Abstract
Objective
This study aims to explore the impact of the triglyceride-glucose (TyG) index and TyG index combined with body mass index (BMI) on risk of incident EC and to estimate the proportion of risk of incident EC mediated by BMI through the TyG index.
Methods
Using the UK Biobank cohort, we included a sample of 168,431 participants for primary analysis. The risk of incident EC was evaluated using multivariable Cox proportional-hazards models. We applied a 2-stage regression approach to investigate the mediation effects. Polygenic risk scores were derived from established single nucleotide polymorphisms associated with EC.
Results
The study found that a higher TyG index is associated with an increased risk of incident EC (hazard ratio [HR]=1.322; p<0.001, per 1 standard deviation [SD] increment), and subgroup analysis revealed a significant trend in the overweight and obesity subgroups (p-value for the quartile trend <0.001). The TyG-BMI index exhibited a notable association with risk of incident EC (HR=1.741; p<0.001, per 1 SD increment). Mediation analysis indicated that 11.97% and 7.77% proportion of the effect of BMI on the risk of incident EC is mediated by the TyG index in overweight and obese individuals. High genetic susceptibility could increase the risk of incident EC (HR=1.455; p<0.001), with the relationship between increased TyG index levels.
Conclusion
The TyG index demonstrates a significant association with the risk of incident EC and mediates the influence of BMI, thereby implicating insulin resistance as a potential contributor to the pathogenesis of EC.
Keywords: Obesity, Insulin Resistance, Mediation Analysis, Endometrial Cancer, UK Biobank
Synopsis
The triglyceride-glucose index, a measure of insulin resistance, can be used to evaluate the incident endometrial cancer (EC) risk. The triglyceride-glucose index is associated with higher incident EC risk and mediates the effect of body mass index. Findings highlight insulin resistance's potential role in EC pathogenesis.
INTRODUCTION
Endometrial cancer (EC) is one of the most common gynecologic cancers. It is estimated that in 2024 there would be 67,880 new cases and 13,250 new deaths in the US [1]. With the development of socio-economic factors and changes in people's lifestyles, the incidence and associated mortality rates of EC have been increasing annually, which is an important global health burden [2].
The risk of incident EC is positively correlated with body mass index (BMI). Among the twenty most common tumor types, EC is notably one of the cancers most strongly linked to obesity; specifically, every 5 kg/m2 increase in BMI correlates with a 54%–60% heightened risk of EC [3,4]. Three biological mechanisms may potentially mediate the relationship between excess BMI and EC: increased bioavailability of estrogen and alterations in sex hormone metabolism; systemic proinflammatory milieu; and insulin resistance and bioavailability of insulin like growth factor I (IGF-1) [5,6]. The inflammation and adipokine hypothesis considered that obesity could create a proinflammatory microenvironment dominated by high circulating levels of C-reactive protein, interleukin-6 and tumor necrosis factor-α, and a relative deficiency of protective immune cell in the endometrium, which might contribute to the carcinogenesis of EC [7]. Obesity could also promote the conversion of androgens into estrogen through aromatase [8]. The unopposed estrogen excess could stimulate the proliferation of endometrium, which was the leading theory behind endometrial carcinogenesis. Insulin resistance and hyperinsulinemia could promote endometrial proliferation by reducing the levels of circulating binding proteins for estrogen and IGF-1, thereby increasing their bioavailability. This results in the activation of the PI3K-AKT-mTOR signaling pathway through direct (estrogen) and indirect (IGF-1) mechanisms [9,10]. While the influence of BMI on EC appears largely mediated by sex hormone metabolism, accumulating evidence suggests a substantial contribution from pathways associated with insulin resistance [10]. However, inconsistencies persist regarding the degree of the impact of insulin resistance on EC risk [11,12,13]. These discrepancies partially arise from differences in racial composition and sample size across studies. Furthermore, the interactive and mediating effects of insulin resistance in conjunction with BMI have not been fully addressed.
The TyG index, the logarithmized product of fasting triglyceride and glucose levels, has been proposed as a measure of insulin resistance. The TyG index shows a strong correlation with the gold standard for assessing insulin resistance, with a moderate Pearson correlation coefficient between the TyG index and euglycemic-hyperinsulinemic clamp test [14]. Consequently, the TyG index is considered a promising and cost-effective alternative for assessing insulin resistance in large-scale studies due to its simplicity and availability compared with fasting insulin or homeostatic model assessment-insulin resistance (HOMA-IR) [15,16,17,18,19]. Further, recent studies have shown that the TyG index combined with obesity indices such as BMI, is also closely associated with insulin resistance and metabolic syndrome [20]. This further makes it a straightforward and clinically useful surrogate marker for evaluating insulin resistance.
Given the established association between obesity and EC, a plausible intermediary mechanism is insulin resistance. Thus, it is pertinent to investigate the association between the TyG index and the risk of incident EC, as well as the degree to which the effects of BMI are mediated by the TyG index. This study aims to explore the impact of the TyG index and TyG index combined with BMI on the risk of incident EC through a large prospective cohort study and to estimate the proportion of EC incidence risk mediated by BMI through the TyG index. We further assessed the extent to which a level of TyG index was associated with an increased the risk of incident EC with a different genetic risk defined by the polygenic risk score (PRS), to further explore the genetic factors and the TyG index on the risk of incident EC.
MATERIALS AND METHODS
1. Study population
The UK Biobank (UKB) is a large prospective cohort, enrolling over 500,000 participants from the general population across the United Kingdom during 2006 to 2010 [21]. Participants were asked to complete a touch-screen questionnaire regarding their lifestyle, medical history, and previous exposures. Biological samples and physical measurements including height, weight, serum and plasma levels of glucose and triglycerides (including fasting time before sampling) were also taken in all assessment centers. All participants provided written informed consent and the study was approved by the North West Multi-Centre Research Ethics Committee (REC reference: 21/NW/0157). Additionally, all studies were designed in accordance with the Declaration of Helsinki, and all participants provided informed consent prior to inclusion in the study.
In this analysis, after excluding participants who lacked data for fasting triglyceride and glucose levels, had undergone a hysterectomy or had unclear hysterectomy status at baseline, and were diagnosed with cancer before baseline, we included a sample of 168,431 participants for primary analysis and 150,165 white participants for genetic risk analysis (Fig. S1). The work has been reported in line with the STROCSS criteria [22].
2. Data measures
BMI was calculated directly from weight and height records [weight (kg)/height (m)2]. The TyG index was calculated as ln [triglycerides (mg/dL) × blood glucose (mg/dL)/2] [23]. And TyG-BMI represents the multiplication of the TyG index and BMI. TyG index and TyG-BMI index were grouped into quartiles same as previous literatures [20,24].
We defined cancer diagnoses using the International Classification of Diseases, 10th revision codes: EC (C54.1). Incident cancer information was obtained from the national cancer registries, which centralize data collected from various regional cancer centers around the UK.
3. PRS calculation
The PRS was derived from data from data obtained in a widely analyzed genome-wide association study (GWAS) of EC [25], incorporating a total of 7 single nucleotide polymorphisms (SNPs) (Table S1). SNPs with minor allele frequencies below 1% or with low informativeness (information scores under 0.8) were excluded. Mismatched, duplicated, and ambiguous SNPs were also removed. The PRS for each participant was calculated using PRSice-2 software version 2.3.5, which combined the UKB SNP database with GWAS summary statistics while adjusting for sex and ten principal genetic components. PRS values were categorized into low, intermediate, and high groups based on tertiles.
4. Statistical analyses
Participants were followed from the date of baseline assessment to the date of cancer diagnosis, death, loss to follow up or August 31, 2022, whichever occurred first. Hazard ratios (HRs) for EC risks per 1- standard deviation [SD] increase in each TyG index were estimated using multivariable Cox proportional-hazards models. The Schoenfeld residual test was conducted to assess the validity of the Cox proportional hazards model assumption, confirming no violations. The models were first adjusted for age, and were further adjusted for ethnicity, center, smoking status, alcohol intake frequency, education, Townsend deprivation index and with and without adjustment for BMI category according to the World Health Organization (WHO) classification [26]. Covariates used in each model are listed in the footnote of the corresponding tables. We further divided the TyG index into quartiles (Q1–Q4). To calculate p-values for trend over quartiles, the Wald test of a linear association of a value’s quartile with the risk of incident EC was used. The TyG-BMI index, which came from TyG modified by multiplying with BMI were further divided into quartiles as well and estimated using multivariable Cox proportional-hazards models to confirm the results. In analyzing the combined effects of the TyG index and BMI, we only included participants with a BMI over 18.5 (no underweight). We then classified the participants into 12 subgroups, using those with normal weight and a low TyG index (Q1) as the reference group. Subgroup analyses were stratified by age (<50 years, ≥50–60 years, ≥60 years), BMI (underweight [<18.5 kg/m2], normal weight [≥18.5 and <25 kg/m2], overweight [≥25 and <30 kg/m2], and obese [≥30 kg/m2]), smoking status (never, former, current, not available [NA]), alcohol intake frequency (never, <1 time/week, 1–4 time/week, and daily, NA), and Townsend deprivation index (≤ median and > median).
The p-values for interactions were evaluated through likelihood ratio tests by comparing Cox proportional hazards regression models with or without the cross-product terms for each assessed factor. Restricted cubic spline models were uesd to explore potential non-linear relationships between TyG index or TyG-BMI index and the risks of incident EC.
We used the 2-stage regression method proposed by Vander Weele to assess the mediation and estimate total, direct, and indirect effects between BMI, TyG index, and EC risk [27]. Two regression models are applied to the data: one for modeling the mediator variable and the other for modeling the outcome variable. The parameter estimates and standard errors (SE) from these 2 separate models are combined using specific formulas to obtain estimates and SEs for the mediation effect size [28]. We assessed mediation by treating BMI both as a continuous and as a categorical variable. First, we fitted a linear regression model for the mediator TyG index, conditional on the exposure BMI and covariates. Second, we fitted a Cox proportional hazards regression model for EC risk on BMI, TyG index and the same covariates as in the model above. Finally, we estimated the desired effects combining the coefficients of these 2 regression models as described before [27]. Using this approach, we decompose the overall effect of BMI on EC incidence into 2 parts: the natural indirect effect size (which reflects the impact of BMI mediated through the TyG index) and the natural direct effect size (the influence of BMI not accounted for by the mediator). We calculated the proportion of the relationship between BMI and the risk of incident EC that is mediated by the TyG index, which indicates the contribution of the natural indirect effect relative to the total effect. This was performed on the log-transformed HR scale (log(indirect association HR)/log(total association HR)). Additionally, we estimated the 95% confidence intervals (CIs) based on SEs derived from the delta method.
In PRS analyses, we evaluated the interaction and joint effect between genetic factors and TyG index on the risk of incident EC. Using a fully adjusted model, including a multiplicative interaction term, TyG index and PRS tertiles were tested for their interaction effect. The association between TyG index and EC stratified by tertiles of PRS was also assessed. We set participants with lowest quartiles of TyG index and low genetic risk as the reference to analyse the association between TyG index and EC genetic risk with 3×4 groups.
To ensure the robustness of our results, several sensitivity analyses were performed. In models including BMI as a linear term, we restricted analyses to participants with a BMI over 18.5 (no underweight) since the assumption of a log-linear association of BMI with cancer risk was violated in the underweight range [11]. We also adjusted for some potential metabolism associated covariates including history of diabetes, hypertension, disorders of lipoprotein metabolism and healthy diet as well, which has been described previously [29]. Women specific covariates including duration of oral contraceptive use, use of hormone replacement therapy, age at menarche, and menopause status, were also adjusted. The waist circumference index was also adjusted as waist circumference enables a further refinement of the adverse health risk characterized by BMI [30]. Finally, we compared associations of BMI with the risk of incident EC, both adjusted and unadjusted for TyG index. This approach, known as the difference method for mediation analysis, was employed to demonstrate the robustness of our results [11].
All analyses were performed in R 4.1.3. Significance was defined as a 2-sided p-value <0.05.
RESULTS
1. Baseline characteristics of participants
The baseline characteristics of 168,431 women were shown in Table 1. Compared to those without incident EC, participants with EC were more likely to be older, overweight and obesity, and higher level of TyG index.
Table 1. Baseline characteristics of 168,431 women in UK Biobank between 2006 and 2010.
| Category | Overall | Non-incident EC | Incident EC | p-value | |
|---|---|---|---|---|---|
| No. of participants | 168,431 | 167,162 | 1,269 | ||
| Age (yr) | <0.001 | ||||
| <50 | 46,878 (27.8) | 46,751 (28.0) | 127 (10.0) | ||
| ≥50 and <60 | 59,072 (35.1) | 58,621 (35.1) | 451 (35.5) | ||
| ≥60 | 62,481 (37.1) | 61,790 (37.0) | 691 (54.5) | ||
| BMI (kg/m2) | <0.001 | ||||
| <18.5 | 1,349 (0.8) | 1,346 (0.8) | 3 (0.2) | ||
| ≥18.5 and <25 | 69,079 (41.0) | 68,809 (41.2) | 270 (21.3) | ||
| ≥25 and <30 | 60,678 (36.0) | 60,271 (36.1) | 407 (32.1) | ||
| ≥30 | 37,325 (22.2) | 36,736 (22.0) | 589 (46.4) | ||
| Ethnic | 0.140 | ||||
| European | 158,574 (94.1) | 157,379 (94.1) | 1,195 (94.2) | ||
| Smoking status | <0.001 | ||||
| Never | 102,034 (60.6) | 101,200 (60.5) | 834 (65.7) | ||
| Former | 51,275 (30.4) | 50,909 (30.5) | 366 (28.8) | ||
| Current | 14,581 (8.7) | 14,517 (8.7) | 64 (5.0) | ||
| Miss | 541 (0.3) | 536 (0.3) | 5 (0.4) | ||
| Alcohol intake frequency | <0.001 | ||||
| Never | 14,868 (8.8) | 14,726 (8.8) | 142 (11.2) | ||
| <1 time/week | 46,078 (27.4) | 45,654 (27.3) | 424 (33.4) | ||
| 1–4 time/week | 79,632 (47.3) | 79,106 (47.3) | 526 (41.4) | ||
| Daily | 27,718 (16.5) | 27,543 (16.5) | 175 (13.8) | ||
| Miss | 135 (0.1) | 133 (0.1) | 2 (0.2) | ||
| Education | <0.001 | ||||
| College or university degree | 56,916 (33.8) | 56,516 (33.8) | 400 (31.5) | ||
| Others | 85,114 (50.5) | 84,523 (50.6) | 591 (46.6) | ||
| Miss | 26,401 (15.7) | 26,123 (15.6) | 278 (21.9) | ||
| Townsend | −1.35±3.03 | −1.35±3.03 | −1.27±3.04 | 0.315 | |
| TyG index | 8.53±0.53 | 8.53±0.53 | 8.74±0.53 | <0.001 | |
| TyG index category | <0.001 | ||||
| Q1 | 42,108 (25.0) | 41,924 (25.1) | 184 (14.5) | ||
| Q2 | 42,108 (25.0) | 41,865 (25.0) | 243 (19.1) | ||
| Q3 | 42,107 (25.0) | 41,781 (25.0) | 326 (25.7) | ||
| Q4 | 42,108 (25.0) | 41,592 (24.9) | 516 (40.7) | ||
Values are presented as mean ± standard deviation or number (%).
BMI, body mass index; EC, endometrial cancer; TyG, triglyceride-glucose.
2. TyG index and TyG-BMI index and risk of EC
When TyG index was treated as a linear term in the statistical model (Table 2), TyG index was associated with the risk of incident EC (HR=1.322; 95% CI=1.252–1.396; p<0.001 and HR=1.135; 95% CI=1.070–1.203; p<0.001, respectively, per 1 SD increment), adjusting for relevant confounders with or without BMI (model 2 and model 3). The quartile analyses of the association between TyG index and the risk of incident EC provided similar results (Q4 vs. Q1: HR=1.357; 95% CI=1.130–1.629) and the p-value of the quartile trend was <0.001. What’s more, when considering the TyG-BMI index, we found that TyG-BMI index was also associated with the risk of incident EC (HR=1.741; 95% CI=1.670–1.814; p<0.001, per 1 SD increment, model 2). The quartile analyses of TyG-BMI index also showed a trend for the risk of incident EC (p<0.001). And compared to the lowest quartiles of TyG-BMI index, the highest quartiles were significantly associated with a higher risk of incident EC (HR=4.407; 95% CI=3.631–5.349; p<0.001). Further subgroup analysis showed that in the overweight and obesity subgroup, the TyG index showed a trend for the risk of incident EC (p=0.042 and p<0.001, respectively) (Table S2). In obese populations, the highest quartiles of TyG index were associated with a greater risk of incident EC (HR=1.576; 95% CI=1.112–2.235; p=0.011). In subgroup analysis stratified by other covariates, we found no interaction between TyG index/TyG-BMI index and age, smoking status, alcohol intake frequency and Townsend deprivation index (Table S3).
Table 2. Risk of endometrial cancer by TyG index.
| Category | Model 1 | Model 2 | Model 3 (with BMI) | ||||
|---|---|---|---|---|---|---|---|
| HR (95% CI) | p-value | HR (95% CI) | p-value | HR (95% CI) | p-value | ||
| TyG index (continuous, per 1 SD increase) | 1.330 (1.261–1.403) | <0.001 | 1.322 (1.252–1.396) | <0.001 | 1.135 (1.070–1.203) | <0.001 | |
| Q1 | Ref | Ref | |||||
| Q2 | 1.133 (0.935–1.375) | 0.203 | 1.138 (0.938–1.380) | 0.191 | 0.976 (0.803–1.185) | 0.804 | |
| Q3 | 1.412 (1.175–1.696) | <0.001 | 1.413 (1.175–1.699) | <0.001 | 1.055 (0.874–1.274) | 0.577 | |
| Q4 | 2.120 (1.783–2.519) | <0.001 | 2.104 (1.766–2.506) | <0.001 | 1.357 (1.130–1.629) | 0.001 | |
| Test for trend | <0.001 | <0.001 | <0.001 | ||||
| TyG-BMI index (continuous, per 1 SD increase) | 1.721 (1.654–1.789) | <0.001 | 1.741 (1.670–1.814) | <0.001 | |||
| Q1 | Ref | Ref | |||||
| Q2 | 1.365 (1.092–1.705) | 0.006 | 1.389 (1.112–1.736) | 0.004 | |||
| Q3 | 1.900 (1.542–2.341) | <0.001 | 1.952 (1.583–2.407) | <0.001 | |||
| Q4 | 4.318 (3.570–5.222) | <0.001 | 4.407 (3.631–5.349) | <0.001 | |||
| Test for trend | <0.001 | <0.001 | |||||
Data are presented as follow: model 1: adjusted for baseline age; model 2: adjusted for baseline age, ethnicity (white European, mixed, South Asian, black, others), center, smoking status (never, former, current, NA), alcohol intake frequency (never, <1 time/week, 1–4 time/week, and daily, NA), education (university or college degree, others, NA), Townsend deprivation index; model 3: adjusted for the same variables as in model 2, plus additionally for BMI category (underweight [<18.5 kg/m2], normal weight [≥18.5 and <25 kg/m2], overweight [≥25 and <30 kg/m2], and obese [≥30 kg/m2]).
BMI, body mass index; CI, confidence interval; HR, hazard ratio; NA, not available; SD, standard deviation; TyG, triglyceride-glucose.
In the sensitivity analysis, we restricted analyses to participants with a BMI over 18.5 and adjusted for metabolism associated covariates and women specific covariates, yielding the similar results (Tables S4, S5, S6, S7, S8). In the restricted cubic spline models, no non-linear associations were found between TyG index or TyG-BMI index and the risk of incident EC (p for non-linearity >0.05) (Fig. 1).
Fig. 1. Dose-response relationships between TyG index or TyG-BMI index and risk of endometrial cancer. The red solid lines represent hazard ratios and the corresponding 95% CIs (shaded areas) estimated using restricted cubic spline Cox regression models. The dashed lines represent the estimates obtained from models assuming linear relationships. The mean of TyG index or TyG-BMI index was used as the reference value. The p-values for non-linearity were from likelihood ratio tests comparing the spline models with linear models. The TyG index model was adjusted by the full model (baseline age, ethnicity, center, smoking status, alcohol intake frequency, education, Townsend deprivation index and BMI category) and the TyG-BMI index model was adjusted by the full model except for BMI.
BMI, body mass index; CI, confidence interval; HR, hazard ratio; TyG, triglyceride-glucose.
3. The joint and interaction effect of BMI and TyG index on EC risk
We found no interaction between BMI and TyG index, when treating BMI both as a continuous and as a categorical variable (Tables S2 and S9).
The joint effect of BMI and TyG index with the risk of incident EC was evaluated (Fig. 2A). Compared to those with the lowest value of each, participants with the highest BMI and TyG index experienced a 4.5-fold higher risk of incident EC (HR=4.558; 95% CI=3.576–5.810). Overall, higher levels of TyG index were associated with an increased risk of incident EC and this association was more obvious in the obesity group.
Fig. 2. The joint effect of BMI category or genetic risk and TyG index category with incident EC. The joint association of BMI category or genetic risk and TyG index on EC risk was examined by the multivariable Cox proportional hazard regression analysis.
BMI, body mass index; EC, endometrial cancer; HR, hazard ratio; NA, not available; TyG, triglyceride-glucose.
*Adjusted for baseline age, ethnicity (white European, mixed, South Asian, black, others), center, smoking status (never, former, current, NA), alcohol intake frequency (never, <1 time/week, 1–4 time/week, and daily, NA), education (university or college degree, others, NA), Townsend deprivation index.
†Adjusted for baseline age, ethnicity (white European, mixed, South Asian, black, others), center, smoking status (never, former, current, NA), alcohol intake frequency (never, <1 time/week, 1–4 time/week, and daily, NA), education (university or college degree, others, NA), Townsend deprivation index and BMI.
4. BMI and risk of EC mediated by TyG index
BMI was associated with an increased risk of incident EC as expected (HR=1.717; 95% CI=1.641–1.797) (Table 3). When including BMI as a linear term in mediation analyses and restricted to individuals with BMI values over 18.5 (Table 3), proportion of the effect of BMI mediated by TyG index for EC was 5.55%. When including BMI as a categorical term according to the WHO classification (Table 3), increased risk of incident EC of overweight and obese individuals was mediated by TyG index through 11.97% and 7.77%, respectively. The increased risk of underweight (BMI <18.5) was not mediated by TyG index. The results of the mediation analysis were corroborated through additional analyses utilizing the traditional difference method for mediation analysis. After additionally adjusting for TyG index, we found notably decreased association of BMI with EC, further proving the mediating role of TyG index (Table S9).
Table 3. Decomposition of the total effect of BMI on endometrial cancer risk into natural direct and indirect effect mediated by the triglyceride-glucose index.
| Category | Total effect HR (95% CI) | Natural indirect effect HR (95% CI) | Natural direct effect HR (95% CI) | Proportion mediated % (95% CI) |
|---|---|---|---|---|
| BMI (per 5 kg/m2 increase) | 1.717 (1.641–1.797) | 1.030 (1.011–1.050) | 1.667 (1.599–1.737) | 5.55 (2.00–9.09) |
| Underweight | 0.563 (0.180–1.758) | 0.958 (0.935–0.976) | 0.588 (0.188–1.836) | - |
| Normal weight | Ref | |||
| Overweight | 1.596 (1.361–1.873) | 1.058 (1.031–1.085) | 1.510 (1.290–1.767) | 11.97 (5.17–18.77) |
| Obesity | 3.860 (3.278–4.544) | 0.105 (SE=0.025) | 3.476 (2.974–4.061) | 7.77 (4.03–11.51) |
BMI, body mass index; CI, confidence interval; HR, hazard ratio; SE, standard error.
5. The joint and interaction effect of genetic factor and TyG index on EC risk
The risk of incident EC increased from the low to high genetic risk group (Table S10). HR for high genetic susceptibility was 1.455 (95% CI=1.257–1.684; p<0.001) compared with low genetic risk. We found no interaction between TyG index or TyG-BMI index and genetic risk. In the subgroup analysis, we found a trend for the risk of incident EC except for the intermediate genetic group. The risk-increasing effect of TyG index on EC were significantly higher in those with low genetic risk (HR=1.552; 95% CI=1.036–2.245) (Table S11). We further evaluated the joint effect of genetic factor and TyG index with the risk of incident EC (Fig. 2B). Compared to those with the lowest value of each, participants with high TyG index and high genetic risk experienced a 2-fold higher risk of incident EC (HR=2.325; 95% CI=1.625–3.325). The association between higher TyG index levels and increased risk of incident EC was largely consistent across different genetic risk groups.
DISCUSSION
The study found that a higher TyG index is associated with an increased risk of incident EC, and subgroup analysis revealed a significant trend in the overweight and obesity subgroups. The TyG-BMI index also showed a significant association with the risk of incident EC. Despite the absence of an interaction between BMI and the TyG index, a strong joint effect was observed. Participants with both high BMI and high TyG index experienced a 4.5-fold increase in the risk of incident EC. Mediation analysis indicated that 11.97% and 7.77% proportion of the effect of BMI on the risk of incident EC is mediated by the TyG index in overweight and obese individuals. High genetic susceptibility could increase the risk of incident EC, and participants with both high TyG index and high genetic risk had a 2-fold higher risk. The association between higher TyG index levels and increased risk of incident EC is largely consistent across different genetic risk groups. In different genetic risk groups, the relationship between elevated TyG index levels and a higher risk of incident EC was largely uniform.
Obesity is one of the primary risk factors for EC, along with diabetes, hypertension, and metabolic syndrome [31]. Our study is consistent with previous research demonstrating a potential link between insulin resistance and EC. Women with high fasting insulin levels, elevated C-peptide levels, and increased HOMA-IR values have a significantly higher risk of EC [10]. Another case-cohort study also found that hyperinsulinemia was an independent risk factor for EC [32]. One retrospective cohort study has confirmed that the TyG index was significantly higher in patients with EC or endometrial atypical hyperplasia than in those with normal endometrium [12]. The TyG index could be considered to be a representative biomarker of insulin resistance and had been widely used to explore the causal relationship between insulin resistance and various diseases [16,17,33]. By taking into account both triglyceride and glucose levels, it may offer a better understanding of the underlying risk factors for EC development.
Besides the established link between obesity, insulin resistance, and EC, our results suggested that a portion of the effect of BMI on the risk of incident EC is mediated by the TyG index. Specifically, we found that 11.97% and 7.77% of the effect of BMI on EC risk is mediated by the TyG index in overweight and obese individuals, respectively. When analyzing participants with a BMI greater than 18.5 as a linear term in mediation analyses, the proportion of the effect mediated by the TyG index for EC was reduced to 5.55%. This finding contrasts with a previous study that did not identify a mediation effect of the TyG index on the risk of incident EC [11]. However, the mediation effect was relatively small in our study, which suggested that other biological mechanisms may play a more significant role in connecting obesity to cancer. In addition to the 3 traditional hypotheses, other potential biological mechanisms include microenvironmental and cellular perturbations, such as vascular disruptions, epithelial-mesenchymal transition, and endoplasmic reticulum stress; factors arising from ectopic fat deposition; and altered intestinal microbiome may mediate the potential mechanism of obesity-induced EC [34]. Given the complexity of these interactions, further studies focusing on the underlying biological mechanisms that mediate the effect of BMI on EC risk are warranted. Such research could provide valuable insights into targeted prevention and treatment strategies for obesity-related EC.
In this study, a panel of 7 SNPs was employed to construct EC-related PRS to evaluate the modifying effect of genetic effect on incident EC. Our findings indicated that high genetic susceptibility was associated with an increased risk of incident EC. Participants with both a high TyG index and high genetic risk faced a 2-fold higher likelihood of EC incidence. Additionally, subgroup analysis found that participants had a higher increased risk of incident EC among those with low genetic risk. Although no significant interaction was found between them, the results still underscore the potential advantages of preventing insulin resistance. The absence of a significant interaction suggests that the effects of genetic factors and the TyG index on EC risk are largely independent and additive. However, the presence of a strong combined effect highlights their complementary role in risk stratification, which indicates that managing metabolic health could be beneficial for women.
Our study has several notable strengths. The prospective design of the UKB study, combined with an extensive follow-up period and a large sample size, substantially strengthens the reliability of our findings. Furthermore, the UKB provides a wealth of data, enabling us to account for numerous confounding factors, particularly those strongly associated with women’s health. Our research not only offers a thorough evaluation of the EC risk linked to the TyG index but also delves into the mediating role of BMI in this relationship, thereby contributing deeper insights into potential risk factors for EC. However, we acknowledge several limitations. Firstly, due to the restriction of the study population mostly confined to European ancestry people, this study may fail to show the heterogeneity among different racial groups and limit the generalizability of our findings to other ethnic groups. What’s more, the lack availability of complete data on covariates may potentially influence the results. For example, due to the limitations of the data available in the UKB database, we failed to restrict our analyses to individuals with a fasting status of 8 hours or more in the sensitivity analysis, which have been used in one study11. Lastly, the UKB database only includes clinical biomarker information collected at baseline, preventing us from examining whether changes in the TyG index over time might influence the risk of incident EC.
To conclude, we found a higher TyG index associated with an increased risk of incident EC. A strong joint effect between BMI and the TyG index was also observed. Participants with both high BMI and high TyG index experienced a 4.5-fold increase in the risk of incident EC. Mediation analysis indicated that 11.97% and 7.77% proportion of the effect of BMI on the risk of incident EC is mediated by the TyG index in overweight and obese individuals. High genetic susceptibility could increase the risk of incident EC, and the association between higher TyG index levels and the risk of incident EC was largely consistent across different genetic risk groups.
ACKNOWLEDGEMENTS
The authors thank the investigators and participants in UK Biobank for their contributions to this study. The authors thank Ms. Xiaolin Tang for support in facilitating the funding process.
Footnotes
Funding: This study was supported by the Gansu Province Health Industry Scientific Research Plan Project (GSWSKY-2019-25).
Conflict of Interest: No potential conflict of interest relevant to this article was reported.
Data Availability: The data supporting the findings of this study are derived from the UK Biobank Resource under Application No: 93810, which was approved for an extended scope of research. Researchers interested in accessing these datasets may do so by registering and applying through the official UK Biobank portal. For further information, please reach out to the corresponding author.
- Conceptualization: M.R.
- Formal analysis: Y.Z., S.Y.
- Investigation: Y.Z., S.Y.
- Methodology: Y.Z., S.Y.
- Resources: Z.T., M.R.
- Writing - original draft: Y.Z., S.Y.
- Writing - review & editing: M.R., S.Y.
SUPPLEMENTARY MATERIALS
The association of risk loci with EC, as reported by Timothy H. T. Cheng et al. [25] in a Genome-Wide Association Studies study involving a total of 6,542 EC cases and 36,393 controls
Risk of incident endometrial cancer by TyG index, stratified by body mass index category
Risk of incident endometrial cancer by TyG index and TyG-BMI index, stratified by other covariates
Sensitivity analysis through analyses restricted to participants with a BMI ≥18.5 (i.e. no underweight)
Sensitivity analysis through analyses adjusted for women specific covariates
Sensitivity analysis through analyses adjusted for chronic disease
Sensitivity analysis through analyses adjusted for healthy diet
Sensitivity analysis through analyses adjusted for waist circumference
Associations of BMI with incident endometrial cancer risk, with and without adjusting for TyG index
Association of genetic risk with endometrial cancer incidence in UK Biobank
Risk of incident endometrial cancer by TyG index/TyG-BMI index, stratified by cancer site genetic risk
Flow chart of study design.
References
- 1.Siegel RL, Giaquinto AN, Jemal A. Cancer statistics, 2024. CA Cancer J Clin. 2024;74:12–49. doi: 10.3322/caac.21820. [DOI] [PubMed] [Google Scholar]
- 2.Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71:209–249. doi: 10.3322/caac.21660. [DOI] [PubMed] [Google Scholar]
- 3.Renehan AG, Soerjomataram I, Tyson M, Egger M, Zwahlen M, Coebergh JW, et al. Incident cancer burden attributable to excess body mass index in 30 European countries. Int J Cancer. 2010;126:692–702. doi: 10.1002/ijc.24803. [DOI] [PubMed] [Google Scholar]
- 4.Crosbie EJ, Zwahlen M, Kitchener HC, Egger M, Renehan AG. Body mass index, hormone replacement therapy, and endometrial cancer risk: a meta-analysis. Cancer Epidemiol Biomarkers Prev. 2010;19:3119–3130. doi: 10.1158/1055-9965.EPI-10-0832. [DOI] [PubMed] [Google Scholar]
- 5.Calle EE, Kaaks R. Overweight, obesity and cancer: epidemiological evidence and proposed mechanisms. Nat Rev Cancer. 2004;4:579–591. doi: 10.1038/nrc1408. [DOI] [PubMed] [Google Scholar]
- 6.Larson EA, Dalamaga M, Magkos F. The role of exercise in obesity-related cancers: current evidence and biological mechanisms. Semin Cancer Biol. 2023;91:16–26. doi: 10.1016/j.semcancer.2023.02.008. [DOI] [PubMed] [Google Scholar]
- 7.Naqvi A, MacKintosh ML, Derbyshire AE, Tsakiroglou AM, Walker TDJ, McVey RJ, et al. The impact of obesity and bariatric surgery on the immune microenvironment of the endometrium. Int J Obes. 2022;46:605–612. doi: 10.1038/s41366-021-01027-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Gjorgoska M, Rizner TL. Integration of androgen hormones in endometrial cancer biology. Trends Endocrinol Metab. 2022;33:639–651. doi: 10.1016/j.tem.2022.06.001. [DOI] [PubMed] [Google Scholar]
- 9.Nead KT, Sharp SJ, Thompson DJ, Painter JN, Savage DB, Semple RK, et al. Evidence of a causal association between insulinemia and endometrial cancer: a mendelian randomization analysis. J Natl Cancer Inst. 2015;107:djv178. doi: 10.1093/jnci/djv178. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Hernandez AV, Pasupuleti V, Benites-Zapata VA, Thota P, Deshpande A, Perez-Lopez FR. Insulin resistance and endometrial cancer risk: a systematic review and meta-analysis. Eur J Cancer. 2015;51:2747–2758. doi: 10.1016/j.ejca.2015.08.031. [DOI] [PubMed] [Google Scholar]
- 11.Fritz J, Bjørge T, Nagel G, Manjer J, Engeland A, Häggström C, et al. The triglyceride-glucose index as a measure of insulin resistance and risk of obesity-related cancers. Int J Epidemiol. 2020;49:193–204. doi: 10.1093/ije/dyz053. [DOI] [PubMed] [Google Scholar]
- 12.Shi H, Guo F, Zheng K, Li R, Zhou H. Triglyceride-glucose index (TyG index) and endometrial carcinoma risk: a retrospective cohort study. Int J Gynaecol Obstet. 2024;164:298–304. doi: 10.1002/ijgo.15038. [DOI] [PubMed] [Google Scholar]
- 13.Gao S, Wang J, Wang T, Wang J. Is insulin resistance a high-risk factor for postmenopausal endometrial cancer: insights from the triglyceride glucose (TyG) index and the metabolic score for insulin resistance (METS-IR) Int J Womens Health. 2024;16:2355–2363. doi: 10.2147/IJWH.S500936. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Guerrero-Romero F, Simental-Mendía LE, González-Ortiz M, Martínez-Abundis E, Ramos-Zavala MG, Hernández-González SO, et al. The product of triglycerides and glucose, a simple measure of insulin sensitivity. Comparison with the euglycemic-hyperinsulinemic clamp. J Clin Endocrinol Metab. 2010;95:3347–3351. doi: 10.1210/jc.2010-0288. [DOI] [PubMed] [Google Scholar]
- 15.Khan SH, Sobia F, Niazi NK, Manzoor SM, Fazal N, Ahmad F. Metabolic clustering of risk factors: evaluation of triglyceride-glucose index (TyG index) for evaluation of insulin resistance. Diabetol Metab Syndr. 2018;10:74. doi: 10.1186/s13098-018-0376-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Yang Z, Gong H, Kan F, Ji N. Association between the triglyceride glucose (TyG) index and the risk of acute kidney injury in critically ill patients with heart failure: analysis of the MIMIC-IV database. Cardiovasc Diabetol. 2023;22:232. doi: 10.1186/s12933-023-01971-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Che B, Zhong C, Zhang R, Pu L, Zhao T, Zhang Y, et al. Triglyceride-glucose index and triglyceride to high-density lipoprotein cholesterol ratio as potential cardiovascular disease risk factors: an analysis of UK biobank data. Cardiovasc Diabetol. 2023;22:34. doi: 10.1186/s12933-023-01762-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Duan M, Zhao X, Li S, Miao G, Bai L, Zhang Q, et al. Metabolic score for insulin resistance (METS-IR) predicts all-cause and cardiovascular mortality in the general population: evidence from NHANES 2001–2018. Cardiovasc Diabetol. 2024;23:243. doi: 10.1186/s12933-024-02334-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Sergi D, Spaggiari R, Dalla Nora E, Angelini S, Castaldo F, Omenetto A, et al. HOMA-IR and TyG index differ for their relationship with dietary, anthropometric, inflammatory factors and capacity to predict cardiovascular risk. Diabetes Res Clin Pract. 2025;222:112103. doi: 10.1016/j.diabres.2025.112103. [DOI] [PubMed] [Google Scholar]
- 20.Dang K, Wang X, Hu J, Zhang Y, Cheng L, Qi X, et al. The association between triglyceride-glucose index and its combination with obesity indicators and cardiovascular disease: NHANES 2003–2018. Cardiovasc Diabetol. 2024;23:8. doi: 10.1186/s12933-023-02115-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Sudlow C, Gallacher J, Allen N, Beral V, Burton P, Danesh J, et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med. 2015;12:e1001779. doi: 10.1371/journal.pmed.1001779. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Mathew G, Agha R, Albrecht J, Goel P, Mukherjee I, Pai P, et al. STROCSS 2021: strengthening the reporting of cohort, cross-sectional and case-control studies in surgery. Int J Surg. 2021;96:106165. doi: 10.1016/j.ijsu.2021.106165. [DOI] [PubMed] [Google Scholar]
- 23.Alizargar J, Bai CH, Hsieh NC, Wu SV. Use of the triglyceride-glucose index (TyG) in cardiovascular disease patients. Cardiovasc Diabetol. 2020;19:8. doi: 10.1186/s12933-019-0982-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Huo RR, Liao Q, Zhai L, You XM, Zuo YL. Interacting and joint effects of triglyceride-glucose index (TyG) and body mass index on stroke risk and the mediating role of TyG in middle-aged and older Chinese adults: a nationwide prospective cohort study. Cardiovasc Diabetol. 2024;23:30. doi: 10.1186/s12933-024-02122-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Cheng TH, Thompson DJ, O’Mara TA, Painter JN, Glubb DM, Flach S, et al. Five endometrial cancer risk loci identified through genome-wide association analysis. Nat Genet. 2016;48:667–674. doi: 10.1038/ng.3562. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Shiwaku K, Anuurad E, Enkhmaa B, Kitajima K, Yamane Y. Appropriate BMI for Asian populations. Lancet. 2004;363:1077. doi: 10.1016/S0140-6736(04)15856-X. [DOI] [PubMed] [Google Scholar]
- 27.VanderWeele TJ. Causal mediation analysis with survival data. Epidemiology. 2011;22:582–585. doi: 10.1097/EDE.0b013e31821db37e. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Fritz J, Brozek W, Concin H, Nagel G, Kerschbaum J, Lhotta K, et al. The triglyceride-glucose index and obesity-related risk of end-stage kidney disease in Austrian adults. JAMA Netw Open. 2021;4:e212612. doi: 10.1001/jamanetworkopen.2021.2612. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Yang Z, Shen Y, Zhang T, Tang X, Mao R. Associations of biological age accelerations and genetic risk with incident endometrial cancer: a prospective analysis in UK Biobank. Int J Surg. 2025;111:512–519. doi: 10.1097/JS9.0000000000001966. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Cerhan JR, Moore SC, Jacobs EJ, Kitahara CM, Rosenberg PS, Adami HO, et al. A pooled analysis of waist circumference and mortality in 650,000 adults. Mayo Clin Proc. 2014;89:335–345. doi: 10.1016/j.mayocp.2013.11.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Makker V, MacKay H, Ray-Coquard I, Levine DA, Westin SN, Aoki D, et al. Endometrial cancer. Nat Rev Dis Primers. 2021;7:88. doi: 10.1038/s41572-021-00324-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Gunter MJ, Hoover DR, Yu H, Wassertheil-Smoller S, Manson JE, Li J, et al. A prospective evaluation of insulin and insulin-like growth factor-I as risk factors for endometrial cancer. Cancer Epidemiol Biomarkers Prev. 2008;17:921–929. doi: 10.1158/1055-9965.EPI-07-2686. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Su J, Li Z, Huang M, Wang Y, Yang T, Ma M, et al. Triglyceride glucose index for the detection of the severity of coronary artery disease in different glucose metabolic states in patients with coronary heart disease: a RCSCD-TCM study in China. Cardiovasc Diabetol. 2022;21:96. doi: 10.1186/s12933-022-01523-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Avgerinos KI, Spyrou N, Mantzoros CS, Dalamaga M. Obesity and cancer risk: emerging biological mechanisms and perspectives. Metabolism. 2019;92:121–135. doi: 10.1016/j.metabol.2018.11.001. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
The association of risk loci with EC, as reported by Timothy H. T. Cheng et al. [25] in a Genome-Wide Association Studies study involving a total of 6,542 EC cases and 36,393 controls
Risk of incident endometrial cancer by TyG index, stratified by body mass index category
Risk of incident endometrial cancer by TyG index and TyG-BMI index, stratified by other covariates
Sensitivity analysis through analyses restricted to participants with a BMI ≥18.5 (i.e. no underweight)
Sensitivity analysis through analyses adjusted for women specific covariates
Sensitivity analysis through analyses adjusted for chronic disease
Sensitivity analysis through analyses adjusted for healthy diet
Sensitivity analysis through analyses adjusted for waist circumference
Associations of BMI with incident endometrial cancer risk, with and without adjusting for TyG index
Association of genetic risk with endometrial cancer incidence in UK Biobank
Risk of incident endometrial cancer by TyG index/TyG-BMI index, stratified by cancer site genetic risk
Flow chart of study design.