“Body fat distribution and cardio-metabolic risk factors in South African men and women”

Abstract

Background: An analysis of pooled population-based studies conducted by the noncommunicable disease (NCD) risk factor collaboration Africa working group found that estimates of adiposity and diabetes prevalence in South Africa (SA) were higher than the global average. Specifically, in the mixed-ancestry population, central obesity rates were high (87.9% and 42.2% as defined by IDF criteria) in women and men respectively. Furthermore, the mixed-ancestry population of SA present with a high prevalence of metabolic syndrome (Mets) (62%) and type-2 diabetes mellitus (28.2%), placing this population at high risk for cardio-vascular disease (CVD). Visceral adipose tissue (VAT) accumulation is a known risk factor for cardio-metabolic disease. Typically, waist circumference (WC) is the accepted proxy of VAT, however, WC and other anthropometry measurements cannot distinguish between VAT and abdominal subcutaneous adipose tissue (SAT). Imaging modalities such as computed tomography (CT) and Magnetic Resonance Imaging (MRI) are considered the gold standard for VAT and abdominal SAT differentiation, but are not readily available in the clinical setting. Whole body composition studies using dual-energy x-ray absorptiometry (DXA) have also proved to be relatively accurate in quantifying VAT and abdominal SAT. Furthermore, DXA is also able to quantify other regional body fat compartments such as gynoid fat (gluteofemoral), trunk fat, arm and leg fat, all at a substantially lower radiation dose than CT. Therefore, the overall objective of this thesis was to validate DXA as a method of quantifying central adiposity, explore the relationship between body fat distribution and cardio-metabolic risk and determine the ability of DXA compared to anthropometry to identify participants with MetS. Methods: The study sample included a total of 46 men and 207 women, all self-described mixed-ancestry volunteers aged 20 years and older who were part of the Cape Town Vascular and Metabolic Health (VMH) parent study. Participants underwent anthropometric measurements, oral glucose tolerance test (OGTT), lipid measurements, DXA and CT scan examinations. Pearson correlation coefficients and Bland-Altman analysis were used to determine agreement between DXA and CT measurements. MetS was quantified using the Joint Interim Statement (JIS) criteria. Robust regression analyses were used to investigate associations between body fat distribution and cardio-metabolic risk factors. The area under the curve (AUC) was used to assess the performance of VAT area and anthropometry in detecting any two components of MetS (excluding WC). Optimal WC and VAT area cut-points were derived to compare their performance for diagnosing MetS and to compare to internationally recognised cut-points Results: The mean age in the 132 women in whom VAT and abdominal SAT were measured using single slice CT and DXA was 55 and ranged from 45 to 64 years. DXA and CT- derived measurements of abdominal VAT and SAT were significantly correlated in the overall sample (r=0.872 and r=0.966, both p<0.001 respectively) and within body mass index (BMI) categories. In the overall sample, the mean difference (DXA-CT estimates) was 75.3 cm2 (95% CI: 68.8-81.8 cm2, p≤0.0001) for VAT and 54.7 cm2 (47.1-62.3 cm2, p≤0.0001) for SAT. Within increasing BMI categories, the variance between the two modalities was fixed for VAT (p=0.359 for obese), whereas the variance for SAT was heteroscedastic (p≤0.0001). In the cross-sectional study, which included 207 mixed-ancestry SA women and 46 men, the men had lower body fat mass compared women (26.5 vs. 44.0%), but had more central and less peripheral fat (both p<0.001). Post-menopausal women had greater % fat mass, (FM), WC and VAT, and less gynoid % FM than pre-menopausal women (all p≤0.004). After adjusting for age and sex, VAT accounted for greatest variance in insulin resistance (R2=0.27, p≤0.01), while trunk %FM and leg %FM accounted for greatest variance in triglyceride (R2=0.13, p≤0.01) and high-density lipoprotein cholesterol concentrations (R2=0.14, p≤0.01). The highest AUC for the prediction of MetS in the 204 women was recorded for VAT, followed by waist-to height-ratio (WHtR) and WC (AUC, 0.767, 0.747 and 0.738 respectively), but these did not differ significantly (all p>0.192). In contrast, VAT was significantly better than BMI (p=0.028), hip (p=0.0004) and a body shape index (ABSI) (p<0.0001). Conclusions: In conclusion, this thesis has for the first time validated DXA as a method of quantifying visceral adiposity, explored the relationship between body fat distribution and cardio-metabolic risk and determined the ability of DXA compared to anthropometry to identify mixed-ancestry participants with MetS. Although we found that DXA overestimated VAT compared to the gold standard, CT, the variance was fixed in the obese category. These findings suggest that DXA is a valid measure of VAT and abdominal SAT in the obese group which is the group most at risk for cardio-metabolic diseases. Furthermore, we have demonstrated the value of the whole body DXA scan which quantified regional fat depots and showed that central fat was the most significant correlate of cardio-metabolic risk and lower body fat was associated with reduced risk. Finally, we have demonstrated in this population that DXA-derived VAT had no advantage in discriminating MetS than WC, and therefore confirm that WC can be used as a marker of MetS provided population specific cut-offs are derived.