A Nomogram for Predicting Coronary Artery Calcification Using Achilles Tendon Cross-Sectional Area, Mechanical Parameters, and TCBI

Abstract

Objective

To develop a nomogram for predicting coronary artery calcification (CAC) in atherosclerosis patients using Achilles tendon cross-sectional area (CSA), tendon mechanical parameters, and a combined nutrition score.

Methods

This case-control study included atherosclerosis patients from January 2021 to January 2025. Data collected included Achilles tendon CSA, lipid content, Young’s modulus, and the triglyceride–total cholesterol–body weight index (TCBI). Multivariable logistic regression was used to identify predictors of CAC. A nomogram was constructed using R, and its clinical utility was evaluated.

Results

A total of 279 patients were analyzed and randomly assigned to training (n=195) and validation (n=84) sets. Multivariable analysis identified CSA [OR=1.23 (1.13–1.33)], Young’s modulus [OR=0.17 (0.05–0.58)], and TCBI [OR=2.26 (1.30–3.91)] as independent predictors of CAC (all P<0.05). The prediction model was: logit(P)= 5.22+0.20*CSA-1.76*Young’s modulus+0.81*TCBI. The AUC values were 0.91 (0.87–0.95) for the training set and 0.84 (0.76–0.92) for the validation set. The Hosmer–Lemeshow test yielded P=0.962 and 0.837, and calibration curves showed good concordance. Decision curve analysis indicated net clinical benefit across risk thresholds of 2%–98% in the training set and 10%–98% in the validation set.

Conclusion

The nomogram incorporating Achilles tendon CSA, mechanical parameters, and TCBI effectively predicts CAC in atherosclerosis patients. The underlying mechanisms of abnormal lipid deposition and chronic inflammation provide a theoretical basis for targeted therapeutic strategies.

Background

Coronary atherosclerosis is a progressive cardiovascular disease caused by multiple key risk factors and characterized by atherosclerotic plaque development, luminal narrowing, and myocardial ischemia. Although percutaneous coronary intervention (PCI) techniques have become increasingly sophisticated, PCI still faces challenges in managing complex lesions. Among these, coronary artery calcification (CAC) is frequently present in patients with advanced or complex coronary artery disease, increasing the technical difficulty of intervention, procedural risks, treatment costs, and the incidence of adverse cardiovascular events following PCI^1,2 (Onnis C et al, 2024; Visinoni et al, 2024). The pathogenesis of atherosclerosis involves complex interactions between lipid metabolism, inflammatory responses, and vascular wall remodeling. Recent studies have highlighted the role of environmental factors such as toxin exposure in dyslipidemia. For instance, a cross-sectional study from Hunan Province demonstrated that microcystin exposure was significantly associated with altered blood lipid profiles and an increased risk of dyslipidemia, providing epidemiological evidence linking environmental toxins to lipid metabolism disturbances ³ (Feng et al, 2023). Additionally, novel therapeutic approaches targeting plaque stabilization have been explored. For example, a recent study demonstrated that HPDA/Zn nanoparticles function as a CREB inhibitor and can be used for ultrasound imaging and stabilization of atherosclerotic plaques, offering a potential theranostic strategy for atherosclerosis management ⁴ (Chen et al, 2022). Accumulating evidence indicates that the likelihood of developing CAC increases with age, with the highest prevalence observed among individuals aged 60-80 years. Consequently, most previous studies on CAC have focused on elderly populations ⁵ (Shahraki et al, 2023). However, other studies have reported that younger patients with high Agatston scores have an even greater risk of mortality compared to those aged over 75 years, underscoring the importance of early detection, prevention, and intervention to reduce future adverse cardiovascular events ⁶ (Garg et al, 2024). Multiple risk factors are involved in the pathogenesis of CAC and may serve as potential targets for prevention. Notably, calcified plaque formation begins before modifiable risk factors reach clinical thresholds and may accelerate as early as 40-45 years of age. Therefore, early risk assessment in target populations and proactive control of risk factors to optimal levels are essential to prevent and delay the progression of CAC.

Overweight and obesity are independent predictors of atherosclerosis and ischemic heart disease. Beyond the degree of fat accumulation, the ectopic distribution of lipids in non-typical metabolic sites-such as the liver, kidneys, skeletal muscle, and epicardium-can trigger more complex and pronounced pathophysiological responses ⁷ (Kitahara et al, 2022). Increased Achilles tendon thickness is a distinctive phenotypic feature of familial hypercholesterolemia (FH) and has been widely used for its diagnosis and for assessing disease severity. Recent studies have revealed that Achilles tendon thickness and lipid content are associated with atherosclerosis in patients with FH^8,9 (Ratliff et al, 2024; Tada et al, 2022). A large prospective Danish study involving 12,745 participants with a 33-year follow-up found that the presence of xanthomas increased the risk of myocardial injury by 48%, ischemic heart disease by 39%, severe atherosclerotic disease by 75%, and all-cause mortality by 14%, while cardiovascular disease incidence in FH patients was 3.2 times higher than in non-FH individuals ¹⁰ (Christoffersen et al, 2011). The Achilles tendon, the thickest and strongest tendon in the human body, results from the merging of the gastrocnemius and soleus tendons. Its thickening is most closely associated with abnormal serum cholesterol metabolism^11-13 (Lin et al, 2022; Fujiwara et al, 2022; Isawa et al, 2023). Moreover, metabolic abnormalities related to nutritional status, such as hyperlipidemia and obesity, are well-established classic risk factors for coronary atherosclerosis. The triglycerides-total cholesterol-body weight index (TCBI), proposed by Japanese researchers in recent years, represents a novel nutritional indicator that is simpler, more convenient, and more broadly applicable than traditional indices including the Geriatric Nutritional Risk Index and the Prognostic Nutritional Index. The latest evidence suggests a potential association between TCBI and CAC ¹⁴ (Shao et al, 2023).

Risk prediction models serve as predictive assessment tools that help screen for high-risk factors and identify high-risk populations ¹⁵ (Xiong et al, 2021). These models statistically quantify the relationship between multiple predictive factors and the probability of an event, which can include disease onset, occurrence of complications, treatment outcomes, or disease progression. Predictive factors are the key variables that influence the likelihood of the event. Developing risk prediction models not only aids healthcare professionals in estimating the probability and risk of an event occurring within a given timeframe, but also helps objectively and effectively identify the main factors influencing that event. This provides a foundation for designing personalized and precise intervention strategies. Recent methodological advances have demonstrated the utility of nomogram-based prediction models across various disease contexts, including cardiovascular conditions. For example, a rigorous nomogram model was developed and validated for predicting portal vein tumor thrombosis in hepatocellular carcinoma, highlighting the importance of systematic variable selection and internal validation in model construction ¹⁶ (Liu et al, 2024). These approaches provided a methodological framework for our study. Here, we developed a nomogram model for predicting CAC in patients with atherosclerosis, incorporating tendon cross-sectional area, tendon mechanical parameters, and TCBI. This model aims to better identify high-risk individuals, providing practical significance and research value for the risk assessment and management of CAC in atherosclerotic patients.

Methods

Study Design and Patients

A case-control study was conducted, including patients with atherosclerosis who visited the hospital between January 2021 and January 2025. Patients were evaluated by CCTA or CAG, with CAC defined as CT value >130 HU on CCTA (Agatston method) or mild to severe calcification on CAG. Participants were randomly assigned to a training set and a validation set in a 7:3 ratio using the train_test_split function. All enrolled patients underwent baseline assessments and relevant examinations prior to treatment, which were used as the baseline values for subsequent analysis.

Inclusion Criteria: (1) Patients meeting the diagnostic criteria for coronary atherosclerosis ¹⁷ (Lawton et al, 2021); (2) Patients who underwent coronary CT angiography (CCTA) or coronary angiography (CAG); (3) Age >18 years and ≤85 years.

Exclusion Criteria: (1) Patients with congenital heart disease, rheumatic heart disease, valvular disease, or other cardiac conditions; (2) Patients with diseases significantly affecting nutritional status or tendon metabolism, such as diabetes mellitus (which may independently alter tendon structure via advanced glycation end-products) or familial hypercholesterolemia; (3) Patients with aortic dissection or aortic aneurysm; (4) Patients with diagnosed psychiatric disorders or cognitive impairment; (5) Patients with chronic consumptive diseases, such as chronic atrophic gastritis, chronic colitis, chronic hepatitis, hyperthyroidism, or diabetes; (6) Patients with a history of cardiac surgery or severe chronic kidney disease; (7) Patients who had previously undergone CAG and stent placement for coronary lesions.

Study Tools

Patient Baseline Characteristics

Clinical data for all enrolled patients were collected through the electronic medical record system, including age, sex, smoking history, alcohol consumption, hypertension history, and other relevant information.

Coronary artery imaging records, obtained via CCTA or CAG during hospitalization, were reviewed to classify the number of affected coronary vessels: Single-vessel disease, Two-vessel disease, Multi-vessel disease. Notably, left main coronary artery involvement was categorized as two-vessel disease, regardless of the condition of the left anterior descending or circumflex arteries. If right coronary artery disease was also present concurrently, the patient was classified as having multi-vessel disease.

Achilles Tendon MRI Assessment

Dixon MRI was used to measure the Achilles tendon CSA, thickness, lipid content, and water content. Participants were asked to refrain from strenuous exercise for 48 hours prior to the MRI scan. During imaging, participants lay supine in a 1.5 Tesla MRI scanner (GE Signa Twin-Speed, USA) with the ankle positioned at approximately 20 degrees. Straps were used to gently stabilize the lower leg and foot to prevent motion artifacts, and participants were instructed to remain completely still throughout the scan. Axial and sagittal scans were performed using the 3-point Dixon method. The scan parameters included a slice thickness of 1.2 mm, no interslice gap, a total of 18 slices covering 3.6 cm, and an in-plane spatial resolution of 0.53 mm. Four image types were generated: in-phase, out-of-phase, pure fat, and pure water images. MRI analysis was performed by a trained researcher using Horos MR imaging software (v4.0.1, Horos Project, USA), who was blinded to the group assignments. All MRI analyses were performed by a single trained researcher who had undergone standardized protocol training, and measurements were repeated twice with the average value used for analysis, to minimize intra-observer variability.

Assessment of Tendon Mechanical Properties

Based on data measured from an ultrasound diagnostic device and an isokinetic dynamometer, the tendon (Achilles tendon) Young’s modulus and stiffness parameters were calculated. All ultrasound measurements were performed by the same experienced ultrasonographer, and each parameter was measured three times with the mean value used for analysis. All biomechanical tests were performed on the participant’s dominant leg. Participants sat on an isokinetic dynamometer (Biodex 3, Medical Systems, USA) with the hip flexed at 85°, the knee fully extended (180°), and the ankle positioned neutrally (0°). The foot was tightly fixed to the dynamometer footplate so that the ankle joint coincided with the rotational axis of the dynamometer arm. To reduce body motion during contractions, stabilization straps were secured over the torso, and the arms were either crossed on the chest or rested on the thighs. After positioning, participants first performed 2-3 gradual isometric plantarflexions, progressively increasing the force to complete a slow ramped contraction. This was followed by a 20-second maximal voluntary contraction of the ankle plantarflexors, during which torque was recorded. Achilles tendon length was measured using the ultrasound device; images were frozen and a caliper was used to record tendon length, allowing calculation of tendon length changes (△L) under the measured torque. Each participant underwent three trials. The testing system was calibrated prior to testing, and procedures were strictly followed to familiarize participants with the test and encourage maximal effort. Additionally, the moment arms of forces at the ankle joint were measured. The Achilles tendon moment arm was defined as the perpendicular distance from the center of rotation (center of the talus) to the tendon. While the participant remained stationary on the testing platform, this distance was measured using the ultrasound device. Each participant underwent three measurements, and the average value was used to represent the moment arm.

The Achilles tendon Young’s modulus and stiffness were calculated. The Young’s modulus, also known as Young’s modulus, was calculated using the formula: E=(F·L)/(S·ΔL), where L is the initial tendon length, F is the force applied to the tendon by the triceps surae, S is the tendon cross-sectional area (CSA), and ΔL is the change in tendon length during plantarflexion. Stiffness was calculated using the formula: K=F/ΔL. Each participant’s tendon Young’s modulus and stiffness were measured three times, and the mean of the three trials served as the value for analysis.

Assessment of Laboratory Biochemical Markers

All laboratory biochemical assessments were performed by the hospital’s clinical laboratory. After an overnight fast of at least 8 hours, participants provided fasting venous blood samples. Serum fasting blood glucose (FBG) was measured using the hexokinase method. Apolipoprotein AI (Apo AI), apolipoprotein B (Apo B), and lipoprotein(a) [Lp(a)] were measured by immunoturbidimetry. Low-density lipoprotein cholesterol (LDL-C) and high-density lipoprotein cholesterol (HDL-C) were measured using direct assays. The TCBI was calculated as: TCBI= Serum triglycerides (TG, mg/dL)×Total cholesterol (TC, mg/dL)×Body weight (BW, kg)/1000, where TC is converted as 1 mmol/L = 38.67 mg/dL.

Assessment Criteria for CAC

CCTA is currently the most important non-invasive method for assessing CAC. Calcified lesions appear as high-density white areas on CT images, typically with a CT value >130 Hounsfield units (HU) according to the Agatston method, and this technique has high sensitivity and specificity. CAG is the most commonly used tool for diagnosing CAC. Based on CAG results, the severity of coronary calcification can be classified as: none, mild, moderate or severe. Participants with a coronary CT value >130 HU or mild to severe calcification on CAG were categorized as the CAC group, whereas individuals without calcification comprised the non-CAC group.

Ethical Considerations

(1) Ethical approval for the study was granted by the Ethics Committee of the First Affiliated Hospital of Soochow University (Approval No.: 2025LSNO.795), and the study followed the relevant provisions of the Declaration of Helsinki. (2) Confidentiality was strictly maintained: all issues involving privacy and personal safety were handled in full compliance with confidentiality principles. (3) Beneficence: the study aims to provide scientific evidence for the risk assessment and management of CAC in patients with atherosclerosis.

Statistical Analysis

All data were analyzed using SPSS 27.0 and GraphPad Prism 9.5. Categorical variables were expressed as [n (%)] and compared using the chi-square test. Continuous variables were first assessed for normality using the Shapiro-Wilk test. Variables conforming to a normal distribution were presented as mean ± standard deviation (SD), and differences between two groups were compared using the independent-samples t-test. LASSO regression with 10-fold cross-validation was performed for variable selection prior to logistic regression. Binary logistic regression was applied to assess the associated factors. A nomogram model was developed with the “ms” package in R (via SPSS R integration). The model’s discriminative ability was examined via the receiver operating characteristic (ROC) curve. The events per variable (EPV) was calculated to assess model overfitting. Additionally, 10-fold cross-validation was performed to evaluate model stability. The calibration was determined through the Hosmer-Lemeshow (H-L) goodness-of-fit test and a calibration curve generated from 1,000 bootstrap resamples. The clinical utility of the model was assessed via decision curve analysis. A two-sided significance level of α = 0.05 was applied.

Results

Demographics of Patients With Atherosclerosis

A total of 279 patients meeting the inclusion and exclusion criteria were retrospectively analyzed and randomly divided into a training set (n = 195) and a validation set (n = 84) at a 7:3 ratio. Modeling was performed using the training set, which included 102 patients in the CAC group and 93 patients in the non-CAC group. As shown in Table 1, significant differences were observed between the two groups in CSA (p < 0.001), Young’s modulus (p < 0.001), LDL-C (p = 0.017), and TCBI (p < 0.001).

Table 1.

Baseline Characteristics and Comparative Analysis

Variables	Total (n = 279)	Test (n = 84)	Train (n = 195)	P	CAC (n = 102)	N-CAC (n = 93)	P
Age, years	56.62±13.55	58.79±13.74	55.68±13.39	0.079	54.33±14.10	57.16±12.47	0.141
CSA, mm2	80.03±13.03	80.96±10.50	79.63±13.99	0.381	88.29 ± 9.84	70.13±11.48	<0.001
Achilles tendon thickness, mm	6.64 ± 1.36	6.63 ± 1.24	6.65 ± 1.41	0.925	6.83 ± 1.55	6.45 ± 1.23	0.059
Lipid, AU	3.30 ± 0.63	3.38 ± 0.76	3.27 ± 0.57	0.229	3.28 ± 0.58	3.26 ± 0.55	0.841
Water content, AU	4.87 ± 0.99	4.85 ± 0.99	4.87 ± 0.99	0.885	4.93 ± 0.86	4.81 ± 1.11	0.421
Young’s modulus, MPa	752.51±75.30	748.98±80.55	754.03±73.08	0.609	705.65±48.06	807.10±57.30	<0.001
Tendon stiffness, N/mm	362.51 ± 60.78	358.48 ± 65.87	364.25 ± 58.55	0.468	356.53 ± 55.78	372.72 ± 60.62	0.054
FBG, mmol/L	6.32 ± 1.42	6.50 ± 1.28	6.24 ± 1.47	0.159	6.21 ± 1.47	6.27 ± 1.47	0.798
Apo AI, mmol/L	1.33 ± 0.29	1.33 ± 0.29	1.33 ± 0.29	0.950	1.34 ± 0.31	1.32 ± 0.27	0.633
ApoB, mmol/L	1.41 ± 0.34	1.42 ± 0.30	1.40 ± 0.35	0.765	1.38 ± 0.34	1.43 ± 0.37	0.372
Lp(a), mg/L	438.07± 125.93	432.54± 116.36	440.4± 130.05	0.631	456.34± 128.50	423.03 ± 130.19	0.074
LDL-C, mmol/L	5.18 ± 1.31	5.20 ± 1.44	5.17 ± 1.25	0.847	5.37 ± 1.24	4.94 ± 1.23	0.017
HDL-C, mmol/L	1.90 ± 0.43	1.83 ± 0.44	1.92 ± 0.42	0.102	1.89 ± 0.44	1.96 ± 0.39	0.224
TCBI	1866.69± 146.19	1870.08± 143.63	1865.22± 147.62	0.800	1907.79± 157.84	1818.54± 119.97	<0.001
Gender, n (%)	​	​	​	0.104	​	​	0.128
Female	103 (36.92)	25 (29.76)	78 (40.00)	​	46 (45.10)	32 (34.41)	​
Male	176 (63.08)	59 (70.24)	117 (60.00)	​	56 (54.90)	61 (65.59)	​
Smoking, n (%)	​	​	0.282	​	​	0.611	​
No	225 (80.65)	71 (84.52)	154 (78.97)	​	82 (80.39)	72 (77.42)	​
Yes	54 (19.35)	13 (15.48)	41 (21.03)	​	20 (19.61)	21 (22.58)	​
Alcoholism, n (%)	​	​	​	0.082	​	​	0.606
No	225 (80.65)	73 (86.90)	152 (77.95)	​	81 (79.41)	71 (76.34)	​
Yes	54 (19.35)	11 (13.10)	43 (22.05)	​	21 (20.59)	22 (23.66)	​
Hypertension, n (%)	​	​	0.343	​	​	0.937	​
No	197 (70.61)	56 (66.67)	141 (72.31)	​	74 (72.55)	67 (72.04)	​
Yes	82 (29.39)	28 (33.33)	54 (27.69)	​	28 (27.45)	26 (27.96)	​
Coronary vascular number, n (%)	​	​	​	0.551	​	​	0.756
1	63 (22.58)	22 (26.19)	41 (21.03)	​	20 (19.61)	21 (22.58)	​
2	191 (68.46)	56 (66.67)	135 (69.23)	​	73 (71.57)	62 (66.67)	​
≥3	25 (8.96)	6 (7.14)	19 (9.74)	​	9 (8.82)	10 (10.75)	​

Results of Correlation and Multivariate Analyses

Correlation analysis of all study variables was performed using Pearson’s correlation coefficient, and the results are shown in Figure 1. No strong correlations were observed among the continuous variables. Variables that showed statistically significant differences in Table 1 were incorporated into the model as independent variables, using CAC as the outcome (N-CAC=0, CAC=1). Multivariate analysis was then performed using a backward stepwise logistic regression approach (Wald method). As shown in Table 2, CSA [OR (95% CI) = 1.23 (1.13–1.33)], Young’s modulus [OR (95% CI) = 0.17 (0.05-0.58)], and TCBI [OR (95% CI) = 2.26 (1.30–3.91)] were independently associated with CAC in patients with atherosclerosis (P < 0.05).

Figure 1.

Correlations between continuous variables. Pearson correlation coefficients among carotid plaque characteristics, lipid profiles, FBG, and TCBI

Table 2.

Results of Multivariate Logistic Regression Analysis

Parameter	Multivariable
	β	S. E	Wald χ2	P	OR (95% CI)
CSA	0.20	0.04	25.00	<0.001	1.23 (1.13 ∼ 1.33)
Young’s modulus	-1.76	0.62	8.05	0.005	0.17 (0.05 ∼ 0.58)
TCBI	0.81	0.28	8.36	0.004	2.26 (1.30 ∼ 3.91)

Construction of the Predictive Model

To exclude the confounding effects of traditional risk factors, this study included age, sex, smoking, hypertension, and other variables together with other candidate indicators in the LASSO logistic regression. The optimal λ value was determined through 10-fold cross-validation. The results showed that under the lambda.1se criterion, only three variables were retained: CSA, Young’s modulus, and TCBI, while traditional risk factors such as age, sex, hypertension, and smoking were automatically excluded by the LASSO regression (coefficients compressed to zero), indicating that these tendon mechanical parameters and nutritional metabolic indicators have predictive value for atherosclerosis independent of traditional cardiovascular risk factors (Figure 2). Variance inflation factor (VIF) analysis indicated that the continuous variables CSA, tendon Young’s modulus, and TCBI had VIFs of 1.225, 1.260, and 1.048, respectively, all close to 1, suggesting no multicollinearity. The predictive model was established using the following formula: logit(P)=5.22+0.20*CSA-1.76*Young’s modulus+0.81*TCBI, A nomogram was generated via the “ms” package in R from the predictive model. As shown in Figure 3, the total nomogram score ranges from 0 to 220 points, corresponding to a predicted probability of CAC in patients with atherosclerosis between 0.1 and 0.9. With increasing CSA and TCBI and decreasing Young’s modulus, the model score increases, leading to a higher total score and an increased probability of CAC.

Figure 2.

Figure 1 LASSO regression for variable selection. (A) Coefficient paths of 19 candidate variables versus log (λ). Vertical dashed lines indicate the optimal λ values selected by 10-fold cross-validation: lambda.1se (λ = 0.1946). (B) Cross-validation curve for binomial deviance versus log(λ)

Figure 3.

Nomogram for predicting CAC in patients with atherosclerosis. Locate the patient’s value on each variable axis to obtain the corresponding Points, sum to obtain Total Points, and project downward to determine the predicted probability of CAC

Validation of the Predictive Model

As shown in Figure 4A and B, the area under the ROC curve (AUC) was 0.91 (0.87–0.95) for the training set and 0.84 (0.76-0.92) for the validation set, indicating that the predictive model for CAC in patients with atherosclerosis demonstrated excellent discriminative ability. Validation of the predictive model” subsection: Added “The EPV was 12.75 (102 CAC events/8 predictor variables), exceeding the recommended minimum of 10, indicating adequate model stability. The calibration of the nomogram model was assessed using the Hosmer-Lemeshow (H-L) test in combination with calibration curves. The H–L test results showed P = 0.962 and 0.837. In the calibration plots, the “Ideal” line represents perfect prediction, the “Bias-corrected” line represents the calibration curve obtained via bootstrap resampling, and the “Apparent” line represents the original nomogram model curve. The closer the calibration curve aligns with the ideal line, the better the performance of the model. Figure 5A and B show that the calibration curves for both the training and validation sets closely matched the ideal line. The mean AUC from 10-fold cross-validation was 0.89 (95% CI: 0.84–0.94), consistent with the training set AUC, further supporting model robustness. Decision curve analysis (Figure 6A and B) was employed to assess the model’s clinical utility. In the plots, the “None” line represents a strategy in which no patients are managed using the risk model, yielding a net benefit of 0; the “All” line represents a strategy in which all patients are managed using the model, resulting in negative net benefit; and the “Nomogram model” line represents the net benefit achieved using the predictive model developed herein. As shown in Figure 5, when the risk thresholds ranged from 2%-98% in the training set and 10%-98% in the validation set, the “Nomogram model” line remained above both the “None” and “All” lines. Based on the decision curve analysis, we propose a 20% risk threshold as a clinically actionable cut-off for recommending further CCTA evaluation. At this threshold, the nomogram model demonstrated a net benefit of 0.12 in the training set and 0.09 in the validation set. The results indicated that the model provides a significant additional clinical net benefit and demonstrating its practical clinical applicability.

Figure 4.

ROC curves of the CAC predictive model. (A) Training set; (B) Validation set. The model achieved an AUC of 0.91 (95% CI: 0.87–0.95) in the training set and 0.84 (95% CI: 0.76–0.92) in the validation set

Figure 5.

Calibration curves of the CAC predictive model. (A) Training set; (B) Validation set. The dashed line represents ideal calibration. Hosmer-Lemeshow P values were 0.962 and 0.837 for the training and validation sets, respectively

Figure 6.

Decision curves of the CAC predictive model. (A) Training set; (B) Validation set. The Model (blue curve) indicates the net benefit of using the predictive model across threshold probabilities, compared with treating all patients (red) or no patients (green)

Discussion

Coronary artery calcification is one of the key pathological changes in coronary atherosclerosis. Abnormal deposition of calcium salts and inflammation induced by lipid oxidation are contributing factors to CAC formation. Moreover, CAC serves as a marker of atherosclerotic plaque burden in the coronary arteries ¹⁸ (Lu et al, 2024). Previous studies have identified advanced age, sex, hypertension, hyperglycemia, dyslipidemia, chronic kidney disease, and smoking as risk factors for CAC ¹⁹ (Aldana-Bitar et al, 2023). CCTA can non-invasively diagnose CAC, but calcification often interferes with accurate stenosis assessment; CAG remains the gold standard yet has limited sensitivity for non-luminal calcifications ²⁰ (Sandhu et al, 2023). Given these diagnostic challenges, we developed a non-invasive nomogram for estimating CAC risk—the first to our knowledge that combines Achilles tendon-specific parameters with a novel nutritional index.

Lipid management is a crucial aspect of care for patients with atherosclerosis. CSA, representing the Achilles tendon CSA, has been reported to correlate with body weight rather than height. Various rehabilitation activities or physical training can induce biomechanical adaptations in the Achilles tendon, including increased CSA, tendon stiffness, and Young’s modulus ²¹ (Corrigan et al, 2022). It is generally believed that the force-transmission mechanisms induced by cyclic strain affect connective tissue homeostasis, thereby regulating adaptive processes and influencing the tendon’s response to mechanical stimuli ²² (Cone et al, 2023). In our study, CSA emerged as an independent predictor of CAC (OR = 1.23). This association likely reflects systemic lipid accumulation: as the Achilles tendon represents a primary site for tendon xanthomas, its cross-sectional enlargement may serve as a “window” into long-term cholesterol burden. Notably, we observed that tendon lipid and water content did not differ significantly between CAC and non-CAC groups, possibly due to widespread statin use in our cohort, which effectively controls lipid levels and reduces new lipid deposition within tendons ²³ (Squier et al, 2025).

Young’s modulus was inversely associated with CAC, lower tendon elasticity correlated with higher CAC risk. We propose the following pathway: chronic dyslipidemia promotes lipid deposition not only in vascular walls but also in connective tissues such as the Achilles tendon. Accumulated lipids and their oxidative products induce local inflammation, disrupt collagen cross-linking, and reduce tendon elasticity (lower Young’s modulus). Simultaneously, the same inflammatory and oxidative stress milieu drives vascular smooth muscle cell transdifferentiation into osteoblast-like cells, promoting calcium salt deposition in coronary plaques. Thus, reduced tendon elasticity and vascular calcification may represent parallel manifestations of a systemic connective tissue remodeling process driven by chronic lipid overload and inflammation. Vascular stiffening, a hallmark of atherosclerosis, may further exacerbate this process through shared mechanobiological signaling pathways that alter extracellular matrix homeostasis in both tendons and vessels ²⁴ (Nakamura et al, 2022).

TCBI, the third component of our nomogram, is a novel nutritional index based on three easily obtainable clinical parameters: TG, TC, and BW. It has garnered attention due to its simplicity of calculation ²⁵ (Sudo et al, 2024). In this study, high TCBI emerged as an independent predictor of CAC in patients with atherosclerosis. This may be explained by the fact that TCBI integrates lipid levels and body weight, both established contributors to coronary disease, while also reflecting overall nutritional status associated with systemic inflammation, insulin resistance, and lipid metabolism. Obesity induces insulin resistance and oxidative stress; excessive nutrient accumulation in adipose tissue promotes inflammatory mediator secretion, accelerating endothelial dysfunction, which is closely linked to plaque formation. Chronic inflammation within plaques drives calcification via macrophage-derived cytokines that promote vascular smooth muscle cell transdifferentiation into osteoblast-like cells, which then secrete calcium salts that deposit within plaques. The pro-inflammatory and dyslipidemic state reflected by high TCBI thus collectively promotes CAC occurrence and progression. From a clinical standpoint, TCBI offers a practical, cost-effective metric that can be calculated from routine blood tests and body weight measurements without additional cost or radiation exposure, making it particularly suitable for primary care and resource-limited settings.

The clinical utility of our nomogram extends beyond risk prediction to actionable patient management. In our decision curve analysis, the nomogram provided net benefit across risk thresholds of 2%–98% in the training set and 10%–98% in the validation set, supporting its flexibility for different clinical scenarios. For instance, at a 20% risk threshold (which we propose for further CCTA evaluation), the model’s net benefit was 0.12 in the training set and 0.09 in the validation set. Practically, this means that for every 100 patients evaluated using our nomogram, approximately 10–12 true CAC cases could be identified earlier than with conventional risk assessment alone. Moreover, the model’s non-invasive nature—requiring only ultrasound, MRI, and routine blood tests—makes it suitable for serial assessments, potentially enabling dynamic monitoring of CAC risk over time. Peng et al developed a nomogram incorporating serum NLRP1 levels and traditional risk factors in 626 CAD patients, achieving AUCs of 0.881 in the training set and 0.825 in the validation set ²⁶ (Peng J et al, 2024). The CAC-prob model, using four routine clinical predictors (age, sex, hypertension/diabetes, and low HDL-C), demonstrated an ordinal C-index of 0.78 in an external validation study ²⁷ (Wongyikul P et al, 2025). Our nomogram, integrating Achilles tendon biomechanical parameters with the novel nutritional index TCBI, achieved an AUC of 0.91 in the training set and 0.84 in the validation set—comparable to or exceeding these existing models. The findings suggest that the predictive model developed herein exhibits good discrimination and strong clinical applicability for assessing CAC risk in patients with atherosclerosis. Notably, our model offers the advantage of incorporating tissue-specific biomarkers reflecting systemic lipid metabolism and connective tissue health, providing pathophysiological insights beyond conventional risk factor-based approaches.

The novelty of this study is threefold. First, while Achilles tendon xanthomas are recognized as a hallmark of familial hypercholesterolemia, their utility beyond FH diagnosis has been underexplored. We demonstrate that CSA of the Achilles tendon independently predicts CAC in a general atherosclerosis population, suggesting that tendon structural changes reflect systemic lipid burden even in non-FH patients. Second, we introduce Young’s modulus—a measure of tendon elasticity—as a novel predictor of CAC. This extends the concept of “vascular–tendon coupling,” wherein vascular stiffening due to atherosclerosis may parallel degenerative changes in connective tissues through shared mechanobiological pathways. Third, we incorporate TCBI, a simple composite index of triglycerides, total cholesterol, and body weight, as a readily obtainable surrogate for nutritional and metabolic status. By integrating these three domains—tendon structure, tendon mechanical property, and systemic metabolism—our nomogram offers a multifaceted risk assessment tool that captures both local tissue pathology and global metabolic dysregulation. However, the study has certain limitations. First, the single-center, retrospective case-control design of this study with a relatively modest sample size precludes causal inferences; our findings should be interpreted as identifying associations rather than causative relationships. Internal validation was performed using a random split-sample approach with 70% training and 30% validation. While this approach is commonly used, it reduces statistical power compared to full-sample analysis with bootstrapping. Therefore, the results require confirmation through external validation in independent, preferably multicenter, cohorts. Larger, multicenter studies are required to confirm and refine the predictive model. Second, body weight and lipid parameters are influenced by numerous factors. Although as many confounding factors as possible were controlled for, some unaccounted influences may still exist. Third, the exclusion of patients with diabetes mellitus, while methodologically necessary to minimize confounding, may limit the generalizability of our findings to the broader atherosclerosis population, given the high prevalence of diabetes in this patient group and its known association with both tendon pathology and CAC. Future multicenter studies should specifically evaluate the performance of this model in diabetic subpopulations. Fourth, while we attempted to minimize variability through standardized protocols and repeated measurements, we did not formally assess inter-observer reliability or intra-observer variability for MRI and ultrasound measurements. Future studies should include formal reproducibility analyses to strengthen the reliability of these operator-dependent techniques. Finally, different exercise intensities can have varying effects on the Achilles tendon and tendon parameters. This study did not categorize participants based on exercise intensity, and future research could explore how tendon and Achilles tendon parameters change under different exercise regimens and their impact on atherosclerosis.

Conclusion

In summary, this study demonstrates that a non-invasive predictive model based on Achilles tendon lipid content, tendon mechanical parameters, and a novel nutritional index can effectively assess the risk of CAC in patients with atherosclerosis. This model equips clinicians with a precise assessment tool and introduces new measures for atherosclerosis prevention and care.

ORCID iD

Juan Chen https://orcid.org/0009-0004-5917-0773

Ethical Considerations

This study was reviewed and approved by the Ethics Committee of The First Affiliated Hospital of Soochow University.

Consent to Participate

Written informed consent was obtained from all participants, and all procedures followed the principles of the Declaration of Helsinki.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.*