Vascular ultrasound-based risk stratification model for atherosclerotic cardiovascular disease in patients with type 2 diabetes mellitus

Abstract

Background

This study aimed to investigate the ability of an ultrasound-based risk stratification model integrating carotid intima thickness (CIT) and carotid-femoral pulse wave velocity (cfPWV) to aid in risk stratification and assessment of atherosclerotic cardiovascular disease (ASCVD) in patients with type 2 diabetes mellitus (T2DM), thereby providing an objective basis for identifying high-risk individuals and informing individualized management strategies.

Methods

A total of 105 patients with T2DM were enrolled in this study. According to the 10-year ASCVD risk score, patients were further classified into T2DM patients with low-to-moderate burden of other cardiovascular risk factors and T2DM patients with high burden of other cardiovascular risk factors. CIT was measured using high-resolution ultrasound to assess vascular structure, while cfPWV was evaluated using the automatic measurement of arterial stiffness (AMAS) system to assess vascular function. Logistic regression and least absolute shrinkage and selection operator (LASSO) regression analyses were performed to identify independent risk factors of high ASCVD risk. Based on these risk factors, individual discriminative models and a nomogram were constructed. Receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA) were used to evaluate model performance, and differences among models were assessed using the DeLong test.

Results

CIT, cfPWV, and estimated glomerular filtration rate (eGFR) were identified as independent risk factors of high 10-year ASCVD risk in patients with T2DM. The areas under the curve (AUCs) for the CIT model, cfPWV model, eGFR model, combined CIT–cfPWV model, and the nomogram were approximately 0.781, 0.808, 0.797, 0.831, and 0.875, respectively. The constructed nomogram demonstrated excellent discrimination, calibration, and clinical applicability.

Conclusions

CIT and cfPWV show strong potential for identifying T2DM patients at high ASCVD risk as estimated by the China-PAR model. Incorporating these parameters into vascular evaluation may aid in risk stratification and provide a robust basis for individualized clinical intervention strategies. Prospective studies are needed to validate their prognostic value for future ASCVD events.

Background

Diabetes mellitus represents a metabolic condition marked by chronic elevations in blood glucose levels, arising from inadequate insulin production or diminished responsiveness of peripheral tissues to insulin. In 2021, an estimated 537 million adults worldwide were living with diabetes, and this prevalence is forecasted to reach 783 million by 2045 [1]. The metabolic disturbances associated with diabetes can lead to various multisystem complications, among which atherosclerotic cardiovascular disease (ASCVD) represents the most serious clinical condition. Substantial evidence indicates that individuals with type 2 diabetes mellitus (T2DM) face a markedly higher risk—about 1.5 to 2.5 times for morbidity and roughly double for mortality—compared with those without diabetes [2, 3]. Consequently, ASCVD risk assessment has become a critical component of comprehensive clinical management in patients with T2DM. T2DM patients at different risk levels exhibit significant differences in ASCVD incidence, disease progression, and prognosis, and therefore require differentiated management strategies. T2DM patients with a high burden of additional cardiovascular risk factors generally require intensified lifestyle interventions combined with pharmacological therapy under clinical supervision to control multiple risk factors, whereas those with a lower burden of such factors are primarily managed through lifestyle modification [4]. Therefore, identifying simple, reliable, and robust risk indicators for ASCVD risk assessment and stratification in patients with T2DM is essential for effectively recognizing those with a high burden of additional risk factors and providing a solid basis for individualized clinical intervention strategies. In elderly patients with T2DM, the accumulation of multiple cardiovascular risk factors often coexists with polypharmacy, frailty syndrome, and increased susceptibility to adverse drug effects [5]. In such populations, not all aggressive medical treatments are justified, underscoring the need for precise risk stratification to guide individualized therapy. Therefore, a non-invasive, reproducible approach that accurately identifies those who would truly benefit from intensive intervention is of particular clinical value.

Carotid intima-media thickness (CIMT) was previously considered a preferred indicator to detecte early structural deterioration of the arterial wall and for assessing ASCVD risk. However, according to the latest European Society of Cardiology guidelines, CIMT-based screening for ASCVD risk assessment in asymptomatic individuals is no longer recommended [6]. This recommendation may be attributed to the fact that CIMT represents a composite measurement of both intimal and medial arterial layers. Thickening of the medial layer primarily reflects vascular aging and blood pressure–related changes, whereas atherosclerosis predominantly affects the intimal layer [7]. In contrast, measurement of carotid intima thickness (CIT) alone can more specifically reflect the early pathological changes associated with atherosclerosis. Nevertheless, conventional carotid ultrasound probes (< 20 MHz) used in clinical practice lack sufficient resolution to clearly differentiate the intimal and medial layers, thereby limiting the widespread application of CIT. The newly developed 24-MHz high-frequency ultrasound probe, based on high-resolution imaging technology, offers superior tissue resolution and penetration, enabling clear visualization of the intima and media for precise carotid artery assessment.

Arterial stiffness is a quantitative indicator which represents the elastic properties of the vascular wall and serves as an important objective parameter for assessing early functional abnormalities of the arteries [8]. Clinically, noninvasive assessment systems based on pulse wave velocity have become reliable tools for evaluating arterial stiffness. Among these methods, carotid–femoral pulse wave velocity (cfPWV) is widely acknowledged as the gold standard for evaluating arterial stiffness [9]. Numerous researches have robustly demonstrated that cfPWV serves as an independent risk factor of coronary heart disease [10], and provides independent prognostic value for future ASCVD risk in both healthy individuals and patients with various diseases [11–13]. Recently, a novel doppler ultrasound–based technology, the automatic measurement of arterial stiffness (AMAS) system, has been introduced into clinical practice [14]. This system automatically calculates pulse wave transit time, significantly reducing measurement duration and improving accuracy, thereby providing a methodological foundation for this study.

Unlike traditional risk assessment approaches, CIT and cfPWV enable a more precise evaluation of atherosclerosis by quantitatively assessing structural and functional vascular alterations. The combination of these two parameters allows for a multidimensional assessment of vascular health, more accurately reflecting cardiovascular pathological changes in patients with diabetes and thereby optimizing ASCVD risk evaluation. However, despite their recognized potential in ASCVD risk stratification, most existing studies have focused on evaluating a single parameter, with limited research investigating the combined application of vascular structural and functional indices, particularly in patients with T2DM. Therefore, this study aimed to integrate CIT and cfPWV into an ultrasound-based vascular assessment and to develop a risk stratification model to explore their potential as objective indicators for ASCVD risk stratification and assessment in patients with T2DM, thereby providing objective evidence for the early identification and individualized management of individuals at high ASCVD risk.

Materials and methods

Subjects

This cross-sectional clinical study recruited 105 patients with T2DM diagnosed and managed at our hospital. The inclusion criteria were as follows: (1) age ≥ 20 years; (2) diagnosis of T2DM according to the 1999 WHO diagnostic criteria for diabetes [15], and (3) ability to undergo ultrasound-based CIT and cfPWV measurements. The exclusion criteria included: (1) a history of ASCVD or other major cardiovascular diseases, such as frequent arrhythmias or peripheral arterial occlusive disease; (2) implantation of a cardiac pacemaker; (3) severe organ dysfunction, including advanced hepatic or renal insufficiency or malignant tumors; (4) recent history of surgical procedures; and (5) acute diabetic complications, such as diabetic ketoacidosis or hyperosmolar coma. Ultimately, 65 patients were classified into the T2DM patients with low-to-moderate burden of other cardiovascular risk factors and 40 patients into the T2DM patients with high burden of other cardiovascular risk factors. This observational study was approved by the Ethics Committee of our hospital (Approval No. PJKT2024-087), and the requirement for written informed consent was waived. All procedures were conducted in accordance with the principles of the Declaration of Helsinki.

Clinical and laboratory data

Information was sourced from patient medical records for analysis. Clinical data included sex, age, waist circumference, body mass index (BMI), smoking status, duration of diabetes, use of lipid-lowering and antihypertensive medications, history of hypertension, and family history of ASCVD; information on geographic region and type of residence was also recorded. A family history of ASCVD was defined as a history of myocardial infarction or stroke in first-degree relatives (parents or siblings). Waist circumference was measured at the end of expiration with the abdomen relaxed, using a measuring tape placed horizontally 0.5–1 cm above the umbilicus. Laboratory parameters included fasting plasma glucose (FPG), glycated hemoglobin A1c (HbA1c), total cholesterol (TC), triglycerides (TG), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), serum creatinine (SCR), serum uric acid (SUA), and estimated glomerular filtration rate (eGFR).

Measurement of vascular ultrasound parameters: CIT and cfPWV

CIT was characterized as the vertical distance between the anterior boundary of the lumen–intima junction and the corresponding intima–media boundary on the far arterial wall. Measurements were obtained by an Aplio i900 Diagnostic Ultrasound System equipped with a 24 MHz transducer (Fig. 1). Patients were examined in the supine position with full exposure of the carotid artery region, and image acquisition was synchronized with the electrocardiogram (ECG). CIT was measured in the posterior wall of the carotid artery at 1.5 cm, 2.0 cm, and 2.5 cm proximal to the carotid bifurcation in plaque-free areas. Each measurement was recorded at the peak of the R-wave on the ECG. The final CIT value was calculated as the arithmetic mean of six measurements obtained bilaterally (three sites on each side).

Fig. 1.

Measurement of CIT by high-resolution ultrasound imaging technology. The white arrow indicates the carotid bifurcation

The cfPWV assessment was conducted with a VINNO ultrasound system equipped with the AMAS system (Fig. 2). During cfPWV measurement, participants were placed in the supine position with ECG electrodes attached. Once the participant’s heart rate stabilized, pulsed-wave doppler mode was applied to capture flow velocity waveforms from both the right common carotid artery and the right common femoral artery, each obtained approximately 1–1.5 cm proximal to their respective bifurcations. Dynamic data covering at least 20 consecutive cardiac cycles were collected and stored. The body surface projection points of both arteries were marked, and the straight-line distance between them was measured using a flexible measuring tape, with the average of three measurements recorded. The stored velocity waveforms were subsequently retrieved, and the “Carotid-R Time” and “Femoral-R Time” functions were applied, respectively. The AMAS system automatically calculated the pulse wave transit time relative to the ECG R-wave across 10 cardiac cycles. Afterward, the “Carotid-femoral PWV” function was selected, the previously measured arterial distance was input, and the software computed and displayed the final cfPWV value automatically.

Fig. 2.

Measurement of cfPWV by AMAS system

Reproducibility analysis of CIT measurements

To assess the reproducibility of CIT measurements, 30 randomly selected patients were independently evaluated by two sonographers to determine inter-observer variability. To evaluate intra-observer variability, the same observer repeated the measurements on two separate occasions in a subset of 20 patients. Reproducibility was quantified using the intraclass correlation coefficient (ICC) with corresponding 95% confidence intervals. An ICC greater than 0.75 was considered indicative of good reliability.

Measurement of 10-year ASCVD risk

The 10-year ASCVD risk was computed for every participant via the prediction for ASCVD risk in China (China-PAR) model (available at https://www.cvdrisk.com.cn). This risk assessment tool incorporates the following variables: sex, age, geographic region (northern or southern China), current residence (urban or rural), HDL-C, TC, waist circumference, systolic blood pressure(SBP), diastolic blood pressure (DBP), presence of diabetes, use of antihypertensive medication, family history of ASCVD, and smoking status. Based on the calculated risk scores, the T2DM cohort was stratified into T2DM patients with low-to-moderate burden of other cardiovascular risk factors (< 9.9%) and T2DM patients with high burden of other cardiovascular risk factors (≥ 10.0%) [16]. The resulting risk stratification (low-to-moderate vs. high risk) was used as a surrogate endpoint in this cross-sectional study to evaluate the discriminative performance of vascular ultrasound parameters.

Selection of independent risk factors

First, univariate logistic regression analysis was performed to identify variables potentially associated with a high 10-year ASCVD risk in patients with T2DM (P < 0.1). Using the lambda value corresponding to the minimum mean squared error plus one standard error (λ.1se) as the selection criterion, the optimal penalty parameter (λ) was determined through 10-fold cross-validation. Least absolute shrinkage and selection operator (LASSO) logistic regression analysis was then applied to further select clinical features with non-zero coefficients.

Construction and performance evaluation of discriminative models

Based on vascular ultrasound parameters, logistic regression was used to construct the CIT model, the cfPWV model, and the combined CIT–cfPWV model, respectively, for identifying individuals at high ASCVD risk. Multivariable logistic regression analysis was performed using the collected clinical and laboratory data to identify independent clinical risk factors and to establish a clinical risk stratification model. All models were designed to distinguish patients with type 2 diabetes mellitus (T2DM) who had a higher burden of additional cardiovascular risk factors from those with a lower burden. Model performance was evaluated using receiver operating characteristic (ROC) curve analysis. The DeLong test was applied to compare differences between models, and the area under the curve (AUC) was calculated. In addition, integrated discrimination improvement (IDI) and net reclassification improvement (NRI) were calculated to further assess improvements in discriminative performance among the models. A nomogram was constructed based on vascular ultrasound parameters and independent clinical risk factor. ROC analysis was used to evaluate the discriminative performance of the nomogram, including sensitivity and specificity at different threshold values. Decision curve analysis (DCA) was performed to assess the clinical utility of the models across a range of threshold probabilities, and the area under the DCA curve (AUDC) was calculated for each model.

Statistical analysis

Statistical analyses were performed using R software (version 4.2.0; https://www.R-project.org) and Python (version 3.10.11; https://www.python.org/). Continuous variables with a normal distribution were expressed as mean ± standard deviation (SD) and compared between groups using the independent-samples t-test. Continuous variables with a non-normal distribution were presented as median (interquartile range, IQR) and compared using the Mann–Whitney U test. Categorical variables were compared using the chi-square (χ²) test. ROC curve analysis was used to evaluate the discriminative performance of each model. Sensitivity, and specificity were calculated based on the optimal cutoff value, and calibration curves were applied to assess model agreement. A two-sided P value < 0.05 was considered statistically significant.

Results

Clinical and imaging characteristics

Comparisons of clinical and laboratory characteristics between the T2DM patients with low-to-moderate burden of other cardiovascular risk factors and the T2DM patients with high burden of other cardiovascular risk factors showed statistically significant differences in CIT, cfPWV, age, BMI, waist circumference, SBP, DBP, eGFR, SCR, duration of diabetes, current residence and history of hypertension (all P < 0.05) (Table 1). The inter-observer ICC for CIT measurements was 0.821 (95% CI: 0.660–0.910), and the intra-observer ICC was 0.766 (95% CI: 0.584–0.937), indicating excellent reproducibility. Spearman correlation analysis demonstrated significant positive correlations between CIT (r = 0.755) and cfPWV (r = 0.826) with the 10-year ASCVD risk score in patients with T2DM (both P < 0.001) (Fig. 3).

Table 1.

Characteristics of the study population according to 10-year ASCVD risk

Variables	T2DM patients with low-to-moderate burden of other cardiovascular risk factors (n = 65)	T2DM patients with high burden of other cardiovascular risk factors (n = 40)	Statistic	P
Age (years)	52.00(46.00, 55.00)	59.00 (54.50, 65.00)	-4.651b	< 0.001
Sex			0.269c	0.604
Male	34 (52.31)	23 (57.50)
Female	31 (47.69)	17 (42.50)
BMI (kg/m2)	23.38 (21.30, 24.72)	25.11 (22.99, 27.10)	-2.257b	0.024
Waist (cm)	86.02 ± 10.41	90.88 ± 7.79	t=-2.544a	0.012
SBP (mmHg)	128.08 ± 14.61	146.95 ± 18.14	-5.856a	< 0.001
DBP (mmHg)	78.94 ± 11.32	84.68 ± 11.67	-2.493a	0.014
FPG (mmol/L)	7.75 (5.69, 9.79)	7.36 (5.61, 9.16)	-0.472b	0.637
HbAlc (%)	10.10 (7.50, 11.70)	9.49 (6.95, 10.15)	-1.904b	0.057
HDL-C (mmol/L)	1.16 ± 0.31	1.16 ± 0.29	-0.019a	0.985
LDL-C (mmol/L)	3.52 ± 1.12	3.72 ± 1.16	-0.909a	0.366
TC (mmol/L)	5.46 ± 1.37	5.81 ± 1.25	-1.325a	0.188
TG (mmol/L)	1.62 (1.13, 2.06)	1.75 (1.35, 2.34)	-1.082b	0.279
eGFR (ml/min)	101.72 ± 15.76	84.59 ± 13.87	5.657a	< 0.001
SUA (µmol/L)	316.30 (254.00, 350.60)	351.85 (275.75, 417.98)	-1.881b	0.060
SCR (µmol/L)	63.00 (50.00, 75.00)	76.50 (64.75, 89.25)	-3.307b	< 0.001
diabetes duration (years)	4.00 (1.00, 8.00)	8.50 (2.75, 12.00)	-2.797b	0.005
CIT (mm)	0.29 (0.23, 0.34)	0.41 (0.34, 0.55)	-5.326b	< 0.001
cfPWV (m/s)	8.35 (7.07, 9.83)	11.37 (10.56, 12.34)	-6.249b	< 0.001
China-PAR risk (%)	4.90 (2.40, 6.60)	12.25 (11.08, 14.40)	-8.576b	< 0.001
atherosclerotic plaques			3.675c	0.055
No	49 (75.38)	23 (57.50)
Yes	16 (24.62)	17 (42.50)
Current residence			5.018c	0.025
Rural	39 (60.00)	15 (37.50)
Urban	26 (40.00)	25 (62.50)
Current smoking			0.006c	0.939
No	54 (83.08)	33 (82.50)
Yes	11 (16.92)	7 (17.50)
Lipid-lowering agents			0.463c	0.496
No	43 (66.15)	29 (72.50)
Yes	22 (33.85)	11 (27.50)
Hypertension			9.161c	0.002
No	51 (78.46)	20 (50.00)
Yes	14 (21.54)	20 (50.00)
Family history of ASCVD			0.223c	0.636
No	60 (92.31)	35 (87.50)
Yes	5 (7.69)	5 (12.50)

Notes: BMI, body mass index; cfPWV, carotid-femoral pulse wave velocity; China-PAR, the prediction for ASCVD risk in China; CIT, carotid intima thickness; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; FPG, fasting plasma glucose; HbA1c, Hemoglobin A1c; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; SBP, systolic blood pressure; SCR, serum creatinine; SUA, serum uric acid; TC, total cholesterol; TG, triglycerides; a: t-value; b: Z-value; c: χ²-value

Fig. 3.

Scatter plots showing the correlations between CIT (A) and cfPWV (B) with the 10-year ASCVD risk score

Analysis of factors associated with high 10-year ASCVD risk and feature selection

The clinical variables that showed statistically significant differences were further included in univariate logistic regression analysis. The results indicated that CIT, cfPWV, age, BMI, waist circumference, eGFR, SCR, duration of diabetes, current residence and hypertension were potential risk factor of high 10-year ASCVD risk in patients with T2DM (all P < 0.05) (Table 2).

Table 2.

Univariate logistic regression analysis for high 10-year ASCVD risk

Variables	β	SE	z	OR (95% CI)	P
Age (years)	0.103	0.028	3.628	1.109 (1.053, 1.178)	< 0.001
Sex
Male	0.000			reference
Female	-0.210	0.405	-0.518	0.811 (0.363, 1.789)	0.604
Urban	0.916	0.413	2.217	2.500 (1.124, 5.718)	0.027
BMI	0.133	0.060	2.222	1.142 (1.019, 1.291)	0.026
Waist (cm)	0.055	0.023	2.384	1.057 (1.012, 1.109)	0.017
FPG (mmol/L)	-0.011	0.056	-0.191	0.989 (0.882, 1.104)	0.848
HbAlc (%)	-0.146	0.081	-1.806	0.864 (0.733, 1.008)	0.071
HDL-C (mmol/L)	0.013	0.667	0.019	1.013 (0.267, 3.751)	0.985
LDL-C (mmol/L)	0.163	0.179	0.910	1.177 (0.829, 1.685)	0.363
TC (mmol/L)	0.203	0.155	1.313	1.226 (0.908, 1.677)	0.189
TG (mmol/L)	0.036	0.104	0.343	1.036 (0.833, 1.280)	0.732
eGFR (ml/min)	-0.081	0.019	-4.353	0.922 (0.886, 0.953)	< 0.001
SUA (µmol/L)	0.003	0.002	1.497	1.003 (0.999, 1.007)	0.135
SCR (µmol/L)	0.032	0.011	2.902	1.032 (1.011, 1.056)	0.004
CIT (×10− 2mm)	0.103	0.023	4.523	1.109(1.064, 1.164)	< 0.001
cfPWV (m/s)	0.886	0.171	5.172	2.426 (1.790, 3.529)	< 0.001
diabetes duration (years)	0.122	0.041	2.967	1.130 (1.046, 1.230)	0.003
atherosclerotic plaques
No	0.000			reference
Yes	0.817	0.430	1.898	2.264 (0.976, 5.319)	0.058
Current residence
Rural	0.000			reference
Urban	0.916	0.413	2.217	2.500 (1.124, 5.718)	0.027
Current smoking
No	0.000			reference
Yes	0.040	0.532	0.076	1.041 (0.352, 2.914)	0.939
Lipid-lowering agents
No	0.000			reference
Yes	-0.299	0.441	-0.679	0.741 (0.305, 1.736)	0.497
Hypertension
No	0.000			reference
Yes	1.293	0.437	2.958	3.643 (1.565, 8.753)	0.003
Family history of ASCVD
No	0.000			reference
Yes	0.539	0.667	0.808	1.714 (0.448, 6.567)	0.419

Notes: BMI, body mass index; cfPWV, carotid-femoral pulse wave velocity; CIT, carotid intima thickness; eGFR, estimated glomerular filtration rate; FPG, fasting plasma glucose; HbA1c, Hemoglobin A1c; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; SCR, serum creatinine; SUA, serum uric acid; TC, total cholesterol; TG, triglycerides

Variables with P < 0.1 identified in the univariate logistic regression analysis were further subjected to LASSO regression for feature selection. When the optimal penalty parameter was set at λ.1se = 0.089, three optimal features were retained: CIT, cfPWV, and eGFR (Fig. 4).

Fig. 4.

LASSO regression parameter optimization. Cross-validated mean squared error (A) and coefficient paths of variables (B)

Multivariable binary logistic regression analysis demonstrated that CIT, cfPWV, and eGFR were independently associated with high 10-year ASCVD risk in patients with T2DM (all P < 0.05), indicating that these variables are independent risk factors. Notably, higher values of CIT and cfPWV were positively associated with an increased probability of ASCVD risk, suggesting that both parameters act as risk factors. In contrast, higher eGFR was associated with a lower risk of ASCVD, indicating a protective effect (Table 3).

Table 3.

Multivariable logistic regression analysis for high 10-year ASCVD risk

Variables	β	SE	z	OR (95% CI)	P
CIT (×10− 2mm)	0.057	0.028	2.074	1.059 (1.007,1.124)	0.038
cfPWV (m/s)	0.872	0.233	3.737	2.391 (1.587,4.015)	< 0.001
eGFR(ml/min)	-0.074	0.023	-3.282	0.929 (0.884,0.967)	0.001

Notes: cfPWV, carotid-femoral pulse wave velocity; CIT, carotid intima thickness; eGFR, estimated glomerular filtration rate

Discriminative value of individual independent predictors for high ASCVD risk

To evaluate the ability of individual indicators to discriminate high-risk individuals with T2DM, ROC curve analyses were performed for the CIT model, cfPWV model, combined CIT–cfPWV model, and the clinical eGFR model. The optimal cutoff value for the CIT model was 0.369 mm, yielding an AUC of 0.781 (95% CI: 0.685–0.868), with a sensitivity of 95% and a specificity of 49%. For the cfPWV model, the optimal cutoff value was 10.34 m/s, with an AUC of 0.808 (95% CI: 0.720–0.889), sensitivity of 83%, and specificity of 72%. The optimal cutoff value for eGFR was 74.02 mL/min/1.73 m², resulting in an AUC of 0.797 (95% CI: 0.707–0.878), with sensitivity and specificity of 88% and 68%, respectively. The combined CIT–cfPWV model demonstrated an AUC of 0.831 (95% CI: 0.742–0.905), with a sensitivity of 85% and a specificity of 71%. These findings indicate that eGFR, CIT, and cfPWV all exhibit good discriminative performance for identifying T2DM patients at high 10-year ASCVD risk, with cfPWV outperforming the other two individual risk factors. Furthermore, The combined CIT–cfPWV model achieved a higher AUC than the individual models and demonstrated superior sensitivity and specificity, highlighting its potential clinical applicability (Table 4; Fig. 5A).

Table 4.

Discriminative performance of all models (CIT, cfPWV, eGFR, CIT + cfPWV, and the nomogram)

Variables	AUC (95% CI)	cutoff	sensitivity	specificity
CIT	0.781(0.685, 0.868)	0.369 mm	95%	49%
cfPWV	0.808 (0.720, 0.889)	10.34 m/s	83%	72%
eGFR	0.797 (0.707, 0.878)	74.02	88%	68%
CIT+cfPWV	0.831 (0.742, 0.905)	-	85%	71%
Nomogram	0.875 (0.800, 0.936)	-	80%	83%

Notes: cfPWV, carotid-femoral pulse wave velocity; CIT, carotid intima thickness; eGFR, estimated glomerular filtration rate

Fig. 5.

Performance evaluation of ASCVD risk stratification models. (A) ROC curves of all models, including the CIT model, cfPWV model, eGFR model, combined CIT–cfPWV model, and the nomogram. (B) Calibration curve of the nomogram. (C) DCA of the nomogram

Construction and performance evaluation of the ASCVD risk stratification nomogram

A nomogram was constructed based on the independent risk factorsCIT, cfPWV, and eGFR (Fig. 6). In the nomogram, Points represent the score assigned to each variable, Total Points indicate the sum of all variable-specific scores for an individual patient, and High 10-year ASCVD Risk denotes the model‑estimated probability of high 10-year ASCVD risk. The nomogram demonstrated excellent discriminatory performance for distinguishing between T2DM patients with low-to-moderate burden of other cardiovascular risk factors and those with high burden, with an AUC of 0.875 (95% CI: 0.800–0.936), sensitivity of 80%, specificity of 83%, and a concordance index (C-index) of 0.875, outperforming all individual models (Fig. 5A). For example, a patient’s CIT was 0.31 mm, cfPWV was 13.7 m/s, and eGFR was 87.24 mL/min/1.73 m². The corresponding scores for these variables were 14.1, 84.3, and 61.1, respectively, yielding a total score of 159.5, which corresponded to an estimated probability of approximately 93.8% for high ASCVD risk. To comprehensively assess model performance, the nomogram was systematically compared with the CIT model, cfPWV model, combined CIT–cfPWV model, and the clinical eGFR model. DeLong’s test demonstrated that the AUC of the nomogram was significantly higher than that of the CIT model (Z = − 2.858, P = 0.004), the cfPWV model (Z = − 1.998, P = 0.046), and the combined CIT–cfPWV model (Z = − 2.002, P = 0.045). Further quantitative analyses showed a significant improvement in discriminative performance of the nomogram, with a NRI of 0.188 and an IDI of 0.116 compared with the CIT model; an NRI of 0.083 and IDI of 0.010 compared with the cfPWV model; and an NRI of 0.073 and IDI of 0.059 compared with the combined model. Regarding calibration, the calibration curve demonstrated good agreement between predicted and observed risks, and the Hosmer–Lemeshow goodness-of-fit test (χ² = 0.745, P = 0.863) further confirmed satisfactory model calibration (Fig. 5B). DCA revealed an AUDC of 0.199, exceeding that of all other models. Across a wide range of threshold probabilities (0.03–0.92), the nomogram yielded a higher net benefit than the “treat-all” and “treat-none” strategies, highlighting its potential clinical applicability (Fig. 5C).

Fig. 6.

Nomogram for predicting high 10-year ASCVD risk

Discussion

Over recent decades, the incidence of ASCVD has steadily increased, emerging as the predominant cause of mortality in individuals with T2DM. To enhance longevity and quality of life in this population, the primary prevention of ASCVD emphasizes risk assessment and stratification as essential initial steps in improving overall health [17]. In the present study, CIT and cfPWV were incorporated into ultrasound-based vascular assessment in patients with T2DM to explore their value in ASCVD risk evaluation. The main findings demonstrated that cfPWV exhibited superior discriminatory ability compared with CIT in identifying individuals classified as high 10-year ASCVD risk by the China-PAR model, and that the combined use of CIT, cfPWV, and eGFR yielded the highest discriminative performance. The nomogram constructed based on these three parameters showed excellent discrimination, calibration, and clinical applicability.

The China-PAR model was developed by Chinese investigators using epidemiological data from four large prospective cohorts representative of the Chinese population and represents the first nationally tailored ASCVD risk prediction tool in China [18]. This model estimates the probability of a first ASCVD event within 10 years among individuals free of ASCVD at baseline. Multiple independent cohort studies have demonstrated that the China-PAR model exhibits significantly better calibration performance than Western risk prediction models, with excellent results in both internal and external validations [19, 20]. Based on robust supporting evidence, the China-PAR model has been endorsed by national guidelines as the standard tool for ASCVD risk assessment in the Chinese population [4]. Therefore, in this cross-sectional study, the 10-year ASCVD risk estimated by the China-PAR model was used as a surrogate endpoint.

In the present study, higher eGFR was identified as a protective factor against high 10-year ASCVD risk in patients with T2DM. A potential mechanism underlying this association is that eGFR serves as an important indicator of renal function, and reduced eGFR may activate the renin–angiotensin–aldosterone system (RAAS), which promotes collagen deposition and arterial stiffening, thereby contributing to the development of ASCVD [20]. However, some studies have reported that higher eGFR levels are independently associated with cardiovascular events. A meta-analysis including 46 cohorts from Europe, North America, South America, Asia, and Oceania demonstrated that both excessively high and low eGFR levels were associated with increased risks of cardiovascular and all-cause mortality [21]. Elevated eGFR may be associated with frailty and reduced muscle mass, leading to decreased serum creatinine production and an overestimation of eGFR rather than reflecting truly preserved renal function. Reliance on eGFR alone for cardiovascular risk assessment remains controversial.

Our findings demonstrated that both CIT and cfPWV are strongly linked to ASCVD risk and may serve as independent risk factors for a high 10-year ASCVD risk in patients with T2DM. Similar findings have been reported in previous investigations. A meta-analysis encompassing 19 studies reported that each 1 m/s rise in cfPWV was linked to an approximately 12% increase in the likelihood of developing ASCVD [22]. Similarly, Xu et al. [23] reported that CIT correlates with traditional atherosclerotic risk factors and significantly improves the diagnostic accuracy for ischemic stroke beyond conventional risk models. Recent studies have further supported the use of CIT as a more sensitive and accurate biomarker of atherosclerosis compared with conventional CIMT [24, 25]. Collectively, these findings underscore the potential of CIT and cfPWV to improve ASCVD risk stratification. The observed associations between elevated CIT and cfPWV values and high ASCVD risk in diabetic patients may be explained by underlying pathophysiological mechanisms. The pathological process of atherosclerosis primarily affects the arterial intima, and CIT is regarded as an early morphological marker that appears before CIMT thickening and plaque formation [26]. Because of the superficial location of the carotid artery, CIT measured by high-resolution ultrasound serves as an ideal “window” for assessing systemic atherosclerosis. Meanwhile, cfPWV is widely recognized as the gold standard for assessing arterial stiffness and reflects overall vascular dysfunction [8]. Increased arterial stiffness raises central pulse pressure and myocardial oxygen demand, thereby promoting left ventricular hypertrophy and myocardial ischemia [27]. In patients with T2DM, chronic hyperglycemia further aggravates arterial stiffness through the formation of advanced glycation end products, endothelial dysfunction, and oxidative stress. These pathological processes synergistically accelerate ASCVD progression in patients with T2DM [28]. The CIT measurement method demonstrated good intra- and inter-observer reproducibility (ICC > 0.75) supporting the reliability of this novel technique and its potential for routine clinical application, consistent with previous reports on intima thickness measurements based on high-resolution ultrasound [29]. Collectively, the noninvasiveness, practicality, and reproducibility of ultrasound-based techniques highlight the potential clinical utility of incorporating CIT and cfPWV into routine ASCVD risk assessment in patients with T2DM.The findings of this study provide a feasible approach for applying CIT and cfPWV in the dynamic assessment of ASCVD risk in clinical practice. As noninvasive and reproducible indicators of vascular structure and function, CIT and cfPWV can be used not only for baseline risk stratification but also for dynamic monitoring during follow-up. If progressive thickening of CIT or a sustained increase in cfPWV is observed during follow-up, it should be considered a sign of vascular disease progression, even if traditional risk factors have not reached conventional intervention thresholds, thereby indicating the need to intensify comprehensive management strategies, such as optimizing blood pressure, glycemic, or lipid control [30, 31]. This dynamic risk assessment strategy, based on vascular structural and functional parameters, facilitates a shift in ASCVD management from “static stratification based on traditional risk factors” to “personalized dynamic management based on vascular disease progression,” thereby enabling clinical decision-making that more closely reflects the patient’s actual vascular health status. Particularly in elderly patients with T2DM, not all individuals benefit from intensive treatment due to factors such as multimorbidity, polypharmacy, and frailty syndrome [5]. In this context, dynamic monitoring of CIT and cfPWV may help identify truly high-risk individuals who require intensified intervention, while avoiding overtreatment in low-risk patients, thereby supporting the goals of precision medicine.

The optimal cutoff values identified in this study—10.15 m/s for cfPWV and 0.35 mm for CIT—are consistent with findings reported in previous studies and current expert consensus. For instance, a cfPWV threshold of 10 m/s has been widely recommended for predicting cardiovascular events [29], whereas CIT values of approximately 0.36 mm have been associated with major adverse cardiovascular events and plaque vulnerability [32, 33]. These consistencies support the validity and clinical relevance of our findings. Notably, the discriminative ability of the structural vascular parameter was lower than that of the functional parameter, which may partly result from the inherent limitations of ultrasound-based measurements. cfPWV reflects global arterial stiffness, whereas CIT captures only localized structural changes in the arterial wall. Moreover, CIT values may vary between vascular segments or even among different sites within the same vessel. Therefore, relying on either CIT or cfPWV alone may underestimate the total atherosclerotic burden.

Patients with T2DM face a significantly higher 10-year risk of ASCVD than the general population. However, previous studies have shown that conventional prediction models designed for the general population, such as the Framingham model, tend to underestimate ASCVD risk in individuals with T2DM [34]. This highlights the need to develop tailored risk stratification models specifically for this high-risk population. In 2020, Shi et al. [35] established a stroke risk model based on a cohort of 4,335 patients with T2DM from Shanghai. Their integrated model incorporated 11 clinical and biochemical variables, including age, diabetes duration, blood pressure, eGFR, and LDL-C, and demonstrated strong predictive performance, with AUC values of 0.884 in the training set and 0.909 in the validation set. Subsequently, Sun et al. [36] developed a cardiovascular complication risk model for patients with T2DM using a community-based chronic disease cohort from the Beijing-Tianjin-Hebei region, with both coronary heart disease and stroke defined as composite endpoints. The model predictors included age, smoking, hypertension, dyslipidemia, diabetes duration, family history of ASCVD, FPG, and serum bilirubin. Internal validation demonstrated good discriminative ability, with AUCs of 0.803 in the training set and 0.820 in the test set.

Unlike previous studies, the present study integrated clinical risk factors with CIT and cfPWV to construct a nomogram, which demonstrated excellent discriminatory performance (AUC = 0.875). This improvement is likely attributable to the model’s ability to dynamically assess both vascular structure and function, thereby overcoming the limitations of traditional models that rely solely on static parameters such as age, diabetes duration, and dyslipidemia. By combining structural and functional vascular indicators, our model may more accurately capture the overall vascular health status than approaches based solely on conventional risk factors. Methodologically, our findings are strengthened by the use of advanced, reproducible techniques, including a 24-MHz high-resolution ultrasound probe for precise CIT measurement and the AMAS system for cfPWV assessment. Compared with conventional manual techniques, the AMAS system integrates built-in quality control algorithms—such as the automatic exclusion of cycles affected by heart rate variability—and automated pulse wave transit time detection, substantially enhancing measurement accuracy, reproducibility, and efficiency. Although no standardized protocol currently exists for CIT assessment, the use of a 24-MHz high-resolution ultrasound probe in this study enabled precise evaluation of intima–media thickness. Measurement variability was further minimized by applying fixed reference points and multi-site measurements (1.5 cm, 2.0 cm, and 2.5 cm proximal to the carotid bifurcation along the posterior wall). Collectively, the non-invasiveness, practicality, and reproducibility of ultrasound-based techniques highlight the potential clinical utility of incorporating CIT and cfPWV into routine ASCVD risk assessment among patients with T2DM. Although the measurement of CIT and cfPWV requires specialized ultrasound equipment and trained operators, these techniques offer the advantage of providing direct, non-invasive assessment of vascular structure and function, complementing traditional risk factor-based approaches. Integrating these parameters into routine ultrasound examinations may facilitate a more comprehensive assessment of cardiovascular risk without substantially increasing examination time.

Several limitations of this study should be acknowledged. First, the ASCVD risk estimations were based on cross-sectional data, which limits the ability to establish causal relationships among CIT, cfPWV, and ASCVD risk. Second, the relatively small sample size, combined with the single-center design and absence of external validation, may limit the generalizability and robustness of the proposed nomogram. Third, although multiple clinical variables were incorporated, unmeasured confounders—such as inflammatory biomarkers and lifestyle factors—may also contribute to ASCVD risk. Fourth, the primary outcome of this study was the 10-year ASCVD risk score estimated by the China-PAR model, rather than prospectively documented ASCVD events. Although the China-PAR model has been well-validated in the Chinese population, its use as a surrogate endpoint in a cross-sectional design limits our ability to establish causal relationships or to conclude that CIT and cfPWV can predict future ASCVD events. Future prospective cohort studies with long-term follow-up are needed to validate the prognostic value of these vascular parameters. Despite these limitations, our findings highlight the potential of integrating vascular structural and functional parameters into risk stratification models, providing new insights into individualized ASCVD risk assessment and management in patients with T2DM.

Conclusions

This study provides a novel and practical framework for improving ASCVD risk stratification in patients with T2DM by combining ultrasound-derived indicators of vascular structure (CIT) and function (cfPWV). The proposed model demonstrates good discriminatory performance for identifying individuals at high ASCVD risk as estimated by the China-PAR model, offering a non-invasive, reproducible, and clinically feasible approach. These findings highlight the value of incorporating vascular ultrasound parameters into routine cardiovascular risk assessment to enable more precise and individualized management of patients with T2DM.

Abbreviations

ASCVD

Atherosclerotic cardiovascular disease

T2DM

Type 2 diabetes mellitus

CIT

Carotid intima thickness

cfPWV

Carotid-femoral pulse wave velocity

LASSO

Least absolute shrinkage and selection operator

DCA

Decision curve analysis

CIMT

Carotid intima-media thickness

AMAS

Automatic measurement of arterial stiffness

BMI

Body mass index

FPG

Fasting plasma glucose

HbA1c

Hemoglobin A1c

SBP

Systolic blood pressure

DBP

Diastolic blood pressure

TC

Total cholesterol

TG

Triglycerides

LDL-C

Low-density lipoprotein cholesterol

HDL-C

High-density lipoprotein cholesterol

SCR

Serum creatinine

SUA

Serum uric acid

eGFR

Estimated glomerular filtration rate

ECG

Electrocardiogram

ROC

Receiver operating characteristic

China-PAR

The prediction for ASCVD risk in China

Author contributions

Cuie Chen, Lijuan Liu and Yiyun Xu contributed to the study conception and design. Cuie Chen was responsible for the manuscript drafting. Cuie Chen, Qiuxiao Xu and Hong Xu were responsible for the data collection. Xueling He, Minhao Lin, Yuan Li and Yinhua Li assisted in the data analysis. Cuie Chen and Lijuan Liu revised the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by Zhanjiang Science and Technology Bureau, China [grant number 2025A502038].

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study protocol received approval from the Ethics Committee of Affiliated Hospital of Guangdong Medical University (No. PJKT2024-087), which waived the requirement for written informed consent due to the observational nature of the research. All procedures were conducted in accordance with the principles of the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.