Abstract
Background
Obesity, particularly visceral obesity, is a significant risk factor for the onset of diabetes mellitus. Visceral adiposity index (VAI) is a novel predictor of visceral obesity that aims to explore its relationship with the incident diabetes.
Materials and methods
This retrospective cohort study included 15,453 Japanese adults who did not have diabetes at the start of the study. Univariate analysis and Cox proportional hazards regression were employed to determine the independent link between VAI and diabetes. The Kaplan-Meier curve was used to investigate the difference in diabetes risk among subjects with varying VAI levels. Restricted cubic spline (RCS) curve was utilized to identify the non-linear relationship between VAI and diabetes risk.
Results
Elevated VAI values are linked to an increased risk of incident diabetes. Individuals with high VAI values had higher levels of blood pressure, blood glucose, blood lipids, and a higher prevalence of fatty liver, with a larger proportion of males and more severe smoking and alcohol consumption habits. On the contrary, individuals with exercise habits showed lower VAI values. The Cox regression analysis and Kaplan-Meier curves revealed a positive correlation between high VAI values and incident diabetes. Analyses of RCS curves also demonstrated a non-linear relationship.
Conclusions
VAI is a valuable indicator for predicting diabetes risk and can be responsive to visceral obesity. The relationship between VAI and diabetes onset events is non-linear, and high VAI levels above 0.75 are associated with an increased risk of diabetes.
Keywords: Visceral adiposity index, Incident diabetes, Obesity, Non-linear
Introduction
Diabetes (T2DM) is a debilitating disease that affects millions of individuals worldwide. The reported global cases of diabetes amount to 537 million, of which nearly 90% accounts for T2DM. The incidence of T2DM is on the rise with no indication of decline, and projected estimates suggest that the number of cases could reach 783 million by 2045 [1].
There is significant evidence to indicate that adiposity, particularly intra-abdominal and visceral adiposity, is the foremost factor that leads to the onset of metabolic diseases, especially insulin resistance and diabetes [2–5]. Adiposity and T2DM are significant public health concerns that are associated with severe morbidity and higher mortality rates1. Visceral adipose tissue is distributed in the chest, abdomen and pelvis, and subcutaneous tissue. Visceral adiposity can cause compensatory hypersecretion of insulin, leading to the development of insulin resistance and ultimately resulting in β-cell exhaustion. This cascade of events can lead to a shift from pre-diabetes to the onset of type 2 diabetes mellitus [6]– [7].
However, precise quantification of this type of adiposity has presented significant challenges. Although magnetic resonance imaging is the most accurate method, it is costly and time-consuming, rendering it impractical for routine clinical applications [8]. Therefore, there is an urgent need to develop a cost-effective tool to assess visceral adiposity [9]. Body mass index (BMI), the commonly used parameter for assessing obesity, cannot effectively distinguish between subcutaneous and visceral obesity, whereas waist circumference (WC) is a more reliable indicator of abdominal obesity [7]– [10]. Visceral adiposity is frequently associated with dyslipidemia, characterized by high triglyceride (TG) and low high-density lipoprotein cholesterol (HDL-C) levels, which has been strongly linked to the onset of type 2 diabetes [11–15]. To address this, Marco C et al. proposed a novel sex-specific index called the visceral adiposity index (VAI), which incorporates WC, BMI, TG, and HDL-C as parameters [16].
Several studies have established a correlation between VAI and diabetes, with higher VAI scores indicating an elevated risk of T2DM. However, these studies have also indicated that while VAI can predict future diabetes independently, its discriminatory power is not superior to that of basic anthropometric measures such as BMI, waist circumference, and waist-to-height ratio [17–22]. The majority of the studies investigating the association between VAI and T2DM were either cross-sectional, had small sample sizes, short follow-up durations, or did not examine non-linear relationships or conduct subgroup analysis. Thus, a secondary analysis of a population-based longitudinal study was conducted to better understand the relationship between VAI and T2DM in a significant cohort of Japanese adults and to assess its predictive value for T2DM.
Materials and methods
Data source and study population
The present study is a cross-sectional analysis based on data from the NAGALA cohort and explores the correlation between VAI and T2DM. The subjects’ data utilized in this analysis were sourced from the Dryad database, uploaded by Hamaguchi et al. The terms and conditions of the Dryad database allow for full utilization of the data under different scientific hypotheses [23]. A detailed description of the study design and procedures of the NAGALA cohort has been published elsewhere [24]. In summary, the NAGALA cohort is a longitudinal study initiated by the Murakami Memorial Hospital in 1994 as part of a general population physical examination program aimed at identifying chronic diseases and their associated risk factors. The data from 15,453 examiners recruited by the NAGALA cohort between 1994 and 2015 were analyzed in this study. The inclusion criteria specified that subjects exhibiting the following characteristics were excluded: (1) absence of covariates; (2) diagnosed or self-reported viral/alcoholic hepatitis or diabetes at baseline; (3) taking medication; (4)having impaired fasting glucose at baseline; (5)incomplete HDL-C. As ethical review and informed consent forms have already been obtained by the previous cohort study conducted by Murakami Memorial Hospital, the present study does not require a fresh application for ethical approval or informed consent. All procedures of the study strictly adhere to the principles of the Helsinki Declaration.
Data collection and measurement
As reported in the previous study [18], demographic information was collected through a standardized questionnaire, and factors such as gender, age, drinking/smoking status, exercise habits, medication use, and disease status were reported by the subjects. Trained professionals measured anthropometric indicators, including height, weight, waist circumference, and blood pressure, using standard methods.
For biochemical marker assessment, venous blood samples were collected from subjects after an 8-hour fast by experienced medical personnel. Biochemical markers such as liver enzymes (gamma-glutamyl transferase (GGT), aspartate aminotransferase (AST), alanine aminotransferase (ALT)), lipid indicators (HDL-C, TG, total cholesterol (TC)), and blood glucose indicators (hemoglobin A1c (HbA1c), fasting plasma glucose (FPG)) were measured using an automated biochemical analyzer. Fatty liver was determined by experienced gastroenterologists using abdominal ultrasound, and identified through several sonograms showing deep attenuation, hepatorenal echo contrast, vascular blurring, and liver brightness.
Visceral adiposity index
VAI was calculated by previous reported formula [12]. The units of WC and BMI were cm and Kg/m2, respectively, and both TG and HDL-C were expressed in mmol/L:
Males: VAI =[WC/(39.68 + 1.88*BMI)] × (TG/1.03)*(1.31/HDL-C)
Females: VAI=[WC/(36.58 + 1.89*BMI)]*(TG/0.81)*(1.52/HDL-C)
Diagnosis of incident diabetes
Incident diabetes was defined as HbA1c ≥ 6.5%, FPG ≥ 7 mmol/L [25] or self-reported.
Statistical analysis
Data analysis was performed using IBM SPSS Statistics 25.0 software (SPSS Inc. Chicago IL, USA) and R version 4.2.3 (www.r-project.org). Continuous variables, which included skewed and normally distributed variables, were reported as median (interquartile range) and mean ± standard deviation, respectively. Categorical variables were presented as frequencies and percentages. The baseline characteristics of all participants were explored based on the quartiles of VAI and the occurrence of diabetes, and differences between groups were compared using one-way analysis of variance (ANOVA), Kruskal-Wallis H test, or chi-square test. The Kaplan-Meier method was utilized to compare the survival and cumulative event rates between different VAI groups. The Receiver Operating Characteristic (ROC) curve is generated, and the Area Under the Curve (AUC) is computed in order to assess the predictive performanceof VAI, WC, and BMI. Subsequently, the DeLong test was employed to compare the differences in AUC values. Additionally, the log-rank test was employed to compare the Kaplan-Meier hazard ratios (HR) of adverse events. Multivariable Cox regression analysis was also employed to investigate the association between VAI and the risk of diabetes. Three models were constructed: crude Model, Model I, and Model II, to evaluate the association between VAI and diabetes risk. As VAI was a continuous variable, the nonlinear relationship between VAI and diabetes was determined using a Restricted Cubic Spline (RCS) with four knots at 0, 2, 4, and 6, after adjusting for various confounding variables, including gender, age, ethanol consumption, habit of exercise, smoking status, ALT, AST, GGT, SBP, DBP, TC, FPG, and HbA1c. In addition, subgroup analysis was conducted based on the Cox proportional hazard model, taking into account factors including gender, age, WC, BMI, smoking status, and drinking status. To ensure clinical relevance, continuous variables such as age, WC, and BMI were converted into categorical variables, using established cut-off points: age (< 60 years old, ≥ 60 years old), WC (< 90 cm in men or 80 cm in women, ≥ 90 cm in men or 80 cm in women), and BMI (< 25 kg/m2, ≥ 25 kg/m2). The statistical significance of results was determined by P < 0.05 in two-tailed tests.
Results
Baseline characteristics of the participants
The baseline characteristics of the study population were described in Table 1, including clinical measurements, biochemical tests, and other parameters. A total of 15,453 individuals were categorized into four groups based on their VAI value, with a mean age of 43.71 ± 8.90 years and a male participation rate of 54.48%. Individuals in the highest VAI quartile group had significantly higher levels of ethanol consumption, anthropometric indexes (including systolic blood pressure (SBP), diastolic blood pressure (DBP), BMI, and WC), ALT, AST, GGT, TG, TC, HbA1c, FPG and VAI (P < 0.001). The proportion of male, fatty liver, and smoking rosed with the increasing of VAI quartiles (P < 0.001). However, individuals with higher VAI value exhibited lower levels of HDL-C (P = 0.003) and lower rates of the habit of exercise (P < 0.001). It is worth noting that the proportion of fatty liver, which is considered a significant manifestation of visceral obesity, increased rapidly from 13.1% to 31.1% with each higher VAI quartile (P < 0.001).
Table 1.
Baseline characteristics of population with different VAI quartile
| VAI | Q1 (≤ 0.504) | Q2 (0.504 to ≤ 0.756) | Q3 (0.756 to ≤ 1.170) | Q4 (> 1.170) | P value |
|---|---|---|---|---|---|
| Participants | 3861 | 3863 | 3859 | 3870 | |
| Gender | < 0.001 | ||||
| Female | 2192 (56.8%) | 1980 (51.3%) | 1646 (42.7%) | 1216 (31.4%) | |
| Male | 1669 (43.2%) | 1883 (48.7%) | 2213 (57.3%) | 2654 (68.6%) | |
| Age (years) | 43.70 ± 8.86 | 43.65 ± 8.88 | 43.63 ± 8.82 | 43.85 ± 9.03 | 0.690 |
| SBP (mmHg) | 110.33 ± 13.69 | 112.49 ± 14.34 | 115.10 ± 14.91 | 120.04 ± 15.09 | < 0.001 |
| DBP (mmHg) | 68.29 ± 9.72 | 70.13 ± 9.98 | 72.17 ± 10.36 | 75.71 ± 10.45 | < 0.001 |
| BMI (kg/m 2 ) | 21.78 ± 3.07 | 22.01 ± 3.13 | 22.23 ± 3.12 | 22.45 ± 3.15 | < 0.001 |
| WC (cm) | 74.90 ± 9.05 | 75.92 ± 8.88 | 76.99 ± 9.09 | 78.05 ± 9.11 | < 0.001 |
| ALT (IU/L) | 16 (12, 21) | 16 (12, 22) | 17 (13, 23) | 18 (13, 25) | < 0.001 |
| AST (IU/L) | 17 (14, 21) | 17 (14, 21) | 17 (14, 21) | 17 (14, 22) | < 0.001 |
| GGT (IU/L) | 14 (11, 21) | 14 (11, 21) | 15 (12, 23) | 16 (12, 24) | < 0.001 |
| HDL-C (mmol/L) | 1.67 ± 0.42 | 1.51 ± 0.38 | 1.40 ± 0.36 | 1.27 ± 0.34 | < 0.001 |
| TG (mmol/L) | 0.40 (0.30, 0.49) | 0.61 (0.51, 0.73) | 0.88(0.72, 1.06) | 1.48 (1.17, 1.94) | < 0.001 |
| TC (mmol/L) | 5.21 ± 0.85 | 5.15 ± 0.86 | 5.11 ± 0.87 | 5.03 ± 0.86 | < 0.001 |
| HbA1c (%) | 5.16 ± 0.30 | 5.16 ± 0.32 | 5.17 ± 0.33 | 5.19 ± 0.34 | < 0.001 |
| FPG (mmol/L) | 5.10 ± 0.41 | 5.12 ± 0.42 | 5.17 ± 0.42 | 5.25 ± 0.39 | < 0.001 |
| VAI | 0.366 ± 0.10 | 0.625 ± 0.07 | 0.937 ± 0.12 | 1.864 ± 0.84 | < 0.001 |
| Fatty liver | 505 (13.1%) | 627 (16.2%) | 753 (19.5%) | 852 (31.1%) | < 0.001 |
| Ethanol consumption(g/wk) | 1 (0, 60) | 1 (0, 60) | 1 (0, 60) | 2.8 (0, 70) | < 0.001 |
| Drinking status | < 0.001 | ||||
| No/little | 3188 (82.6%) | 2998 (77.6%) | 2912 (75.5%) | 2700 (69.8%) | |
| Light | 354 (9.2%) | 449 (11.6%) | 467 (12.1%) | 486 (12.6%) | |
| Moderate | 245 (6.3%) | 309 (8.0%) | 343 (8.9%) | 462 (11.9%) | |
| Heavy | 74 (1.9%) | 107(2.8%) | 137 (3.6%) | 222 (5.7%) | |
| Smoking status | < 0.001 | ||||
| Non | 2804 (72.6%) | 2448 (63.4%) | 2108 (54.6%) | 1665 (43.0%) | |
| Past | 544 (14.1%) | 708 (18.3%) | 801 (20.8%) | 899 (23.2%) | |
| Current | 513 (13.3%) | 707 (18.3%) | 950 (24.6%) | 1306 (33.7%) | |
| Habit of exercise | 0.003 | ||||
| No | 3144 (80.7%) | 3190 (82.6%) | 3200 (82.9%) | 3243 (83.8%) | |
| Yes | 747 (19.3%) | 673 (17.4%) | 659 (17.1%) | 627 (16.2%) |
Abbreviations: VAI, visceral adiposity index; SBP, systolic blood pressure; DBP, diastolic blood pressure; BMI, body mass index; WC, waist circumference; ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT, gamma-glutamyl transferase; HDL-C, high-density lipoprotein cholesterol; TG, triglycerides; TC, total cholesterol; HbA1c, hemoglobin A1c; FPG, fasting plasma glucose.
Association between VAI and incident diabetes
As shown in Tables 2 and 373 participants developed diabetes over a median follow-up period of 5.39 years. The individuals in the incident diabetes group exhibited significantly higher levels of ethanol consumption, SBP, DBP, ALT, GGT, TG, HbA1c, FPG and VAI (P < 0.001). In addition, the proportion of male, heavy drinking and smoking was also higher (P < 0.001). The levels of AST, HDL-C and TC showed no significant difference between the incident diabetes group and the diabetes-free group (P = 0.061, 0.662 and 0.622). However, individuals with incident diabetes in these populations exhibited a lower BMI, smaller WC, lower proportion of fatty liver (P < 0.001), and lower rates of the habit of exercise(P = 0.009).
Table 2.
Baseline characteristics of population between diabetes-free group and incident diabetes group
| VAI | Diabetes-free group | Incident diabetes group | P value |
|---|---|---|---|
| Participants | 15,081 | 372 | |
| Gender | < 0.001 | ||
| Female | 6999 (46.4%) | 35 (9.4%) | |
| Male | 8082 (53.6%) | 337 (90.6%) | |
| Age (years) | 43.71 ± 8.89 | 43.66 ± 9.32 | 0.910 |
| SBP (mmHg) | 114.31 ± 14.91 | 121.86 ± 15.30 | < 0.001 |
| DBP (mmHg) | 71.44 ± 10.47 | 77.08 ± 10.04 | < 0.001 |
| BMI (kg/m 2 ) | 22.14 ± 3.13 | 21.15 ± 2.68 | < 0.001 |
| WC (cm) | 76.52 ± 9.14 | 74.18 ± 7.46 | < 0.001 |
| ALT (IU/L) | 17 (13, 23) | 18 (15, 24) | < 0.001 |
| AST (IU/L) | 17 (14, 21) | 18 (14, 22) | 0.061 |
| GGT (IU/L) | 15 (11, 22) | 16.5 (13, 24) | < 0.001 |
| HDL-C (mmol/L) | 1.46 ± 0.40 | 1.45 ± 0.37 | 0.662 |
| TG (mmol/L) | 0.72 (0.49, 1.11) | 1.20 (0.86, 1.93) | < 0.001 |
| TC (mmol/L) | 5.13 ± 0.87 | 5.10 ± 0.82 | 0.622 |
| HbA1c (%) | 5.16 ± 0.32 | 5.53 ± 0.36 | < 0.001 |
| FPG (mmol/L) | 5.15 ± 0.41 | 5.61 ± 0.36 | < 0.001 |
| VAI | 0.939 ± 0.70 | 1.352 ± 0.95 | < 0.001 |
| Fatty liver | 2708 (18.0%) | 29 (7.8%) | < 0.001 |
| Ethanol consumption(g/wk) | 1 (0, 60) | 12.48 (1, 90) | < 0.001 |
| Drinking status | < 0.001 | ||
| No/little | 11,532 (76.5%) | 266 (71.5%) | |
| Light | 1717 (11.4%) | 39 (10.5%) | |
| Moderate | 1322 (8.8%) | 37 (9.9%) | |
| Heavy | 510 (3.4%) | 30 (8.1%) | |
| Smoking status | < 0.001 | ||
| Non | 8881 (58.9%) | 144 (38.7%) | |
| Past | 2875 (19.1%) | 77 (20.7%) | |
| Current | 3325 (22.0%) | 151 (40.6%) | |
| Habit of exercise | 0.009 | ||
| No | 12,459 (82.6%) | 288 (77.4%) | |
| Yes | 2622 (17.4%) | 84 (22.6%) |
Abbreviations: VAI, visceral adiposity index; SBP, systolic blood pressure; DBP, diastolic blood pressure; BMI, body mass index; WC, waist circumference; ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT, gamma-glutamyl transferase; HDL-C, high-density lipoprotein cholesterol; TG, triglycerides; TC, total cholesterol; HbA1c, hemoglobin A1c; FPG, fasting plasma glucose.
ROC analyses of predictive capacity of different indices related to obesity for incident diabetes are presented in Fig. 1. VAI (AUC = 0.652) demonstrates stronger prediction performance than WC (AUC = 0.584) and BMI (AUC = 0.597)(P<0.001).
Fig. 1.
ROC analyses of predictive capacity of different indices related to obesity for incident diabetes
According to Table 3, the incidence of incident diabetes was 399.14 per 100,000 person-years. Furthermore, the incidence rate of diabetes increased significantly from 213.21 to 700.43 per 100,000 person-year with each higher VAI quartile (P < 0.001). The multivariate Cox regression analysis was also used to explore the association of AVI with the risk of diabetes. In the Crude Model, no covariates were adjusted. In Model 1, age and sex were adjusted. In Model 2, gender, age, ethanol consumption, smoking status, habit of exercise, SBP, DBP, ALT, AST, GGT, TC, HbA1c, and FPG were adjusted. In all three models, there was a positive correlation between VAI and incident diabetes (P < 0.001). At the same time, when VAI was categorized into quartiles and reinserted into these three models, the adjusted hazard ratio (HR) (95% CI) for the Q4 group was also found to be higher than that of the Q1 group (P < 0.001).
Table 3.
Relationship between AVI and incident diabetes in different models
| VAI | Participants (n) |
Diabetes events (n) | Incidence rate (per 100,000 person-year) | Crude Model (HR, 95% CI, P) | Model I (HR, 95%CI, P) |
Model II (HR, 95%CI, P) |
|---|---|---|---|---|---|---|
| Total | 15,453 | 373 | 399. 14 | 1.49 (1.38, 1.61) < 0.001 | 1.49 (1.38, 1.61) < 0.001 | 1.30 (1.18, 1.43) < 0.001 |
| Q1 | 3861 | 47 | 213.21 | ref | ref | ref |
| Q2 | 3863 | 65 | 277.70 | 0.31 (0.22, 0.42) < 0.001 | 0.43 (0.31, 0.60) < 0.001 | 0.51 (0.36, 0.71) < 0.001 |
| Q3 | 3859 | 88 | 375.08 | 0.40 (0.30, 0.53) < 0.001 | 0.51 (0.38, 0.67) < 0.001 | 0.57 (0.43, 0.77) < 0.001 |
| Q4 | 3870 | 172 | 700.43 | 0.54 (0.42, 0.70) < 0.001 |
0.60 (0.47, 0.78) < 0.001 |
0.65 (0.50, 0.84) < 0.001 |
|
P for trend |
< 0.001 | < 0.001 | < 0.001 | < 0.001 |
Crude Model: we did not adjust for other covariants
Model I: we adjusted for gender, age.
Model II: we adjusted for gender, age, ethanol consumption, smoking status, habit of exercise, SBP, DBP, ALT, AST, GGT, TC, HbA1c, and FPG. Abbreviations: VAI, visceral adiposity index; HR, hazard ratio; CI, confidence interval; Ref, Reference
Kaplan-Meier curves for the probability of diabetes-free survival were presented in Fig. 2. A significant difference was observed among the four VAI groups in terms of diabetes risk (P < 0.001). The probability of diabetes-free survival gradually declined as VAI levels increased, indicating that individuals in the highest VAI group were at the greatest risk of developing diabetes.
Fig. 2.
Kaplan–Meier event-free survival curve of incident diabetes based on VAI quartiles
In Fig. 3, the RCS analyses were used to explore the nonlinear correlation between VAI and incident diabetes, and revealed a nonlinear association between VAI and diabetes after adjusting for gender, ethanol consumption, habit of exercise, smoking status, age, DBP, SBP, AST, GGT, ALT, TC, FPG, and HbA1c. The spline model indicates diabetes risks started to increase rapidly after a VAI level of 0.755, while there was no correlation between VAI levels and incident diabetes when VAI was less than 0.755 (P = 0.035).
Fig. 3.
RCS model showing the nonlinear association between the VAI and incident diabetes
Further subgroup analyses were performed on several factors, including gender, age, WC, BMI, smoking status, and alcohol consumption (Fig. 4). Notably, a stronger association between VAI and diabetes risk was observed in participants with female, age ≥ 60 years old, WC < 90 cm in men or < 80 cm in women, BMI < 25 kg/m2, past-smoking, and no/little-drinking. Conversely, a weaker association between VAI and diabetes risk was observed in participants with male, age < 60 years old, WC ≥ 90 cm in men or ≥ 80 cm in women, BMI ≥ 25 kg/m2, never or current-smoking, and little, moderate or heavy-drinking.
Fig. 4.
Subgroups analysis based on the Cox proportional hazard model
Discussion
This retrospective study investigated the association between VAI and the risk of incident diabetes in Japan. The findings indicate that VAI is a sensitive indicator of visceral obesity, and higher levels of VAI are associated with an increased risk of diabetes. The association between VAI levels and diabetes onset was found to be nonlinear, with a significant increase in risk observed when VAI was greater than 0.75.
Diabetes is a chronic disease that affects millions of people worldwide. Risk factors for diabetes include age, genetics, poor nutrition and obesity, lack of exercise, hypertension, and gestational diabetes. Non-alcoholic fatty liver disease has also been associated with increased risk of diabetes [1, 25, 26]. It is crucial to identify, address, and effectively manage diabetes risk factors to prevent or delay the onset of type 2 diabetes. However, in the present study, individuals with incident diabetes in these populations exhibited a lower BMI, smaller WC, and lower proportion of fatty liver, which may be surprising given that obesity and fatty liver are well-established risk factors for diabetes. On the other hand, there was a higher proportion of individuals with exercise habits within the incident diabetes group, which is a known protective factor against diabetes. Therefore, it is imperative to increase the sample size for further studies to ensure the robustness and generalizability of the results.
Obesity and the accumulation of visceral fat are significant contributors to the development of diabetes mellitus [2–5, 21]. A comprehensive study of Chinese adults determined that BMI and WC were strongly associated with an increased risk of developing type 2 diabetes, but that the amount of visceral fat present was an even stronger indicator of risk [27]. Visceral fat specifically refers to the fat that accumulates around the organs in the abdominal cavity, and it can have profound impacts on overall health. Excessive visceral fat has been shown to contribute to the development of diabetes by producing cytokines that impede insulin action, promoting insulin resistance and hyperglycemia [2–7]. Additionally, visceral fat can contribute to other metabolic disorders such as dyslipidemia, hypertension, and cardiovascular disease [28–30]. Recent studies have highlighted the close relationship between visceral fat and diabetes, with one investigation demonstrating a significant correlation between increased visceral fat and type 2 diabetes risk [31]. The ROC curve analysis in this study also made clear that relying solely on BMI or waist circumference as diagnostic indicators is insufficient, and that visceral fat content (VAI) offers a more precise means of assessing diabetes risk.
The VAI is a novel, precise tool for assessing the degree of visceral fat accumulation in the human body, and represents a valid measure for predicting an individual’s risk of developing cardiovascular disease and diabetes [16–35]. It is calculated by factoring in a combination of variables, such as body weight, waist circumference, height, and levels of TG and HDL-C. The VAI boasts several advantages, including ease of use and affordability. A higher VAI index indicates increased amounts of cardiac fat deposits, which have been linked to higher incidences of both diabetes and cardiovascular disease. The reference values for VAI vary depending on age and sex, and studies suggest that even lifestyle and diet can influence these values. The present study also found that the top 25% of individuals with highest VAI values had higher levels of blood pressure, blood glucose, blood lipids, BMI, and WC. Additionally, this group had a higher prevalence of fatty liver, with a larger proportion of males and more severe smoking and alcohol consumption habits. On the contrary, individuals with exercise habits showed lower VAI values. These may explain why VAI has weaker predictive ability for incident diabetes in individuals with higher BMI or larger WC.
To establish the validity of VAI as an indicator for diabetes risk, this study employed three COX regression models that included VAI both as a continuous and categorical variable. The results clearly indicated that VAI is indeed a reliable indicator of the likelihood of diabetic development. Adjusting for a range of other variables, including gender, age, smoking status, alcohol consumption, physical activity, SBP, DBP, ALT, AST, GGT, TC, FPG, and HbA1c, the RCS curve analysis indicated a non-linear relationship between VAI and diabetes (P = 0.035). The curve slope significantly increased at a VAI index of 0.75, which suggests that this point should be closely scrutinized by medical professionals. Additionally, subgroup analysis revealed that VAI’s predictive capacity is strongly influenced by factors such as gender, age, and lifestyle behaviors [16–22, 35]. These findings are consistent with previous research. In comparison to a large-scale, 6-period cross-sectional study of US adults from the NHANES database, the Japanese population exhibited generally lower VAI scores. However, both investigations revealed a positive, nonlinear relationship between VAI and diabetes (incidence or prevalence), with a more pronounced correlation in females. The direction of the effect was consistent across studies, and the strength of the association was comparable. It is worth noting that the NHANES study was a cross-sectional prevalence analysis and did not provide clear cutoff values [36].
Despite its valuable findings, this study had several limitations due to its retrospective nature. Firstly, it was not possible to continuously monitor indicators such as BMI, WC, TG, and HDL-C levels in the study population, and there may have been unmeasured or uncontrolled confounding variables, such as income, education, dietary habits, medication use, and family history of diabetes. Secondly, the exclusion of individuals with excessive alcohol consumption, viral hepatitis, or drug abuse means that the study’s results may not be applicable to a broader population. Thirdly, imaging methods were not employed to calculate visceral fat, and the VAI’s accuracy in gauging visceral fat was not evaluated. Lastly, the study was unable to continuously track changes in blood glucose levels in the study subjects, making it difficult to accurately determine the correlation between VAI and fluctuations in blood glucose.
Conclusion
This retrospective study provides evidence of the association between VAI and the risk of incident diabetes in a Japanese population. The findings suggest that VAI is a sensitive indicator of visceral obesity and an important predictor of diabetes risk. However, further studies are needed to validate these findings and establish the utility of VAI as a diagnostic tool in clinical practice. The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.
Acknowledgements
The authors express gratitude to the participants and researchers who dedicated their time and effort during the data collection phase of the NAGALA cohort. We also thank the Dryad platform for allowing us to access and download the NAGALA dataset.
Author contributions
Conception and design of the research: ZNY, JW. Acquisition of data: ZNY, JW. Analysis and interpretation of the data: ZNY, JW. Statistical analysis: ZNY, JW. Obtaining financing: ZNY, XWZ. Writing of the manuscript: ZNY, WXZ. Critical revision of the manuscript for intellectual content: JW, WXZ. All authors have read and approved the final manuscript submitted.
Funding
The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.
Data availability
The datasets used in this study are available at online repositories, and their names and accession numbers are provided below: https://datadryad.org/stash/dataset/doi:10.5061%2Fdryad.8q0p192.
Declarations
Conflict of interests
The authors declare no competing interests.
Ethics approval
The studies involving human participants were reviewed and approved by the Institutional Review Board of the Murakami Memorial Hospital.
Informed consent
The patients/participants provided their written informed consent to participate in this study.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.IDF Diabetes Atlas|Tenth, Edition. [(accessed on 20 February 2023)]. Available online: https://diabetesatlas.org/
- 2.S.E. Kahn, R.L. Hull, K.M. Utzschneider, Mechanisms linking obesity to insulin resistance and type 2 diabetes. Nature. 444(7121), 840–846 (2006). 10.1038/nature05482 [DOI] [PubMed] [Google Scholar]
- 3.C. Rattarasarn, Dysregulated lipid storage and its relationship with insulin resistance and cardiovascular risk factors in non-obese Asian patients with type 2 diabetes. Adipocyte. 7(2), 71–80 (2018). 10.1080/21623945.2018.1429784 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.J.H. Lee, J.H. Choi, K.B. Jeong et al., Safety and utility of rush immunotherapy with aqueous allergen extracts for treatment of respiratory allergies. J. Korean Med. Sci. 36(3), e18 (2021). Published 2021 Jan 18. 10.3346/jkms.2021.36.e18 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.American Diabetes Association Professional Practice Committee. 8. Obesity and Weight Management for the Prevention and Treatment of Type 2 Diabetes: Standards of Care in Diabetes-2025. Diabetes Care, 2025;48(1 Suppl 1):S167-S180. 10.2337/dc25-S008 [DOI] [PMC free article] [PubMed]
- 6.C. Iacobini, G. Pugliese, C. Blasetti Fantauzzi, M. Federici, S. Menini, Metabolically healthy versus metabolically unhealthy obesity. Metabolism. 92, 51–60 (2019). 10.1016/j.metabol.2018.11.009 [DOI] [PubMed] [Google Scholar]
- 7.M.E. Piché, A. Tchernof, J.P. Després, Obesity Phenotypes, Diabetes, and cardiovascular diseases. Circ. Res. 126(11), 1477–1500 (2020). 10.1161/CIRCRESAHA.120.316101 [DOI] [PubMed] [Google Scholar]
- 8.Ł. Nogajski, M. Mazuruk, M. Kacperska et al., Epicardial fat density obtained with computed tomography imaging - more important than volume? Cardiovasc Diabetol. 2024;23(1):389. Published 2024 Oct 29. 10.1186/s12933-024-02474-x [DOI] [PMC free article] [PubMed]
- 9.T. Baum, C. Cordes, M. Dieckmeyer et al., MR-based assessment of body fat distribution and characteristics. Eur. J. Radiol. 85(8), 1512–1518 (2016). 10.1016/j.ejrad.2016.02.013 [DOI] [PubMed] [Google Scholar]
- 10.R. Ross, I.J. Neeland, S. Yamashita et al., Waist circumference as a vital sign in clinical practice: a consensus statement from the IAS and ICCR working group on visceral obesity. Nat. Rev. Endocrinol. 16(3), 177–189 (2020). 10.1038/s41574-019-0310-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Z. Chen, H. Hu, M. Chen et al., Association of Triglyceride to high-density lipoprotein cholesterol ratio and incident of diabetes mellitus: a secondary retrospective analysis based on a Chinese cohort study. Lipids Health Dis. 2020;19(1):33. Published 2020 Mar 4. 10.1186/s12944-020-01213-x [DOI] [PMC free article] [PubMed]
- 12.T.K. Lim, H.S. Lee, Y.J. Lee, Triglyceride to HDL-cholesterol ratio and the incidence risk of type 2 diabetes in community dwelling adults: A longitudinal 12-year analysis of the Korean genome and epidemiology study. Diabetes Res. Clin. Pract. 163, 108150 (2020). 10.1016/j.diabres.2020.108150 [DOI] [PubMed] [Google Scholar]
- 13.D. Zheng, H. Li, F. Ai et al., Association between the triglyceride to high-density lipoprotein cholesterol ratio and the risk of type 2 diabetes mellitus among Chinese elderly: the Beijing longitudinal study of aging. BMJ Open. Diabetes Res. Care. 8(1), e000811 (2020). 10.1136/bmjdrc-2019-000811 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.D.J. Rader, Effect of insulin resistance, dyslipidemia, and intra-abdominal adiposity on the development of cardiovascular disease and diabetes mellitus. Am. J. Med. 120(3 Suppl 1), S12–S18 (2007). 10.1016/j.amjmed.2007.01.003 [DOI] [PubMed] [Google Scholar]
- 15.J.P. Després, Intra-abdominal obesity: an untreated risk factor for type 2 diabetes and cardiovascular disease. J. Endocrinol. Invest. 29(3 Suppl), 77–82 (2006) [PubMed] [Google Scholar]
- 16.M.C. Amato, C. Giordano, M. Galia et al., Visceral adiposity index: a reliable indicator of visceral fat function associated with cardiometabolic risk. Diabetes Care. 33(4), 920–922 (2010). 10.2337/dc09-1825 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.M. Zhang, L. Zheng, P. Li et al., 4-Year trajectory of visceral adiposity index in the development of type 2 diabetes: A prospective cohort study. Ann. Nutr. Metab. 69(2), 142–149 (2016). 10.1159/000450657 [DOI] [PubMed] [Google Scholar]
- 18.de C.M. Oliveira, J.L. Pavani, C. Liu et al., Comparing different metabolic indexes to predict type 2 diabetes mellitus in a five years follow-up cohort: the Baependi heart study. PLoS One. 17(6), e0267723 (2022). Published 2022 Jun 3. 10.1371/journal.pone.0267723 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.R. Nusrianto, D.L. Tahapary, P. Soewondo, Visceral adiposity index as a predictor for type 2 diabetes mellitus in Asian population: A systematic review. Diabetes Metab. Syndr. 13(2), 1231–1235 (2019). 10.1016/j.dsx.2019.01.056 [DOI] [PubMed] [Google Scholar]
- 20.P.J. Liu, F. Ma, H.P. Lou, Y. Chen, Visceral adiposity index is associated with Pre-Diabetes and type 2 diabetes mellitus in Chinese adults aged 20–50. Ann. Nutr. Metab. 68(4), 235–243 (2016). 10.1159/000446121 [DOI] [PubMed] [Google Scholar]
- 21.F. Shen, C. Guo, D. Zhang, Y. Liu, P. Zhang, Visceral adiposity index as a predictor of type 2 diabetes mellitus risk: A systematic review and dose-response meta-analysis. Nutr. Metab. Cardiovasc. Dis. 34(4), 811–822 (2024). 10.1016/j.numecd.2023.04.009 [DOI] [PubMed] [Google Scholar]
- 22.M.J. Park, M. Kang, S.Y. Jang et al., Sex- and age-specific body composition indices as predictors of new-onset type 2 diabetes mellitus in koreans: a nationwide cohort study. BMJ Open. 15(3), e093598 (2025). Published 2025 Mar 18. 10.1136/bmjopen-2024-093598 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.T. Okamura et al., Data from: ectopic fat obesity presents the greatest risk for incident type 2 diabetes: a population-based longitudinal study. 2019. Dryad Dataset. 10.5061/dryad.8q0p192 [DOI] [PubMed]
- 24.T. Okamura, Y. Hashimoto, M. Hamaguchi, A. Obora, T. Kojima, M. Fukui, Ectopic fat obesity presents the greatest risk for incident type 2 diabetes: a population-based longitudinal study. Int. J. Obes. (Lond). 43(1), 139–148 (2019). 10.1038/s41366-018-0076-3 [DOI] [PubMed] [Google Scholar]
- 25.American Diabetes Association. 2. Classification and Diagnosis of Diabetes: Standards of Medical Care in Diabetes-2021. Diabetes Care, 2021;44(Suppl 1):S15-S33. 10.2337/dc21-S002 [DOI] [PubMed]
- 26.A.P. Kengne, J.W. Beulens, L.M. Peelen et al., Non-invasive risk scores for prediction of type 2 diabetes (EPIC-InterAct): a validation of existing models. Lancet Diabetes Endocrinol. 2(1), 19–29 (2014). 10.1016/S2213-8587(13)70103-7 [DOI] [PubMed] [Google Scholar]
- 27.Q. Wu, B. Li, Y. Wang et al., Association of weight-adjusted-waist index with type 2 diabetes mellitus in Chinese urban adults: a cross-sectional study. Front. Endocrinol. (Lausanne). 16, 1460230 (2025). Published 2025 Feb 10. 10.3389/fendo.2025.1460230 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.B.M. Hamooya, L. Siame, L. Muchaili, S.K. Masenga, A. Kirabo, Metabolic syndrome: epidemiology, mechanisms, and current therapeutic approaches. Front. Nutr. 12, 1661603 (2025). Published 2025 Sep 3. 10.3389/fnut.2025.1661603 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.M.B. Schulze, N. Stefan, Metabolically healthy obesity: from epidemiology and mechanisms to clinical implications. Nat. Rev. Endocrinol. 20(11), 633–646 (2024). 10.1038/s41574-024-01008-5 [DOI] [PubMed] [Google Scholar]
- 30.P. Mathieu, P. Poirier, P. Pibarot, I. Lemieux, J.P. Després, Visceral obesity: the link among inflammation, hypertension, and cardiovascular disease. Hypertension. 53(4), 577–584 (2009). 10.1161/HYPERTENSIONAHA.108.110320 [DOI] [PubMed] [Google Scholar]
- 31.Y. Zhang, W. Gao, B. Li et al., The association between the visceral obesity indices and the future diabetes mellitus risk: A prospective cohort study. Diabetes Obes. Metab. 27(8), 4490–4498 (2025). 10.1111/dom.16492 [DOI] [PubMed] [Google Scholar]
- 32.X. Zhou, S. Wang, D. Yu, T. Niu, A prediction nomogram for mild cognitive impairment in type 2 diabetes mellitus based on the Chinese visceral adiposity index. Sci Rep. 2025;15(1):32393. Published 2025 Sep 9. 10.1038/s41598-025-17241-5 [DOI] [PMC free article] [PubMed]
- 33.R. Wang, J. Liu, G. Fang, J. Shi, C. Zhang, Y. Huang, Association between visceral adiposity index and cardiovascular disease: A systematic review and meta-analysis. Nutr. Metab. Cardiovasc. Dis. Published Online July. 2 (2025). 10.1016/j.numecd.2025.104216 [DOI] [PubMed]
- 34.S. Kang, S. Chen, Z. Lin et al., Association between Chinese visceral adiposity index and heart failure in diabetic patients. Diabetes Obes. Metab. Published Online August. 22 (2025). 10.1111/dom.70053 [DOI] [PubMed]
- 35.D. Zooravar, S. Shojaei, A. Mousavi, P. Soltani, B.S. Amiri, H. Radkhah, Novel lipid biomarkers and microvascular complications in patients with diabetes mellitus: A systematic review and Meta-analysis. Clin. Med. Insights Endocrinol. Diabetes. 18, 11795514251365301 (2025). Published 2025 Aug 16. 10.1177/11795514251365301 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.H. Zhou, T. Li, J. Li, X. Zhuang, J. Yang, The association between visceral adiposity index and risk of type 2 diabetes mellitus. Sci Rep. 2024;14(1):16634. Published 2024 Jul 18. 10.1038/s41598-024-67430-x [DOI] [PMC free article] [PubMed]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets used in this study are available at online repositories, and their names and accession numbers are provided below: https://datadryad.org/stash/dataset/doi:10.5061%2Fdryad.8q0p192.