as a useful anthropometric assessment for low HDL, as well as TG, and all anthropometric prediction of T2DM and MetS. These results were in measures in obese cases (BMI, WHR, WHtR, BAI, accordance with two prospective studies showing that and VAI). Interestingly, among obesity indices, the WC and WHtR performed equally well in their ability highest positive correlation was found between VAI to predict T2DM in Pima Indians [22–24]. and parameters of MetS. In this study, we attempted to point out the association In agreement with our finding, Amato et al. [15], between MetS and WHtR and to compare between among an age-stratified Caucasian Sicilian WHtR and BMI. Our results revealed that BMI had a population, found that the cut-off points of VAI strong positive correlation with parameters of MetS were proved to be strongly associated with MetS; compared with WHtR. they explained that VAI represents physical (BMI and WC) and metabolic parameters (TG and HDL On the contrary, study by Hsieh et al. [25] found that cholesterol), and may indirectly reflect other WHtR was a practical and simple anthropometric nonclassical risk factors, that is, cytokines and measurement for identifying patients of both sexes plasma-free fatty acids. with higher metabolic risk. Association of new obesity indices Rashad and Emad 627

In contrast to our results, previous studies found that risk factors associated with the development of obesity- WHtR was the most predictive measure of T2DM, related comorbidities. Further future multicenter followed by BMI [26–30]. The diverse results studies with bigger sample size are needed to summarized above contributed to differences in race/ validate our findings. ethnicity. Financial support and sponsorship Although the BMI could be used as a diagnostic test of Nil. obesity and its related metabolic complications, we still indeed need other obesity indices to overcome the Conflicts of interest limitations of BMI with regard to measurement of There are no conflicts of interest. excess fat. Accordingly, we analyzed our data by ROC to estimate the sensitivity and specificity of VAI as well References as BAI. Our results detected that the cutoff values of 1 Kirkendoll K, Clark PC, Grossniklaus D, Igho-Pemu P, Mullis R, Dunbar SB. BMI, VAI, and BAI were 28.82, 7.31, and 27.58 and Metabolic syndrome in African Americans: views on making lifestyle – the AUC values were 0.958 (95% CI=0.931–0.985), changes. J Transcult Nurs 2010; 21:104 113. = – 2 Yamaoka K, Tango T. Effects of lifestyle modification on metabolic 0.952 (95% CI 0.923 0.980), and 0.770 (95% syndrome: a systematic review and meta-analysis. BMC Med 2012; CI=0.703–0.837), respectively. In addition, the 10:138–148. 3 Walther G, Obert P, Dutheil F, Chapier R, Lesourd B, Naughton G, et al. sensitivity and specificity of BMI were 94.7 and Metabolic syndrome individuals with and without type 2 diabetes mellitus 99.9%, those of VAI were 74.4 and 99.9%, and present generalized vascular dysfunction: cross-sectional study. those of BAI were 83.3 and 58%. Arterioscler Thromb Vasc Biol 2015; 35:1022–1029. 4 Anderson PJ, Critchley JA, Chan JC, Cockram CS, Lee ZS, Thomas GN, Tomlinson B. Factor analysis of the metabolic syndrome: obesity vs insulin The main finding in the current study is that BMI is resistance as the central abnormality. Int J Obes Relat Metab Disord 2001; 25:1782–1788. still the most powerful diagnostic tool for obesity. 5 Mottillo S, Filion KB, Genest J, Joseph L, Pilote L, Poirier P, et al. The Although, in certain conditions where there are metabolic syndrome and cardiovascular risk a systematic review and meta- limitations of using BMI, for example, heart failure analysis. J Am Coll Cardiol 2010; 56:1113–1132. 6 Kassi E, Pervanidou P, Kaltsas G, Chrousos G. Metabolic syndrome: and ascites, we can use other obesity indices; VAI and definitions and controversies. BMC Med 2011; 9: 48–61. BAI could be used to discriminate cardiovascular risk 7 Finucane MM, Stevens GA, Cowan MJ, Danaei G, Lin JK, Paciorek CJ, et among obese patients. Interestingly, the power of VAI al. National, regional, and global trends in body-mass index since 1980: systematic analysis of health examination surveys and epidemiological was sensitive and specific in parallel to BMI in the studies with 960 country-years and 9 · 1 million participants. Lancet 2011; diagnosis of metabolic risk among T2DM patients. 377:557–567. 8 Badran M, Laher I. Obesity in Arabic-speaking countries. J Obesity 2011. Our study explored that after adjusting for the 24:1–9. traditional risk factors, the logistic regression analysis 9 Frühbeck G. Obesity: screening for the evident in obesity. Nat Rev test found that the odds ratio of VAI value was the best Endocrinol 2012; 8:570–572. 10 Ng M, Fleming T, Robinson M, Thomson B, Graetz N, Margono C, et predictor of T2DM in comparison with BMI and BAI al. Global, regional, and national prevalence of overweight and among obese patients, P value less than 0.05. obesity in children and adults during 1980–2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet 2014; 384:766–781. et al. In a study by Amato [16] among an age-stratified 11 Jackson AS, Stanforth PR, Gagnon J, Rankinen T, Leon AS, Rao DC, et al. Caucasian Sicilian population, the cut-off points of The effect of sex, age and race on estimating percentage body fat from : The Heritage Family Study. Int J Obes Relat Metab VAI were proved to be strongly associated with MetS. Disord 2002; 26:789–796. 12 Cornier MA, Després JP, Davis N, Grossniklaus DA, Klein S, Lamarche B, et al. Assessing adiposity: a scientific statement from the American Heart On the contrary, Bennasar-Veny and colleagues Association. Circulation 2011; 124:1996–2019. reported that BAI does not overcome the limitations 13 InterAct C, Langenberg C, Sharp SJ, Schulze MB, Rolandsson O, Overved of BMI and that it is not a good tool to measure K, et al. Long-term risk of incident type 2 diabetes and measures of overall and regional obesity: the EPIC-InterAct case cohort study. PLoS Med 2012; metabolic risk in the Caucasian population. They 9:e1001230. concluded that BAI is less useful not only than BMI 14 Browning LM, Hsieh SD, Ashwell M. A systematic review of waist-to-height but also than other adiposity indexes such as WHtR, ratio as a screening tool for the prediction of cardiovascular disease and diabetes: 0.5 could be a suitable global boundary value. Nutr Res Rev WHR, and WC [31]. 2010; 23:247–269. 15 Amato MC, Giordano C, Galia M, Criscimanna A, Vitabile S, Midiri M, et al. Visceral adiposity index: a reliable indicator of visceral fat function associated with cardiometabolic risk. Diabetes Care 2010; Conclusion 33:920–922. In conclusion, unusual anthropometric measurements 16 Bergman RN, Stefanovski D, Buchanan TA, Sumner AE, Reynolds JC, such as BAI and VAI, which represent physical (BMI Sebring NG, et al. A better index of body adiposity. Obesity (Silver Spring) 2011; 19:1083–1089. and WC) and metabolic parameters (TG and HDL 17 Alberti KGM, Zimmet P, Shaw J, Group I. The metabolic syndrome − a new cholesterol), may indirectly reflect other nonclassical worldwide definition. Lancet 2005; 366:1059–1062. 628 The Egyptian Journal of Internal Medicine, Vol. 31 No. 4, October-December 2019

18 Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of 25 Hsieh SD, Yoshinaga H, Muto T. Waist-to-height ratio, a simple and low-density lipoprotein cholesterol in plasma, without use of the preparative practical index for assessing central fat distribution and metabolic risk in ultracentrifuge. Clin Chem 1972; 18:499–502. Japanese men and women. Int J Obes Relat Metab Disord 2003; – 19 Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner 27:610 616 RC. Homeostasis model assessment: insulin resistance and b-cell function 26 MacKay MF, Haffner SM, Wagenknecht LE, D’Agostino RB Jr, Hanley AJ. from fasting plasma glucose and insulin concentrations in man. Prediction of type 2 diabetes using alternate anthropometric measures in a Diabetologia 1985; 28:412–419. multi-ethnic cohort: the insulin resistance study. Diabetes – 20 Pradeep AR, Nagpal K, Karvekar S, Patnaik K. Levels of lipocalin-2 in Care 2009; 32:956 958. crevicular fluid and tear fluid in chronic periodontitis and obesity subjects. J 27 He YH, Chen YC, Jiang GX, Huang HE, Li R, Li XY. Evaluation of Investig Clin Dent 2015; 19:1–7. anthropometric indices for metabolic syndrome in Chinese adults aged – 21 Gharipour M, Sarrafzadegan N, Sadeghi M, Andalib E, Talaie M, Shafie D. 40 years and over. Eur J Nutr 2012; 51:81 87. Predictors of metabolic syndrome in the Iranian population: waist circumference, 28 Bosy-Westphal A, Geisler C, Onur S, Korth O, Selberg O, Schrezenmeir body mass index, or waist to hip ratio? Cholesterol 2013; 24:2013. J. Value of body fat mass vs anthropometric obesity indices in the 22 Beydoun MA, Kuczmarski MT, Wang Y, Mason MA, Evans MK, Zonderman assessment of metabolic risk factors. Int J Obes (Lond) 2006; – AB. Receiver-operating characteristics of adiposity for metabolic 30:475 483. syndrome: the Healthy Aging in Neighborhoods of Diversity across the 29 Sayeed M, Mahtab H, Latif Z, Khanam P, Ahsan K, Banu A, et al. Waist-to- Life Span (HANDLS) study. Public Health Nutr 2011; 14:77–92. height ratio is a better obesity index than body mass index and waist-to-hip 23 Tulloch-Reid MK, Williams DE, Looker HC, Hanson RL, Knowler WC. Do ratio for predicting diabetes, hypertension and lipidemia. Bangladesh Med measures of body fat distribution provide information on the risk of type 2 Res Counc Bull 2003; 29:1. diabetes in addition to measures of general obesity? Comparison of 30 Kahn HS. Choosing an index for abdominal obesity: an opportunity for anthropometric predictors of type 2 diabetes in Pima Indians. Diabetes epidemiologic clarification. J Clin Epidemiol 1993; 46:491–494. – Care 2003; 26:2556 2561. 31 Bennasar-Veny M, Lopez-Gonzalez AA, Tauler P, Cespedes ML, Vicente- 24 Warne DK, Charles MA, Hanson RL, Jacobsson LT, McCance DR, Knowler Herrero T, Yañez A, et al. Body adiposity index and cardiovascular health WC. Comparison of body size measurements as predictors of NIDDM in risk factors in Caucasians: a comparison with the body mass index and Pima Indians. Diabetes Care 1995; 18:435–439. others. PLoS One 2013; 8:e63999.