Introduction

Arterial stiffening significantly impacts cardiovascular health by increasing pulsatile load, raising left ventricular workload and impairing tissue perfusion. Such changes contribute to a heightened risk of cardiovascular and age-related diseases. Arterial stiffening is characterized by increased arterial wall stiffness, thickening, and lumen dilatation. These structural alterations reduce arterial compliance and buffering capacity, exacerbate pulsatile load, and enhance the penetration of pulsatile energy into the microvasculature of target organs. Through its elastic reservoir function, the aorta plays a pivotal role in modulating pulsatile load and maintaining hemodynamic stability.

With growing concern about arterial stiffening and pulsatile hemodynamics in cardiovascular disease, recent studies have emphasized the role of aortic dilatation in modulating pulsatile load and left ventricular afterload1,2,3,4. Historically, aortic dilatation has been recognized as a manifestation of target organ damage associated with arterial stiffening and hypertension, both linked to an elevated risk of cardiovascular events5,6. However, more recent evidence suggests that aortic dilatation also serves as an adaptive, compensatory response to increased arterial stiffness and elevated blood pressure. Campos Arias et al. proposed a concept of age-dependent regulatory function of the aorta, where increased aortic stiffness was not always paralleled by elevated aortic impedance, because the hemodynamic impact of aortic stiffening may be overcome by aortic dilatation4. Supporting this, a community-based study found that individuals with smaller aortic diameters and higher pulse wave velocity (PWV) faced greater cardiovascular risk3. These findings align with the water hammer equation, Zc = ρ × PWV/πr2 (where ρ is blood density and r is the aortic radius), indicating that aortic dilatation may compensate for the pulsatile overload from arterial stiffening. Our previous study further demonstrated that aortic dilatation mitigates the impact of pulsatile load on left ventricular remodeling, validating its compensatory role2. Aortic dilatation represents a continuum of vascular remodeling in response to arterial stiffening and hemodynamic stress. Early-stage dilatation serves a compensatory function by buffering increased pulsatile load, while progressive dilatation reflects the cumulative effects of vascular aging and chronic hemodynamic burden. This dynamic spectrum highlights the complex and evolving nature of aortic remodeling in cardiovascular health and disease.

Interestingly, while hypertension and arterial stiffening generally promote aortic dilatation, diabetes mellitus exerts a distinct effect by limiting aortic dilatation. This phenomenon was first observed in patients with aneurysms. A study reported that type 2 diabetes is associated with a reduced incidence of abdominal aortic aneurysm7,8. Further corroborating these findings, the European Society of Cardiology (ESC) noted in a 2019 consensus document that patients with diabetes exhibit a lower prevalence of aortic aneurysm or dissection than age-matched individuals without diabetes9. However, limited studies are exploring this phenomenon in populations without aortic aneurysms. Previous studies offer conflicting findings, with a Western study suggesting a protective effect of diabetes against aortic dilatation while another involving Chinese participants with hypertension found no significant correlation10,11. Our prior research involving 1556 elderly Chinese patients with hypertension identified a significant negative correlation between diabetes and aortic diameter2. These findings suggest that the relationship between diabetes and aortic diameter and its influence on pulsatile hemodynamics warrants further investigation.

Given that diabetes is a known risk factor for arterial stiffening and cardiovascular disease, arterial stiffening is associated with an increase in aortic diameter. This raises an important question: how should we explain the negative correlation between diabetes and aortic diameter10,12,13? We hypothesize that aortic wall remodeling in diabetes restricts aortic dilatation, which compromises its buffering function and elevates cardiovascular risk. This study aims to investigate the relationships between diabetes, aortic diameter, pulsatile load, and their collective impact on left ventricular remodeling in an elderly Chinese population.

Methods

Study population

This single-center, retrospective observational study enrolled consecutive patients aged 60 and above without aortic aneurysm or dissection who underwent brachial-ankle pulse wave velocity (baPWV) measurement at the Department of Geriatrics, First Hospital of China Medical University, between December 2018 and November 2022. Patients were excluded if they had incomplete data, body mass index (BMI) outside the range of 18.5–28 kg/m2, heart failure with reduced ejection fraction or acute heart failure, chronic kidney disease (CKD) stage IV–V, atrial fibrillation, and any acute medical condition. Trained physicians collected demographic and clinical data, which were recorded in the hospital’s electronic information system. Morning fasting blood samples were collected for laboratory analysis. This study was conducted in accordance with the Declaration of Helsinki and approved by the ethical review board of The First Hospital of China Medical University (approval number: [2024] No.130). Informed consent was obtained from all participants. All methods were carried out in accordance with relevant guidelines and regulations.

Evaluation of arterial stiffness

BaPWV was used to assess arterial stiffness, utilizing the OMRON Vascular Profiler (VP-1000; OMRON Healthcare, Kyoto, Japan). Calibrated blood pressure cuffs with integrated sensors were applied bilaterally to the brachial and ankle arteries following a standardized five-minute rest in the supine position. Two automated baPWV measurements were obtained, and the average PWV from the right side was recorded to ensure standardization.

24-Hour ambulatory blood pressure monitoring

Blood pressure (BP) monitoring was performed over 24 h using an ambulatory device (Meditech ABPM-05; Meditech Ltd., Budapest, Hungary) with a cuff applied to the right brachial artery. BP was measured every 30 min from 6:00 AM to 10:00 PM and hourly from 10:00 PM to 6:00 AM. Average 24-h systolic blood pressure (SBP), diastolic blood pressure (DBP), and heart rate (HR) were recorded. Pulse pressure (PP) was calculated as SBP minus DBP and mean arterial pressure (MAP) was calculated using the formula MAP = 1/3 × SBP + 2/3 × DBP. Monitoring data were deemed valid if more than 80% of the scheduled readings were successfully obtained.

Echocardiographic measurement

All transthoracic echocardiography (Philips iE33; Philips Healthcare, Eindhoven, The Netherlands) was performed at rest in the left lateral decubitus position following standard protocols14. Each scan was reviewed by at least two experienced ultrasound experts. Left ventricular (LV) end-diastolic diameter, interventricular septal thickness (IVST), and LV posterior wall thickness (LVPWT) were obtained from parasternal long-axis views using 2D-guided M-mode imaging. LV mass (LVM) was calculated by the Devereux-modified cube formula and indexed to body surface area. End-diastolic LV volume and ejection fraction were derived from biplane Simpson’s method. Aortic diameters were measured at end-diastole from parasternal long-axis 2D images using the leading-edge to leading-edge convention. Measurements were obtained at the sinuses of Valsalva, sinotubular junction, and proximal ascending aorta; the latter was defined as AoR in this study.

Diastolic function was assessed by pulsed-wave Doppler of mitral inflow (E, A waves) at the leaflet tips in the apical four-chamber view. Tissue Doppler imaging (TDI) was performed at the septal mitral annulus to obtain early diastolic velocity (e′), and the E/ e′ ratio was calculated to estimate LV filling pressure.

Definitions and grouping

Diabetes mellitus (DM) was defined as fulfilling any of the following criteria: (1) glycated hemoglobin (HbA1c) ≥ 6.5%; (2) random plasma glucose ≥ 11.1 mmol/L; (3) fasting plasma glucose (FPG) ≥ 7.0 mmol/L; or (4) 2-h plasma glucose ≥ 11.1 mmol/L following an oral glucose tolerance test. Hypertension was defined as either office SBP ≥ 140 mmHg or office DBP ≥ 90 mmHg at rest, or the current use of antihypertensive medication. CKD was defined as an estimated glomerular filtration rate (eGFR) of less than 90 mL/min/1.73 m2, calculated using the CKD-EPI equation. Arterial stiffening was classified as PP ≥ 60 mmHg15. Body surface area (BSA) was calculated with the Mosteller formula. Both LVM and AoR were normalized to BSA. Left ventricular hypertrophy (LVH) was defined as an LVMI ≥ 115 g/m2 for men or ≥ 95 g/m2 for women, or IVST (interventricular septal thickness) or LVPWT (left ventricular posterior wall thickness) > 11 mm for men and > 10.5 mm for women15,16.

Statistical analysis

The normality of continuous variables was assessed using the Shapiro–Wilk test. The Mann–Whitney U test was applied for variables that did not meet normality assumptions. Descriptive statistics are reported as mean ± standard deviation for continuous variables and as frequencies and percentages for categorical variables. Continuous variables were compared between groups using the Student’s t-test or ANOVA, and categorical variables were compared using the chi-square test. Univariate linear regression analysis was first conducted to examine the individual associations between variables. Multivariate linear regression analysis was then used to evaluate the independent effects of each variable while adjusting for potential confounders. Mediation analysis was conducted to explore the mechanisms among the independent variable, dependent variable, and mediator variable. The bootstrap method was used for significance testing with 1,000 resamples, applied through the SPSS PROCESS macro version 4.3. Analysis of covariance (ANCOVA) was employed to compare means across groups, adjusting for relevant covariates.

All statistical tests were two-tailed, with a significance level set at a P-value of 0.05. All analyses were conducted using IBM SPSS version 26 (SPSS Inc, Chicago, IL, USA).

Results

Population characteristics

A total of 375 patients were initially assessed, with 104 excluded based on predefined criteria, resulting in a final sample of 271 patients included in the study (Supplementary Fig. 1).

This study analyzed patients aged 60–100 years (mean age 74.06 ± 9.71 years), predominantly male (83.39%). Hypertension and diabetes prevalences were 77.12 percent and 45.02 percent, respectively. Twenty-four-hour mean SBP averaged 128.27 ± 13.57 mmHg, DBP 69.54 ± 9.09 mmHg, and MAP 89.13 ± 9.05 mmHg. HbA1c was 6.27 ± 1.03 percent, and FBG averaged 5.89 ± 1.44 mmol/L, respectively. Total cholesterol, low-density lipoprotein cholesterol (LDL-C) and triglycerides measured 4.08 ± 0.98, 2.42 ± 0.88 and 1.35 ± 0.82 mmol/L, respectively.

Table 1 summarizes the population characteristics, categorized into two groups based on diabetes status. There was no significant age difference between the two groups (74.47 ± 10.02 vs. 73.54 ± 9.27 years old, P = 0.463), but the group with diabetes had a higher proportion of males (77.18% vs.90.98%, P = 0.002), greater BSA (1.79 ± 0.15 vs. 1.84 ± 0.14 m2, P = 0.002) and marginally higher smoking prevalence (34.23% vs. 36.89%, P > 0.05) . Patients with diabetes had higher FPG (5.15 ± 0.54 vs. 6.80 ± 1.66 mmol/L, P < 0.001), HbA1c (5.73 ± 0.40% vs. 6.94 ± 1.18%, P < 0.001) and triglycerides (1.52 ± 0.95 vs. 1.21 ± 0.68 mmol/L, P < 0.001), but lower high-density lipoprotein cholesterol (HDL-C) (1.34 ± 0.90 vs. 1.13 ± 0.33 mmol/L, P = 0.003) and slightly lower LDL-C (2.52 ± 0.84 vs. 2.31 ± 0.92 mmol/L; P = 0.050).

Table 1 The main characteristics of the study population stratified by diabetes mellitus (DM) status.
Full size table

Absolute aortic diameter did not differ between patients with and without diabetes (36.17 ± 3.83 vs. 35.72 ± 3.56 mm; P > 0.05). After adjustment for BSA, AoRi (20.37 ± 2.67 vs. 19.49 ± 2.16 mm/m2, P = 0.003) was smaller in patients with diabetes. Patients with diabetes also exhibited higher left atrial diameter (LAD), IVST, LVPWT, LVMI and E/e’(P < 0.05). Patients with diabetes exhibited a widened PP (55.85 ± 12.09 vs. 62.37 ± 12.08 mmHg, P < 0.001) despite similar baPWV (1.80 ± 0.42 vs. 1.83 ± 0.42 m/s, P = 0.525), with no significant differences in 24-h mean DBP or 24-h mean MAP between the groups(P > 0.05).

Correlation between aortic diameter and glycemic traits

In both univariate and multivariate analyses, diabetes, FBG, and HbA1c were inversely associated with aortic diameter. Diabetes correlated with smaller aortic diameter (univariate: B = − 0.886, P = 0.004; multivariate: B = − 0.984, P = 0.039). FBG (univariate: B = − 0.286, P = 0.006; multivariate: B = − 0.275, P = 0.017) and HbA1c (univariate: B = − 0.327, P = 0.025; multivariate: B = − 0.479, P = 0.007) showed similar associations (Table 2).

Table 2 Associations between AoRi with glycemia traits.
Full size table

Mediation analysis of aortic diameter between baPWV, PP, and LVMI relationships

In mediation models (Fig. 1), arterial stiffness (baPWV) positively associated with AoRi (path a = 0.942, P = 0.015), AoRi negatively associated with PP (path b = − 0.990, P < 0.001). The direct effect of baPWV on PP was positive (c′ = 3.336, P = 0.024), while the total effect was not statistically significant (c = 2.404, P = 0.109). The indirect effect mediated via AoRi was − 0.933 (95%CI − 2.133, − 0.028).

Fig. 1
Fig. 1
Full size image

The mediation effect models for AoRi between baPWV, PP, LVMI, and diabetes. (A) The mediation effect model for baPWV, AoRi and PP. (B) The mediation effect model for PP, AoRi and LVMI. (C) The mediation effect model for DM, AoRi and LVMI. Age, gender, smoking status, triglycerides, total cholesterol, creatinine, uric acid, 24-h mean MAP, 24-h mean HR, hypertension, use of antidiabetic medication, insulin, and antihypertensive medication were adjusted in the mediation effect model. AoRi aortic diameter normalized to BSA, BaPWV brachial-ankle pulse wave velocity, PP 24-h mean pulse pressure, LVMI left ventricular mass index, DM diabetes mellitus.

PP demonstrated a total effect on LVMI of 0.308 (P = 0.007) and a direct effect of 0.439 (P < 0.001), with an indirect attenuation via AoRi (− 0.131, 95%CI − 0.230, − 0.058).

Mediation analysis of diabetes, aortic diameter, and PP

When diabetes status was the independent variable, the effect on AoRi was negative (path a = − 0.984, P = 0.039), and AoRi negatively influenced PP (path b = − 0.843, P < 0.001). The direct effect of diabetes on PP was 3.903 (P = 0.031); the total effect was 4.732 (P = 0.010), with the indirect effect via aortic diameter index accounting for the difference.

Subgroup and chain mediation analyses

Forest plots (Fig. 2) show stronger inverse correlations between glycemic traits AoRi in elderly and subgroup with hypertension. In patients with hypertension, chain mediation (Fig. 3) demonstrated that diabetes had a significant total effect on LVMI (P = 0.043) but a non-significant direct effect after adjustment (P = 0.138). The specific indirect pathway Diabetes → AoRi → PP → LVMI was significant (0.230, 95% CI 0.013 to 0.592).

Fig. 2
Fig. 2
Full size image

Forest plot of glycemic traits association with AoRi calculated using a linear regression model. (A) Forest plot of diabetes association with AoRi calculated using a linear regression model. (B) Forest plot of FPG association with AoRi calculated using a linear regression model. (C) Forest plot of HbA1c association with AoRi calculated using a linear regression model. Age, gender, smoking status, triglycerides, total cholesterol, creatinine, uric acid, 24-h mean MAP, 24-h mean HR, hypertension, use of antidiabetic medication, insulin, and antihypertensive medication were adjusted in multivariable regression. DM diabetes mellitus, FBG fasting blood glucose, HbA1c glycated hemoglobin, AoRi aortic diameter normalized to BSA, CKD chronic kidney disease, LVH left ventricular hypertrophy.

Fig. 3
Fig. 3
Full size image

The chain mediation effect model for DM, AoRi, PP and LVMI. Age, gender, smoking status, 24-h mean MAP, and 24-h mean HR were adjusted in the mediation effect model. DM diabetes mellitus, AoRi aortic diameter normalized to BSA, PP 24-h mean pulse pressure, LVMI left ventricular mass index.

Comparison of aortic diameter stratified by glycemic traits

Among patients with hypertension, those with diabetes had smaller AoRi than those without diabetes (19.45 ± 2.11 vs. 20.43 ± 2.67 mm/m2, P = 0.004). ANCOVA adjustment increased this difference (19.41 ± 0.30 vs. 20.48 ± 0.30 mm/m2; P < 0.001) (Supplementary Fig. 2). AoRi decreased progressively across HbA1c tertiles (P for trend < 0.001).

Discussion

Our findings demonstrate that diabetes, FPG and HbA1c were significantly negatively associated with aortic diameter even in a general, elderly Chinese population. Through both basic and chain mediation models, we confirmed that restricted aortic dilatation in diabetes increases pulsatile load, which in turn exacerbates LVH. Notably, these mechanisms were most pronounced in participants with hypertension, underscoring the synergistic impact of elevated blood pressure and hyperglycemia on aortic compliance and cardiac structure.

Aortic dilatation plays a dual role, representing a continuum from adaptive to maladaptive remodeling rather than distinct stages. Initially, it compensates for increased arterial stiffness by expanding reservoir capacity during systole, thereby buffering pulsatile load. Over time, however, this dilatation reflects chronic hemodynamic overload. Substantial evidence has linked aging and hypertension to a positive association with increased aortic diameter17. While some studies associate larger aortic diameter with elevated LVMI and incident heart failure, others find smaller diameters predict higher cardiovascular risk17,18,19. Studies focusing on pulsatile hemodynamics suggest that aortic dilatation may compensate for impaired aortic elastic reservoir function by increasing the local storage capacity during systole3,20. We reconcile these findings by proposing a time-dependent framework: early compensatory dilatation can buffer pulsatile stress, but persistent hypertension burden eventually deplete vascular reserve, leading to maladaptive remodeling. This model is supported by observations that individuals with elevated PWV but limited aortic expansion exhibit the widest PP and the highest cardiovascular risk3. Nevertheless, it remains inherently difficult to clearly separate compensatory from pathological dilatation, as these processes overlap continuously without a clear boundary between beneficial and detrimental remodeling.

Some early studies have reported a negative correlation between diabetes and aortic diameter, but they largely focused on patients with thoracic or abdominal aortic aneurysms7. Patients with diabetes not only have a lower prevalence of aortic aneurysms, but those with aneurysms also have a lower postoperative re-rupture rate9. These findings suggest that diabetes may significantly slow the progression of aortic dilatation. In recent years, a growing number of studies have explored the relationship between diabetes and aortic diameter in various populations12,21. Several small-scale studies in populations with end-stage renal disease have identified the same negative correlation between diabetes and aortic diameter22. Furthermore, large studies in the general population have further confirmed this inverse association23,24. Additionally, we found that both short-term blood glucose levels (FPG) and long-term blood glucose levels (HbA1c) were significantly negatively correlated with aortic diameter. In a prediction model for ascending aortic diameter based on a training cohort of 30,018 participants, diabetes was included as a key clinical variable. Although its independent effect was small, it had a significant combined impact when considered alongside age and blood pressure23. This aligns with our finding that the correlation coefficient between diabetes and aortic diameter is greater in the context of hypertension.

The impact of diabetes on aortic diameter, whether protective or detrimental to cardiovascular health, remains an issue of ongoing debate25. Diabetes is closely associated with arterial stiffening, correlating with elevated PWV and PP, and increasing the risk of heart failure with preserved ejection fraction (HFpEF) through its adverse impact on pulsatile hemodynamics26,27. Patients with diabetes in our study also exhibited higher LVMI and impaired diastolic function, both recognized features of HFpEF. While we observed associations suggesting that diabetes may contribute to increased pulsatile load potentially by restricting aortic dilatation, as supported by our mediation models, we refrain from making definitive conclusions regarding the development of HFpEF or the predictive value of aortic diameter due to the absence of longitudinal follow-up data. Precision medicine approaches in HFpEF emphasize targeting underlying pathophysiological mechanisms, particularly in diabetic patients who often exhibit a restrictive HFpEF phenotype characterized by severe diastolic dysfunction, elevated left ventricular filling pressures, and concentric hypertrophy28. PP serves as an independent predictor of HFpEF29, and increased pulsatile load associated with reduced aortic dilatation may play a role in HFpEF pathogenesis, especially in diabetes; however, this hypothesis requires further investigation in prospective studies.

Some viewpoints attribute the negative relationship between diabetes and aortic diameter to the protective effects of anti-diabetic medications30. However, our study and others have found a significant negative correlation between blood glucose traits and aortic diameter independent of anti-diabetic treatment, indicating that anti-diabetic medications can only partially explain this relationship. In our analysis, although anti-diabetic treatment was adjusted as a covariate, we did not perform separate analyses for different classes of anti-diabetic drugs (e.g., Metformin, Thiazolidinediones, Dipeptidyl Peptidase-4 inhibitors, Glucagon-Like Peptide-1 receptor agonists, Sodium-Glucose Cotransporter-2 inhibitors) due to limited sample size and statistical power. Therefore, the potential differential effects of these drug classes on aortic remodeling remain unclear. Future studies with larger cohorts should stratify patients according to specific anti-diabetic therapies to elucidate their distinct impacts on aortic structure and compliance. Additionally, increased arterial wall thickness in individuals with diabetes mellitus may partly contribute to their smaller aortic diameter31, which warrants further investigation.

Some studies pay attention to gender differences in aortic remodeling. A study based on UK biobank data showed females had greater hypertension-related reduction in aortic compliance32. In our study, females also showed a stronger correlation between blood glucose traits and aortic diameter, with a statistically significant interaction. However, due to the smaller proportion of females in our study population, this finding requires further validation in future research.

This study has several limitations. First, its cross-sectional design precludes any inference of causality between diabetes, aortic remodeling, and subsequent cardiovascular events; longitudinal data are required to determine temporal relationships and causative effects. Second, the relatively modest sample size limited statistical power, restricting detailed adjustment for individual classes of antihypertensive and antidiabetic therapies, as well as for potential confounders. Although we included medication use as a covariate, we could not stratify analyses by specific drug classes. Third, because our cohort consisted predominantly of older, male, ethnically Chinese participants, the external validity of our findings is constrained; extrapolation to younger individuals, females, or other ethnic groups should be undertaken with caution. Finally, while we measured fasting plasma glucose and glycated hemoglobin as markers of glycemic control, additional biomarkers—such as advanced glycation end products and inflammatory mediators—may provide deeper insight into the molecular mechanisms of aortic extracellular matrix remodeling. Future prospective studies with larger, more diverse populations and serial measurements are needed to evaluate specific therapeutic regimens’ effects on aortic diameter, PWV, and the development of HFpEF, as well as to elucidate underlying pathophysiological pathways under hyperglycemic and oxidative stress conditions.

Conclusions

Our results highlight a previously underrecognized mechanism by which diabetes contributes to cardiovascular remodeling. In an elderly, predominantly male Chinese population, particularly among individuals with hypertension, diabetes and elevated blood glucose levels were significantly associated with smaller ascending aortic diameters. Impaired aortic dilatation increased pulsatile load, accelerating left ventricular hypertrophy and potentially promoting HFpEF. These findings highlight aortic remodeling as a distinct pathological process in diabetes and suggest that preserving aortic compliance may offer a novel therapeutic target. Future research should elucidate the molecular and biomechanical mechanisms underlying these changes and develop interventions to mitigate cardiovascular risk in patients with diabetes.

What is already known on this topic

Arterial stiffening and impaired aortic compliance are key drivers of cardiovascular morbidity. While diabetes reduces the risk of aortic aneurysm, its role in altering aortic structure and disrupting hemodynamic compensation in elderly patients without aneurysms remains unclear.

What this study adds

This study identifies impaired ascending aortic dilatation as a novel mechanism by which diabetes exacerbates pulsatile load and promotes left ventricular hypertrophy, independent of arterial stiffness severity.

How this study might affect research, practice or policy

These findings suggest that targeting aortic compliance could offer a new therapeutic strategy to prevent myocardial hypertrophy and heart failure with preserved ejection fraction (HFpEF) in elderly patients with diabetes and hypertension.