Associations of quantitative computed tomography measured abdominal and ectopic fat with impaired glucose metabolism in overweight/obese Chinese adults

Abstract

BACKGROUND

Obesity and ectopic fat deposition are well-established correlates of metabolic disorders, including type 2 diabetes mellitus (DM). However, the respective contributions of distinct adipose depots—namely, visceral adipose tissue (VAT), hepatic fat fraction (HFF), and subcutaneous adipose tissue (SAT)—to impaired glucose metabolism remain insufficiently elucidated. Furthermore, whether the associations between these fat depots and glucose metabolism differ according to sex remains poorly understood.

AIM

To evaluate the sex-specific associations of abdominal and ectopic fat with impaired glucose metabolism in overweight/obese Chinese adults using quantitative computed tomography.

METHODS

This study included 205 participants (100 males, 105 females) with a body mass index ≥ 24 kg/m², aged between 18 and 50 years. All participants underwent a standard 75-g oral glucose tolerance test. Subjects were classified into three categories according to glucose status: Normal glucose tolerance (NGT), impaired glucose tolerance (IGT), and DM, in both sexes. Abdominal computed tomography was performed to quantify abdominal fat and fat deposition within the liver, pancreas, and skeletal muscle. Impaired glucose metabolism (IGM) was defined as the presence of IGT or DM. All data were analyzed separately by sex.

RESULTS

HFF was found to be significantly higher in IGM patients than in NGT subjects in both sexes. After adjusting for age and dyslipidemia, the odds ratios for IGM prevalence in the highest tertile of HFF were 5.24 (95%CI: 1.64–16.73) in males and 6.45 (95%CI: 2.12–19.63) in females. Further analysis demonstrated that VAT was also significantly elevated in IGM patients compared with NGT subjects in both sexes. For VAT, the odds ratios for IGM were 6.56 (95%CI: 1.55–27.86) in males and 3.29 (95%CI: 1.17–9.29) in females. SAT levels were significantly lower in participants with DM than in those with NGT or IGT. In males, the odds ratio for IGM associated with SAT was 0.13 (95%CI: 0.03–0.60).

CONCLUSION

The results demonstrated that both HFF and VAT were independently associated with IGM in Chinese overweight/obese subjects of both sexes. In male subjects, higher SAT was associated with lower odds of IGM. These findings suggest that hepatic fat accumulation and visceral adiposity are closely related to glucose metabolism disorders, and that sex differences in abdominal fat distribution may contribute to heterogeneity in glucose metabolism among Chinese overweight/obese individuals.

Key Words: Obesity; Fat distribution; Impaired glucose metabolism; Type 2 diabetes; Metabolic disease

Core Tip: This cross-sectional study of overweight/obese Chinese adults utilized quantitative computed tomography to evaluate the sex-specific associations of abdominal and ectopic fat depots with impaired glucose metabolism (IGM). Hepatic fat fraction (HFF) and visceral adipose tissue (VAT) were independently associated with IGM in both sexes, with odds ratios ranging from 5.24 to 6.45 for HFF and from 3.29 to 6.56 for VAT. Notably, higher subcutaneous adipose tissue (SAT) was associated with lower odds of IGM exclusively in males (odds ratio = 0.13). These findings underscore that sex differences in abdominal fat distribution may contribute to heterogeneity in glucose metabolism disorders.

INTRODUCTION

Numerous studies have established that obesity is associated with an increased risk of metabolic disorders, including type 2 diabetes mellitus (T2DM), dyslipidemia, hypertension, cardiovascular disease, and various types of cancer[1]. Accumulating evidence suggests that excess adiposity in specific body depots may exert differential effects on the pathogenesis of diabetes. In particular, abdominal adiposity—especially visceral adipose tissue (VAT)—has been consistently linked to an elevated risk of diabetes in most prior studies[2,3]. Compared with subcutaneous adipose tissue (SAT), VAT is more metabolically active and releases a broader array of pro-inflammatory cytokines, contributing to systemic low-grade inflammation and insulin resistance[4]. However, the role of SAT in diabetes development remains debated[5,6].

Currently, obesity is commonly defined using body mass index (BMI). Nevertheless, it has been observed that certain individuals classified as obese by BMI may not develop metabolic disorders, at least over an extended period. This subgroup has been termed "metabolically healthy obesity"[7]. This phenomenon may be partially explained by the "spillover theory"[8], which posits that when adipose tissue ceases to expand in cell size or number, excess lipids accumulate in ectopic sites such as the skeletal muscle, pancreas, and liver, where they promote insulin resistance, β-cell dysfunction, and diabetes[9]. This hypothesis is further supported by single-gene studies of extreme obesity and lipodystrophy—two apparently opposing conditions[10,11]. Nonetheless, the relative contribution of each specific fat depot to the pathogenesis of diabetes remains insufficiently characterized.

Compared with non-Asian populations, Asian individuals generally exhibit lower BMI levels; however, they tend to have higher adiposity per unit BMI, which may predispose them to an increased risk of T2DM even at relatively low BMI thresholds[12]. Chinese imaging studies have documented sex-specific heterogeneity in abdominal fat distribution and its associated metabolic risks. For instance, a recent study by Chen et al[13], based on the Shanghai Nicheng cohort comprising 12137 adults aged 45–70 years, examined the associations of abdominal visceral and subcutaneous fat with glucose metabolism and found that SAT exerted a protective effect against newly diagnosed diabetes in women, but not in men. In a nationwide quantitative computed tomography (Q-CT) study of 59429 Chinese adults, abdominal adiposity varied substantially by age and sex, underscoring the importance of sex-stratified analyses[14]. Similarly, a study of 1449 Chinese adults from the Yinchao community in Chengdu reported stronger associations between VAT and glucose-related metabolic risk than those observed for SAT[15]. In contrast, a Japanese study demonstrated that both VAT and SAT were positively associated with diabetes risk, particularly among women[16].

Notably, although the aforementioned studies measured visceral and subcutaneous adipose areas using magnetic resonance imaging or computed tomography (CT), they did not assess ectopic fat depots. Moreover, previous investigations have typically used directly measured visceral and subcutaneous adipose areas as independent variables in association analyses, without accounting for variations in body size. Given that these measures can differ considerably among individuals with the same BMI, the present study concurrently quantified abdominal and ectopic fat depots—including VAT, SAT, total adipose tissue (TAT), and hepatic, intramuscular, and pancreatic fat fractions—using Q-CT in Chinese adults with overweight or obesity. Furthermore, we innovatively designed this study to explore whether the metabolic relevance of SAT differs by sex, after adjusting for visceral and ectopic fat deposition—a dimension that has rarely been simultaneously examined in prior Chinese research.

MATERIALS AND METHODS

Study participants

Participants were recruited from the endocrinology outpatient clinics of Qingdao Municipal Hospital between March 2022 and September 2023. Over this period, potentially eligible outpatients were screened, and a total of 227 individuals were initially enrolled. Of these, 22 were subsequently excluded due to incomplete oral glucose tolerance test (OGTT), Q-CT, or laboratory data, yielding a final analytic sample of 205 participants (100 males and 105 females). As this was a cross-sectional study, all assessments were conducted at the enrollment visit; therefore, no longitudinal follow-up was performed, and no participants were lost after inclusion. The study protocol adhered to the principles of the Declaration of Helsinki and was approved by the Ethics Review Committee of Qingdao Municipal Hospital (2022 LSZ No. 90). Written informed consent was obtained from all participants.

All participants completed a questionnaire covering current and past illnesses and medication use, and underwent physical examinations and biochemical tests. The inclusion criteria were age 20–50 years and BMI ≥ 24 kg/m². The exclusion criteria were as follows: (1) Secondary obesity due to Cushing’s syndrome, hypothyroidism, or drug-induced causes; (2) Liver injury of known etiology, including alcohol abuse, autoimmune liver disease, or viral hepatitis; (3) Use of anti-diabetic or anti-obesity medications within the preceding 6 months; and (4) Type 1 diabetes.

Measurements

A standardized 75-g OGTT was performed for all participants. Blood samples were collected at fasting and at 30, 60, and 120 minutes after glucose ingestion for measurement of glucose and insulin levels.

Anthropometric indicators, including weight, height, and systolic and diastolic blood pressure, were measured by trained investigators. Height was recorded without shoes to the nearest centimeter; weight was measured with light clothing to the nearest 0.1 kg. BMI was calculated as weight in kilograms divided by height in meters squared (kg/m²). Blood pressure was measured twice at 3-minute intervals using a mercury sphygmomanometer from the right arm in the sitting position after 5 minutes of rest, and the mean value was used for analysis.

Laboratory assays

Blood glucose and insulin were measured using an enzymatic method and a radioimmunoassay kit (Roche Diagnostics, Germany), respectively. Fasting serum triglyceride (TG), total cholesterol (TC), and high-density lipoprotein cholesterol (HDL-C) were measured by colorimetric methods using an autoanalyzer (Modular E170, Roche). Low-density lipoprotein cholesterol (LDL-C) was calculated using Friedewald’s formula.

Definitions

Overweight and obesity were defined as BMI ≥ 24 kg/m² and ≥ 28 kg/m², respectively[17]. Glucose status was classified according to the 1999 World Health Organization diagnostic criteria: Diabetes mellitus (DM) was defined as fasting plasma glucose (FPG) ≥ 7.0 mmol/L and/or 2-hour postprandial glucose (2hPG) ≥ 11.1 mmol/L; impaired glucose tolerance (IGT) as 7.8 mmol/L ≤ 2hPG < 11.1 mmol/L; and normal glucose tolerance (NGT) as FPG < 6.1 mmol/L and 2hPG < 7.8 mmol/L during OGTT[18]. Hypertension was defined as systolic blood pressure ≥ 140 mmHg and/or diastolic blood pressure ≥ 90 mmHg. Dyslipidemia was defined as the presence of any of the following: TG ≥ 2.26 mmol/L, TC ≥ 6.22 mmol/L, LDL-C ≥ 4.14 mmol/L, or HDL-C < 1.03 mmol/L.

Assessment of β-cell function and insulin resistance based on OGTT

In stages of IGT and new-onset diabetes, β-cell impairment is characterized by deficiency in early-phase insulin secretion. In this study, we used the disposition index (DI_early) to assess β-cell function, as previously described[19]. Insulin resistance was evaluated using the homeostatic model assessment of insulin resistance (HOMA-IR): HOMA-IR = Gluc₀ × Ins₀/22.5. Early-phase insulin secretion was estimated using the insulinogenic index: (Ins₃₀ − Ins₀)/ (Gluc₃₀ − Gluc₀)/HOMA-IR[20], where Gluc₀, Ins₀, Gluc₃₀, and Ins₃₀ denote plasma glucose and insulin levels at 0 and 30 minutes after oral glucose challenge, respectively. DI_early was calculated as early-phase insulin secretion adjusted for HOMA-IR[21].

Q-CT measurements of fat distribution

A standard abdominal CT scan was performed for each participant using a CT scanner (GE Revolution) with a Mindways Q-CT Pro calibration phantom (Mindways, Austin, TX, United States) positioned under the back. Abdominal adipose tissue assessment followed a standard protocol as previously described[14].

Abdominal total cross-sectional area (CSA), total adipose tissue (TAT) area, and visceral adipose tissue (VAT) area were determined at the level of the 2^nd lumbar vertebra. TAT and VAT areas were automatically measured on 1.25-mm thick slices using the Tissue Composition Module of the software (Mindways, Austin, TX, United States). Subcutaneous adipose tissue (SAT) area at the same level was calculated by subtracting VAT area from TAT area. TAT, VAT, and SAT areas were adjusted for CSA to represent relative fat contents.

Liver fat assessment was performed as previously described[22]. Briefly, each participant underwent Q-CT scanning with a five-rod calibration phantom placed beneath the spine (Model 3 phantom; Mindways Software, Austin, TX, United States). Scan parameters were 120 kVp; 150-200 mAs; FOV, 500 mm; and slice thickness, 1-1.25 mm. A slice of the liver where the right branch of the portal vein enters was selected. Three regions of interest (ROIs), each of 250 mm², were placed on that slice in the periphery of the left lobe, right anterior lobe, and right posterior lobe, avoiding major blood vessels and bile ducts. The fat percentage of the ROIs was calculated using Mindways Software, and the averaged value was used as the hepatic fat fraction (HFF).

Similarly, skeletal muscle fat assessment was performed in the psoas muscle (PM). The CSA of PM and the area of intramuscular fat at the level of the 2^nd lumbar vertebra were calculated by Mindways Software. Muscle fat fraction (MFF) was defined as the area of intramuscular fat divided by the CSA of PM.

Pancreatic fat assessment was also performed using the Q-CT phantom. Three circular ROIs, each approximately 100 mm², were selected in the pancreatic head, avoiding intrapancreatic vessels and ducts. Mindways Software was used to calculate the fat percentage of each ROI, and the mean value of these measurements was used as the pancreatic fat fraction (PFF)[23].

Statistical analysis

All analyses were conducted using SPSS version 22 (IBM, Armonk, NY, United States). Normally distributed variables are presented as the mean ± SD, and skewed variables as median with interquartile range. Differences in continuous variables among groups were compared using one-way analysis of variance or the Kruskal-Wallis test, as appropriate. Differences in categorical variables were evaluated using the χ² test. Binary logistic regression analysis was performed to estimate the associations between adiposity variables and the risk of impaired glucose metabolism (IGM). VAT, SAT, and HFF in male participants, and VAT and HFF in female participants were included in the primary binary logistic regression analysis using visual binning. SAT was not included in the primary model for females because no significant between-group difference was observed.

Although participants were classified into NGT, IGT, or DM for descriptive purposes, the primary regression outcome was binary, combining IGT and DM as IGM and using NGT as the reference. Therefore, primary analyses were not performed separately within the three relatively small glucose-status subgroups. To avoid assuming a linear dose-response relationship and to ensure adequate sample sizes within each subgroup for sex-stratified analyses, HFF, VAT, and SAT were categorized into sex-specific tertiles, with cutoff points at the 33.3^rd and 66.7^th percentiles. These thresholds were derived independently for men and women from the analytical cohort using the Visual Binning procedure in SPSS. The lowest tertile served as the reference category, and the resulting cutoff values are presented in Table 1 and Supplementary Table 1. Blood glucose status (NGT or IGM) was used as the dependent variable. Variables that were clinically relevant or had a P value < 0.05 in univariate analysis were included as covariates in the multivariate models; ultimately, age and dyslipidemia were entered. To further explore the potential effects of adiposity variables on glucose metabolism, Spearman correlation tests were performed to assess relationships between these variables and DI_early, a key factor in the pathogenesis of T2DM. All statistical tests were two-sided, and statistical significance was set at P < 0.05.

Table 1 Binary logistic regression analysis of risk factors for impaired glucose metabolism in both sexes.

Item	Crude		Model 1		Model 2
	OR (95%CI)	P value	OR (95%CI)	P value	OR (95%CI)	P value
Men
HFF
Q1 ≤ 13.30	1.00 (ref)		1.00 (ref)		1.00 (ref)
Q2: 13.30-21.30	3.50 (1.27-9.64)	0.015	3.32 (1.17-9.40)	0.024	3.45 (1.18-10.13)	0.024
Q3 > 21.30	6.50 (2.17-19.44)	0.001	5.85 (1.89-18.16)	0.002	5.24 (1.64-16.73)	0.005
VAT
Q1 ≤ 0.26	1.00 (ref)		1.00 (ref)		1.00 (ref)
Q2: 0.26-0.33	2.14 (0.79-5.83)	0.135	1.80 (0.62-5.18)	0.278	1.59 (0.54-4.70)	0.406
Q3 > 0.33	11.67 (3.28-41.49)	< 0.001	8.54 (2.07-35.31)	0.003	6.56 (1.55-27.86)	0.011
SAT
Q1 ≤ 0.24	1.00 (ref)		1.00 (ref)		1.00 (ref)
Q2: 0.24-0.35	0.18 (0.05-0.65)	0.008	0.20 (0.05-0.75)	0.017	0.21 (0.05-0.79)	0.207
Q3 > 0.35	0.09 (0.02-0.31)	< 0.001	0.11 (0.03-0.49)	0.004	0.13 (0.03-0.60)	0.008
Women
HFF
Q1 ≤ 8.30	1.00 (ref)		1.00 (ref)		1.00 (ref)
Q2: 8.30-17.50	1.90 (0.70-5.16)	0.211	2.12 (0.76-5.93)	0.153	2.08 (0.73-5.96)	0.173
Q3 > 17.50	5.02 (1.79-14.05)	0.002	5.67 (1.95-16.49)	0.001	6.45 (2.12-19.63)	0.001
VAT
Q1 ≤ 0.20	1.00 (ref)		1.00 (ref)		1.00 (ref)
Q2: 0.20-0.26	1.28 (0.48-3.41)	0.619	1.25 (0.47-3.36)	0.653	1.25 (0.45-3.44)	0.667
Q3 > 0.26	3.08 (1.13-8.41)	0.028	3.02 (1.10-8.29)	0.032	3.29 (1.17-9.29)	0.024

Sex-specific tertile cutoffs were established at the 33.3% and 66.7% of each adiposity measure derived from the analytic sample. Q1, representing the lowest tertile, served as the reference category. Model 1 adjusted for age; model 2 adjusted for age and dyslipidemia. VAT: Visceral adipose tissue; HFF: Hepatic fat fraction; SAT: Subcutaneous adipose tissue; OR: Odds ratio; ref: Reference.

RESULTS

Clinical and biological characteristics of the participants at baseline

A total of 205 eligible participants (100 males and 105 females) were included in this study. Given the well-documented sex differences in fat distribution, all analyses were performed separately by sex. Participants of both sexes were classified into NGT, IGT, and DM subgroups. Baseline characteristics are summarized in Table 2. In both sexes, BMI was similar across glycemic subgroups and did not differ significantly. In contrast, age showed notable differences among subgroups stratified by glucose status in both men and women. Hypertension prevalence did not differ significantly across groups; however, a significant between-group difference was observed for dyslipidemia prevalence in males, whereas no such difference was found in females. Glucose metabolism parameters, including FPG, 2hPG, and HOMA-IR, progressively increased in patients with IGT and DM compared with the NGT group in both sexes. Notably, insulin secretion, as estimated by DI_early, declined significantly in patients with IGT and DM relative to the NGT group in both sexes. Moreover, statistically significant differences in DI_early were also observed between DM and IGT patients, with insulin secretion decreasing stepwise across the three groups in accordance with the severity of glucose intolerance. Regarding lipid profiles, TG levels were elevated and HDL-C levels were reduced in patients with IGT or DM compared with those with NGT. No significant differences were found in TC or LDL-C levels among subgroups in either sex.

Table 2 Baseline characteristics of the participants grouped by glucose metabolism status in both sexes, mean ± SD/median (interquartile range).

Variable	Men				Women
	NGT (n = 39)	IGT (n = 26)	DM (n = 35)	ANOVA/Kruskal-Wallis test (P value)	NGT (n = 54)	IGT (n = 27)	DM (n = 24)	ANOVA/Kruskal-Wallis test (P value)
Age (years)	28.95 ± 8.94	32.46 ± 9.16	37.17 ± 8.43b	< 0.001	29.56 ± 7.89	33.37 ± 6.46a	30.46 ± 9.57	0.130
BMI	35.34 ± 4.60	34.80 ± 4.41	33.32 ± 6.30	0.252	32.26 ± 4.19	34.69 ± 4.85a	33.19 ± 4.82	0.080
FPG (mmol/L)	5.10 (4.85-5.36)	5.99 (5.60-6.37)b	7.61 (6.69-8.87)b,d	< 0.001	5.08 (4.87-5.38)	5.53 (5.14-5.91)b	6.72 (5.04-7.46)b	< 0.001
2hPG (mmol/L)	5.74 (5.27-7.13)	8.88 (8.04-10.16)b	13.19 (12.06-17.09)b,d	< 0.001	6.51 (6.06-7.19)	8.70 (8.30-9.23)b	12.48 (11.96-15.50)b	< 0.001
TC (mmol/L)	4.80 (4.30-5.49)	5.39 (4.74-5.78)	4.92 (4.44-5.64)	0.153	4.72 (4.31-5.51)	4.73 (4.33-5.84)	4.78 (4.15-5.25)	0.648
TG (mmol/L)	1.42 (1.16-2.07)	1.91 (1.61-2.87)	2.1 (1.56-3.61)b	0.005	1.19 (0.93-1.59)	1.66 (1.28-2.19)b	1.40 (1.11-2.25)	0.006
LDL-C (mmol/L)	3.08 (2.55-3.58)	3.23 (2.90-3.71)	2.94 (2.64-3.43)	0.205	2.82 (2.65-3.27)	2.88 (2.63-3.54)	3.05 (2.44-3.32)	0.765
HDL-C (mmol/L)	1.23 (1.07-1.35)	1.09 (1.00-1.34)	1.08 (0.88-1.20)a	0.027	1.29 (1.04-1.46)	1.13 (0.94-1.22)	1.05 (0.94-1.13)b	0.004
DIearly	8.98 (6.35-12.46)	2.97 (1.85-4.58)b	0.73 (0.37-2.17)b,d	< 0.001	8.73 (5.56-10.82)	5.01 (3.50-7.31)a	1.63 (1.01-5.18)b	< 0.001
HOMA-IR	5.82 (3.55-8.22)	7.46 (6.08-11.48)	9.29 (5.31-15.65)a	0.010	4.40 (2.85-7.68)	6.03 (4.48-8.56)	6.98 (4.65-10.21)	0.038
Hypertension	6	7	14	0.059	7	7	7	0.172
Dyslipidemia	10	11a	20c	0.022	11	10	9	0.160

^aP < 0.05 vs normal glucose tolerance.

^bP < 0.01 vs normal glucose tolerance.

^cP < 0.05 vs impaired glucose tolerance.

^dP < 0.01 vs impaired glucose tolerance.

NGT: Normal glucose tolerance; IGT: Impaired glucose tolerance; DM: Diabetes mellitus; BMI: Body mass index; FPG: Fasting plasma glucose; 2hPG: 2-hour postprandial glucose; TC: Total cholesterol; TG: Triglyceride, HDL-C: High-density lipoprotein cholesterol; LDL-C: Low-density lipoprotein cholesterol; HOMA-IR: Homeostatic model assessment of IR; ANOVA: Analysis of variance.

Fat distribution variables based on Q-CT measurements

Fat distribution variables were determined by Q-CT, and the results are presented in Table 3. In both sexes, HFF was significantly higher in participants with DM than in those with NGT. In males, VAT was significantly increased in both IGT and DM groups compared with the NGT group, whereas SAT was significantly decreased in the DM group relative to the NGT group, suggesting that VAT and SAT may exert opposing effects on glucose metabolism. In females, VAT was significantly elevated in the DM group compared with the NGT group, while SAT did not differ significantly among subgroups. Other variables, including TAT, MFF, and PFF, showed no significant differences across subgroups in either sex.

Table 3 Fat distribution variables based on quantitative-computed tomography measurements, mean ± SD/median (interquartile range).

Variable	NGT	IGT	DM	ANOVA/Kruskal-Wallis test (P value)
Men	n = 39	n = 26	n = 35
HFF	12.6 (10.00-21.60)	21.15 (17.05-26.9)b	16.80 (13.13-24.48)	0.006
VAT	0.25 ± 0.07	0.30 ± 0.08a	0.32 ± 0.07b	< 0.001
SAT	0.34 ± 0.09	0.30 ± 0.10	0.23 ± 0.08b,d	< 0.001
TAT	0.59 ± 0.07	0.60 ± 0.11	0.55 ± 0.10	0.111
MFF	0.05 (0.03-0.08)	0.06 (0.03-0.10)	0.05 (0.04-0.08)	0.620
PFF	15.30 (11.10-19.80)	14.00 (11.58-17.80)	17.25 (12.83-22.93)	0.468
Women	n = 54	n = 27	n = 24
HFF	9.95 (5.58-15.08)	14.80 (7.83-22.50)	23.70 (12.55-29.13)b	< 0.001
VAT	0.22 ± 0.06	0.24 ± 0.05	0.25 ± 0.06a	0.060
SAT	0.39 ± 0.09	0.38 ± 0.09	0.36 ± 0.08	0.644
TAT	0.61 ± 0.10	0.62 ± 0.09	0.62 ± 0.10	0.761
MFF	0.06 (0.03-0.10)	0.07 (0.06-0.11)	0.07 (0.05-0.09)	0.259
PFF	15.50 (13.40-18.60)	15.50 (13.23-18.40)	17.40 (12.10-19.60)	0.548

^aP < 0.05 vs normal glucose tolerance.

^bP < 0.01 vs normal glucose tolerance.

^dP < 0.01 vs impaired glucose tolerance.

VAT: Visceral adipose tissue; SAT: Subcutaneous adipose tissue; HFF: Hepatic fat fraction; MFF: Muscle fat fraction; PFF: Pancreatic fat fraction; NGT: Normal glucose tolerance; IGT: Impaired glucose tolerance; ANOVA: Analysis of variance; TAT: Total adipose tissue; DM: Diabetes mellitus.

Multivariate logistic regression analysis of the associations between adiposity variables and IGM

Multivariate logistic regression analysis was performed to estimate the associations between adiposity variables and the risk of IGM. In males, HFF, VAT, and SAT were included as independent variables. In females, HFF and VAT were included, whereas SAT was excluded from the primary model because it showed no significant between-group differences. In males (Table 1, Model 2), after adjusting for age and dyslipidemia, the odds ratios for IGM prevalence in the highest tertile compared with the lowest tertile were 5.24 (95%CI: 1.64–16.73) for HFF, 6.56 (95%CI: 1.55–27.86) for VAT, and 0.13 (95%CI: 0.03–0.60) for SAT. These findings indicate that higher HFF and VAT were associated with increased odds of IGM, whereas higher SAT was associated with decreased odds of IGM in male participants. In females (Table 1), the odds ratios for IGM prevalence in the highest vs lowest tertile were 6.45 (95%CI: 2.12–19.63) for HFF and 3.29 (95%CI: 1.17–9.29) for VAT. In a sensitivity analysis for female participants, SAT was additionally included in the tertile-based logistic regression model. After adjusting for age and dyslipidemia, SAT was not independently associated with IGM in females (Supplementary Table 1).

Correlation of fat depots with β-cell function

β-cell dysfunction and insulin resistance are two major predisposing factors for type 2 diabetes in individuals with overweight or obesity. To elucidate the role of each fat depot in the pathogenesis of diabetes, we examined the relationships between these variables and DI_early using Spearman correlation analysis. As shown in Table 4, in males, DI_early was negatively correlated with VAT and HFF, and positively correlated with SAT. In females, DI_early was negatively correlated only with VAT.

Table 4 Correlation between fat depots and -cell function.

Fat depot	Men		Women
	ρ	P value	ρ	P value
HFF	-0.232	0.032	-0.113	0.271
VAT	-0.386	< 0.001	-0.252	0.015
SAT	0.357	< 0.001	0.093	0.375
TAT	0.049	0.661	-0.073	0.487
MFF	-0.095	0.389	-0.095	0.357
PFF	0.054	0.628	0.006	0.956

DI_early as dependent variable. VAT: Visceral adipose tissue; HFF: Hepatic fat fraction; SAT: Subcutaneous adipose tissue; TAT: Total adipose tissue; MFF: Muscle fat fraction; PFF: Pancreatic fat fraction.

DISCUSSION

Although abdominal adiposity and ectopic fat deposition have consistently been linked to metabolic disorders such as T2DM, findings across different sex, ethnic, and BMI groups remain inconsistent. In this sex-stratified Q-CT study of Chinese adults with overweight or obesity, VAT and HFF were consistently associated with IGM in both sexes, whereas a larger SAT depot was related to lower odds of IGM only in men. Collectively, these results strengthen the view that visceral and ectopic fat distribution provides metabolic information extending beyond BMI, and further suggest that the relationship between SAT and glucose metabolism is modified by sex and context. Given the cross-sectional design, these findings should be interpreted as associational rather than causal.

Abdominal adiposity, in particular VAT, has been shown to be closely associated with an increased risk of diabetes[2,24]. A proposed mechanism is that visceral fat produces and secretes pro-inflammatory cytokines, resulting in systemic inflammation and insulin resistance. In addition, a higher VAT area was shown to be associated with a lower first-phase insulin secretion rate in a cross-sectional study of 43 black South African women without T2DM and obesity[25]. In this study, Spearman correlation analysis was performed to assess the relationships between adiposity variables and DI_early and HOMA-IR. The results showed that VAT was negatively correlated with DI_early in both sexes, confirming that VAT is a crucial risk factor for β-cell dysfunction. In males, SAT was positively correlated with DI_early, suggesting that SAT may have a protective effect on β-cell function. The reason for this sex difference is unknown. We propose that reduced fat storage capacity of subcutaneous adipose tissue in overweight/obese males may increase VAT content, and in turn increase the risk of β-cell damage and IGM. In this study, no association between HOMA-IR and VAT or SAT was found in either sex (results not shown). It is possible that insulin resistance does not increase proportionally with changes in abdominal adiposity in this population, or that HOMA-IR is not an ideal indicator of insulin resistance.

In this study, our results pointed to a central role of hepatic fat deposition in the pathogenesis of IGM in Chinese overweight or obese individuals, which is consistent with findings from previous studies. A Korean study evaluating 29836 non-diabetic obese subjects found that those without fatty liver were protected from developing T2DM[26]. In a large American cohort study of 51463 adults without diabetes at baseline, stratified by BMI and metabolic dysfunction-associated steatotic liver disease (MASLD), lean MASLD participants (with normal BMI) showed a higher risk of incident diabetes than overweight/obese participants without MASLD after a follow-up of 4.0 years[27]. These studies strongly suggest that hepatic fat deposition may play a more important role than BMI in the development of diabetes. In a Chinese population, the association of lean MASLD with incident type 2 diabetes showed sex differences, with the effect of MASLD appearing more pronounced in women than in men, although the mechanism remains unknown[28]. A study from Japan showed that even transient remission of MASLD reduced the risk of incident diabetes in a Japanese population[29]. All these studies highlighted a central role of liver fat deposition in the pathogenesis of T2DM. How the accumulation of liver fat leads to IGM and drives the development of diabetes remains elusive. It has been shown that liver fat content is strongly correlated with peripheral and hepatic insulin resistance. In the liver, adipocytes secrete various pro-inflammatory cytokines, resulting in systemic inflammation. In addition, fat-derived diacylglycerol activates the protein kinase C pathway, which in turn interferes with insulin signaling and leads to hepatic insulin resistance[30]. Hepatic insulin resistance impairs insulin-mediated suppression of glucose production, which results from increased gluconeogenesis and decreased hepatic glycogen synthesis. In the short term, excess hepatic glucose output may promote insulin secretion, leading to more severe hyperinsulinemia, obesity, peripheral insulin resistance, and finally hyperglycemia. β-cell dysfunction is a key factor predisposing to diabetes in people with overweight/obesity. During the stages of IGT and new-onset diabetes, β-cell damage is characterized by early-phase insulin secretion deficiency. Consistent with this notion, our results showed that early-phase insulin secretion, estimated by DI_early, was significantly lower in IGT and DM patients than in NGT controls in both sexes. This observation is consistent with prior studies that have implicated liver fat accumulation in the pathogenesis of insulin resistance and β-cell dysfunction. Recently, a link between liver fat content and β-cell dysfunction has been proposed. A recent study showed that in patients with MASLD, upregulation of hepatic and serum calcium-binding protein S100 (also known as calcyclin) impairs insulin secretion from pancreatic β-cells via interaction with the receptor for advanced glycation end products, revealing a novel liver-pancreas axis[31]. In our study, we found that DI_early was negatively associated with HFF in both sexes, although the association did not reach statistical significance in females. This observation is consistent with prior studies that have implicated liver fat accumulation in the pathogenesis of insulin resistance and β-cell dysfunction; however, the precise mechanism remains to be further investigated.

In addition to liver fat deposition, accumulation of pancreatic fat has also been suggested to have an adverse effect on insulin secretion[32]. However, controversy still exists[23]. A recent large-scale Mendelian randomization study showed that higher pancreatic fat was associated with a higher risk of type 2 diabetes, but no causal evidence was found between pancreatic fat and type 2 diabetes[24]. We cannot rule out the possible adverse effects of pancreatic fat on β-cell function; however, our findings may suggest that fat accumulation in the pancreas is not significant at the early stage of diabetes pathogenesis in overweight/obese Chinese subjects. Similarly, no significant differences in MFF were found among subgroups in either sex.

Findings from this study support the idea that fat distribution may play a more important role than obesity per se in the pathogenesis of metabolic disorders. Recent genetic studies have revealed that fat distribution is at least in part genetically determined. A study by Ji et al[33] identified 14 alleles associated with higher body fat and higher BMI but lower risk of type 2 diabetes, heart disease, and hypertension. These individuals had higher subcutaneous fat but lower liver fat and a lower visceral-to-SAT ratio. These findings may partially explain why some individuals with obesity are metabolically healthy[7]. In our study, SAT was significantly lower in male patients with IGM than in those with NGT, whereas in female participants, SAT was not included in the primary logistic regression model because no significant difference was observed across glucose status subgroups, indicating a sex disparity in fat distribution. The inverse SAT–IGM association in men should not be interpreted as intrinsic protection; rather, a larger subcutaneous depot may merely reflect preserved lipid-storage capacity, as functional SAT expands via adipocyte hyperplasia to buffer excess energy and limit ectopic lipid deposition. When this capacity is exceeded, adipocyte hypertrophy, hypoxia, inflammation, enhanced lipolysis, and an adverse adipokine profile may promote insulin resistance and β-cell stress. Relative to VAT, SAT generally exhibits lower inflammation and a more favorable adipokine profile, including higher adiponectin, although these advantages may be lost with SAT dysfunction[34]. Sex hormones and depot-specific adipocyte biology may modulate this association, with estrogen promoting subcutaneous over visceral fat and greater insulin sensitivity, whereas men tend to accumulate more visceral fat[35]. Consequently, a larger CSA-standardized SAT may better reflect preserved lipid buffering capacity and reduced ectopic deposition in overweight/obese men, while in women, menopausal status and estrogen exposure may alter the SAT–glucose metabolism link. Nevertheless, as these factors were not measured, such mechanisms remain speculative.

In summary, our results showed that both HFF and VAT were independently associated with IGM in Chinese overweight/obese subjects of both sexes. In addition, lower SAT content was associated with IGM in male subjects, suggesting that sex-specific fat distribution may play a role in glucose metabolism. Given the cross-sectional design of this study, these findings should be interpreted as associations rather than causal relationships.

Strengths and limitations

Our study has several strengths. First, overweight/obese adults of both sexes were included and all analyses were sex-stratified, enabling exploration of potential sex-specific differences in fat distribution and its relationship with glucose metabolism. Second, BMI was similar across glucose status subgroups, reducing confounding by overall obesity. Third, individuals with recent use of anti-diabetic or anti-obesity medications were excluded, minimizing treatment-related effects on glucose and fatdistribution parameters. Fourth, abdominal and ectopic fat depots, including VAT, SAT, HFF, MFF, and PFF, were quantified by Q-CT, providing a more direct and standardized assessment of fat distribution than BMI alone. Finally, all participants underwent a standard 75-g OGTT, allowing accurate glucose status classification and assessment of early-phase insulin secretion.

Several limitations should also be acknowledged. First, this was a cross-sectional study, and therefore causal relationships between fat distribution and IGM cannot be established. The observed associations should not be interpreted as evidence that specific fat depots directly cause IGM. Prospective studies with longitudinal follow-up are needed to determine whether VAT, HFF, or SAT can predict the future development of IGM or diabetes. Second, the sample size was relatively modest, particularly after stratification by sex and glucose status subgroup, which may have limited statistical power to detect certain associations, such as the association between SAT and IGM in female participants. Third, the study was restricted to Chinese overweight/obese adults aged 18–50 years, potentially limiting the generalizability of the findings to other ethnic groups, age ranges, or individuals with normal BMI. Larger prospective studies with more comprehensive covariate assessment are therefore warranted to validate our findings.

CONCLUSION

Visceral adiposity and hepatic fat accumulation were independently associated with IGM in Chinese overweight/obese subjects of both sexes. This study identifies potential biological markers associated with an elevated risk of IGM in this population, which may have clinical implications for targeted intervention. These findings may offer useful imaging-based markers to identify overweight/obese individuals with a higher probability of IGM, although prospective studies are required to determine whether these fat depots predict incident diabetes.

ACKNOWLEDGEMENTS

The authors thank all the participants and staff involved in this research.