In this prospective cohort study, we confirmed that obesity is an important risk factor of type 2 diabetes among a Chinese population, and found the increase of visceral adiposity was positively associated with metabolic disorders and diabetes risk with clear dose–response relationships. Compared to other body fatness indices, VAI was observed to be better in identifying the risk of diabetes than BMI, WC and WHtR.
The strong relationship between obesity and diabetes has been reported by many studies. Some researchers even use the term “diabesity” to describe their close associations. Yang SL et al. found that people with large WC has 3.79-fold risk of diabetes than those whose WC were normal. Yang WY et al. reported that the prevalence of diabetes in overweight group was 43% higher than that of normal people, and about 70% diabetes patients whose BMI were more than 25 kg/m2. Clinical trials showed that even 5% weight loss was sufficient to prevent most obese subjects from impaired glucose tolerance and developing diabetes, especially for abdominal obesity.
The main harm caused by obesity is visceral adipose accumulation. He et al. found that among men with normal WC, the incidence of MS in visceral obese people was significantly higher than that of normal group. Fox et al. reported that both subcutaneous abdominal adipose tissue (SAT) and VAT were associated with increased odds ratio (OR) of MS. In women, the OR for VAT (OR = 4.7) was stronger than that for SAT (OR = 3.0); similar difference was shown for men (OR for VAT = 4.2; OR for SAT = 2.5). The molecular mechanism underlying this is unclear yet. It has been suggested that compared with subcutaneous fat, high visceral fat produces more free fatty acid, thus will increase the risk of IR and diabetes[8, 28]. Fontana et al. reported that visceral adipose can secrete a large number of inflammatory cytokines, cells and adipokines, which may play important roles in the occurrence of IR and diabetes. Liu et al. observed that visfatin, an adipokine which mainly produced by visceral adipose had the insulin-like effect, and has been proved to aggravate IR. Moreover, as Masuzaki found in transgenic animals experiments, when 11 beta hydroxysteroid dehy-drogenase type 1(11βHSD-1) excessively expressed in fat cells, it would cause visceral adiposity and a series of metabolic disorders, which indicates that 11βHSD- 1 enzyme may have the same molecular basis with visceral obesity and metabolic disorders. However, the mechanism between visceral adiposity and metabolic disorders still needs to be further elucidated.
Although visceral fat has been found independently associated with IR and diabetes, and could be used to estimate the risk of diabetes, it’s not easy to measure body VAT volumes. MRI and CT have been considered as the gold standard for VAT measurements, but it is obviously they are not suitable for large epidemiological studies because of the high cost and inconvenience. Different from direct VAT determination, VAI could be easily conducted in large-scale epidemiological studies and has been suggested as a useful surrogate marker of VAT. VAI includes physical (BMI and WC) and metabolic (TG and HDL-C) parameters, it may indirectly reflects other non-classical risk factors, i.e. altered production of adipokine, increased lipolytic activity and plasma-free fatty acids. Recently reported by Al-Daghri et al., VAI was negatively related with adiponectin value, this was the first report for the direct relations of VAI with adipose tissue secretion. Some researches have proved that VAI could be used to predictive individual risk of IR, MS, acromegaly, cardiovascular disease (CVD) and diabetes[14, 32–34]. Consistent with previous researches[14, 31], our study also indicated that VAI is a useful surrogate marker to identify the risk of diabetes; individuals with high VAI were accompanied with increased risk of metabolic disorders and diabetes. The risk of getting diabetes at the highest VAI group was 2.55 folds higher as compared to the lowest VAI group.
According to the results of ROC analysis, the largest AUC was observed for VAI (AUC = 0.620, p < 0.001), following by WC, WHtR and BMI. Except VAI, AUC for other body fatness indices were all close to 0.5, which means the relatively lower predictive discriminatory power. Moreover, we observed positive trends between diabetes risk and VAI levels (HR = 1.38, 95% CI: 1.19-1.60), similar association was observed for WC (HRadj = 1.48, 95% CI: 1.04-2.10) and WHtR (HRadj = 1.43, 95% CI: 1.03-1.97), whereas only an insignificant adjusted HR was observed for BMI (HRadj = 1.07, 95% CI: 0.83-1.36). These findings suggested that VAI may be better in identifying diabetes risk. However, the ability of VAI in identifying diabetes risk was not found to be superior to WHtR in previous research which conducted among Tehran people. This difference might be attributed to different study populations. It has reported that compared with Caucasians, Asians may have significantly higher risk of type 2 diabetes and CVDs despite substantially lower BMI. Wang et al. found that Chinese men tend to have higher levels of FPG and DBP and significantly lower levels of HDL-C than Europeans. Similar result was also found for Chinese women by Lear et al.. In additional, several researches have proved that Chinese people may have higher VAT levels than European with a given body size[38, 39]. Variations in lifestyle may also have effect on the relationship between VAT and diabetes risk among different populations besides hereditary factors. To date, few published studies have explored relationship between VAI and diabetes risk and compared different body fatness indices with diabetes risk. Thus, the efficiency of VAI in predicting the risk of chronic diseases including diabetes still needs further verification from studies in different areas and different ethnicities.
Some limitations of our study need to be discussed. Firstly, because of the relative short time of follow-up and hence the relatively small number of new diabetes cases, the statistical power to examine the associations between VAI and diabetes may be insufficient. Secondly, IR was not measured in the present analysis therefore we were not able to directly examine the associations between VAI and IR. Thirdly, we did not examine 2-hour postprandial glucose, which may lead to the under diagnosis of some diabetic patients. Finally, residual confounders, an innate limitation of observational studies, could not be eliminated thus may increase the possibility that uncontrolled or inadequately measured confounders affected our results.
Despite these limitations, our study confirmed that obesity is an important risk factor of diabetes among Chinese population and found visceral adiposity is more important in disease occurrence. Compared to other indices for body fatness measurements, VAI is a good and convenience surrogate marker for visceral adipose measurement and could be used in identifying the risk of diabetes in large-scale epidemiologic studies.