Association of the triglyceride–glucose index with the severity of lower extremity arterial disease across glucose metabolic states: a single-center cross-sectional study of 656 patients with angiographically confirmed disease
- 1Department of Endocrinology, Cardiovascular and Cerebrovascular Disease Hospital, General Hospital of Ningxia Medical University
- 2Department of Vascular Surgery, General Hospital of Ningxia Medical University
Abstract
Objective: To examine the cross-sectional association between the triglyceride–glucose (TyG) index and the severity of lower extremity arterial disease (LEAD) in hospitalized patients with angiographically confirmed LEAD, to test whether this association differs across Glucose Metabolic States, and to quantify the ability of the TyG index to discriminate prevalent ischemic tissue loss (Rutherford grades 5–6).
Methods: This single-center, retrospective, cross-sectional study included 656 consecutive patients with angiographically confirmed LEAD who were hospitalized in the Department of Vascular Surgery at the General Hospital of Ningxia Medical University from January 2023 through December 2024. Diagnosis was confirmed by digital subtraction angiography (DSA). Patients were classified into four groups according to their Glucose Metabolic States: normal glucose tolerance (NGT, n = 150), impaired glucose regulation (IGR, n = 148), diabetes with a duration of <10 years (DM<10y, n = 174), and diabetes with a duration of ≥10 years (DM≥10y, n = 184). Glucose and lipid parameters and LEAD severity (Rutherford classification) were assessed during the same hospitalization. Spearman correlation, multinomial logistic regression (with grade 1 mild claudication as the reference), ordinal logistic regression with a TyG × Glucose Metabolic States interaction term, proportional-odds assessment by cumulative binary splits, ROC analysis with bootstrap internal validation (1,000 resamples), and decision-curve analysis were performed. Sensitivity analyses used continuous glycemic measures instead of the categorical groups, excluded implausibly low fasting glucose values, and added liver enzymes to the adjustment set. The primary outcome was ischemic tissue loss, defined as Rutherford grades 5–6 (n = 205, 31.2%).
Results: The TyG index increased progressively across Glucose Metabolic States categories (8.79 ± 0.51, 9.15 ± 0.48, 9.57 ± 0.56, and 9.67 ± 0.52 for NGT, IGR, DM < 10 years, and DM ≥ 10 years, respectively; P < 0.001) and was strongly correlated with Rutherford grade (Spearman r = 0.677, P < 0.001). In a multinomial logistic regression model adjusted for age, sex, smoking, hypertension, BMI, and fatty liver disease (all variance inflation factors < 1.5), the TyG index remained associated with each severity category above the reference, with adjusted odds ratios per 0.5-unit increment of 2.51 (grade 2), 2.83 (grade 3), 7.56 (grade 4), 9.62 (grade 5), and 20.45 (grade 6). The interaction between TyG and Glucose Metabolic States was statistically significant (likelihood-ratio χ² = 95.63, df = 3, P < 0.001); stratum-specific odds ratios per 0.5-unit increment were 1.51 (95% CI 1.07–2.13) in IGR, 2.50 (1.90–3.29) in DM < 10 years, and 8.14 (5.23–12.67) in DM ≥ 10 years, whereas the NGT-stratum estimate was unstable (quasi-complete separation) and could not be interpreted. When the categorical groups were replaced with a continuous glycemic measure, the interaction with fasting plasma glucose remained significant but was in the opposite direction (χ² = 9.36, df = 1, P = 0.002): the TyG–severity association was stronger at lower fasting glucose levels (OR per 0.5 unit 4.81 at 4.8 mmol/L) than at higher fasting glucose levels (3.46 at 8.3 mmol/L). For discriminating Rutherford grades 5–6, the AUC was 0.833 (95% CI 0.796–0.870); the Youden-optimal cutoff was 9.42 (sensitivity 85.9%, specificity 71.0%, positive predictive value 57.3%, negative predictive value 91.7%, LR + 2.96, LR− 0.199), and this cutoff was stable across bootstrap resamples. Decision-curve analysis showed a positive net benefit relative to treat-all or treat-none strategies across threshold probabilities of approximately 0.09–0.57.
Conclusion: In this cohort of patients with angiographically confirmed LEAD, the TyG index was independently associated with LEAD severity. The strength of this association appeared to vary across Glucose Metabolic States, but the apparent gradient from normal glucose tolerance to long-standing diabetes was not reproduced when glucose was modelled continuously, suggesting that part of the categorical pattern is attributable to the grouping itself (i.e., mathematical coupling and range restriction). The TyG index showed moderate ability to discriminate prevalent ischemic tissue loss. Because exposure and outcome were assessed during the same hospitalization, these findings represent a cross-sectional association rather than a predictive one; the proposed cutoff values were derived from the same sample in which they were evaluated and require external validation before they can be considered for clinical use.
Related
- Association of the triglyceride-glucose index with dynamic bidirectional transitions of cardiovascular-kidney-metabolic syndrome: a multistate Markov model study based on CHARLS
- Association between triglyceride glucose–relative fat mass index and obstructive sleep apnea risk : A cross-sectional study from NHANES
- Association between triglyceride-glucose index and total trabecular bone score in US adults : A cross-sectional study
- Association of the triglyceride-to-high-density lipoprotein cholesterol ratio with kidney stone status: cross-sectional and machine-learning analyses
- Association between triglyceride-glucose and fibrosis-4 indices and nephrolithiasis: a cross-sectional case-control study.
BibTeX
@article{zhou2026association,
title = {{Association of the triglyceride–glucose index with the severity of lower extremity arterial disease across glucose metabolic states: a single-center cross-sectional study of 656 patients with angiographically confirmed disease}},
author = {Xiaoli Zhou and Dan Zhao and Hong Lei and Gang Zhao and Liping Sha and Xixuan Lu and Haiyan Liang},
journal = {Frontiers in Endocrinology},
year = {2026},
month = oct,
doi = {10.3389/fendo.2026.1957246},
url = {https://www.frontiersin.org/articles/10.3389/fendo.2026.1957246/full},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Association of the triglyceride–glucose index with the severity of lower extremity arterial disease across glucose metabolic states: a single-center cross-sectional study of 656 patients with angiographically confirmed disease
AU - Xiaoli Zhou
AU - Dan Zhao
AU - Hong Lei
AU - Gang Zhao
AU - Liping Sha
AU - Xixuan Lu
AU - Haiyan Liang
PY - 2026
DA - 2026/10/23
JO - Frontiers in Endocrinology
DO - 10.3389/fendo.2026.1957246
UR - https://www.frontiersin.org/articles/10.3389/fendo.2026.1957246/full
AB - Objective: To examine the cross-sectional association between the triglyceride–glucose (TyG) index and the severity of lower extremity arterial disease (LEAD) in hospitalized patients with angiographically confirmed LEAD, to test whether this association differs across Glucose Metabolic States, and to quantify the ability of the TyG index to discriminate prevalent ischemic tissue loss (Rutherford grades 5–6). Methods: This single-center, retrospective, cross-sectional study included 656 consecutive patients with angiographically confirmed LEAD who were hospitalized in the Department of Vascular Surgery at the General Hospital of Ningxia Medical University from January 2023 through December 2024. Diagnosis was confirmed by digital subtraction angiography (DSA). Patients were classified into four groups according to their Glucose Metabolic States: normal glucose tolerance (NGT, n = 150), impaired glucose regulation (IGR, n = 148), diabetes with a duration of <10 years (DM<10y, n = 174), and diabetes with a duration of ≥10 years (DM≥10y, n = 184). Glucose and lipid parameters and LEAD severity (Rutherford classification) were assessed during the same hospitalization. Spearman correlation, multinomial logistic regression (with grade 1 mild claudication as the reference), ordinal logistic regression with a TyG × Glucose Metabolic States interaction term, proportional-odds assessment by cumulative binary splits, ROC analysis with bootstrap internal validation (1,000 resamples), and decision-curve analysis were performed. Sensitivity analyses used continuous glycemic measures instead of the categorical groups, excluded implausibly low fasting glucose values, and added liver enzymes to the adjustment set. The primary outcome was ischemic tissue loss, defined as Rutherford grades 5–6 (n = 205, 31.2%). Results: The TyG index increased progressively across Glucose Metabolic States categories (8.79 ± 0.51, 9.15 ± 0.48, 9.57 ± 0.56, and 9.67 ± 0.52 for NGT, IGR, DM < 10 years, and DM ≥ 10 years, respectively; P < 0.001) and was strongly correlated with Rutherford grade (Spearman r = 0.677, P < 0.001). In a multinomial logistic regression model adjusted for age, sex, smoking, hypertension, BMI, and fatty liver disease (all variance inflation factors < 1.5), the TyG index remained associated with each severity category above the reference, with adjusted odds ratios per 0.5-unit increment of 2.51 (grade 2), 2.83 (grade 3), 7.56 (grade 4), 9.62 (grade 5), and 20.45 (grade 6). The interaction between TyG and Glucose Metabolic States was statistically significant (likelihood-ratio χ² = 95.63, df = 3, P < 0.001); stratum-specific odds ratios per 0.5-unit increment were 1.51 (95% CI 1.07–2.13) in IGR, 2.50 (1.90–3.29) in DM < 10 years, and 8.14 (5.23–12.67) in DM ≥ 10 years, whereas the NGT-stratum estimate was unstable (quasi-complete separation) and could not be interpreted. When the categorical groups were replaced with a continuous glycemic measure, the interaction with fasting plasma glucose remained significant but was in the opposite direction (χ² = 9.36, df = 1, P = 0.002): the TyG–severity association was stronger at lower fasting glucose levels (OR per 0.5 unit 4.81 at 4.8 mmol/L) than at higher fasting glucose levels (3.46 at 8.3 mmol/L). For discriminating Rutherford grades 5–6, the AUC was 0.833 (95% CI 0.796–0.870); the Youden-optimal cutoff was 9.42 (sensitivity 85.9%, specificity 71.0%, positive predictive value 57.3%, negative predictive value 91.7%, LR + 2.96, LR− 0.199), and this cutoff was stable across bootstrap resamples. Decision-curve analysis showed a positive net benefit relative to treat-all or treat-none strategies across threshold probabilities of approximately 0.09–0.57. Conclusion: In this cohort of patients with angiographically confirmed LEAD, the TyG index was independently associated with LEAD severity. The strength of this association appeared to vary across Glucose Metabolic States, but the apparent gradient from normal glucose tolerance to long-standing diabetes was not reproduced when glucose was modelled continuously, suggesting that part of the categorical pattern is attributable to the grouping itself (i.e., mathematical coupling and range restriction). The TyG index showed moderate ability to discriminate prevalent ischemic tissue loss. Because exposure and outcome were assessed during the same hospitalization, these findings represent a cross-sectional association rather than a predictive one; the proposed cutoff values were derived from the same sample in which they were evaluated and require external validation before they can be considered for clinical use.
SN - 1664-2392
ER -