Optimal Waist Circumference Cut-Off Point for Multiple Risk Factor Aggregation : Results from the Maracaibo City Metabolic Syndrome Prevalence Study

1Endocrine and Metabolic Diseases Research Center, The University of Zulia, 20th Avenue, Maracaibo 4004, Venezuela 2Clinical Pharmacology Unit, José Maŕıa Vargas School of Medicine, Central University of Venezuela, Caracas 1051, Venezuela 3Lipid and Atherosclerosis Unit, Department of Medicine, Carlos III Institute of Health, IMIBIC/Reina Sofia University Hospital/University of Córdoba and CIBER Obesity and Nutrition Physiopathology (CIBEROBN), Córdoba 230002, Spain


Introduction
Metabolic Syndrome (MS) is a cluster of endocrine-metabolic disturbances which increases the risk of developing Cardiovascular Disease (CVD) and type 2 diabetes mellitus (T2DM) [1].In recent years its prevalence has increased worldwide, especially in Latin America where it has become a public health issue, fueled by the sociocultural backdrop and psychobiologic habits pertaining to this region [2,3].In an attempt to "harmonize" MS diagnostic criteria, organizations such as the International Diabetes Federation, National Heart, Lung, Blood Institute, American Heart Association, and others (IDF/NHLBI/AHA-2009) agreed on an integrated set of diagnostic criteria that may be widely utilized worldwide, and whose clinical application would allow for the identification of subjects in risk of further cardiometabolic disease [4].
Despite efforts to standardize criteria for this entity, waist circumference (WC) is still the most controversial variable in the MS definition due to the absence of specific cut-off points for Latin-Americans, Sub-Saharan Africans, Eastern Mediterranean, and Middle East (Arab) populations.Therefore, when the International Diabetes Federation exposed their criteria for MS diagnosis in 2005 (IDF-2005) [5], they suggested that when using this criterion in South American countries, researchers should rely on cut-off points from the South Asia (90 cm for men and 80 cm for women) until more data would be obtained from such countries.Such point of view is maintained in the IDF/NHLBI/AHA-2009, supporting the notion that these WC references might not reflect the real epidemiologic behavior of obesity.
Given the latest recommendations by the IDF/NHLBI/ AHA-2009 consensus, several research groups have started to select population-specific WC cutpoints that would properly aid in the detection of cardiometabolic disease, such as the work from Mansour et al. [6] conducted in a rural Iraqi setting (WC in women: 99 cm and in men: 97 cm), Delavari et al. [7] in Iran (WC in women: 89 cm and in men: 91 cm), and Motala et al. [8] in a rural community from South Africa (WC in women: 86 cm and in men: 92 cm).The overall approach [6][7][8] was the application of ROC curve construction to determine cutpoints for WC and metabolic component aggregation in order to define individuals who are healthy or sick.
We previously investigated the prevalence of obesity using Body Mass Index (BMI) and WC as surrogates for both, total and abdominal adiposity [9].The results were quite alarming with 33.3% obesity and 34.8% overweight when BMI classification was used, and 74.2% obesity when applying the IDF/NHLBI/AHA-2009 criteria, suggesting the possibility that the latest consensus might be associated with overestimation of obesity in our population, with particular ethnic, genetic background and sociocultural traits which would influence such prevalence [9].Consequently, due to the lack of properly determined and populationspecific WC cutpoints in Venezuela, we decided to determine gender-specific WC cutpoints to diagnose abdominal obesity and enhance accuracy of MS identification in an obesityburdened locality such as ours.[10] is a cross-sectional study undertaken in the city of Maracaibo-Venezuela, the second largest city in the country with 2.500.000inhabitants, with the purpose of identifying and analyzing MS and cardiovascular risk factors in the adult population of Maracaibo.A total of 2,230 subjects were enrolled, out of which 1,902 were selected after the exclusion of individuals without serum insulin levels determination, and those with a previous diagnosis of T2DM because current pharmacological and nonpharmacological intervention would alter the variables evaluated in this investigation.The study was approved by the Bioethics Committee of the Endocrine and Metabolic Diseases Research Center, University of Zulia, and all participants signed a written consent before being interrogated and physically examined by a trained team.

Clinical Analysis.
All participants were subject to a complete physical examination, physical evaluation, and anthropometric measurements.The information that was obtained from anamnesis included socioeconomic data, education and working status, ethnicity, and smoking and drinking habit.Assessment of blood pressure was done using a calibrated mercury sphygmomanometer, with patients previously rested (during at least 15 minutes) in a sitting position with both feet touching the floor; the arm was positioned at heart level; a proper sized cuff was used for the procedure.WC was evaluated using calibrated measuring tapes in accordance with the USA National Institutes of Health protocol [11].For the quantification of Body Mass Index (BMI), the [weight/height 2 ] formula was utilized and results were expressed in kg/m 2 [12].

Laboratory
Analysis.An antecubital blood sample was taken following an overnight fasting of 8-12 hours in order to measure the following: fasting blood glucose, total cholesterol, triacylglycerides (TAG), and HDL-C with an enzymatic colorimetric technique and a computerized analyzer (Human, Magdeburg, Germany).Serum fasting insulin was determined using an ultrasensitive ELISA double-sandwich method (DRG Instruments GmbH, Germany, Inc).Insulin resistance (HOMA2-IR index) was calculated using the software HOMA Calculator v2.2.2 supplied by the Oxford Center for Diabetes Endocrinology and Metabolism available at https://www.dtu.ox.ac.uk/homacalculator/download.php which requires fasting insulin and glycemia for proper program calculation.

Clinical Definitions.
Presence of insulin resistance (IR) in our population was defined as HOMA2-IR ≥ 2 [13].Diagnosis of MS was done using the IDF/NHLBI/AHA-2009 consensus [4].Multiple risk factor aggregation (MRFA) was defined as ≥2 of the following: high blood pressure (Systolic Blood Pressure ≥ 130 mmHg and/or Diastolic Blood Pressure ≥ 85 mmHg; or history of antihypertensive usage), impaired fasting glucose (≥100 mg/dL or presence of diagnosis of T2DM), high TAG (≥150 mg/dL or presence of treatment for this disorder), and low HDL-C (<40 mg/dL in males and <50 mg/dL in females or presence of treatment for this disorder), similar to the definition used in the South African community by Motala et al. [8] and the state of risk factor clustering proposed by Hara et al. [14].

Statistical Analysis. Statistical analysis was done using the Statistical Package for the Social Sciences (SPSS) v19
for Windows (IBM Inc.Chicago, IL).Normal distribution of continuous variables was evaluated by using Geary's test.For normally distributed variables, the results were expressed as arithmetic mean ± SD.Variables without a normal distribution were logarithmically transformed and normal distribution later corroborated.Differences between arithmetic means were assessed using Student's -test (when two groups were compared) or ANOVA (when three or more groups were compared).Qualitative variables were expressed as absolute and relative frequencies, while  2 test was applied for association determination, considering the results statistically significant when  < 0.05.
The Receiving Operating Characteristic (ROC) curves were constructed to expect presence of MRFA or ≥2 metabolic components for MS [4], with the exception of WC.ROC curves were plotted for each gender using R Project software for Statistical Computing.Several indexes were calculated to assess the optimal cut-off point on the curve,

Results
A total of 1,902 subjects were studied: 52.2% females and 47.8% males.The arithmetic mean of age was 38.7 ± 15.0 years; whereas the mean WC was 90.7 ± 13.7 cm in females, and 98.2 ± 15.9 cm in males,  < 0.0001.The prevalence of MS in the general population was 39.7%, with 36.8% among females and 42.9% among males.The prevalence of MRFA (≥2 MS factors) was 43.3%, with 39.8% in females and 47.1% in males.The most prevalent risk factors were abdominal obesity (74.5%) and low HDL-C levels (56.5%), seemly more frequent in the female group; the rest of the metabolic features of the population sample are described in Table 1.
On the other hand, clinical characteristics and biochemical results are observed in Table 2. ROC curves were plotted for both genders to determine the cut-off values of WC in relation to MRFA (Figure 1).According to the ROC curves, the cut-offs yielding the best predictive values for such status was 90.25 cm for women, with 68.4% sensitivity and 65.8% specificity (AUC: 0.716; CI 95%, 0.684-0.748),and 95.15 cm, with 71.1% sensitivity and 67.4% specificity (AUC: 0.746; CI 95%, 0.715-0.778)for men.Likewise, Table 3 shows the cut-off points for each ethnic group in our sample, while abdominal obesity prevalence comparison according to IDF/NHLBI/AHA-2009 and our cutpoints are observed in   Youden Index = Sensitivity + Specificity − 1. LR+ = (positive likelihood ratio).BMI = Body Mass Index.* One way-ANOVA = 0.076.* * One way-ANOVA = 0.011: Post Hoc Tukey: mixed race versus Amerindian:  = 0.018; Hispanic whites versus Amerindian:  = 0.015.
Figure 2, with a resulting decline in percentages obtained in the overall sample and according to sex ( < 0.05).
The main metabolic risk factors associated with a high WC are shown in Table 4, being IR the variable with the greatest association (OR: 3.10; CI 95%: 2.37-3.83; < 0.01), followed by high blood pressure (OR: 2.38; CI 95%: 1.87-3.05; < 0.01).Table 5 displays the results for MRFA using the proposed WC cut-off points.The results showed that subjects with obesity determined by our WC cut-off points have 2.5 times more risk for metabolic risk aggregation (OR 2.56; CI 95%: 2.05-3.20; < 0.01).This risk is slightly higher than that obtained with IR (OR 2.30; CI 95%:

Discussion
The purpose of this investigation was to evaluate metabolic components and determine appropriate WC cutpoints that are well-matched for this particular population, which has been described as obesity-prone [9] and highly sedentary [18], with high prevalence of hypertension [19] and low grade inflammation [20].The results demonstrate that IDF/NHLBI/AHA-2009 WC cut-off points overestimate obesity and that the obtained local WC cutpoints are more suited for obesity diagnosis and MRFA.Reports of an alarming rise on MS have been published in Latin America in the last decade [2,3,21].When analyzing the epidemiologic behavior of its components, low HDL-C and abdominal obesity are common results in regional studies [3,9].Although the former has been linked to a plausible "American" genetic component [22], the great prevalence of central obesity may be distorted by the application of nonethnic-specific cut-offs points to our population, such as those proposed by IDF using unclear methodology from Sub-Asian demographics and information [4,5].
The cut-off points in this report are higher than those suggested in the latest MS consensus (90 cm for men and 80 cm for women) and are comparable to those obtained by Aschner et al. [23] in 457 individuals from diverse Latin American cities, as ascertained by measurement of visceral adipose tissue through imaging techniques; they reported a cut-off point of 94 cm for men and 90-92 cm for women [23].In this regard, other South American countries, such as Colombia and Cuba, have published their findings.Gallo et al. [24] determined a WC cut-off point that could discriminate the presence of IR in a Colombian population (using ROC curves and Youden Index for cut-off point selection); they proposed a cut-off point of 92 cm for men (sensitivity 82.2% and specificity 70.14%) and 84 cm for women (sensitivity 78.15% and specificity 73.98%).On the other hand, Popa et al. [25] use individuals from the Cuban region of Holguín-Granma, selecting a WC cut-off point that could determine the presence of 2 or more MS components (also applying ROC curves and Youden Index as selection determinant); they published a cut-off point of 97 cm for men (sensitivity 55% and specificity 91%) and 85 cm for women (sensitivity 87% and specificity 72%) [25].
The significance of determining WC cut-off points according to local sociocultural characteristics is so important that differences in this anthropometrical parameter are not only observed within Latin-American populations, but also between the different ethnic groups in this study [26], demonstrating that several factors can influence the presence of obesity in our sample, and also in our country [26,27].Nevertheless, in spite of the differences observed between cutpoints in each ethnic group, especially between Afro-Venezuelans and Amerindians which show the lowest prevalence of obesity in our city, they represent the minority groups in our population and their results do not modify the results obtained in the general population.Consequently, ethnicspecific cut-off points need to be determined in such groups, as well as the description of genetic and environmental factors that could influence the observed differences with studies focused on these minority groups.
There are advantages to using our new cut-offs to define abdominal obesity.First off, we appropriately identify subjects with obesity, which seems to be overdiagnosed in our community when using previously established cutpoints [9] (Figure 2).Secondly, their association with all components of MS in our population as assessed by multivariate analysis aids in the evaluation of cardiometabolic risk, especially with HBP as previously described by Olinto et al. [28] in Brazilian women.The proposed mechanisms in this relationship are IR and chronic hyperinsulinemia, acting as principal mediators [29].However, in our study this relation is independent of IR, which coincides with the Olivetti Heart Study [30], a research that suggests that the neuroendocrine abnormalities observed in the sympathetic nervous system are the link between abdominal obesity and elevated high blood pressure.Moreover, in the last decade, new molecules such as adiponectin and leptin have been associated with hypertension, yet more research is needed to truly understand their association with other major players, such as low grade inflammation in obesity-induced hypertension [31].Nevertheless, IR is the variable typically related to our cutpoints for WC, not only as a component independently related to central obesity but also as a risk factor for the aggregation of MS components, associated with similar likelihood to that entailed by high WC values in this aspect.This harmonizes with the proposal of this anthropometric parameter as a surrogate for the evaluation of IR [32] and it differs with the results published by Onat et al. [33] in 1,534 Turkmen, observing lack of accordance between the presence of insulin resistance and components of MS.
In regards to aging, several reports have suggested that MS prevalence increases with age, reaches a plateau, and decreases in frequency as sarcopenia and low fat mass develop [34,35].However, this phenomenon is not observed in our city, where a steady increase in MS diagnosis is observed in the general population and according to gender [36], coinciding with the proposal that there is a close relationship between age and MS development, associated with defective insulin signaling, hormonal disorders, mitochondrial and endothelial dysfunction, oxidative stress and dysregulation of renin-angiotensin-aldosterone system, and major disorders that seem to coincide and converge with aging [37,38].
It is important to note the limitations ensued by the cross-sectional design of this study, which is why further longitudinal research is necessary to confirm these findings.Moreover, future analyses from our research group will refine the selection methods of metabolic health/sick subjects through data mining techniques that include metabolic and inflammation biomarkers, instead of classic MS components, hypothesizing an enhanced curve plotting with much better sensitivity and specificity.Moreover, the application of visceral adipose index offers a potential alternative to assessment of WC in communities where other imaging methods (such as ultrasound and axial computer tomography) are not available at larger scales.
In conclusion, given the importance of determining population-specific cut-offs, we suggest a WC cut-off point of 90 cm for women and 95 cm for men.As shown in the multivariate analysis, these cut-off points are associated with multiple risk aggregation and presence of IR, serving as

Figure 1 :
Figure 1: ROC curves constructed to determine waist circumference cut-off points for the detection of multiple risk factor aggregation in the general population according to gender, Maracaibo, 2014.

Table 1 :
General characteristics of the general population according to gender, Maracaibo, 2014.

Table 3 :
Waist circumference cutpoints for abdominal obesity and BMI according to ethnic groups, Maracaibo, 2014.
a Confidence interval (95%).b Significance Level.c Adjusted for gender, age groups, ethnic groups, high blood pressure, impaired fasting glucose, high triacylglycerides, low HDL-C, and insulin resistance.d Criteria from the IDF/AHA/NHLBI-2009 consensus.e Estimated through the HOMA2-IR model; cut-off value: ≥2.

Table 5 :
Logistic regression model for multiple risk factor aggregation † , Maracaibo, 2014.Adjusted for gender, age groups, ethnic groups, high waist circumference, and insulin resistance.d Defined as ≥90.25 for females and ≥95.15 cm for males.