Association between triglyceride glucose index and endometriosis in adults in the United States: A comprehensive study from the National Health and Nutrition Examination Survey (NHANES)

Endometriosis represents a complex medical disorder characterized by a range of syndromes resulting from the abnormal implantation of endometrial stroma and glands outside the uterine cavity. Its most significant clinical manifestations are debilitating pain and infertility [ 1 ]. The pathogenesis of endometriosis remains a topic of considerable debate. The latest research evidence indicates that hormones and immune factors create an inflammatory environment that promotes the onset and development of endometriosis [ 2 , 3 ]. Additionally, the activation of inflammatory signal transduction pathways is closely related to insulin resistance (IR) [ 4 ]. Nevertheless, the association between IR and endometriosis remains elusive. Studies have established that obesity is a prominent etiological factor of IR [ 5 ]. Hyperlipolysis, characterized by increased lipolysis, is a pivotal source of pro-inflammatory cytokines [ 6 ]. Obesity and IR are associated with low-grade chronic systemic inflammation. Adipose tissue promotes levels of pro-inflammatory factors, such as tumor necrosis factor-alpha (TNF-α), interleukin-1 (IL-1), and interleukin 6 (IL-6). These pro-inflammatory cytokines stimulate major inflammatory NFκB and JNK pathways in cells [ 7 ]. Consequently, in the presence of IR, inflammatory factors are produced. The development of endometriosis is associated with abnormalities in autoimmune functions, especially the function of immune cells, such as neutrophils, macrophage differentiation, T cells, and B cells [ 8 ]. In endometriosis, macrophages and mast cells have been found to release chemokines such as TNF and IL-6 [ 9 , 10 ]. These immune cells can promote endometriotic cell proliferation and neovascularization in lesions [ 11 , 12 ]. If the phagocytosis of cells by macrophages is weakened, the growth inhibition of ectopic endometrium is lifted. In addition, mast cells are important immune cells involved in allergic reactions in the body. The number of mast cells in ectopic endometrial tissue is substantially higher than that in normal endometrial tissue. More important differentiation factors of mast cells are observed in the abdominal cavity of patients with endometriosis [ 13 , 14 ]. Furthermore, NK cells and T cells play an essential role in endometriosis. This suggests that the abnormal immune response mediated by immune cells caused by IR is a key factor in endometriosis. The continuous release of proinflammatory mediators constitutes a risk factor for endometriosis. Accordingly, we hypothesize that IR is crucial in the pathophysiological development of endometriosis. The triglyceride-glucose (TyG) index, an established biomarker of IR, can predict the risk of diabetes, hypertension, and coronary heart disease across diverse ethnic populations. It has been extensively studied as an indicator of IR, with significant associations with the prognosis in patients with acute coronary syndrome, heart failure, and other circulatory diseases [ 15 , 16 ]. Moreover, research has revealed that elevated levels of the TyG index are significantly correlated with the risk of gestational diabetes mellitus and gestational hypertension [ 17 ]. Nonetheless, the relationship between the TyG index and endometriosis remains inadequately explored. Thus, this study was to elucidate the potential association between the TyG index and endometriosis, based on the comprehensive NHANES database.

The characteristics of the included subjects are detailed in Table 1 . 2508 subjects aged between 30 and 46 years were examined for fasting triglycerides and fasting glucose and answered questionnaires about endometriosis. Age, birthplace, age of the family reference person at the time of screening, fasting triglyceride, whether at least one ovary was removed, and TyG index were significantly different. Notes: 1 Median (IQR); n (unweighted) (%). 2 Wilcoxon rank-sum test for complex survey samples; chi-squared test with Rao & Scott’s second-order correction. Three multivariate weighted logistic regression models were employed to examine the association between the TyG index and endometriosis ( Table 2 ). A significant positive association was revealed between the TyG index and endometriosis in the unadjusted model (Model 1) and partially adjusted model (Model 2). After adjustment for full covariates (Model 3), this positive association remained robust (P = 0.038). The prevalence of endometriosis was notably increased in the highest TyG index group compared with the lowest TyG index group. Furthermore, among the covariates, age, birthplace, and whether an ovary had been removed had significant effects on the risk of endometriosis. Notes: Model 1: Unadjusted; Model 2: Adjusted for age, body mass index (BMI); Model 3: Adjusted for age, body mass index (BMI), Education level, country of birth, HH Ref Person Gender, HH Ref Person Age, HH Ref Person Marital Status, age of menarche, number of pregnancies and whether at least one ovary was removed. * p<0.05, p<0.01 *** p<0.001. As a continuous variable, the TyG index had a positive correlation with endometriosis. OR was 1.676 (95%: 1.293–2.172) in Model 1. 1.642 (95%: 1.224–2.204) in Model 2, and 1.52 (95%: 1.024–2.258) in Model 3, which was still significantly positive. Moreover, RCS analysis ( Fig 2 ) revealed a non-linear correlation between the TyG index and endometriosis in the studied population (Model 1: P = 0.1137; Model 2: P = 0.1474; Model 3: P = 0.0711). The risk of endometriosis increased rapidly with the rise of the TyG index, especially the TyG index > 8.5. As illustrated in Fig 3 , the subgroup analysis elucidated that the TyG index retained a statistically significant association with the risk of endometriosis across subgroups stratified by age (≤40 years (OR [95% CI]: 1.628(1.086,2.441); P = 0.019) and >40 years (OR [95% CI]: 1.59(1.101,2.296); P = 0.014)) and BMI (OR [95% CI]: 2.426(1.441,4.084); P = 0.001). The TyG index remained significantly associated with endometriosis in all study populations. IVW methods observed positive causal inference of IR to endometriosis (OR, 1.628; 95% ci, 0.061–0.913; p = .024) ( Table 3 ). The sensitivity analysis revealed no heterogeneity (p for Cochran’s Q, 0.206) or pleiotropy (p for MR‐Egger intercept, 0.177) among these SNPs ( Table 4 ). Abbreviations: IVW, inverse-variance weighted; EMS, endometriosis; IR, insulin resistance.

TyG index has a robust positive correlation with endometriosis. With the increase of the TyG index, the risk of endometriosis also increases significantly. Age and birthplace are significant factors influencing the relationship.

All data employed in this study are accessible through the NHANES website ( https://www.cdc.gov/nchs/nhanes/ ). The NHANES is a population-based cross-sectional survey aimed at collecting detailed information on the health and nutritional status of adults and children in the United States. This survey integrates structured interviews and rigorous physical examinations. The dataset encompasses demographic particulars, dietary profiles, clinical examination findings, laboratory assays, questionnaire responses, and limited-access data. The findings from this survey are instrumental in assessing the prevalence of major diseases and associated risk factors, as well as in evaluating national parameters such as height, weight, and blood pressure. Only the study period of 1999–2006 had variables for endometriosis. This study analyzed data from the most recent four NHANES survey cycles from 1999 to 2006. Initially, the dataset comprised 41,474 subjects. However, in the process of data collection, 35,917 subjects were systematically excluded due to the absence of pertinent endometriosis data, 3,017 subjects due to lacking essential fasting glucose measurements, and an additional 32 subjects due to lacking fasting triglyceride data. Finally, 2,508 subjects were included. A detailed screening procedure is visually depicted in Fig 1 . Studies involving human participants were reviewed and approved by the NCHS Research Ethics Review Board (ERB). In accordance with national legislation and institutional requirements, written informed consent was not required for participation in this study. Triglyceride levels were assessed exclusively during the morning examination within the NHANES. This study included a cohort of individuals aged≥ 12 years who had fasted for at least 8.5 hours but less than 24 hours, yielding measurable triglyceride concentrations and non-zero fasting sample weights. Glucose concentration was measured utilizing the hexokinase method—an enzymatic assay for endpoint determination, involving the correction of sample blanks. Fasting blood glucose was tested in all participants in the morning after a fasting period of 9 hours. Data concerning endometriosis were sourced from the reproductive health questionnaire (RHQ). The RHQ is an intricate and comprehensive instrument that includes numerous questions, which are categorized by age or skip patterns based on responses to diverse reproductive health issues. Subjects were defined as having endometriosis if they received a diagnosis from a doctor or other professionals. Specific information on other covariate data, such as age, education level, country of birth, BMI, weight, and marital status can be obtained at www.cdc.gov/nchs/nhanes/ . The Mendelian randomization (MR) analysis for IR and endometriosis was obtained from the ieu database with the ids ebi-a-GCST005179 and ebi-a-GCST90018839, respectively. The ieu database provides 346,363,816,988 genetic associations from 50,040 GWAS summary datasets. MR was performed using the TwoSampleMR R software package with IR as the exposure and endometriosis as the outcome. Due to the small number of single nucleotide polymorphisms (SNPs) associated with IR or endometriosis, we selected instrumental variables with P values < 1x10 -5 , which were independent (R2 cutoff of 0.1, distance window, 10,000 KB). Continuous variables were presented as the standard deviation (SD) of the mean or the mean of the interquartile range (IQR) and categorical variables were presented as frequencies and percentages (%). Statistical comparisons for continuous variables were conducted using Student’s t-test, whereas categorical variables were compared using Pearson’s chi-square test or Fisher’s exact test. According to the official NHANES guidelines, weights from 4-year fasting data were used from 1999 through 2002 (WTSAF4YR) and weights from 2-year fasting data were used from 2003 through 2006 (WTSAF2YR). TyG index was categorized into four quartiles and three models were established via weighted logistic regression. Model 1 remained unadjusted, Model 2 was partially adjusted (adjusted for age and BMI), and Model 3 was fully adjusted (adjusted for age, BMI, educational attainment, country of birth, gender, and age of the reference person, marital status of the reference person, age at menarche, parity, and whether at least one ovary was removed). To avoid multicollinearity, for the variables in the multifactor model, we calculated the variance inflation factor (VIF) for each variable in Model 3 and excluded variables with VIF > 5. A restricted cubic spline (RCS) model was employed to assess the non-linear correlation between the TyG index and the risk of endometriosis. Furthermore, subgroup analyses were performed by age and BMI, and forest plots were generated. All analyses were performed utilizing R 4.3.0 ( https://www.r-project.org ). P<0.05 denoted statistical significance. The fixed‐effects inverse‐variance weighted (IVW) method was utilized to obtain the primary MR estimation. In the presence of horizontal pleiotropy in the SNPs, MR‐Egger regression and weighted median were conducted to test the robustness of the results. Isn the sensitivity analysis, IVW methods with Cochran’s Q statistics and MR‐Egger intercept were utilized to evaluate the heterogeneity and pleiotropy of individual SNPs. Statistical analyses were undertaken using the TwoSampleMR package in R statistical software 4.3.2. (R Foundation, Vienna, Austria), with a two‐tailed p‐value < .05 implying statistical significance.