Metabolic score for visceral fat (METS-VF) is associated with arterial stiffness progression in Type 2 diabetes: a retrospective cohort study of China Authors: Kun Li, Ruili Yin, Dong Zhao, Longyan Yang Published: 2025 ABSTRACT BackgroundArterial stiffness is a risk factor for cardiovascular disease (CVD). There is an interaction effect between arterial stiffness and diabetes and may increase the risk of CVD. It is important to find a sensitive biomarker to estimate the risk of arterial stiffness in Type 2 diabetes mellitus (T2DM).MethodsParticipants with type 2 diabetes mellites registered at the Metabolic management center (MMC) of Beijing Luhe Hospital. The association between METS-VF and the progression of arterial stiffness (AS) was evaluated using a logistical regression model, and the odds ratio (OR) and 95% confidence interval (CI) were reported and visualizing the risk of arterial stiffness by drawing restricted cubic splines. The trajectory of the metabolic score for visceral Fat (METS-VF) was fitted with the latent class mixture model (LCMM). Moreover, the Cross-lagged panel model was performed to detect the bidirectional cross-lagged associations between METS-VF and brachial-ankle pulse wave velocity (baPWV).ResultsA total of 701 participants enrolled in the procedure of data analysis. The OR (95%CI) of METS-VF on arterial stiffness progression was 1.54 (1.02, 2.36). We identified three discrete trajectories of METS-VF, the ORs (95%CI) in three trajectories were 1.00 (reference), 1.56 (1.02,2.40), 3.33 (1.65, 6.80), respectively. And there were significant bidirectional cross-lagged associations between METS-VF and baPWV after adjusting for confounding factors, β1 and β2 were 0.025 (95%CI:0.006,0.045, P = 0.011) and 0.211 (95%:0.109,0.312, P < 0.001), respectively.ConclusionsThere was a causal relationship between METS-VF and the progression of arterial stiffness in T2DM. The series of results provided high-quality evidence that controlling METS-VF may affect the progression of arterial stiffness in T2DM.Clinical trial numberNot applicable.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12902-025-02141-6. Background Type 2 diabetes mellitus (T2DM) is a significant public health problem, especially in low and middle-income countries and areas [1, 2]. The estimated prevalence of diabetes had climbed to 12.4% in mainland China [3]. T2DM is a well-known risk factor for renal dysfunction, cardiovascular disease (CVD), and all-cause mortality [4]. The epidemiologic study indicates that the global impact of diabetes may continue to increase substantially [5]. Arterial stiffness is recognized as an important risk factor for CVD. Recent research evidence indicated an interaction effect between arterial stiffness and diabetes and may increase the risk of CVD [6, 7]. Therefore, it is important to find a sensitive biomarker to estimate the risk of arterial stiffness in T2DM. In the new decade, several studies have explored surrogate markers for insulin resistance, including the triglyceride-glucose (TyG) index, and also found the value of these indicators in predicting the risk of CVD occurrence. However, TyG does not take into account age, gender, and other metabolic states, and thus cannot achieve relatively personalized assessment. The metabolic score for visceral Fat (METS-VF) is an extended indicator on metabolic score for insulin resistance (METS-IR), taking into account age, gender, BMI and other comprehensive indicators [8]. It is a personalized alternative indicator for insulin resistance and can assess the risk of various vascular complications [9–12]. The present study aims to elucidate the longitudinal and causal relationships between METS-VF and arterial stiffness in T2DM. The brachial-ankle pulse wave velocity (baPWV) changes are considered the gold standard for estimating arterial stiffness, which is used to reflect the arterial stiffness progression in this study. All participants of this study have been registered at the National Metabolic Management Center (MMC), an innovation project for the management of metabolic diseases and complications in China [13]. Methods Participants and study design Participants registered at the Metabolic Management Center (MMC) of Beijing Luhe Hospital. The recruitment process consisted of three key steps: structured oral questionnaire interviews, venous blood sample collection, and comprehensive systemic physical examinations. Participants aged 18–80 years and diagnosed with T2DM were included in the present study. T2DM was defined by confirmation of self-reported diabetes that was previously diagnosed by a physician, current use of antidiabetic agents, Fasting Plasma Glucose (FPG) ≥ 7.0 mmol/L, 2-h postload glucose ≥ 11.1 mmol/L, or HbA1c ≥ 6.5% [14]. Participants were excluded if they met any of the following conditions: (1) a confirmed diagnosis of malignant tumor (regardless of treatment status); (2) current pregnancy or lactation; (3) presence of acute diabetic complications (e.g., diabetic ketoacidosis, hyperosmolar hyperglycemic state); (4) missing data for critical variable. Data collection All data were collected by trained nurses and physicians in strict accordance with a standardized protocol [13]. The oral questionnaires were collected and stored by the Electronic Data Capture System. Supplemental material 1 presents the main variable information. Smoking and drinking status. Smoking status was defined as yes if participants had never smoked or had quit smoking for ≥ 12 months; drinking status was recorded as yes if participants consumed alcohol at least once per week (or almost weekly). Education attainment was categorized as < high school and ≥ high school. The physical examination included measurements of height, body weight, and waist circumference. Body mass index (BMI) was calculated using the formula: BMI = weight (kg) / [height (m)]². Blood samples were collected in the morning after overnight fasting. Levels of high-density lipoprotein (HDL), low-density lipoprotein (LDL), and triglycerides were measured using an automatic biochemical analyzer (AU5800, Beckman Coulter, USA). BaPWV was measured to assess arterial stiffness using an automated recording device (BP-203RPE III, for PWV/ABI, Omron Healthcare Co.) Participants were instructed to rest in a supine position for at least 5 min before measurement to ensure stable hemodynamics. BaPWV was calculated as the distance from the bilateral upper arms to the ankles divided by the time interval between the wavefront of the brachial waveform and that of the ankle waveform. Definitions The arterial stiffness progression was defined according to whether baPWV persisted high or increased during follow-up. Supplemental Fig. 1 shows the detail of the method for the definition of arterial stiffness progression. We substituted the quartile of baPWV changes for the change value of baPWV measurements during the follow-up period [15]– [16]. The METS-VF was defined as the following formula:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eqalign{& METS - VF \cr & = 4.466 + 0.011\left[ {{{\left( {Ln\left( {METS - IR} \right)} \right)}^3}} \right] \cr & + 3.239\left[ {{{\left( {Ln\left( {Whtr} \right)} \right)}^3}} \right] + 0.319\left( {Sex} \right) \cr & + 0.594\left( {Ln\left( {Age} \right)} \right) \cr} $$\end{document} And the METS-IR was calculated as:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:METS-IR=\left(Ln\right[\left(2*FPG\right)+TG\left]*BMI\right)/\left(Ln\right(HDL\left)\right).$$\end{document} Statistical analysis All statistical analyses were performed in R software (version 4.2.2, https://www.r-project.org/). The generalized additive models (GAMs) were used to investigate the non-linear association between METS-VF and baPWV. The association between METS-VF and the progression of arterial stiffness was evaluated using a logistical regression model, and the odds ratio (OR) and 95% confidence interval (CI) were reported. We constructed three models with adjustments for major covariables: Model 1 was the crude model; Model 2 was adjusted for age, sex, duration, ideal smoke, drink, education level, TG, SBP, BMI and HDL; Model 3 was adjusted for insulin, oral antidiabetic, Lipid-lowering agents, and the variables in Model 2. In addition, to assess the relationship between METS-VF and the risk of arterial stiffness progression in continuous, we used the restricted cubic spline to estimate the multivariate-adjusted OR (95%CI) and draw the continuous risk plot. Three knots at the 10th, 50th, and 90th percentiles were used for fitting the restricted cubic spline. Moreover, the association between METS-VF and baPWV was also estimated using the logistic model. To identify distinct trajectories of METS-VF over the study period, a latent class mixture model (LCMM) was constructed [17]– [18]. All model fitting procedures were implemented using the lcmm package in R. The Bayesian Information Criterion (BIC) was used to select the optimal number of latent classes (trajectory groups). The cross-lagged panel model (CLPM) was employed to verify the bidirectional relationship between METS-VF and baPWV. This model simultaneously quantifies two key effect sizes: the impact of baseline METS-VF measurements on subsequent baPWV values, and the influence of baseline baPWV measurements on subsequent METS-VF values. Given the relatively large absolute value of baPWV, we apply logarithmic transformation for baPWV. The Lavaan package of R was used to perform the procedure [19]. Model 1 was the crude model; Model 2 was adjusted for age, sex, duration, ideal smoke, drink, education level, SBP, and BMI; Model 3 was adjusted for TG, HDL, insulin, oral antidiabetic, Lipid-lowering agents, and plus the variables in Model 2. Results Demographic and clinical characteristics of participants After data cleaning, a total of 701 participants enrolled in the procedure of data analysis (Fig. 1). The demographic and clinical characteristics of these participants were shown in Table 1. The mean (SD) follow-up duration was 36.07 (8.85) months. Among the 701 participants, the mean (SD) age is 51.57 (11.59) years, and duration of diabetes was 87.98 (80.28) months. Participants comprised 421 (60.06%) males and 280 (39.94%) females. Fig. 1The data cleaning procedure of this study. Abbreviations: HbA1c, glycosylated hemoglobin; FPB, Fasting Plasma Glucose; METS-VF, the metabolic score for visceral Fat; baPWV, the brachial-ankle pulse wave velocity The data cleaning procedure of this study. Abbreviations: HbA1c, glycosylated hemoglobin; FPB, Fasting Plasma Glucose; METS-VF, the metabolic score for visceral Fat; baPWV, the brachial-ankle pulse wave velocity Table 1Demographic and clinical characteristics of participants in this studyALLMaleFemale P N = 701N = 421N = 280Age, mean(SD), year51.57 (11.59)49.33 (11.38)54.94 (11.09)<0.001Duration, mean(SD), month87.98 (80.28)78.24 (75.34)102.54 (85.23)<0.001Ideal Smoke, n (%)197 (28.22%)183 (43.68%)14 (5.02%)<0.001Drink, n (%)289 (41.46%)267 (63.72%)22 (7.91%)<0.001High school or more, n (%)409 (58.35%)276 (65.56%)133 (47.50%)<0.001BMI, mean(SD), kg/m226.76 (3.96)27.11 (3.92)26.23 (3.98)0.004Weight, mean (SD), kg74.47 (14.09)80.05 (13.00)66.08 (11.24)<0.001Height, mean (SD), cm166.48 (8.67)171.72 (6.08)158.61 (5.41)<0.001SBP, mean (SD), mmHg131.75 (18.07)130.98 (17.62)132.91 (18.70)0.170DBP, mean (SD), mmHg79.66 (11.81)81.14 (12.19)77.44 (10.86)<0.001Waist, mean (SD), cm94.19 (10.44)96.22 (9.83)91.14 (10.59)<0.001HbA1c, mean (SD), %8.63 (2.17)8.60 (2.22)8.67 (2.10)0.673baPWV, mean(SD), cm/s1547.24 (301.73)1509.80 (293.40)1603.52 (305.84)<0.001Lipid-lowering agents, n (%)213 (30.82%)124 (29.95%)89 (32.13%)0.601HDL, mean (SD), mmol/L1.17 (0.27)1.12 (0.26)1.26 (0.26)<0.001LDL, mean (SD), mmol/L3.03 (0.88)3.02 (0.85)3.05 (0.92)0.658TC, mean (SD), mmol/L4.77 (1.17)4.72 (1.17)4.84 (1.17)0.188TG, median(IQR), mmol/L1.60 [1.06–2.38]1.68 [1.11–2.58]1.49 [1.01–2.06]0.003FPG, mean (SD), mmol/L9.50 (4.17)9.60 (4.24)9.34 (4.06)0.420GFR, mean (SD), mL/min/1.73m295.76 (19.91)91.21 (20.32)102.58 (17.19)<0.001Insulin, n (%)224 (31.95%)119 (28.27%)105 (37.50%)0.013Oral antidiabetic, n (%)423 (60.34%)266 (63.18%)157 (56.07%)0.071METS-VF6.90 (0.48)6.99 (0.45)6.76 (0.48)<0.001Abbreviations: SBP, systolic blood pressure; HbA1c, glycosylated hemoglobin; HDL, High-density lipoprotein; LDL, Low-density lipoprotein; TC, total cholesterol; TG, triglyceride; FPB, Fasting Plasma Glucose; BMI, Body mass index; METS-VF, the metabolic score for visceral Fat; baPWV, the brachial-ankle pulse wave velocity Table 1Demographic and clinical characteristics of participants in this studyALLMaleFemale P N = 701N = 421N = 280Age, mean(SD), year51.57 (11.59)49.33 (11.38)54.94 (11.09)<0.001Duration, mean(SD), month87.98 (80.28)78.24 (75.34)102.54 (85.23)<0.001Ideal Smoke, n (%)197 (28.22%)183 (43.68%)14 (5.02%)<0.001Drink, n (%)289 (41.46%)267 (63.72%)22 (7.91%)<0.001High school or more, n (%)409 (58.35%)276 (65.56%)133 (47.50%)<0.001BMI, mean(SD), kg/m226.76 (3.96)27.11 (3.92)26.23 (3.98)0.004Weight, mean (SD), kg74.47 (14.09)80.05 (13.00)66.08 (11.24)<0.001Height, mean (SD), cm166.48 (8.67)171.72 (6.08)158.61 (5.41)<0.001SBP, mean (SD), mmHg131.75 (18.07)130.98 (17.62)132.91 (18.70)0.170DBP, mean (SD), mmHg79.66 (11.81)81.14 (12.19)77.44 (10.86)<0.001Waist, mean (SD), cm94.19 (10.44)96.22 (9.83)91.14 (10.59)<0.001HbA1c, mean (SD), %8.63 (2.17)8.60 (2.22)8.67 (2.10)0.673baPWV, mean(SD), cm/s1547.24 (301.73)1509.80 (293.40)1603.52 (305.84)<0.001Lipid-lowering agents, n (%)213 (30.82%)124 (29.95%)89 (32.13%)0.601HDL, mean (SD), mmol/L1.17 (0.27)1.12 (0.26)1.26 (0.26)<0.001LDL, mean (SD), mmol/L3.03 (0.88)3.02 (0.85)3.05 (0.92)0.658TC, mean (SD), mmol/L4.77 (1.17)4.72 (1.17)4.84 (1.17)0.188TG, median(IQR), mmol/L1.60 [1.06–2.38]1.68 [1.11–2.58]1.49 [1.01–2.06]0.003FPG, mean (SD), mmol/L9.50 (4.17)9.60 (4.24)9.34 (4.06)0.420GFR, mean (SD), mL/min/1.73m295.76 (19.91)91.21 (20.32)102.58 (17.19)<0.001Insulin, n (%)224 (31.95%)119 (28.27%)105 (37.50%)0.013Oral antidiabetic, n (%)423 (60.34%)266 (63.18%)157 (56.07%)0.071METS-VF6.90 (0.48)6.99 (0.45)6.76 (0.48)<0.001Abbreviations: SBP, systolic blood pressure; HbA1c, glycosylated hemoglobin; HDL, High-density lipoprotein; LDL, Low-density lipoprotein; TC, total cholesterol; TG, triglyceride; FPB, Fasting Plasma Glucose; BMI, Body mass index; METS-VF, the metabolic score for visceral Fat; baPWV, the brachial-ankle pulse wave velocity Demographic and clinical characteristics of participants in this study Abbreviations: SBP, systolic blood pressure; HbA1c, glycosylated hemoglobin; HDL, High-density lipoprotein; LDL, Low-density lipoprotein; TC, total cholesterol; TG, triglyceride; FPB, Fasting Plasma Glucose; BMI, Body mass index; METS-VF, the metabolic score for visceral Fat; baPWV, the brachial-ankle pulse wave velocity Association between METS-VF and BaPWV in cross-sectional data of baseline The non-linear relationship between METS-VF and baPWV is shown in Supplemental Fig. 2. In the GAM model, baPWV exhibited a continuously increasing trend as METS-VF increased. Moreover, in the logistic model, after adjusted confounding factors, the ORs (95%CI) across increasing tertiles of METS-VF were 1.00 (reference), 2.49 (1.54,4.07), and 8.26 (4.29,16.34), respectively (Supplemental Table 1). Association between METS-VF and the progression of arterial stiffness in cohort The association between METS-VF and the progression of arterial stiffness was estimated by the logistic model (Table 2). The OR (95%CI) of METS-VF on arterial stiffness progression was 1.54 (1.02, 2.36), after adjusted confounding factors. Table 2OR and 95% CI of METS-VF on arterial stiffness progression in the T2DM participantsOR (95%CI) P Model 12.08 (1.48, 2.86)< 0.001Model 21.62 (1.08,2.46)0.021Model 31.54 (1.02, 2.36)0.044Model 1 was the crude model; model 2 was adjusted for age, sex, duration, ideal smoke, drink, education level, TG, SBP, BMI and HDL; model 3 was adjusted for insulin, oral antidiabetic, Lipid-lowering agents and the variables in model 2. Abbreviations: OR, odds ratio; CI, confidence interval Table 2OR and 95% CI of METS-VF on arterial stiffness progression in the T2DM participantsOR (95%CI) P Model 12.08 (1.48, 2.86)< 0.001Model 21.62 (1.08,2.46)0.021Model 31.54 (1.02, 2.36)0.044Model 1 was the crude model; model 2 was adjusted for age, sex, duration, ideal smoke, drink, education level, TG, SBP, BMI and HDL; model 3 was adjusted for insulin, oral antidiabetic, Lipid-lowering agents and the variables in model 2. Abbreviations: OR, odds ratio; CI, confidence interval OR and 95% CI of METS-VF on arterial stiffness progression in the T2DM participants Model 1 was the crude model; model 2 was adjusted for age, sex, duration, ideal smoke, drink, education level, TG, SBP, BMI and HDL; model 3 was adjusted for insulin, oral antidiabetic, Lipid-lowering agents and the variables in model 2. Abbreviations: OR, odds ratio; CI, confidence interval In addition, Fig. 2 illustrates the odds of arterial stiffness progression increased with the increase of METS-VF (P < 0.001), the solid line is multivariate-adjusted OR and the corresponding red background indicates the 95% CI derived from restricted cubic spline regression. Fig. 2The association between METS-VF and the progression of baPWV. The solid line is the multivariate-adjusted OR and the corresponding red background indicates the 95% CI derived from restricted cubic spline. The knots are located at the 10th, 50th, and 90th percentiles for METS-VF The association between METS-VF and the progression of baPWV. The solid line is the multivariate-adjusted OR and the corresponding red background indicates the 95% CI derived from restricted cubic spline. The knots are located at the 10th, 50th, and 90th percentiles for METS-VF Moreover, we did a subgroup analysis in the sex group, the trend of restriction splines was similar in both male and female groups (Supplemental Fig. 3). METS-VF trajectories and their effect on arterial stiffness progression We identified three discrete trajectories of METS-VF during the period of follow-up. The trajectory clusters were named based on the visual change patterns of METS-VF: High level, the METS-VF maintained with high METS-VF level during the period of follow-up; Median group, characterized by maintaining median METS-VF levels throughout follow-up; Low and increased gently group, staring with a low METS-VF level, and increased gently during the period of follow-up (Fig. 3). The demographic and clinical character of participants were shown in Supplemental Table 2. Compared to the participants in the trajectory group of low and increased gently, those in the trajectory group of Median group and High level had longer duration of diabetes and more males, larger BMI. Fig. 3The metabolic score for visceral Fat (METS-VF) trajectory groups during the period of follow-up The metabolic score for visceral Fat (METS-VF) trajectory groups during the period of follow-up The relationship between METS-VF trajectories and their effect on arterial stiffness progression was described in Table 3. Compared to the participants in low and increased trajectory group, those in the trajectory group of Median group and High level have higher odds of arterial stiffness progression. The ORs (95%CI) in three trajectories were 1.00 (reference), 1.56 (1.02, 2.40), 3.33 (1.65, 6.80), respectively. Table 3OR and 95% CI of METS-VF trajectory classes on arterial stiffness progression in the T2DM participantsTrajectory ClassesModel 1Model 2Model 3OR (95%CI) P OR (95%CI) P OR (95%CI) P Low and increased gentlyReference-Reference-Reference-Median group1.38 (1.03,1.92)0.0421.52 (1.01,2.33)0.0481.56 (1.02, 2.40)0.041High level2.38 (1.48,3.84)< 0.0013.29 (1.65,6.65)0.0013.33 (1.65, 6.80)0.001Model 1 was the crude model; model 2 was adjusted for age, sex, duration, ideal smoke, drink, education level, TG, SBP, BMI and HDL; model 3 was adjusted for insulin, oral antidiabetic, Lipid-lowering agents and the variables in model 2. Abbreviations: OR, odds ratio; CI, confidence interval Table 3OR and 95% CI of METS-VF trajectory classes on arterial stiffness progression in the T2DM participantsTrajectory ClassesModel 1Model 2Model 3OR (95%CI) P OR (95%CI) P OR (95%CI) P Low and increased gentlyReference-Reference-Reference-Median group1.38 (1.03,1.92)0.0421.52 (1.01,2.33)0.0481.56 (1.02, 2.40)0.041High level2.38 (1.48,3.84)< 0.0013.29 (1.65,6.65)0.0013.33 (1.65, 6.80)0.001Model 1 was the crude model; model 2 was adjusted for age, sex, duration, ideal smoke, drink, education level, TG, SBP, BMI and HDL; model 3 was adjusted for insulin, oral antidiabetic, Lipid-lowering agents and the variables in model 2. Abbreviations: OR, odds ratio; CI, confidence interval OR and 95% CI of METS-VF trajectory classes on arterial stiffness progression in the T2DM participants Model 1 was the crude model; model 2 was adjusted for age, sex, duration, ideal smoke, drink, education level, TG, SBP, BMI and HDL; model 3 was adjusted for insulin, oral antidiabetic, Lipid-lowering agents and the variables in model 2. Abbreviations: OR, odds ratio; CI, confidence interval The causal relationship between METS-VF and BaPWV The results of cross-lagged panel models between METS-VF and baPWV were shown in Fig. 4, there were significant bidirectional cross-lagged associations between METS-VF and baPWV after adjusting for confounding factors. The standardized correlation coefficient of baseline METS-VF and baPWV of follow-up (β1) was 0.025 (95%CI:0.006–0.045, P = 0.011), and that of baseline baPWV and METS-VF of follow-up (β2) was 0.211 (95%CI:0.109–0.312, P < 0.001). Table 4 shows the details of the CLPM models: the standardized mean square residual (SRMR) was 0.120, and the comparative fit index (CFI) was 0.502. Fig. 4The results of cross-lagged analysis between METS-VF and baPWV. Abbreviations: METS-VF, the metabolic score for visceral Fat; baPWV, the brachial-ankle pulse wave velocity The results of cross-lagged analysis between METS-VF and baPWV. Abbreviations: METS-VF, the metabolic score for visceral Fat; baPWV, the brachial-ankle pulse wave velocity Table 4The results of cross-lagged analysis between METS-VF and BaPWVMETS1→baPWV2baPWV1→METS2METS1→METS2baPWV1→baPWV2R2 of baPWVR2 of METSCFISRMRRMSEAModel 10.028(0.008,0.048)0.116(0.058,0.274)0.834(0.791,0.877)0.775(0.724,0.826)0.5370.6461.0000.0000.000Model 20.029(0.010,0.049)0.219(0.118,0.321)0.500(0.460,0.541)0.647(0.598,0.695)0.5090.5560.5090.1650.351Model 30.025(0.006,0.045)0.211(0.109,0.312)0.500(0.460,0.541)0.644(0.596,0.693)0.5120.5590.5020.1200.280Model 1 was crude model; model 2 was adjusted for age, sex, duration, smoke, drink, education level, SBP, BMI; model 3 was adjusted for TG, HDL, insulin, oral antidiabetic, Lipid-lowering agents and plus the variables in model 2. METS1was the measurement of METS-VF at baseline; METS2 was the measurement of METS-VF at last time of follow-up; baPWV1 was the measurement of baPWV at baseline; baPWV2 was the measurement of baPWV at last time of follow-up Table 4The results of cross-lagged analysis between METS-VF and BaPWVMETS1→baPWV2baPWV1→METS2METS1→METS2baPWV1→baPWV2R2 of baPWVR2 of METSCFISRMRRMSEAModel 10.028(0.008,0.048)0.116(0.058,0.274)0.834(0.791,0.877)0.775(0.724,0.826)0.5370.6461.0000.0000.000Model 20.029(0.010,0.049)0.219(0.118,0.321)0.500(0.460,0.541)0.647(0.598,0.695)0.5090.5560.5090.1650.351Model 30.025(0.006,0.045)0.211(0.109,0.312)0.500(0.460,0.541)0.644(0.596,0.693)0.5120.5590.5020.1200.280Model 1 was crude model; model 2 was adjusted for age, sex, duration, smoke, drink, education level, SBP, BMI; model 3 was adjusted for TG, HDL, insulin, oral antidiabetic, Lipid-lowering agents and plus the variables in model 2. METS1was the measurement of METS-VF at baseline; METS2 was the measurement of METS-VF at last time of follow-up; baPWV1 was the measurement of baPWV at baseline; baPWV2 was the measurement of baPWV at last time of follow-up The results of cross-lagged analysis between METS-VF and BaPWV Model 1 was crude model; model 2 was adjusted for age, sex, duration, smoke, drink, education level, SBP, BMI; model 3 was adjusted for TG, HDL, insulin, oral antidiabetic, Lipid-lowering agents and plus the variables in model 2. METS1was the measurement of METS-VF at baseline; METS2 was the measurement of METS-VF at last time of follow-up; baPWV1 was the measurement of baPWV at baseline; baPWV2 was the measurement of baPWV at last time of follow-up Discussion In the present study, we demonstrated the causal relationship between METS-VF and the risk of arterial stiffness progression by applying trajectory modeling and logistic regression analysis. Moreover, the CLMP revealed significant bidirectional cross-lagged associations between METS-VF and baPWV. Collectively, these findings provide evidence that METS-VF may contribute to the progression of arterial stiffness in individuals with T2DM. Originally developed as a surrogate marker for intra-abdominal fat, METS-VF has demonstrated superior predictive performance compared to other visceral adipose tissue indices and is associated with incident T2DM and cardiometabolic health [12]. In this study, we revealed the potential clinical significance of METS-VF in evaluating complications in patients with T2DM. Given its utility as a biomarker across multiple disease contexts, METS-VF is well-suited for widespread adoption in primary care settings. As reflected in its calculation formula, METS-VF incorporates not only markers of insulin resistance but also adjustments for age, gender, and waist circumference. In contrast to the TyG index, METS-VF more effectively captures interindividual variations in metabolic status [9, 20]. Compared with the linear analysis method, the LCMM could better demonstrate the etiological relationships through phenotypic analysis of subgroups at higher or lower risk [17, 21, 22]. In the present study, we identified three discrete trajectories of METS-VF during follow-up. The results indicated that participants in the medium- and high-trajectory groups exhibited a higher likelihood of arterial stiffness progression compared to those in the low-trajectory group. These findings support the clinical utility of assessing baseline METS-VF, monitoring its trajectory over time, and implementing risk stratification based on METS-VF levels. In particular, patients with elevated METS-VF should receive continuous monitoring for arterial stiffness risk and timely interventional measures, especially in primary healthcare settings. Significant differences in demographic and clinical characteristics were observed across trajectory groups. Participants in the high-level trajectory group were older, had higher BMI, longer diabetes duration, and a higher proportion of males—consistent with known risk factors for CVD and metabolic syndrome [23, 24]. Elevated baseline METS-VF was temporally linked to subsequent increases in arterial stiffness, reinforcing a potential causal role of METS-VF in arterial stiffness progression. Conversely, higher baseline baPWV was also associated with later elevations in METS-VF, suggesting that vascular pathology may play a temporal role in driving metabolic disorders. Similar bidirectional relationships have been reported between the TyG index and baPWV [25]. Given the large absolute values of baPWV, we applied a logarithmic transformation to facilitate interpretation of the CLPM coefficients, which led to overly large value of β2. Therefore, when interpreting the model coefficients, it is necessary to combine the weights and measures of the variables. The potential mechanisms underlying the association between METS-VF and arterial stiffness progression may be outlined as follows. First, METS-VF serves as a reliable surrogate for VAT. The anatomical structure of VAT interacts with insulin resistance to increase the release of free fatty acids directly into the portal circulation, resulting in dysregulation of glucose uptake, glycogen synthesis, and glucose oxidation [26]. Second, impaired endothelial function can lead to dysfunction of capillary dilation, contraction, or sparse distribution, which leads to arterial wall hardening. Arterial stiffness can damage the capillaries, leading to a vicious cycle [27]. The main strength of this study was that the National Metabolic Management Center was a national project that provided standard management of metabolic patients, and the risk factor measurements were collected comprehensively, meaning that the model could be adjusted for more potentially confounding variables. There were some limitations in this study. First, due to insufficient sample size, we were unable to conduct additional subgroup analyses (e.g., by education level or age group). The relatively wide confidence intervals observed in the restricted cubic spline analysis may also be attributed to the sample size (Supplemental Fig. 2). Second, during the COVID-19 pandemic, patients may have reduced their MMC visit frequency, potentially resulting in some participant loss to follow-up. In addition, due to the long course of arterial stiffness, it is difficult for us to observe the rapid progression of the arterial stiffness within the limited follow-up period. Therefore, the progression of arterial stiffness was defined by the changes in the baPWV sequence of patients before and after follow-up. Last, this is a single-center study. The geographical and lifestyle characteristics of the study population may affect the extrapolation of the conclusion. Subsequent research should aim to include more diverse populations to confirm these findings. Conclusions In summary, our study demonstrates that there was a causal relationship between METS-VF and the progression of arterial stiffness in T2DM. The series of results provided high-quality evidence that controlling METS-VF may affect the progression of arterial stiffness in T2DM. The mechanisms underlying the relationship between METS-VF and progression of arterial stiffness need to be elucidated in future studies. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3