The discriminatory power of visceral adipose tissue area vs anthropometric measures as a diagnostic marker for metabolic syndrome in South African women

The International Diabetes Federation estimates the global prevalence of MetS to be around 25% [1, 2]. Previous research in South Africa showed that the mixed ancestry population were a high risk group for MetS [3]. It is therefore essential in a high risk population like this that one is able to identify those at risk and introduce lifestyle modifications.

Typically WC is the accepted proxy of visceral adipose tissue (VAT) and measure of central adiposity, and is used in the clinical diagnosis of MetS [4]. The advantage of WC is that it is quick and easy to measure and does not require technical equipment [5, 6]. Other anthropometric measures of total adiposity and body fat distribution such as BMI, WC, hip circumference (HC), waist to height ratio (WHtR), waist to hip ratio (WHR) and more recently, a body shape index (ABSI) have also been used as indicators of cardiometabolic risk [7]. All have their benefits, as well as their weaknesses. For example, while BMI is most commonly used as an indicator of total adiposity, it does not differentiate between muscle and fat or the location of fat [8]. WHtR, unlike WC thresholds, takes into account body size [8] and has shown closer agreement of values between men and women at all ages [9]. Similarly, the ABSI takes into account body size and is based on the WC adjusted for height and weight [10]. In contrast, WHR, uses the ratio of WC and HC, however the practicality of measuring 2 circumferences may be cumbersome and prone to error [11].

While metabolic abnormalities can be due to differential distribution of adipose tissue and or adipose dysfunction [12], a major limitation of WC, and other anthropometric measurements, is the inability to discriminate between VAT and subcutaneous adipose tissue (SAT) [6, 13]. Findings of the Framingham heart study revealed that both SAT and VAT correlated with metabolic risk factors, but that VAT was more powerfully associated with an unfavourable metabolic risk profile even after accounting for easily measured anthropometric indexes such as BMI and WC [14]. The mechanisms linking VAT accumulation to metabolic complications involve the greater production of proinflammatory cytokines and the greater lipolytic action compared to SAT, with the resultant increase in cytokines and free fatty acid transfer to the hepatic portal system impacting on insulin sensitivity [15].

In the clinical setting, VAT is however difficult to measure as it requires expensive technical equipment for imaging [6, 13, 16]. Although computed tomography (CT) and magnetic resonance imaging (MRI) are considered the gold standard imaging methods for quantifying body fat and its distribution, reliable algorithms have been developed using dual x-ray absorptiometry (DXA) software. These have been validated against CT in women with varying body mass index (BMI) [17, 18]. Numerous studies in various populations have identified associations between DXA-derived VAT area and cardio-metabolic risk factors [19‐22]. Although waist circumferences may be similar, ethnic variations in VAT and subcutaneous (SAT) have been documented amongst different ethnic groups [15]. Few studies have attempted to measure VAT in African populations and more specifically in the mixed ancestry women of South Africa where the prevalence of MetS is high.

The main aim of the study was to compare the ability of DXA-derived VAT area and anthropometry measures of total and central adiposity for diagnosing MetS in this high-risk sample of South African women.

In this study we set out to compare the discriminatory power of DXA-derived VAT area and other anthropometric measurements to diagnose any two components of MetS in a sample of high-risk mixed ancestry women from South Africa. The main finding was that VAT area, WC and WHtR performed similarly, lending support to the current recommendation of using the WC measurement for the diagnosis of MetS.

VAT accumulation is an important predictor of MetS [32] and is more closely associated with MetS risk than SAT due to its greater lipolytic activity and higher inflammatory profile [15]. WC is an internationally recognized surrogate for VAT and one of the five JIS criteria for the diagnosis of MetS [4]. As WC and other anthropometric measurements are however unable to discriminate between VAT and SAT, we set out to determine whether VAT would perform better than these measures and could therefore be used in risk prediction. Notably, we found that DXA- derived VAT did not perform better that WC and some of the other measures of central adiposity in diagnosing MetS in this sample. A possible explanation for this is that abdominal SAT is heterogenous, is of greater volume than VAT and has similar metabolic activity to VAT, thus also impacting on the development of insulin resistance and MetS [12]. The larger volume of SAT vs. VAT is particularly true for women compared to men, with abdominal SAT measured at L4–L5 being roughly five fold greater than VAT in the same location [12]. Moreover, abdominal SAT can be divided at the level of the fascia superficialis into deep (dSAT) and superficial SAT (sSAT), with the dSAT having higher metabolic activity and inflammatory profile than sSAT, and intermediate to VAT [33]. This suggests that the metabolic effects of accumulation of both VAT and abdominal SAT on MetS risk may be additive. Other studies have shown that VAT area performs better than WC, WHtR and WHR in determining MetS risk. For example, the results of the cross-sectional Netherlands epidemiology of Obesity study involving mostly white middle-aged obese women indicated that MRI-derived VAT was most strongly associated with cardiometabolic risk factors followed by WC and WHR [34]. Likewise in a study in Japanese women, CT-derived VAT performed better than WC in predicting MetS [35]. In contrast, and similar to our findings, Evans et al. [36] found that WC, WHtR and a CT-derived measure of VAT performed similarly in predicting MetS in pre-menopausal black and white South African women. These findings may be explained by ethnic-specific associations between adipose tissue distribution and insulin sensitivity. Indeed, studies have shown that VAT was the most significant correlate of insulin sensitivity in white women, whereas in black women, SAT performed similarly or better to VAT [37]. Another possible explanation for VAT not performing significantly better than WC and WHtR in diagnosing MetS may lie in the methodological limitation of imaging to precisely distinguish the various anatomical adipose tissue compartments [38]. Although DXA-derived VAT and SAT have been validated against the gold standard imaging methods such as CT and MRI in other ethnic groups [17, 18], they have not yet been validated in a mixed ancestry population. Additionally, DXA is unable to differentiate between dSAT and sSAT.

When comparing the performance of the VAT area and WC thresholds for detecting any two components of MetS in this sample, we found that the sensitivity of VAT area was higher than WC (73% vs 61%), but the specificity was lower (70% vs 80.8%). The implication of this is that VAT may be more sensitive than WC to detect MetS, but may also over diagnose MetS in screening. Furthermore, the accuracy and PPV of the WC and VAT area, as well as the DOR were very similar, reiterating the view that there may not be any advantage of using the more costly measurement of VAT in the clinical setting.

In contrast to WC and WHtR, we found that VAT area performed better than BMI, HC, WHR and ABSI in diagnosing any two components of MetS. A possible reason for this is that the latter anthropometry measures are essentially measures of total adiposity, which is not as closely associated with MetS as central adiposity [7]. For example, several studies have shown weak correlations for ABSI in predicting MetS [39, 40]. Similarly in a large cross-sectional study in an Iranian population, BMI had the lowest AUC in women for predicting Mets [41]. WC only requires one measurement unlike BMI and ABSI which requires height to be measured which can be challenging in the clinical setting [42].

The optimal VAT area for predicting any 2 components of MetS (other than WC) in this sample was 174 cm² (CI 137.7–181.5). This is higher than the ≥ 163 cm² found in peri and post-menopausal African and Caucasian American women [19]. Our VAT thresholds were also considerably higher than those used to predict metabolic risk variables in black (> 48 cm²) and white (> 107 cm²) pre-menopausal SA women [36]. Notably, recommended VAT cut points for diagnosing any two components of MetS (other than WC) differ by age and ethnicity [19, 35, 36]. Indeed, studies have shown that for the same BMI or WC, black women have less VAT than white women [43]. Further, it is well known that VAT accumulation occurs at menopause [44], and that VAT and total fat mass are independent with regards to metabolic risk [45], supporting the notion that WC corresponding to critical levels of VAT area may be age specific [19, 35]. This is supported by results of the study by Evans et al. [36] which identified considerably lower CT-derived VAT thresholds (> 88 cm²) for MetS risk factors in younger pre-menopausal white South African women than those recommended for peri- and post-menopausal women from our study (174 cm²) and others (≥ 163 cm²) [19].

The strengths of the study are that we derived WC and VAT area cut-points specific to this sample, which enabled us to directly compare the performance of these cut-points to diagnose any 2 components of MetS. Additionally, the two statistical approaches used to derive the cut-points yielded very similar results, an indication that the relationships are consistent. Furthermore, this is the first study to measure DXA-derived VAT area in mixed ancestry South African women. Although CT and MRI are considered the gold-standard for measuring VAT area, DXA has been proven to be an accurate imaging tool for deriving VAT [18]. The limitations of this study are that the sample included mostly post-menopausal women and thus the results cannot be generalised to younger women or men given that there are specific age and sex thresholds for WC that correspond to critical levels of VAT cm² [19, 35]. Additionally, the sample size was relatively small, yielding unstable estimates. A larger representative study is needed to determine whether VAT performs better than WC for diagnosing MetS and if shown to be better, ethnic, sex and age specific cut-points developed.

Our study showed that there is no advantage of measuring VAT over WC in the diagnosis of MetS as VAT area, WC and WHtR performed similarly in predicting two components of MetS in this sample of mixed ancestry South African women. WC is easier to measure in the clinical setting than other anthropometric measures and an universally recognised proxy for central adiposity.