Visceral fat accumulation is more strongly related to insulin resistance than to excess total adiposity. The visceral adiposity index (VAI) has recently been suggested as an indicator of the visceral adiposity measured by magnetic resonance imaging. To evaluate whether the VAI could replace visceral computed tomography (CT) scanning and predict insulin resistance in young women with polycystic ovary syndrome (PCOS) was presented.
Design and Methods
One hundred and eighty Korean women aged 16-41 years who were diagnosed with PCOS using the Rotterdam criteria were included. The VAI was derived from a formula using body mass index, waist circumference, triglycerides, and high-density lipoprotein cholesterol. Visceral adiposity was defined as a visceral fat area (VFA) measuring > 100 cm2 by CT scanning. Insulin sensitivity was evaluated by insulin-mediated glucose utilization (M value), which was obtained using a euglycemic hyperinsulinemic clamp.
The VAI positively correlated with VFA, the visceral-to-subcutaneous fat ratio, and systolic and diastolic blood pressure, and negatively correlated with the M value. In a linear regression analysis, the VAI was an independent determinant of the insulin sensitivity after controlling for age, systolic blood pressure, fasting glucose, fasting insulin, and testosterone levels. In a logistic regression analysis, the VAI odds ratio was 3.5 (95% CI 1.2-9.8) for predicting visceral adiposity after controlling for the various metabolic parameters and testosterone.
The VAI can replace visceral CT scanning as a marker for visceral adiposity, and it predicts insulin resistance in young women with PCOS.
Polycystic ovary syndrome (PCOS) is one of the most common endocrine disorders, affecting 5-10% of the population in women of reproductive age ([1, 2]). It is characterized by hyperandrogenemia, chronic anovulation, and polycystic ovary morphology. Insulin resistance is a key pathophysiologic feature relating to the metabolic dysfunction associated with PCOS, and women with PCOS are more likely to have central obesity, glucose intolerance or type 2 diabetes, dyslipidemia, and hypertension than normally cycling women ().
Central obesity, as indicated by an increased waist circumference (WC) and waist-to-hip ratio, has been adopted as a more accurate predictor of obesity-related metabolic abnormalities and has replaced body mass index (BMI) for defining a clinical diagnosis of metabolic syndrome (). In the presence of an elevated WC, elevated fasting triglycerides (TG) reflect the status of an individual's relative inability to manage and store extra energy in subcutaneous fat depots. Hypertriglyceridemic waist is useful to identify individuals with atherogenic metabolic triad, e.g., fasting hyperinsulinemia, elevated apolipoprotein B, and increased proportion of small dense low-density lipoprotein (). Therefore, hypertriglyceridemic waist phenotype is understood to reflect visceral adiposity and is related to various metabolic abnormalities. As such, it has been proposed as an alternative measure of excessive lipid accumulation and as a better marker for identifying diabetes or cardiovascular risk than BMI or WC in several studies ().
Recently, the visceral adiposity index (VAI), a mathematical model that uses simple anthropometric (BMI and WC) and functional (TG and high-density lipoprotein (HDL) cholesterol) parameters, has been proposed to reflect visceral adiposity and insulin resistance ().
Increased visceral adiposity is characteristic in PCOS and plays a major role in the development of hyperandrogenism via insulin resistance and compensatory systemic hyperinsulinism (). Visceral adiposity and PCOS interact to induce premature atherosclerosis and increased cardiovascular mortality by mechanisms that include low-grade chronic inflammation, local metabolism of sex steroids and cortisol in visceral fat, and the secretion of adipokines such as leptin, which exert direct effects on the ovary ([3, 13, 14]). The Androgen Excess and PCOS (AE-PCOS) Society has recommended the assessment of cardiometabolic risk in women with PCOS in clinical practice (). Therefore, the assessment of visceral adiposity should be performed independent of overall obesity to predict cardiovascular morbidity and plan the therapeutic intervention for PCOS.
The International Diabetes Federation recommends the assessment of visceral fat accumulation using magnetic resonance imaging (MRI) or computed tomography (CT) scanning (). A visceral fat area (VFA) of 100 cm2 was the optimal cutoff value for identifying individuals at risk for obesity-related disorders in several ethnic groups ([17, 18]). However, because performing MRIs and CT scans is extremely expensive, complicated, and poses a radiation hazard, their use cannot be recommended for routine clinical practice. A limited number of studies have observed the clinical implication of VAI as a predictive marker for cardiovascular risk, and only one study examined women with PCOS ().
This study aimed to evaluate whether the VAI could replace visceral CT scanning and predict insulin resistance in young Korean women with PCOS. We also aimed to determine an optimal VAI cutoff value for visceral adiposity in this study population.
Methods and Procedures
We enrolled 180 women with PCOS who attended an outpatient endocrinology or gynecology clinic for menstrual abnormality or infertility between March 2003 and June 2006. The PCOS diagnosis was based on the Rotterdam PCOS consensus criteria put forth by the European Society of Human Reproduction and Embryology (), which are as follows: (1) amenorrhea or oligomenorrhea (<10 menstrual cycles per year), (2) clinical or biochemical hyperandrogenism, and (3) polycystic ovary morphology observed during an ultrasound. A PCOS diagnosis excluded hyperandrogenemia, such as congenial adrenal hyperplasia, hyperprolactinemia, or androgen-secreting neoplasia. Patients with 21-hydroxylase-deficient nonclassic adrenal hyperplasia were excluded using a basal morning 17-hydroxyprogesterone level cutoff of > 2 ng/mL.
All medications known to affect sex hormone metabolism, insulin action, or kinetics were discontinued for at least 3 months before study enrollment. None of the patients were taking antihypertensive, lipid lowering, or hypoglycemic medications. Studies were performed on the second to the fifth day of the menstrual cycle in PCOS women with spontaneous menses and on arbitrary days in PCOS women with amenorrhea.
Subjects' weight and height were measured wearing light clothing and without shoes; BMI was also calculated (kg/m2). WC was measured to the nearest 0.1 cm on bare skin during mid respiration at the narrowest indentation between the 10th rib and the iliac crest.
Non-contrast CT scans were performed in the supine position. CT imaging was obtained using a CT Highlight Advantage (General Electric CO., USA) with a 5-mm thickness slice. A single-slice CT scan of the abdomen was performed at the level of the umbilicus and analyzed for a cross-sectional area of adipose tissue assuming a density of −150 to −50 Hounsfield Units as described by Tokunaga et al. (). The VFA and subcutaneous fat area (SFA) were measured, and the visceral-to-subcutaneous fat ratio (VSR) was calculated. Visceral adiposity was defined as a VFA > 100 cm2.
A standard 75-g oral glucose tolerance test was performed to evaluate glucose tolerance status. Euglycemic hyperinsulinemic clamp tests were also performed to assess insulin sensitivity using the protocol described by DeFronzo et al. (). After a 10-min insulin prime, a constant insulin infusion was performed at a rate of 3 μU/kg/min. During the clamp period, the plasma glucose concentration was maintained at 90 mg/dL by monitoring the glucose level at 5-min intervals. Quantitative estimates of insulin sensitivity, the glucose disposal rate, and insulin-mediated peripheral glucose utilization (M value) were provided by the mean glucose infusion rate (mg/kg/min) in the last 15-min of the 2-h euglycemic hyperinsulinemic clamp study.
Total testosterone was measured using commercial RIA kits (Diagnostic Products Co., Los Angeles, CA), and plasma sex hormone-binding globulin (SHBG) was measured using a specific immunoradiometric assay.
The VAI was calculated using the formula [WC/(36.58 + (1.89 × BMI))] × (TG/0.81) × (1.52/HDL), where the TG and HDL concentrations are expressed in mmol/L as described by Amato et al. ().
Data analysis was performed using SAS version 9.1 (SAS Institute, Cary, NC). All data were expressed as the mean ± SD. The age-adjusted partial correlation coefficients were calculated to determine the strength of the associations. Multiple logistic regression analyses were used to assess the association of VAI and the presence of visceral adiposity, with a VFA > 100 cm2 defined as the outcome variable. Continuous values of age, systolic blood pressure, fasting glucose, total testosterone, and the SHBG level were used as covariates. Because the VAI contains the BMI, WC, and the HDL cholesterol and TG levels, multicolinearity can affect the results when combining these continuous variables with VAI as covariates. Therefore, we created categorical variables including obesity (BMI ≥ 25 kg/m2), central obesity (WC ≥ 80 cm), and dyslipidemia (TG ≥ 150 mg/dL, HDL cholesterol < 50 mg/dL) on the basis of the International Diabetes Federation criteria. A multiple linear regression analysis was used to determine the independent association between the insulin sensitivity (M value) and the VAI after controlling for age, systolic blood pressure, fasting glucose, fasting insulin, testosterone, and the SHBG level. To compare the VAI with other anthropometric measures, the multiple linear regression analysis was repeated with BMI and WC in place of the VAI after controlling for the same covariates.
Because plasma insulin, VAI, TG, and SHBG showed slightly skewed distributions, P values were based on logarithmic data. However, the presented mean values are untransformed data. The VAI cutoff for predicting visceral adiposity was determined using a receiver-operating characteristic (ROC) curve. All P values were two tailed, and statistical significance was defined as a P < 0.05.
Statement of ethics
The institutional review board of Ewha Womans University, Mokdong Hospital approved the study. Informed consent was obtained from all participants. We certify that all applicable institutional and governmental regulations concerning the ethical use of human volunteers were followed during this research.
Table 1 shows the clinical characteristics of 180 study subjects. The mean age was 26 ± 5 years (16-41 years), the BMI was 23.0 ± 4.3 kg/m2 (16.5-39.4 kg/m2), the WC was 74.8 ± 11.0 cm (58.5-115.0 cm), and the VAI was 1.50 ± 1.89 (0.18-14.49).
The VAI showed a significant linear correlation between BMI (r = 0.52, P < 0.0001) and WC (r = 0.51, P < 0.0001). BMI and WC were more strongly correlated with each other than with VAI (r = 0.92, P < 0.0001). Table 2 shows the age-adjusted partial correlation coefficients for VAI, BMI, and WC with metabolic parameters including visceral adiposity and insulin sensitivity. The VAI positively correlated with systolic blood pressure (r = 0.42, P = 0.029), diastolic blood pressure (r = 0.45, P = 0.020), VFA (r = 0.57, P = 0.002), and the VSR (r = 0.60, P = 0.0009). The VAI negatively correlated with the M value (r = −0.55, P = 0.0033) and SHBG (r = −0.46, P = 0.017). BMI positively correlated with VFA (r = 0.80, P < 0.0001) and SFA (r = 0.80, P < 0.0001), and it negatively correlated with the M value (r = −0.47, P = 0.013) and SHBG (r = −0.65, P = 0.0003). WC positively correlated with VFA (r = 0.54, P = 0.0036) and SFA (r = 0.73, P < 0.0001).
The results of the multiple logistic regression analysis for predicting visceral adiposity are shown in Table 3. The VAI odds ratio (OR) was 3.5 (95% CI 1.2-9.8) after adjusting for age, SBP, FPG, testosterone, and SHBG as continuous variables and BMI, WC, TG, and HDL as categorical variables.
Table 3. Association of visceral adiposity (defined as VFA > 100 cm2) with VAI by a multiple logistic regression analysis
Table 4 presents the multiple linear regression analysis for the association between the insulin sensitivity (M value) and the VAI. After controlling for age, systolic blood pressure, fasting glucose, fasting insulin, total testosterone, and SHBG, the VAI predicted insulin sensitivity (β = −0.93, P < 0.0001). When we repeated the multiple linear regression after controlling for age, systolic blood pressure, fasting glucose, fasting insulin, total testosterone, and SHBG, the BMI and WC were not associated with the M value (data not shown).
Table 4. Multiple linear regression analyses for the insulin sensitivity (M value) with the VAI
Model 1 includes age; model 2 includes age, systolic blood pressure, fasting plasma glucose, and fasting plasma insulin; model 3 includes age, systolic blood pressure, fasting plasma glucose, fasting plasma insulin, total testosterone, and SHBG. VAI, fasting plasma insulin, and SHBG were log transformed for this analysis.
The optimal VAI cutoff point to predict visceral adiposity was 1.79 (82.6% sensitivity, 84.7% specificity), and the area under the ROC curve was 0.88 (95% CI 0.81-0.95, Figure 1).
In this study, the VAI was independently associated with visceral adiposity (defined as VFA > 100 cm2 when measured by CT scanning), suggesting that the VAI could be a useful substitute for visceral CT. Women with PCOS have excessive visceral fat accumulation, even if they are of normal weight, and visceral adiposity is associated with more pronounced hyperandrogenism and insulin resistance than subcutaneous or peripheral fat accumulation ([12, 23, 24]). Hyperandrogenism and hyperinsulinemia can interact and trigger metabolic imbalances, resulting in cardiovascular disease (CVD). Visceral adiposity is closely related to CVD via increased adipokine production, proinflammatory activity, and the deterioration of insulin sensitivity (). Therefore, the assessment of visceral adiposity is critical to identify subjects who are at risk for CVD.
Because visceral adipose tissue and not subcutaneous tissue plays a decisive role in the development of CVD, distinguishing visceral adiposity from central or abdominal obesity is critical. Unger and Scherer proposed defining hypoleptinemia as a state of visceral adiposity and “high TG/low HDL cholesterol dyslipidemia,” which were not observed in generalized obesity (). In our study, both BMI and WC strongly correlated with VFA and SFA but not with the VSR, suggesting that these two measurements might reflect total fat accumulation but not visceral adiposity.
We observed that the VAI was an independent determinant for the insulin sensitivity obtained using a euglycemic hyperinsulinemic clamp, the gold standard for assessing insulin resistance. When we used the calculated levels of free testosterone instead of total testosterone and SHBG, free testosterone was also significantly associated with the M value (P < 0.0001) and the VAI (P = 0.0033, data not shown). PCOS women with visceral adiposity present a higher prevalence of menstrual abnormalities and hirsutism, profound alterations of both production and metabolic clearance rates of androgens, and reduced SHBG levels (). At the level of visceral depots, testosterone stimulates lipolysis, increases free fatty acid efflux, and worsens the insulin resistance state (). Eventually, visceral adiposity can have a large effect on altered androgen metabolism and insulin resistance (). However, the role of testosterone in the body composition regulation in women remains undefined. In women with PCOS, additional confounding factors such as hyperinsulinemia and obesity might complicate the ability to regulate physiologic and metabolic consequences of testosterone. A study from healthy, nonobese, premenopausal women without hyperandrogenemia or hyperinsulinemia showed that higher testosterone within normal physiologic range was not related to abdominal adiposity or other metabolic parameters ().
Because we observed the relationship between VAI, hyperandrogenemia, and insulin sensitivity, it can support the visceral adiposity-hyperandrogenism-insulin resistance relationship theory in women with PCOS. For future therapeutic strategies to reduce hyperandrogenism and insulin resistance, it may be important to identify subjects with visceral adiposity independent of overall obesity. We also suggest that a measure of visceral adiposity may be useful in designing future studies better to address the role of fat depots in androgen excess.
Although VAI is simply obtained and well correlated with visceral fat accumulation, there is no definite value with which to diagnose visceral adiposity. In this study, we first provided the optimal VAI cutoff point for visceral adiposity (defined as VFA greater than 100 cm2 by CT scanning) in young Korean women with PCOS and this value was 1.79. Amato et al. reported age-stratified VAI cutoff points ranged from 1.92 to 2.52 to detect the metabolic syndrome in the male and female Caucasian Sicilian population aged 16-99 years old (). The optimal VAI cutoff points were 2.52 (age < 30 years), 2.23 (age 30-41 years), 1.92 (age 42-51 years), 1.93 (age 52-65 years), and 2.00 (age ≥66 years).
In summary, we observed that only the VAI correlated with the VSR and that the VAI was an independent determinant of visceral adiposity (VFA 100 cm2 by CT scanning). These data suggest that the VAI can replace visceral CT scanning with the advantages of a reduced economic burden and a reduced radiation hazard. The VAI also predicted insulin resistance measured using the euglycemic hyperinsulinemic clamp after controlling for metabolic and hyperandrogenemic disturbances. The optimal cutoff point for visceral adiposity was 1.79 (82.6% sensitivity, 84.7% specificity). However, it is necessary to identify the age- and sex-specific cutoff points in the general population as well as in women with PCOS for early diagnosis and individualized therapeutic programs in persons at risk for CVD.