Predictive significance of triglyceride-glucose index in forecasting adverse cardiovascular events among Type 2 diabetes mellitus patients with co-existing metabolic dysfunction-associated steatotic liver disease (MASLD): evidence from two cohort studies Authors: Jing Xu, Jianyong Zhao, Zhijuan Dai, Yue-Chun Li, Chunmei Li Published: 2026 ABSTRACT AbstractBackgroundTriglyceride-glucose (TyG) is a vital marker for assessing cardiovascular risk and insulin resistance (IR) in individuals with Type 2 diabetes mellitus (T2DM) or metabolic dysfunction-associated steatotic liver disease (MASLD). Nevertheless, the prognostic significance of TyG, especially in patients manifesting both conditions, is insufficiently characterized. In this study, the association between TyG and the occurrence of major adverse cardiovascular events (MACEs) in patients concurrently diagnosed with T2DM and MASLD was examined.MethodsIn this retrospective cohort study, 1021 patients diagnosed with both T2DM and MASLD at Wenzhou Hospital from 2019 to 2022 were encompassed. To explore the non-linear association between TyG and outcomes, such as all-cause mortality (ACM) or MACEs, a cox proportional hazards model was employed. Additionally, subgroups were analyzed to assess the consistency across different demographic and clinical subgroups. Furthermore, mediation analyses were conducted to investigate the potential mediating effects of body roundness index (BRI) and estimated glomerular filtration rate (eGFR). Parallel supportive analysis was performed on a cohort of 1343 individuals from NHANES 1999–2018.ResultsOver a median follow-up of 53.4 months, 62 instances of ACM and 101 MACEs were documented. Cox proportional hazards regression analysis revealed that an elevated TyG was significantly associated with an increased risk of mortality (HR = 1.85, 95% CI: 1.25, 2.73) and MACEs (HR = 1.71, 95% CI: 1.26, 2.33). Subgroup analysis revealed that the association between TyG and the risk of MACE is more prominent among individuals below 60. Mediation analysis revealed that eGFR mediated 8.8% of the total effect of TyG on ACM and 10.6% of the effect on MACEs. BRI accounted for 6.2% of the total effect of TyG on ACM. Incorporating TyG into conventional risk models significantly enhanced the predictive accuracy.ConclusionThis study established a strong association between TyG and both ACM and MACEs in patients with T2DM and MASLD in the Wenzhou cohort, with TyG-ACM association independently validated in the NHANES cohort.Graphical abstract Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12933-026-03259-0.Keywords: Major adverse cardiovascular events, Triglyceride-glucose index, MASLD, T2DM, Insulin resistance Research insights What is currently known about this topic? Coexisting T2DM and MASLD elevate cardiovascular risk. Coexisting T2DM and MASLD elevate cardiovascular risk. TyG associated with cardiovascular risk in T2DM patients. TyG associated with cardiovascular risk in T2DM patients. Existing models lack precision for T2DM-MASLD patients. Existing models lack precision for T2DM-MASLD patients. What is the key research question? Whether TyG effectively predicts adverse cardiovascular events among T2DM patients with MASLD. What is new? Based on two large cohorts (Wenzhou and NHANES), TyG is independently related to cardiovascular events in T2DM-MASLD patients. Based on two large cohorts (Wenzhou and NHANES), TyG is independently related to cardiovascular events in T2DM-MASLD patients. Mediation analysis suggested that BRI and eGFR may partially mediate the impact of TyG on ACM and MACEs Mediation analysis suggested that BRI and eGFR may partially mediate the impact of TyG on ACM and MACEs Confirms the clinical significance of incorporating TyG into risk stratification. Confirms the clinical significance of incorporating TyG into risk stratification. How might this study influence clinical practice? It provides a cost-effective, simple method for enhancing risk assessment. Introduction The global prevalence of Type 2 diabetes mellitus (T2DM) has reached epidemic proportions, affecting an estimated 537 million adults worldwide, with projections indicating a rise to 783 million by 2045 [1]. Concurrently, metabolic dysfunction-associated steatotic liver disease (MASLD) has emerged as the most prevalent chronic liver condition, affecting approximately 30% of the global population [2, 3]. These two disorders frequently coexist: recent meta-analyses reveal that nearly two-thirds of individuals with T2DM have concurrent MASLD, with a significant proportion advancing to metabolic dysfunction-associated steatohepatitis (MASH) or advanced fibrosis [4]. This high degree of overlap reflects shared pathophysiological underpinnings and represents a substantial and growing public health challenge. The co-occurrence of T2DM and MASLD markedly amplifies cardiovascular risk beyond that observed with either condition alone. Patients with both disorders exhibit a substantially elevated incidence of major adverse cardiovascular events (MACEs) and all-cause mortality (ACM) compared to those with isolated T2DM or MASLD [5–7]. Despite this well-established risk, existing cardiovascular risk stratification tools lack precision for this specific high-risk population, underscoring the urgent need for refined predictive markers. Over the past decade, TyG has gained prominence as a practical and accessible evaluative tool, garnering substantial scientific interest within IR research. Empirical studies have demonstrated a robust association between TyG and the progression of cardiovascular disease [8, 9]. An elevated TyG often serves as a predictor of unfavorable cardiovascular outcomes, with the trend being evident in both the patients suffering from a variety of chronic conditions and general population [10–12]. Notably, even following statistical adjustments for established cardiovascular risk factors, TyG has substantial prognostic significance [13, 14]. While the clinical utility of TyG for cardiovascular prediction is well-established [15], there is a paucity of research specifically addressing populations concurrently affected by T2DM and MASLD. Previous studies have predominantly focused on populations with either isolated T2DM or MASLD, whereas the current study uniquely examines the coexistence [16–18]. Especially, only one study has explored the association between TyG and mortality among individuals with diabetes or prediabetes complicated by MASLD, utilizing data from NHANES [19]. However, this study aggregated patients with prediabetes and diabetes without distinguishing between diabetes subtypes, and its outcome assessment was limited to mortality outcomes alone, precluding evaluation of non-fatal cardiovascular events. Furthermore, its scope is confined to the American population, casting doubt on the generalizability of its findings to the Chinese demographic. Therefore, conducting an in-depth investigation into the role of TyG in the cardiovascular prognosis of Chinese patients with T2DM complicated by MASLD holds significant importance. In addition to exploring simple correlations, we also endeavor to elucidate the specific mechanisms by which TyG influences cardiovascular risk, including obesity, renal dysfunction. BRI not only associated with overall body weight but also serves as an indicator of central obesity, and it has been consistently linked to hepatic fat accumulation and unfavorable prognoses in MASLD [20, 21]. eGFR represents a readily obtainable metric of renal function. A decline in eGFR is closely associated with the cardiovascular risk and long-term outcomes of T2DM and MASLD [22, 23]. Consequently, BRI and eGFR emerge as biologically plausible mediators connecting TyG to mortality and the risk of MACE. Leveraging clinical data from Wenzhou Hospital, this study seeks to ascertain the predictive capacity of TyG for adverse cardiovascular outcomes within a specific high-risk cohort comprising patients with both T2DM and MASLD. Furthermore, to corroborate our ACM findings across diverse populations, we conducted a re-analysis of the NHANES dataset, concentrating exclusively on individuals with diabetes while excluding those with prediabetes. The insights derived from this analysis have the potential to assist healthcare professionals in promptly identifying individuals at high risk and implementing suitable interventions to enhance patients’ long-term prognosis. Investigating TyG across diverse disease conditions may contribute to the formulation of more holistic strategies for the management and prevention of cardiovascular diseases. Methods Research design This study utilized a retrospective cohort design, encompassing 1,021 patients concurrently diagnosed with both MASLD and T2DM at the Second Affiliated Hospital of Wenzhou Medical University between 2019 and 2022. Figure 1 depicts the patient screening process. The inclusion criteria required participants to be between 18 and 80 years old and to meet the diagnostic criteria for both MASLD and T2DM. This age range was selected to exclude children, who exhibit distinct metabolic profiles, and individuals over 80 years old, who often present with complex comorbidities. This approach ensures a focus on the primary population affected by T2DM and MASLD. Displays the flowchart depicting patient inclusion in the study The exclusion criteria encompassed severe hepatic disease, defined as a Child–Pugh score of 7 or greater; significant renal impairment; specific etiologies of liver disease, including liver cancer, Hepatitis C, or Hepatitis B; excessive alcohol abuse, defined as an intake of 30 g per day or more for males and 20 g per day or more for females; secondary hypertension; unmanageable heart failure; malignancy; cardiogenic shock; incomplete follow-up or baseline data. To evaluate potential selection bias, a comprehensive comparison of baseline characteristics between the included and excluded cohorts (those with incomplete follow-up or TyG data) was performed. Demographic and clinical parameters, including age, BMI, diastolic blood pressure (DBP), systolic blood pressure (SBP), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), TyG, low-density lipoprotein cholesterol (LDL-C), smoking status, alcohol abuse, and prevalence of coronary heart disease (CHD), stroke, demonstrated no statistically significant differences between the two groups (P > 0.05, Supplementary Table S1). TyG is computed using the formula “ln [triglyceride (TG) × fasting plasma glucose (FPG)/2]”, with participants categorized into three tertiles in Wenzhou cohort: T1 (< 8.93), T2 (8.93–9.48), and T3 (> 9.48). Data definition and collection All participants provided blood samples following an overnight fast in the early morning, and these samples were subsequently analyzed on the same day within a standardized laboratory setting. Clinical data were extracted from medical records by trained personnel who remained blinded to the research objectives. The comprehensive data collection encompassed clinical risk indicators (such as DBP, SBP, medical history, and lifestyle behaviors), demographic parameters [including age, sex, and body mass index (BMI)], as well as extensive laboratory evaluations. These biochemical evaluations comprised metabolic biomarkers, hematological parameters, and lipid profiles. The diagnosis of diabetes mellitus adhered to established clinical guidelines [24]. In classifying T2DM, it is imperative to systematically rule out other potential etiologies of diabetes mellitus. Additionally, BRI was determined with a specified formula [25]: eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) 2009 creatinine equation, expressed as: eGFR = 141 × min(SCr/K, 1)a × max(SCr/K, 1)−1.209 × 0.993age × 1.018 [in females]. SCr indicates serum creatinine, K is 0.7 for females and 0.9 for males, a is − 0.329 for females and − 0.411 for males, “min” indicates the minimum of SCr/K or 1, and “max” indicates the maximum of SCr/K or 1 [26]. The diagnostic criteria for hyperlipidemia are delineated as follows: a TG level of 1.7 mmol/L or higher, a HDL-C level below 1.0 mmol/L, a LDL-C level of 3.4 mmol/L or above, or currently undergoing effective lipid-lowering therapy [27]. The defining features of a history of CHD encompass previously confirmed coronary artery bypass grafting, percutaneous coronary intervention (PCI), myocardial infarction (MI), and other coronary artery revascularization procedures, or angiographic evidence of at least one major coronary artery exhibiting a stenosis of 50% or greater [28]. Ultrasonography for MASLD diagnosis All participants underwent abdominal ultrasonography utilizing a Mindray DC-7 system equipped with a 5 MHz convex probe. Ultrasonographic examination was conducted following an 8 to 10-hour fasting period by a single experienced radiologist, who was blinded to the clinical and laboratory parameters to minimize observer bias. In the Wenzhou cohort, the diagnosis of simple liver disease (SLD) was established in accordance with standard echogenic criteria, necessitating the presence of at least two of the following indicators: (1) obscuration of hepatic vascular margins, (2) enhanced hepatorenal echogenicity contrast, and (3) posterior beam attenuation (deep attenuation). Due to the unavailability of conventional ultrasound data in NHANES, MASLD was defined using the ultrasound fatty liver index (USFLI), a validated surrogate marker for hepatic steatosis, where a USFLI cutoff value of ≥ 30 signified SLD [29]. MASLD was characterized by the presence of SLD in patients, along with at least one cardiometabolic risk factor, while excluding other causes of liver fat accumulation, such as specific etiology SLD (including hepatitis B or C or liver cancer) or high alcohol intake (≥ 30 g/d for male and ≥ 20 g/d for female) [30]. In this study, since all participants had T2DM, MASLD was defined as the coexistence of T2DM and SLD. Fibrosis-4 index (FIB-4) was calculated to assess hepatic fibrosis using the following formula: age (years) * AST (U/L) / (platelet count [10^9/L] * √ALT [U/L]) [31]. FIB-4 score of less than 1.30 suggests a low risk of liver fibrosis, whereas a score of 1.30 or greater indicates a moderate to high risk of liver fibrosis. Endpoint events and follow-up Trained medical professionals conducted follow-ups of participants via electronic health records, telephone communication, or outpatient consultations. Follow-ups commenced at discharge and extended until January 2026, participant death, or withdrawal, whichever occurred first. The primary endpoint of this study was the incidence rate of MACE, a composite endpoint including overall mortality, non-fatal MI, non-fatal stroke, cardiovascular-specific mortality, as well as hospitalization due to unstable angina or deterioration of cardiac insufficiency. Non-fatal stroke included cerebral hemorrhage and cerebral infarction, both confirmed by imaging studies. Should participants encountered multiple events during follow-ups, the most severe event would be documented as final outcomes, ranked in order of severity as follows: all-cause mortality, cardiovascular disease (CVD) mortality, non-fatal stroke, and non-fatal MI [14, 32]. The diagnosis of endpoint events would be independently reviewed and confirmed by two senior experts at least. CVD encompass conditions such as congestive heart failure, CHD, angina pectoris, MI, and stroke. Statistical analyses For normally distributed data, continuous variables were expressed as mean ± SD, whereas non-parametric data were presented in terms of the median and interquartile range. Categorical variables were characterized by percentage and frequency, and comparisons were made with chi-square test. Multiple imputation by chained equations (MICE) was performed to address the missing data, assuming that the data were missing at random. Detailed summaries of the missing data counts and the specific imputation methods employed are provided in Supplementary Table S2. Potential confounders were selected using DAGs, and the final variables included age, sex, BMI, smoking, drinking status, SBP, DBP, diabetes duration, LDL-C, and metformin use (Supplementary Fig. S1). The variance inflation factor method was used to test for multicollinearity, with a variance inflation factor of five or higher indicating multicollinearity. None of the variables showed multicollinearity (Supplementary Table S3 and Fig. S2). Subsequently, the Cox proportional hazards model was employed to assess the association between TyG and clinical outcomes, yielding HR and 95% CI. The proportional hazards assumption within the Cox analysis was corroborated by the Schoenfeld residual test, which affirmed its applicability to the endpoints of both ACM and MACEs (see Supplementary Fig. S3). Three multivariate Cox proportional hazards regression models were employed for the analysis: Model 1 (unadjusted), Model 2 (adjusted for sex and age), and Model 3 (comprehensively adjusted for age, sex, BMI, smoking, drinking status, SBP, DBP, diabetes duration, LDL-C, and metformin use). Survival probabilities for each group were depicted through Kaplan-Meier curves, and the differences among groups were assessed via log-rank tests. Potential nonlinear association between TyG and clinical outcomes was examined by employing restricted cubic spline (RCS) regression. Additionally, heterogeneity was evaluated based on comorbidities (e.g., hypertension, and hyperlipidemia), demographic characteristics (age and sex), metabolic parameters (glycated hemoglobin concentration), liver fibrosis (FIB-4 level), and anthropometric measurements (BMI) through subgroup analyses. The interaction effect was quantified through likelihood ratio tests. The incremental predictive value of TyG, relative to traditional risk factors, was assessed employing C-index, continuous net reclassification improvement (NRI), and integrated discrimination improvement (IDI). Subsequently, a causal mediation model was adopted in this study to examine whether the effect of TyG index on MACE and mortality risk is mediated by potential mediators, including eGFR and BRI. First, a mediator model is constructed with potential mediators as the continuous dependent variable and TyG as the independent variable to evaluate the path a coefficient of TyG with respect to potential mediators. Subsequently, an outcome model was constructed with the follow-up duration as the time scale, ACM or MACE as the endpoint events, and TyG and potential mediators as the exposure factors, obtaining the direct coefficient of TyG to the MACE risk and the coefficient of potential mediators to the MACE risk. The mediator model and the outcome model simultaneously controlled the confounding factors of Model 3. Through 1000 non-parametric Bootstrap samples with the “mediation” R package, the direct effect (DE), indirect effect (IE), and total effect (TE) were calculated, and the mediating proportion (IE/TE * 100%) was obtained. Given the observational design and baseline assessment of variables, these assumptions cannot be formally tested and should be interpreted with caution. To evaluate the robustness of the findings, sensitivity analyses were performed from multiple perspectives. First, acknowledging that medication usage can influence TyG, adjustments were made for the administration of hypolipidemic drugs, insulin, and GLP-1 receptor agonists (GLP-1RA). Second, to mitigate reverse causality, participants who died within the first year of follow-up were excluded (n = 11). Third, to minimize the impact of extreme TyG on the primary association, the standardized Z-score method was employed to identify and exclude outliers. Specifically, TyG with an absolute Z-score exceeding 3 was classified as an outlier (n = 7). Fourth, survival analysis was performed using the Fine-Gray competing risks model, with event outcomes categorized as 0 = without MACE, 1 = CVD (primary event, CVD does not encompass mortality.), and 2 = ACM (competing event). Fifth, employing TyG dichotomy (with a median value of 9.18) as the principal matching variable, we conducted a 1:1 propensity score matching (PSM) to minimize the influence of potential confounding factors. Sixth, to further confirm the robustness of the results, the effect of unmeasured confounders on the relationship between TyG and MACE was quantified by calculating the E-values, which provided an estimate of the strength of the confounding factors required to explain the observed association. Finally, in the NHANES cohort, SLD was characterized using USFLI with a threshold value of ≥ 30. NHANES represents a comprehensive national survey encompassing both children and adults across the United States. Initiated in 1999, this survey is conducted biennially. Mortality data, accessible up to December 31, 2019, were sourced from NHANES Public-Use Linked Mortality Files. Notably, NHANES does not gather data pertaining to non-fatal cardiovascular events. Consequently, the parallel supportive analysis conducted through NHANES is confined to the endpoint of all-cause mortality. To examine the association between TyG and the risk of mortality among DM patients with MASLD, three multivariate Cox regression models were employed. Model 1 unadjusted. Model 2 included adjustments for age and sex, while Model 3 further accounted for smoking status, alcohol abuse, physical activity, previous CVD, BMI, SBP, TC, DBP, creatinine, uric acid, ALT, gamma-glutamyl transferase (GGT), albumin, and hemoglobin. The analysis was conducted using R software (version 4.3.2) and the Free Statistics platform (version 2.3.1), with statistical significance determined at a two-tailed p-value of less than 0.05. Results Baseline characteristics In this study, a total of 1021 patients diagnosed with T2DM accompanied by MASLD at Wenzhou Hospital were categorized into three groups based on the tertiles of TyG. Over a median follow-up period of 53.4 (46.4, 61.5) months, 101 participants (9.9%) experienced MACEs, and 62 individuals (6.1%) died. The cohort had a mean age of 57.4 years old, with females representing 39.0% (n = 398) of the population. Table 1 delineates the baseline characteristics stratified by TyG tertiles. Participants in the highest TyG tertile, despite being younger, demonstrated a higher incidence of MACE, increased mortality, a greater prevalence of hypertension, elevated SBP and DBP, more frequent use of GLP-1RA medications compared to those in other tertiles. Table 1Baseline characteristics of participantsCharacteristicsTotalTyG indexP, valueT1T2T3N1021340340341Age, year57.4 ± 15.058.7 ± 13.558.7 ± 15.154.9 ± 16.1< 0.001Female, n (%)398 (39.0)140 (41.2)131 (38.5)127 (37.2)0.562Diabetes duration, year8.4 ± 7.49.0 ± 7.38.6 ± 7.67.8 ± 7.30.104BMI, kg/m225.4 ± 4.425.1 ± 3.825.3 ± 5.025.9 ± 4.30.044SBP, mmHg143.4 ± 25.0138.9 ± 23.2143.7 ± 27.5146.7 ± 24.2< 0.001DBP, mmHg84.1 ± 8.283.1 ± 8.083.6 ± 8.985.6 ± 7.5< 0.001Hypertension, n (%)544 (53.3)167 (49.1)179 (52.6)198 (58.1)< 0.001Hyperlipidemia, n (%)631 (61.8)69 (20.3)227 (66.8)335 (98.2)< 0.001Previous CHD, n (%)43 (4.2)12 (3.5)15 (4.4)16 (4.7)0.731Previous stroke, n (%)126 (12.3)40 (11.7)42 (12.4)44 (12.9)0.891Smoking status, n (%)316 (31.0)102 (30)103 (30.3)111 (32.6)0.733Alcohol consumption, n (%)263 (25.8)88 (25.9)92 (27.2)83 (24.3)0.692Hemoglobin, g/L137.2 ± 17.3135.4 ± 17.1137.2 ± 17.2138.9 ± 17.50.042HbA1c, %9.4 ± 2.19.2 ± 2.19.3 ± 2.19.7 ± 2.00.007FPG, mmol/L6.9 ± 1.95.9 ± 1.46.9 ± 1.67.8 ± 2.1< 0.001Uric acid, µmol/L359.8 ± 112.0323.2 ± 92.9360.8 ± 101.1396.2 ± 127.9< 0.001Albumin, g/dL41.1 ± 4.440.5 ± 3.541.1 ± 4.741.8 ± 5.0< 0.001Creatinine, µmol/L70.3 ± 35.665.5 ± 26.271.4 ± 38.874.0 ± 39.70.006ALT, U/L23.0 (16.0, 37.0)20.0 (14.0, 31.0)23.0 (17.0, 37.0)25.0 (17.0, 41.0)< 0.001AST, U/L21.0 (16.0, 30.0)19.0 (15.0, 28.0)21.0 (16.0, 30.0)23.0 (17.0, 33.0)< 0.001GGT, U/L32.0 (21.0, 57.0)25.0 (17.0, 40.0)32.0 (22.0, 55.0)44.0 (26.0, 78.0)< 0.001TC, mmol/L4.6 (3.8, 5.4)4.2 (3.4, 4.8)4.4 (3.8, 5.1)5.1 (4.3, 5.9)< 0.001TG, mmol/L1.8 (1.3, 2.7)1.2 (1.0, 1.4)1.8 (1.6, 2.2)3.5 (2.6, 5.0)< 0.001HDL-C, mmol/L0.9 (0.8, 1.1)1.0 (0.9, 1.2)0.9 (0.8, 1.0)0.9 (0.8, 1.0)< 0.001LDL-C, mmol/L2.8 (2.1, 3.5)2.6 (2.1, 3.4)2.9 (2.2, 3.6)2.7 (2.0, 3.5)0.022 Discharge prescription, n (%) Hypolipidemic drugs784 (76.8)256 (75.2)258 (75.6)270 (79.3)0.585Metformin880 (86.3)294 (86.5)289 (85)297 (87.4)0.667Insulin747 (73.2)249 (73.2)242 (71.2)256 (75.1)0.517GLP-1RA175 (17.3)43 (12.8)51 (15.2)81 (23.9)< 0.001All-cause mortality, n (%)62 (6.1)9 (2.6)26 (7.6)27 (7.9)0.005CVD mortality44 (4.3)7 (2.1)16 (4.7)21 (6.2)0.028CVD incidence48 (4.7)11 (3.2)16 (4.7)21 (6.2)0.197MACEs, n (%)101 (9.9)19 (5.6)40 (11.8)42 (12.3)0.005TyG9.3 ± 0.78.6 ± 0.39.2 ± 0.210.0 ± 0.5< 0.001Data are presented as mean ± SD for continuous variables and n (%) for categorical variables. TC, total cholesterol; TG, triglyceride; HDL-C, High density lipoprotein cholesterol; LDL-C, Low density lipoprotein cholesterol; TyG, triglyceride-glucose index; BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; HbA1c, glycated hemoglobin A1c; ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT, gamma-glutamyl transferase; FPG, fasting plasma glucose; MACEs, major adverse cardiovascular events; GLP-1RA, glucagon-like peptide-1 receptor agonist; CVD, cardiovascular disease; CHD, coronary heart disease Baseline characteristics of participants Data are presented as mean ± SD for continuous variables and n (%) for categorical variables. TC, total cholesterol; TG, triglyceride; HDL-C, High density lipoprotein cholesterol; LDL-C, Low density lipoprotein cholesterol; TyG, triglyceride-glucose index; BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; HbA1c, glycated hemoglobin A1c; ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT, gamma-glutamyl transferase; FPG, fasting plasma glucose; MACEs, major adverse cardiovascular events; GLP-1RA, glucagon-like peptide-1 receptor agonist; CVD, cardiovascular disease; CHD, coronary heart disease Risk factors for MACEs Table 2 presents the disparities in clinical characteristics and baseline demographic between participants who experienced MACEs and those remained free from such events. Throughout follow-ups, MACEs occurred in 101 patients, accounting for 9.9% of the total. Participants who experienced MACEs exhibited significantly lower levels of albumin, BMI, ALT (P < 0.05), as well as a lower proportions of smoking, alcohol abuse, and metformin use. Conversely, they demonstrated significantly higher values for SBP, duration of diabetes mellitus, FPG, creatinine, AST, TG, and LDL-C, as well as increased prevalence of hypertension, CHD, and stroke. Table 2Baseline clinical characteristics of patients stratified by MACEsIndicatorsTotalNon-MACEsMACEsP valueN1021920101Age, year57.4 ± 15.055.8 ± 14.661.9 ± 10.3< 0.001Female, n (%)398 (39.0)342 (37.2)56 (55.4)< 0.001Diabetes duration, year8.4 ± 7.48.1 ± 7.211.9 ± 8.6< 0.001BMI, kg/m225.4 ± 4.425.7 ± 3.923.3 ± 7.2< 0.001SBP, mmHg143.4 ± 25.0142.7 ± 24.7150.4 ± 27.30.003DBP, mmHg84.1 ± 8.284.1 ± 8.383.9 ± 7.60.777Hypertension, n (%)544 (53.3)462 (50.2)82 (81.2)< 0.001Hyperlipidemia, n (%)631 (61.8)564 (61.3)72 (71.3)0.044Previous CHD, n (%)43 (4.2)31 (3.4)12 (11.9)< 0.001Previous stroke, n (%)126 (12.3)72 (7.8)54 (53.5)< 0.001Smoking status, n (%)316 (31.0)297 (32.3)19 (18.8)0.005Alcohol consumption, n (%)263 (25.8)244 (26.6)19 (18.8)0.090Hemoglobin, g/L137.2 ± 17.3138.1 ± 17.0128.1 ± 17.5< 0.001HbA1c, %9.4 ± 2.19.4 ± 2.19.4 ± 2.30.713FPG, mmol/L6.9 ± 1.96.8 ± 1.77.9 ± 2.6< 0.001Uric acid, µmol/L359.8 ± 112.0359.2 ± 110.3385.4 ± 126.90.048Albumin, g/dL41.1 ± 4.441.3 ± 4.340.0 ± 5.80.009Creatinine, µmol/L70.3 ± 35.669.1 ± 35.180.8 ± 38.00.002ALT, U/L23.0 (16.0, 37.0)23.0 (16.0, 38.0)19.5 (14.0, 29.8)0.021AST, U/L21.0 (16.0, 30.0)21.0 (16.0, 30.0)24.0 (18.0, 34.0)0.007GGT, U/L32.0 (21.0, 57.0)32.0 (21.0, 58.0)32.0 (20.0, 52.0)0.467TC, mmol/L4.6 (3.8, 5.4)4.5 (3.8, 5.4)4.6 (3.5, 5.4)0.260TG, mmol/L1.8 (1.3, 2.7)1.8 (1.3, 2.7)2.1 (1.6, 2.7)< 0.001HDL-C, mmol/L0.9 (0.8, 1.1)0.9 (0.8, 1.1)0.9 (0.8, 1.1)0.971LDL-C, mmol/L2.8 (2.1, 3.5)2.5 (1.9, 3.4)2.8 (2.1, 3.6)0.039 Discharge prescription, n (%) Hypolipidemic drugs784 (76.8)702 (76.4)82 (81.2)0.447Metformin880 (86.3)810 (88.1)70 (69.3)< 0.001Insulin747 (73.2)665 (72.3)82 (81.2)0.055GLP-1RA175 (17.3)161 (17.7)14 (14.0)0.359TyG9.3 ± 0.79.2 ± 0.79.5 ± 0.60.006Data are presented as mean ± SD for continuous variables and n (%) for categorical variables. TC, total cholesterol; TG, triglyceride; HDL-C, High density lipoprotein cholesterol; LDL-C, Low density lipoprotein cholesterol; TyG, triglyceride-glucose index; BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; HbA1c, glycated hemoglobin A1c; ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT, gamma-glutamyl transferase; FPG, fasting plasma glucose; GLP-1RA, glucagon-like peptide-1 receptor agonist; CVD, cardiovascular disease; CHD, coronary heart disease Baseline clinical characteristics of patients stratified by MACEs Data are presented as mean ± SD for continuous variables and n (%) for categorical variables. TC, total cholesterol; TG, triglyceride; HDL-C, High density lipoprotein cholesterol; LDL-C, Low density lipoprotein cholesterol; TyG, triglyceride-glucose index; BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; HbA1c, glycated hemoglobin A1c; ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT, gamma-glutamyl transferase; FPG, fasting plasma glucose; GLP-1RA, glucagon-like peptide-1 receptor agonist; CVD, cardiovascular disease; CHD, coronary heart disease Association between TyG and cardiovascular events Kaplan-Meier curve (Fig. 2) demonstrates that within the upper tertile of TyG, there is a statistically significant proportional increase in the risk of cardiovascular events (log-rank P = 0.001). Throughout follow-ups, individuals with the highest tertile consistently exhibited the greatest cumulative risk of cardiovascular events. Kaplan-Meier curves for A all-cause mortality and B major adverse cardiovascular events (MACEs) stratified by TyG tertiles Table 3 illustrates the quantitative association between TyG and the incidence of MACE in patients with coexisting T2DM and MASLD. As a continuous variable, each 1-unit increase in TyG was associated with a significantly elevated risk of MACE (HR = 1.71, 95% CI: 1.26, 2.33) and ACM (HR = 1.85, 95% CI: 1.25, 2.73) in the fully adjusted Model 3. When analyzed categorically, the risk of cardiovascular events demonstrated a progressive increase with ascending TyG tertiles across all analytical models. In Model 3, specific comparisons for MACE showed HR = 2.37 (95% CI: 1.36, 4.13) for T3 versus T1. For ACM, the comparison between T3 and T1 yielded an HR of 3.12 (95% CI: 1.43, 6.83). Table 3Cox regression analysis for the association between TyG and all-cause mortality and MACEsSubgroupsModel1Model2Model3HR (95%CI)P-valueHR (95%CI)P-valueHR (95%CI)P-value All-cause mortality TyG(continuous)1.63 (1.17, 2.29)0.0041.88 (1.30, 2.70)0.0011.85 (1.25, 2.73)0.002TyG (tertile)T11(Ref)1(Ref)1(Ref)T22.95 (1.38, 6.31)0.0052.58 (1.20, 5.54)0.0152.27 (1.04, 4.95)0.039T33.13 (1.47, 6.65)0.0033.25 (1.52, 6.94)0.0023.12 (1.43, 6.83)0.004P for trend0.0040.0020.004 MACEs TyG(continuous)1.45 (1.11, 1.90)0.0071.72 (1.28, 2.30)< 0.0011.71 (1.26, 2.33)0.001TyG (tertile)T11(Ref)1(Ref)1(Ref)T22.16 (1.25, 3.72)0.0061.88 (1.08, 3.26)0.0241.69 (0.97, 2.96)0.066T32.31 (1.34, 3.97)0.0032.49 (1.44, 4.29)0.0012.37 (1.36, 4.13)0.002P for trend0.0030.0010.002Model 1: UnadjustedModel 2: Adjusted for Model 1 + age + sexModel 3: Adjusted for Model 2 + BMI, smoking status, drinking status, SBP, DBP, diabetes duration, LDL-C, and metformin use Cox regression analysis for the association between TyG and all-cause mortality and MACEs Model 1: Unadjusted Model 2: Adjusted for Model 1 + age + sex Model 3: Adjusted for Model 2 + BMI, smoking status, drinking status, SBP, DBP, diabetes duration, LDL-C, and metformin use Furthermore, RCS analysis (Fig. 3) revealed a linear association between TyG and an increased risk of ACM and MACEs (P for non-linearity = 0.401 and 0.841, corresponding for ACM and MACE, respectively). Restricted cubic spline curve illustrating the association between the TyG index and A all-cause mortality and B MACEs. The solid line represents the estimated probability, while the shaded area denotes the 95% confidence interval. The background histogram illustrates the distribution density of TyG within the study cohort. To ensure model stability and minimize the influence of extreme outliers, the X-axis is restricted to the range between the 1st and 99.5th percentiles of the TyG distribution. A significant linear relationship was observed between TyG and ACM, MACE risk Subgroup analysis As illustrated in Fig. 4, subgroup analyses demonstrated that the association between TyG and ACM remained consistent across various clinical and demographic characteristics, including age, BMI, HbA1c levels, sex, and histories of hypertension, hyperlipidemia, and FIB-4 level (P for interaction > 0.05). Notably, the association between TyG and increased MACE incidence was more pronounced in younger individuals, with HR of 2.73 and 95% CI of 1.39 to 5.68, as depicted in Fig. 4B (P for interaction = 0.029). Stratified analysis of the associations between TyG index and A all-cause mortality and B MACEs. Adjusted for age, sex, BMI, smoking status, drinking status, SBP, DBP, diabetes duration, LDL-C, and metformin use Analysis of ROC curve values for TyG The evaluation of predictive performance for ACM and MACE through ROC curve analysis demonstrated that the baseline risk model attained an area under the curve (AUC) of 0.851 and 0.821, respectively (P < 0.05). Incorporating TyG significantly enhanced AUC to 0.868 for ACM and 0.834 for MACE (all P < 0.05). Comparative ROC curves depicting this improvement are presented in Fig. 5. ROC Curve analysis for TyG index predicted A all-cause mortality and B MACEs. Baseline model: age, sex, BMI, smoking status, drinking status, SBP, DBP, diabetes duration, LDL-C, and metformin use Table 4 demonstrates that the inclusion of TyG substantially enhanced the predictive capability for MACE, as evidenced by a 36.1% increase in NRI (P < 0.001) and a 1.6% increase in IDI (P = 0.008). Similarly, the predictive capability for ACM was significantly improved, with a 34.4% increase in NRI (P = 0.007) and a 1.7% increase in IDI (P = 0.036). Table 4The AUC for each index to discriminate all-cause mortality and MACEsc-statisticP-valueNRI (continuous)P-valueIDI (95%CI)P-value All-cause mortality Baseline0.851 (0.805, 0.898)1.0 (Ref)1.0 (Ref)1.0 (Ref)+TyG0.868 (0.827, 0.909)0.0280.344 (0.091, 0.596)0.0070.017 (0.001, 0.032)0.036 MACEs Baseline0.821 (0.782, 0.860)1.0 (Ref)1.0 (Ref)1.0 (Ref)+TyG0.834 (0.798, 0.870)0.0460.361 (0.159, 0.563)< 0.0010.016 (0.004, 0.029)0.008Baseline model: age, sex, BMI, smoking status, drinking status, SBP, DBP, diabetes duration, LDL-C, and metformin use The AUC for each index to discriminate all-cause mortality and MACEs Baseline model: age, sex, BMI, smoking status, drinking status, SBP, DBP, diabetes duration, LDL-C, and metformin use Decision curve analysis (DCA) The clinical utility of the model was evaluated via DCA. As illustrated in Supplementary Fig. S4, TyG-based model (red line) yielded a consistently higher net benefit compared to both the ‘screen-all’ and ‘screen-none’ strategies across a wide threshold probability range of approximately 0.10 to 0.90. Notably, TyG based model also demonstrated increased net benefit over the basic model (blue line) across the same range. Mediation analysis The mediation analysis indicated that eGFR served as indirect mediators in the association between TyG and both MACE and ACM (refer to Fig. 6). Specifically, the indirect effects of eGFR on ACM and MACE were quantified at 8.8% (95%CI: 0.4%, 21.2%) and 10.6% (95%CI: 4.3%, 17.5%), respectively. In a similar vein, the BRI-mediated indirect effect accounted for 6.2% (95%CI: 1.6%, 11.1%) on ACM. Mediation analyses of BRI and eGFR in the association between TyG and ACM and MACE in Wenzhou cohort. Panels A–B present mediation models for all-cause mortality, with eGFR (Panel A), and BRI (Panel B) as potential mediators. Panels C–D present corresponding mediation models for MACE, with eGFR (Panel C), and BRI (Panel D) as mediators. For each panel, total effects, indirect effects, 95% confidence intervals (CIs), and the proportion mediated are shown. In accordance with the primary regression analyses, A and C models were adjusted for age, sex, BMI, smoking status, drinking status, SBP, DBP, diabetes duration, LDL-C, and metformin use. BMI was omitted in B and D models due to high collinearity with BRI. B and D models were adjusted for age, sex, smoking status, drinking status, SBP, DBP, diabetes duration, LDL-C, and metformin use Sensitivity analysis To thoroughly investigate the robustness of the association between TyG and the risk of MACE in the context of T2DM complicated by MASLD, this study conducted a sensitivity analysis across seven dimensions. Firstly, we incorporated additional adjustments for the administration of hypolipidemic drugs, insulin, and GLP-1RA into the multivariate Cox proportional hazards model (Supplementary Table S4). Secondly, participants who died within the first year after the initiation of follow-up were excluded (Supplementary Table S5). Thirdly, standardized methods were employed to identify and remove outliers in TyG (Supplementary Table S6). Fourth, after accounting for the competing risks between CVD incidence and ACM, we analyzed the correlation between TyG and CVD incidence (Supplementary Table S7). Fifth, PSM was employed to control for potential confounding variables (Supplementary Table S8). The findings from the aforementioned sensitivity analysis align with those of the preliminary analysis. Sixth, E-values were computed to evaluate the influence of unmeasured confounders, with the E-value for the TyG-MACE association was 2.81, while for TyG-ACM it was 3.10 (Supplementary Fig. S5). The relatively large E-values indicate that they were robust. Finally, parallel supportive analysis of TyG-ACM association was performed utilizing independent samples from NHANES 1999–2018. This cohort initially comprised 101,316 participants. Following the exclusion of individuals under 18 years old, those without MASLD and DM, those missing TyG, and those with viral hepatitis or excessive alcohol abuse, a total of 1,343 DM patients with MASLD were included in the analysis. The baseline characteristics of these participants are detailed in Supplementary Table S9. Over a follow-up duration averaging 8.1 ± 5.2 years, a total of 380 deaths (28.3%) were recorded. The results of the Cox regression analysis, as presented in Supplementary Table S10, indicate that while no significant difference was observed in the univariate Model 1 (HR = 1.00, 95% CI 0.88–1.14), a significant positive association between TyG and ACM was identified in both Model 2 (HR = 1.18, 95% CI 1.02–1.37) and Model 3 (HR = 1.20, 95% CI 1.01–1.41). This finding implies the existence of negative confounders, with age serving as a pivotal negative confounding factor in this analysis. Discussion This study is an investigation into the association between TyG and cardiovascular risk in patients concurrently diagnosed with T2DM and MASLD. The principal findings of this study were as follows: (1) Following comprehensive adjustments for confounding variables, TyG maintained a robust association with MACEs, irrespective of whether it was analyzed as a continuous or categorical variable; (2) A distinct linear association was observed between TyG values and the incidence of MACEs, with age serving as a modifying factor; and (3) The incorporation of TyG into conventional risk prediction models resulted in modest yet statistically significant enhancements in predictive accuracy across various statistical metrics; (4) Importantly, the core finding of a positive TyG-ACM association has been independently validated in the NHANES cohort, thereby further bolstering the reliability and generalizability of conclusions. These findings underscore the significant role of TyG as an indicator for IR, contributing to enhanced precision in cardiovascular risk stratification within this specific patient cohort. By leveraging TyG to more precisely identify individuals at high risk, clinicians can institute targeted interventions, thereby optimizing the efficacy of intensified management strategies. TyG, an accessible, straightforward, and easily quantifiable parameter for detecting IR, has been associated with the onset and prognosis of various cardiometabolic diseases [33–35]. Notably, TyG has demonstrated high predictive accuracy for CVD and ACM in adults with prediabetes or diabetes mellitus [33, 36]. Furthermore, in a study conducted by Wei et al., TyG was found to be significantly associated with diabetes mellitus-related mortality among adults with metabolic syndrome [37]. Consistently, TyG has been associated with the onset and progression of MASLD. Notably, Xu et al. established that TyG holds significant clinical value in identifying individuals at high risk for MASLD, demonstrating strong predictive accuracy [38]. Furthermore, Qiao et al. revealed that an elevated TyG is associated not only with heightened risks of CVD but also with ACM in patients with MASLD [39]. The promising predictive capability of TyG for the risk of MASLD has similarly been corroborated in studies conducted in various countries and regions [40–42]. In summary, while prior research has substantiated that TyG serves as a reliable predictor for cardiovascular events, its prognostic significance in patients concurrently suffering from T2DM and MASLD warrants further investigation. To date, only one previous study has explored the association between TyG-related indicators and mortality in patients with diabetes mellitus complicated by MASLD. This study conducted by Zhang et al. on data from NHANES identified a positive association between TyG and both ACM and cardiovascular mortality among patients with MASLD and either DM or prediabetes [19]. However, Zhang et al. aggregated patients with prediabetes, T1DM, and T2DM into a single analytical group, instead of exclusively focusing on confirmed T2DM—a population with substantially elevated cardiovascular risk. Additionally, their outcome assessment was confined to mortality endpoints (all-cause and cardiovascular-specific), precluding evaluation of non-fatal cardiovascular events such as non-fatal MI and stroke. In the current study, the data from Chinese adults diagnosed with both T2DM and MASLD were analyzed to assess the potential association between TyG and the risk of cardiovascular diseases. This study aimed to fill the existing knowledge gap and extend the findings of previous studies. Previous research has identified an association between an elevated TyG and increased ACM, with varying degrees of association reported [43–45]. We performed parallel supportive analysis utilizing the NHANES cohort, employing stricter inclusion criteria (confirmed diagnosis of diabetes and MASLD as defined by USFLI), and the findings aligned with those reported by Zhang et al. Specifically, within the fully adjusted model, we identified a significant positive association between TyG and ACM (HR = 1.20, Supplementary Table S10), a result that is largely consistent with the HR value documented by Zhang et al. (HR = 1.215) [19]. Utilizing progressively refined Cox regression analyses, this study also demonstrated that TyG serves as an independent predictor of MACEs, even after extensive adjustment for confounding variables, which was particularly pronounced in patients under 60 years old with T2DM complicated by MASLD. In individuals younger than 60, heightened metabolic activity may render TyG changes more impactful on the metabolic system. Furthermore, the relatively lower prevalence of chronic diseases in this younger cohort suggests that even minor alterations in TyG could more readily influence their health status. Conversely, in participants aged 60 and above, the presence of multiple chronic conditions may obscure the effect of TyG on mortality due to the complex interplay of these diseases. In addition, incorporating TyG into existing risk assessment models significantly improved the predictive accuracy of various statistical indicators, including C statistic, NRI, IDI, and DCA, highlighting its practical value in clinical applications, surpassing the traditional scope of cardiovascular risk indicators. The interplay between glucose and triglyceride metabolism may exert synergistic effects, thereby accelerating the progression of atherosclerosis [46] which partially accounts for the predictive efficacy of TyG. Although the precise mechanisms connecting an elevated TyG to adverse outcomes in patients with T2DM and MASLD remain not fully elucidated, IR is likely a critical factor. IR initiates a cascade of metabolic disturbances, including dyslipidemia, hyperglycemia, obesity, and hypertension, which collectively compromise vascular systems. These metabolic abnormalities expedite the formation of advanced glycation end products and the generation of free radicals, thereby enhancing oxidative stress and inflammatory responses. This sequence of events ultimately undermines endothelial integrity and vascular health [47]; Moreover, the impairment of insulin signaling due to IR contributes to the progression of atherosclerosis and the development of hypertension [48]. Additionally, IR leads to the hyperactivation of platelets, enhancing the abnormal adhesion capabilities and elevating tissue factor expression via thromboxane A2-mediated pathways [49]. Together, these pathological changes substantially increase the inflammatory responses and risk of arterial thrombosis, thereby offering a mechanistic rationale for TyG’s predictive effectiveness in forecasting cardiovascular events among patients with concurrent T2DM and MASLD. However, the research mechanisms mentioned above are of a speculative nature. In the future, further studies can be conducted on the interactions between these mechanisms to provide new targets for improving patient prognosis. Through an assessment of the mediating effects of BRI and eGFR, the extent to which the association between TyG and the risk of MACE is attributable to obesity and renal function, respectively. IR can facilitate the accumulation of visceral adipose tissue and induce lipotoxicity. The buildup of free fatty acids within skeletal muscle cells may inhibit the IRS-1/PI3K signaling pathway, accelerate the catabolism of muscle proteins, and consequently lead to an increase in BRI [50]. An elevated BRI can initiate a cascade of reactions that heighten the mortality risk associated with MASLD [20]. In addition, IR promotes renal vasoconstriction, sodium retention, and hypertension, all of which are critical risk factors for chronic kidney disease [51, 52]. Furthermore, IR activates oxidative stress and inflammatory pathways, leading to renal cell fibrosis and damage [53]. An elevated IR also stimulates renin-angiotensin-aldosterone system, contributing to increased glomerular pressure, glomerulosclerosis, and endothelial injury [54]. All the aforementioned conditions can lead to a decline in eGFR, a factor that has been established as a prognostic indicator influencing cardiovascular events and mortality rates. The cross-sectional nature of biomarkers, in which the exposure factors and mediating factors were measured at the same time point, and the observational design excluded the establishment of a time or causal sequence. Therefore, the above explanation should be regarded as a biologically reasonable hypothesis rather than an explicit mechanism inference. This study has several strengths. First, employing a retrospective cohort design with a substantial sample size, this study provides robust data support for elucidating the association between TyG and the risk of MACE among patients concurrently diagnosed with T2DM and MASLD. Second, through mediation analysis, the study systematically clarifies the mediating effects of obesity status and renal function on the association between TyG and the risk of mortality, thereby enhancing the comprehension of underlying pathophysiological mechanisms. Third, TyG-ACM association has been replicated in an independent external cohort from NHANES, thereby bolstering the generalizability and reliability of the study’s conclusions across diverse ethnic and geographical contexts. Despite these strengths, this study still harbors several limitations. First, as a single-center retrospective cohort study in China, causal relationships are challenging to establish conclusively. Selection bias and geographical limitations may affect patient inclusion and data acquisition. Additionally, regional treatment practices, including widespread intensive hypoglycemic strategies, could limit generalizability. Caution should thus be exercised when extrapolating these results to other populations. Second, the cross-sectional nature of baseline medication assessment precludes determination of whether drug exposure preceded or followed TyG measurement, complicating causal inference. Third, as an observational study, it is incapable of assessing the causal association between TyG and ACM and MACE in MASLD among T2DM patients. Consequently, further clinical and basic investigations are warranted to corroborate the reliability of current findings. Fourth, due to the data absence of ultrasonographic measurements in NHANES, hepatic steatosis was defined based on USFLI ≥ 30. However, the sensitivity and specificity of USFLI have been validated in previous studies [55, 56]. Additionally, the diagnosis of MASLD in Wenzhou cohort relied on abdominal ultrasonography interpreted by a single radiologist, which may limit diagnostic reproducibility. Fifth, eGFR and BRI data used in the mediation analysis came from the baseline survey, precluding the establishment of a strict temporal sequence; thus, inferences about mediation effects need future validation. Sixth, the scarcity of ACM events may result in diminished statistical power. Seventh, in the absence of standardized clinical cut-off values for TyG, participants were categorized based on tertiles values, which may not adequately reflect clinically relevant thresholds. Eighth, while the study was adjusted for measurable confounders, it was acknowledged that unmeasured lifestyle factors (e.g., dietary intake, physical activity intensity) and medication adherence patterns could theoretically influence the outcomes. Ninth, NHANES study utilized data from diabetic patients without distinguishing between different types of diabetes. Conclusion In summary, this study has demonstrated that TyG serves as an independent predictor for MACEs and mortality among patients concurrently diagnosed with T2DM and MASLD. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1. Acknowledgements The authors thank the staff at the Department of Endocrinology, the Second Affiliated Hospital and Yuying Children’s Hospital of Wenzhou Medical University, and all the patients who participated in the study. Author contributions YCL and CML designed the study; JYZ and ZJD collected biochemical data; JX drafted the manuscript. All authors read and approved the final manuscript. Funding This study was supported by the Wenzhou Municipal Science and Technology Bureau (Y2023360 to Zhijuan Dai). Data availability It can be requested from the author if necessary. Declarations Ethics approval and consent to participate This study was approved by the Ethics Committee of the Second Affiliated Hospital of Wenzhou Medical University (2025-K-270-01, dated February 15, 2025). The author waived the requirement for patients’ written informed consent for the anonymity of patients was strictly confidential and the current work was based on observation. All methods were performed in accordance with the relevant guidelines and regulations. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Jing Xu and Jianyong Zhao have contributed equally to this work. Contributor Information Yue-Chun Li, Email: liyuechun1980@sina.com. Chunmei Li, Email: 250120575@qq.com. References Associated Data Data Availability Statement It can be requested from the author if necessary.