Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2025 Jun 18;17(1):221.
doi: 10.1186/s13098-025-01774-w.

Association of DNA methylation epigenetic markers with all-cause mortality and cardiovascular disease-related mortality in diabetic population: a machine learning-based retrospective cohort study

Yuxin Nong #  1   2 , Huazhen Huang #  3 , Lulu Xu #  4 , Xin Tan  1 , Shuai Xu  1 , Xinyu Zhou  1 , Yiyao Zeng  1 , Yufeng Jiang  1 , Hezi Jiang  1 , Xiangyu Wang  1 , Xian Li  1 , Anchen Xu  1 , Qiaoyi Sun  1 , Hongju Wang  5 , Pinfang Kang  5 , Jili Fan  6 , Xiaohong Bo  6 , Huimin Fan  7   8 , Yafeng Zhou  9   10
Affiliations

Association of DNA methylation epigenetic markers with all-cause mortality and cardiovascular disease-related mortality in diabetic population: a machine learning-based retrospective cohort study

Yuxin Nong et al. Diabetol Metab Syndr. .

Abstract

Background: Diabetes has a large and diverse population, with individuals exhibiting significant heterogeneity in the disease. The factors influencing survival and prognosis are complex, making early intervention in diabetic populations particularly challenging. Research has demonstrated a close relationship between DNA methylation (DNAm) clocks and aging as well as various diseases, showing superior predictive capabilities. However, the relationship between DNAm clocks and long-term survival in diabetic patients, particularly concerning cardiovascular-related mortality, remains unclear.

Methods: We analyzed the data of the diabetes population cohort in the National Health and Nutrition Examination Survey, which was followed for 20 years. We employed eight machine learning (ML) models to analyze 29 potential DNAm derived epigenetic markers and utilized Cox regression models to assess the risks of all-cause mortality and cardiovascular disease-related mortality in the diabetic population. Additionally, we applied restricted cubic spline (RCS) to analyze potential influence trends.

Results: A total of 454 people with diabetes were followed up, with a median follow-up time of 177.6 months. Through machine learning methods, we identified several high-performing DNAm markers, finding that four epigenetic biomarkers, ZhangAge (HR = 2.86, 95% CI: 2.19-3.73, P < 0.001), GrimAge2Mort (HR = 3.06, 95% CI: 2.26-4.14, P < 0.001), TIMP1Mort (HR = 2.95, 95% CI: 2.18-4.01, P < 0.001), and PhenoAge (HR = 2.94, 95% CI: 1.23-3.88, P < 0.001), were significantly associated with all-cause mortality in the diabetic population. Further research indicated that GrimAge2 Mort (HR = 2.86, 95% CI: 1.30-6.29, P = 0.009) and TIMP1Mort (HR = 4.08, 95% CI: 2.17-7.68, P < 0.001) were associated with cardiovascular disease-related mortality. RCS curves demonstrated that the mortality risk for all diabetic patients increased with rising levels of these DNAm epigenetic markers.

Conclusion: We found four DNAm-derived epigenetic markers (ZhangAge, GrimAge2 Mort, TIMP1Mort, PhenoAge) that are associated with all-cause mortality risk in the diabetic population. Further research suggested that GrimAge and PhenoAge influence the risk of cardiovascular-related mortality.

Keywords: All-cause mortality; Cardiovascular disease-related mortality; DNA methylation biomarkers; Diabetes.

PubMed Disclaimer

Conflict of interest statement

Declarations. Ethical approval: Not application. Consent to participate: Not applicable. Consent to publish: All authors agree to publish. Competing interests: The authors declare no competing interests.

Figures

Fig. 1
Fig. 1
ROC curve of machine learning models. XGBoost: eXtreme gradient Boosting gradient boosting, AdaBoost: Adaptive Boosting, GBM: gradient boosting machine, MLP: multilayer perception, SVM: support vector machine
Fig. 2
Fig. 2
Machine learning hybrid matrix
Fig. 3
Fig. 3
Machine learning model variable feature importance based on XGBoost
Fig. 4
Fig. 4
Shapely values distribution of feature variables of machine learning model based on XGBoost
Fig. 5
Fig. 5
Kaplan-Meier survival plots for all-cause mortality
Fig. 6
Fig. 6
Restricted cubic splines for mortality risk in diabetes population
See this image and copyright information in PMC

Similar articles

See all similar articles

References

    1. Global regional, national burden of diabetes. From 1990 to 2021, with projections of prevalence to 2050: a systematic analysis for the global burden of disease study 2021. Lancet. 2023;402:203–34. - PMC - PubMed
    1. Magliano DJ, Boyko EJ. IDF DIABETES ATLAS. 10th ed. Brussels; 2021.
    1. Parker ED, Lin J, Mahoney T, Ume N, Yang G, Gabbay RA, et al. Economic costs of diabetes In the U.S. In 2022. Diabetes Care. 2024;47:26–43. - PubMed
    1. Golovaty I, Ritchie ND, Tuomilehto J, Mohan V, Ali MK, Gregg EW, et al. Two decades of diabetes prevention efforts: A call to Innovate and revitalize our approach to lifestyle change. Diabetes Res Clin Pract. 2023;198:110195. - PMC - PubMed
    1. Ali MK, Pearson-Stuttard J, Selvin E, Gregg EW. Interpreting global trends in type 2 diabetes complications and mortality. Diabetologia. 2022;65:3–13. - PMC - PubMed

LinkOut - more resources