Diabetes mellitus (DM) has become a major worldwide public health burden in the past decade, especially in developing countries [1
]. The China national Diabetes and Metabolic Disorders Study reported prevalence of diabetes was 9.7% among adults in 2010 [2
]. This increased trend of DM is concurrent with rising rate of obesity in China, and excessive body fat has been proven to be a crucial pathogenic factor for insulin resistance [4
]. Excessive body fat disposed in the ectopic tissue, such as visceral adiposity tissue (VAT), may cause dysfunctional adiposity and it plays a vicious role in metabolic diseases [5
]. In addition, body fat distribution is related to metabolic disturbances and metabolic disorders [6
]. Thus, knowing the ability to predict the visceral adiposity index for diabetes risk is greatly needed.
Body mass index (BMI), widely used since the early 1990s worldwide for classifying overweight and obesity, as well as studying obesity associated risks, provides reliable information concerning body weight excess, but does not differentiate fat from lean mass [7
]. Waist circumference (WC) is a simple anthropometric parameter for abdominal adiposity and it reflects visceral obesity better than BMI [10
]. It is a better indicator of obesity associated risks for DM, as shown by our research [11
], though it has limitations in distinguishing VAT from subcutaneous fat mass [13
]. The magnetic resonance imaging (MRI) and computed tomography (CT) are considered the gold standard for body fat determination, but they were less recommended in routine clinical practice due to their unavailability.
The VAI, which is comprised of anthropometric measures like BMI, WC and clinical measures of serum triglycerides (TG) and high-density lipoprotein-cholesterol (HDL-C) levels, was shown to be a better surrogate index than these single anthropometric indices in predicting insulin resistant-related metabolic disorders [14
]. However, evaluating the predicting performance of VAI, which was developed for Caucasians, on people who have diabetes in Chinese population may lead to inaccurate results [16
]. Meanwhile, a body shape index (ABSI), which encompasses waist circumference and BMI, particularly depicts fat distribution [18
], and was shown to be a reliable index of body fat accumulation. However, there is little research on this in China [20
Therefore, we hypothesize that non-invasive, clinically measurable surrogates could be useful in identifying body fat distribution and help predict diabetes risk. We aim to examine the associations between these two indicators and diabetes risk, and to investigate their performance in identifying diabetes compared with BMI and WC in Chinese adults.
2. Materials and Methods
2.1. Study Design and Population
Data collected in the 2009 China Health and Nutrition Survey (CHNS) was used in this study. The CHNS is an ongoing, open, multipurpose household based cohort study since 1989. It adapts a multistage random-cluster sampling process and is conducted in 9 provinces (Liaoning, Heilongjiang, Shandong, Jiangsu, Henan, Hubei, Hunan, Guangxi and Guizhou), which covers approximately 56% of China’s population. It was designed to represent a large set of the population, varying significantly in geography, economic development, public resources and health status. More detailed information about the CHNS has been described elsewhere [21
]. The CHNS was approved by the institutional review committee of the University of North Carolina at Chapel Hill and Chinese Center for Disease Control and Prevention. All participants were required to provide written informed consents before their participation.
Using CHNS data, we assessed the associations between the two indicators and diabetes risk, since fasting blood samples were collected initially in 2009 and the biochemical data collected in 2015 is unavailable currently. A structured questionnaire was used to obtain information on sex, age and lifestyle behaviors. Participants aged ≥18 years and without missing information on physical examination and biochemical measurements were included. Exclusion criteria included pregnancy and no information on age, sex and lifestyle behavior indicators. Ultimately, a total number of 5838 adults with anthropometry and clinical examination information were included in the analysis (Figure S1
2.2. Definition of Key Study Outcome DM
DM was defined as having FPG ≥ 7.0 mmol/L, HbA1c ≥ 6.5%, previous diagnosis of DM, or use of antidiabetic medications. Of the 5838 subjects, 482 (8.3%) had DM.
2.3. Anthropometry and Biochemical Measurements
Weight (kg) and height (m) were measured according to standard methods, and body mass index (BMI) was calculated as weight/height squared (kg/m2). WC was measured to the nearest 0.1 cm at the middle point between the bottom of the rib cage and the uppermost border of the iliac crests at the end of exhalation in standing positions with an inelastic tape. Standard mercury sphygmomanometers were used to measure blood pressure by trained investigator at three different consecutive times at 3–5 min intervals on one visit. All physical examinations were performed following the same protocol at each study visit and study site.
2.4. Biochemical Measurements
Blood was collected from the participants after an at least 8 h overnight fast. The whole blood was centrifuged immediately after collection, and plasma and serum samples were then frozen and stored at −87 °C for future analysis. Samples for fasting plasma glucose (FPG) and glycated hemoglobin (HbA1c) measurements were tested immediately. Serum glucose was measured by the glucose oxidase phenol 4-aminoantipyrine peroxidase (GOD-PAP) methods with a Hitachi 7600 analyzer (Hitachi, Tokyo, Japan). Whole blood HbA1c was measured with a high-performance liquid chromatography system (model HLC-723 G7, Tosoh Corporation, Tokyo, Japan). Fasting insulin concentration was tested using the radioimmunology assay (Gamma counter XH-6020, Xi’an, China). Lipids including total cholesterol (TC), Total triglycerides (TG), low-density lipoprotein cholesterol (LDL-C) and HDL-C, as well as uric acid, were measured using a biochemical auto-analyzer (Hitachi 7600 automated analyzer, Tokyo, Japan). Hypersensitive C-reactive protein (hs-CRP) was determined by the immunoturbidimetric method.
The VAI score was calculated using the specific formula for Chinese population [22
Males: CVAI = −267.93 + 0.68 × age + 0.03 × BMI + 4.00 × WC + 22.00 × log10(TG) − 16.32 × HDL
Females: CVAI = −187.32 + 1.71 × age + 4.23 × BMI + 1.12 × WC + 39.76 × log10(TG) − 11.66 × HDL.
ABSI was calculated as WC/(BMI2/3
) and expressed in m11/6
]. The homeostasis model assessment of insulin resistance (HOMA-IR) was estimated as HOMA-IR = Fasting glucose (mmol/L) × fasting insulin (μIU/mL)/22.5 [23
2.5. Statistical Analysis
Descriptive analyses were presented on the basis of gender-specific quartiles of CVAI and ABSI scores in order to control for the well-known sexual dimorphism in body composition. The characteristics were presented as mean (SD) and median with interquartile range (25–75%) given in parentheses for normalized continuous and skewed variables. Categorical variables were expressed as numbers and percentages. One-way analysis of variance (ANOVA) or the Mann–Whitney U-test was used for comparisons of quantitative variables among groups. Chi-squared test was performed to assess differences in proportions across groups. Partial correlations between two anthropometric indices and metabolic parameters adjusting for age and sex were evaluated with Pearson’s/Spearman’s correlation analysis.
Multivariable logistic regression models were performed to estimate the odds ratios (ORs) and 95% confident interval (95% CI) of DM associated with these two indices in four models for men and women, respectively. Receiver operating characteristic (ROC) curve analyses were used to compare the diagnostic performance of CVAI and ABSI as compared with BMI and WC for DM risk. A user written command cutpt
was used to calculate the Youden index [24
All statistical analyses were conducted using Stata software (version 15.0). A two-tailed statistical measure was used with a p-value of less than 0.05 is considered significant.
This study assessed the associations of the Chinese visceral adiposity index (CVAI) and body shape index (ABSI)—two indicators of adiposity distribution and function—with DM risks among adult Chinese people. We found graded positive associations of CVAI and ABSI scores with diabetes risks in both men and women. Partial correlation analysis found that the CVAI and ABSI were independently related to insulin resistance and lipid profiles. We also showed that the CVAI is superior to ABSI, BMI and waist circumference in predicting diabetes in both genders.
Increase in the prevalence of obesity and physical inactivity was concurrent with a tripling in diabetes incidence and other metabolic diseases over the past few decades in China [26
]. Of the many obesity phenotypes, visceral obesity/fat is more metabolically deleterious than general obesity or subcutaneous fat, and it has been proposed as a marker of adiposity dysfunctional and ectopic fat deposition, which in turn leads to lipotoxicity and insulin resistance. Given that quantitating visceral adipose using CT or MRI is not feasible in large cohort studies and daily investigation, some simple clinical anthropometric indices such as WC and WHtR were used as surrogate indices of visceral adiposity to identify metabolic diseases. However, these classic anthropometric indices have the common shortcoming of their inability to take metabolic measures into consideration, and thus are not robust in various populations.
The present study adopted a developed indicator of visceral adiposity for the Chinese population, which is reliable for evaluating metabolic risk [27
] and highest discriminatory power for dysglycemia in male and females. Consistent with other studies [22
], our results showed that the CVAI score is associated with diabetes risk and is a good clinical index for prediction of visceral fat dysfunction. The associations of obesity indicators, including BMI and WC, with incident diabetes in multiethnic populations worldwide have been confirmed by several meta-analysis studies by the significant pooled estimates of the relative risk [29
]. In our study, the CVAI score was highly correlated to and behaved slightly better in diabetes prediction than BMI and WC in Chinese adults by means of a higher AUC and an overall differentiating ability. Thus, CVAI is a useful clinical obesity indictor for diabetes risk when biospecimen data are available.
Previously, ABSI was developed base on the US National Health and Nutrition Examination Survey (NHANES) 1999–2004 data, which included several ethnicities like Mexican, other Hispanic, white, black, or other. For a given height and weight, ABSI is suitable for reflecting VAT [31
]; however, subsequent research revealed conflicting results regarding its predictability for chronic diseases and mortality [31
]. A study conducted in Dutch adults showed that ABSI was not a suitable index to identify CVD or CVD risk factors [34
]. ABSI was not a better predictor of diabetes, hypertension or dyslipidaemia than WC or BMI in Japanese adults [35
]. By contrast, Zhao and colleagues [18
] recently indicated that ABSI had a better predictive ability than BMI in predicting diabetes in Han Chinese people in Northeast China. In agreement with previous studies [31
], our study indicates that ABSI may not be a better predictor of diabetes than BMI or WC.
Furthermore, we incorporate types of staple foods into the analysis to examine the reason of conflicting findings with Zhao et al. People living the North China mainly consume wheat products. After re-analyzed the ROC separately by rice/wheat intake, our results showed that ABSI had a better predictive ability than BMI in men whose staple food is rice (results not reported), which is consistent with Zhao’s findings that ABSI is better than BMI in predicting diabetes in the Northeast Chinese population.
Interestingly, our study showed that increasing CVAI scores associated stronger with diabetes in women than men, whereas the ABSI had comparable predicting of diabetes risk in men and women. The explanations for these issues remain to be elucidated, and this may be related to gender differences in patterns of visceral fat deposition and regional adipose tissue distribution [38
]. The mechanisms by which visceral adiposity can lead to diabetes may be different in men versus women. Besides, age is a well-established risk factor for diabetes, and women in the 4th CVAI quartile had a higher age than men in this group, hence diabetes risk may be partly explained by age and other related unfavorable traits. We think CVAI and ABSI serve as markers of diabetes risk for both men and women, but women should pay more attention to their visceral fat deposition.
Our results showed that both CVAI and ABSI indicate close correlations with lipid levels, insulin resistance and inflammation in both men and women (p < 0.01). In addition, all four anthropometric indices exhibited the capability to identify individuals with diabetes (all AUC > 0.5 for all), and CVAI together with ABSI are strong and independent risk factors for diabetes (all ORs > 1). However, CVAI demonstrated the strongest prediction ability (AUC = 0.729 in men and 0.794 in women) among all anthropometric indices and ABSI had the weakest association with diabetes risk in both men and women. This might be because CVAI serve as a strong surrogate marker of visceral adiposity dysfunction, while ABSI, similar to BMI and WC, does not differentiate excess central adiposity in Chinese adults. Despite a high AUC value, CVAI has the lowest sensitivity among all the obesity indicators in both genders. It might not be the best indicator for the screening of diabetes.
The cutoffs for BMI and WC in predicting risk of T2DM in Chinese adults remain controversial. In response to WHO’s recommendations on the cutoff values for public health action for Asians, the Working Group on Obesity in China (WGOC) suggested that BMI ≥ 24 signifies overweight in Chinese population [39
]. In the present study, the optimal cut-off values for Chinese men and women were found to be approximately BMI = 25 and 23, and 88.0 and 85.0 cm for WC, respectively. The current recommended cutoffs regarding central obesity in China are WC ≥ 90 cm for men and WC ≥ 90 for women. Compared with WGOC definitions and WC recommendations, our data observed a similar cutoff value for WC in women, but a lower value in men, and a sex difference for BMI. However, these results need be interpreted with caution because these cutoffs were based on data collected in a sample of 2433 men and 3320 women aged 52 ± 14 from 9 provinces in China.
Our study has several limitations. First, the cross-sectional data analyses cannot make causal inferences regarding the relationships between CVAI or ABSI and DM risks. Second, no direct measure of insulin resistance was done, and we were unable to directly assess the association of CVAI/ABSI with insulin resistance. Third, data on 2 h postprandial glucose was unavailable, which might lead to the underdiagnoses of some diabetic subjects. Key strengths of the study include that the data set provide rich related measures and a large sample allows us to assess the various associations and the predicting power of a set of indices.