Diabetes constitutes one of the most formidable public health challenges worldwide, leading to a considerable healthcare and socioeconomic burden that is likely to persist in the foreseeable future.1 Report from the International Diabetes Federation (IDF) in 2021 has shown approximately 537 million individuals suffer from type 2 diabetes mellitus (T2DM) in the world, and this figure may increase by 46 % to 783 million in 2045.2 Patients with diabetes are predisposed to detrimental multi-system impairments, rendering remarkably higher risk of cardiovascular (CV) disease and death than that among non-diabetic patients.3,4 The lifetime CV risk has been documented to be 2–4 times greater in patients with T2DM compared with non-diabetes.5,6 For patients with diabetes, early detection and stratification for the CV risk, based on which targeted interventions could be therefore formulated, is of great importance to reduce the CV morbidity and mortality, improve their quality of life and to reduce global healthcare cost.7

Preponderance of evidence to date has demonstrated that aortic stiffness (AS), a key biomarker of vascular health, is considered as a robust predictor for CV morbidity and all-cause mortality above and beyond conventional CV risk factors, in various populations.8, 9, 10 Pulse wave velocity (PWV) constitutes the most effective non-invasive approach for evaluating the extent of AS. A number of observational studies have found AS was increased in diabetes, by measuring carotid-femoral pulse wave velocity (cfPWV), the “gold standard” method for assessing AS, and the brachial-ankle pulse wave velocity (baPWV) which was mainly used in East Asia population.11,12 Both cfPWV and baPWV were commonly used, they still have inherent differences when evaluating arterial stiffness: cfPWV mainly reflects the stiffness of aorta and acts as the “gold standard” of AS, whereas baPWV mainly reflects the stiffness of lower limb peripheral arteries.

Regarding the predictive role of AS for the risk of CV morbidity and mortality in diabetes patients, the evidence from previous individual studies seems discrepant. Data from many prospective cohort studies suggest that AS is independently associated with future CV outcomes or all-cause mortality in diabetes, whereas other studies do not show this role of AS. Methodological differences or population variations may contribute to the discrepancies. To date, however, there is no overall quantitative estimate of such the predictive role of AS in patients with diabetes specifically. An individual participant meta-analysis published by Ben-Shlomo Y et al. in 20149 has shown that AS, measured by cfPWV, was able to improve the CV risk prediction in standard nondiabetic elderly, where each 1 m/s increase in cfPWV was associated with a 7 % increase of the CV risk. In that study, patients with diabetes were analyzed as a subgroup and the data analysis was modeled as log transformation of cfPWV, rather than m/s or dichotomy classification, which limited the applicability of the results. In addition, considering the timeliness and pertinency of the study before, it is necessary to conduct a special aggregate-data meta-analysis to determine the exact predictive role of AS for CV events and all-cause mortality specialized in the population with diabetes.

Therefore, we conducted this systematic review and meta-analysis with the aim to address the question that can AS measured by PWV predict the future CV events or all-cause mortality in patients with diabetes and calculate the pooled effect size. We hope this study would help to positively influence clinical practice, guidelines, and further related research.