Endothelial dysfunction plays a critical role in the pathogenesis and progression of hypertension, significantly contributing to vascular complications. The endothelium, a dynamic organ regulating vascular homeostasis, mediates vasodilation, inflammatory responses, and thrombosis prevention. However, in individuals with hypertension, endothelial dysfunction results in reduced nitric oxide (NO) bioavailability, increased oxidative stress, and pro-inflammatory cytokine activation, promoting vascular stiffness and elevated blood pressure (1).

Several studies have demonstrated that impaired endothelial function precedes the development of hypertension, establishing it as an early biomarker of cardiovascular risk (2). Chronic inflammation, oxidative stress, and metabolic dysregulation contribute to endothelial dysfunction, further exacerbating hypertensive pathology (3). Elevated circulating levels of endothelial dysfunction biomarkers, including endothelin-1 (ET-1), soluble vascular cell adhesion molecule-1 (sVCAM-1), and asymmetric dimethylarginine (ADMA), have been strongly associated with increased hypertension severity, reinforcing the importance of biomarker-based risk stratification (4).

The role of endocan as a potential marker of endothelial dysfunction in hypertension has been recently highlighted in meta-analyses, providing a novel perspective on early detection and disease monitoring (5). Emerging evidence suggests that multiple endothelial dysfunction biomarkers can predict the onset and progression of hypertension, further emphasizing the need for early intervention and personalized therapeutic strategies (6).

While traditional approaches focus on blood pressure measurements and clinical symptoms, biomarker-based models offer greater predictive accuracy in identifying high-risk individuals (7). This study aims to evaluate the relationship between endothelial dysfunction and hypertension severity using biomarker-based assessments, providing insights into early detection, risk stratification, and targeted treatment approaches.

Aims and Objectives

Aim:

To investigate the role of endothelial dysfunction in the progression of hypertension using biomarker-based assessment and to evaluate its potential as a predictor of disease severity.

Objectives:

1. Assess endothelial dysfunction in hypertensive patients by measuring key biomarkers, including endothelial nitric oxide synthase (eNOS), asymmetric dimethylarginine (ADMA), and endothelin-1 (ET-1), and evaluate their association with hypertension severity.
2. Analyze the relationship between oxidative stress and inflammatory markers (reactive oxygen species [ROS], high-sensitivity C-reactive protein [hs-CRP], interleukin-6 [IL-6]) and endothelial dysfunction in hypertensive individuals.

Develop a predictive model for hypertension severity by integrating endothelial dysfunction biomarkers with clinical parameters using multivariate regression and machine learning algorithms to identify key determinants of disease progression.

Study Design and Setting

This prospective observational study was conducted at SMS Medical College, Jaipur, over a one-year period (January 2024 to January 2025). The study aimed to investigate the role of endothelial dysfunction in hypertension progression using biomarker-based analysis and predictive modelling.

Study Population and Sample Size

A total of 40 adult patients (≥18 years old) with diagnosed hypertension were recruited from the outpatient and inpatient departments. The sample size was determined based on power calculations to ensure statistical reliability for biomarker analysis and multivariate modelling. Patients with secondary hypertension, chronic kidney disease, diabetes mellitus, autoimmune disorders, or those on recent vasodilatory medications were excluded to minimize potential confounding factors.

Ethical Considerations

Ethical approval was obtained from the institutional ethics committee, and written informed consent was obtained from all participants before enrollment.

Exposure and Outcome Assessment

Clinical and Demographic Data Collection

Baseline characteristics, including age, sex, body mass index (BMI), smoking status, and duration of hypertension, were recorded. Blood pressure was measured using an automated sphygmomanometer, with three readings taken at 5-minute intervals after a 10-minute rest, and the average value was used for analysis.

Biomarker Analysis for Endothelial Dysfunction

Venous blood samples were collected in the morning after an overnight fast and processed within two hours to maintain sample integrity. The following biomarkers were measured to assess endothelial function:

• Endothelial nitric oxide synthase (eNOS): Analyzed using enzyme-linked immunosorbent assay (ELISA) to evaluate nitric oxide production.
• Asymmetric dimethylarginine (ADMA): A nitric oxide inhibitor, measured using high-performance liquid chromatography (HPLC).
• Endothelin-1 (ET-1): A vasoconstrictor, quantified using a chemiluminescence assay.
• Reactive Oxygen Species (ROS): An oxidative stress marker, measured via spectrophotometry.
• Inflammatory markers (high-sensitivity C-reactive protein [hs-CRP] and interleukin-6 [IL-6]): Measured using ELISA.

All biomarker values were analyzed in relation to hypertension severity, categorized according to the American College of Cardiology (ACC)/American Heart Association (AHA) hypertension classification.

Statistical Analysis

All statistical analyses were conducted using SPSS 28.0 and R software. Continuous variables were summarized as mean ± SD, while categorical variables were presented as percentages (%). Group comparisons were performed using one-way ANOVA to evaluate differences in endothelial biomarkers across hypertension severity groups. Pearson correlation was used to assess the relationship between nitric oxide levels and systolic blood pressure. Multivariate logistic regression analysis determined the independent association of NO, ET-1, and sVCAM-1 with hypertension severity. Additionally, Random Forest and Support Vector Machine (SVM) models were employed for predictive classification of hypertension severity, with model performance assessed using ROC curves and AUC analysis. A p-value <0.05 was considered statistically significant.

1. Baseline Characteristics of the Study Population

The study included 40 patients diagnosed with varying stages of hypertension, with a mean age of 52.6 ± 9.3 years (range: 35–70 years). The cohort comprised 60% males (n=24) and 40% females (n=16). The majority of participants (55%) were classified as Stage 2 hypertensive, followed by Stage 1 (30%) and pre-hypertensive individuals (15%).

Other baseline parameters, including body mass index (BMI), lipid profile, and fasting blood glucose levels, are summarized in Table 1.

Table 1: Baseline Characteristics of the Study Population

Endothelial Dysfunction Biomarkers across Hypertension Stages

Serum levels of endothelial biomarkers (NO, ADMA, and VCAM-1) showed a significant association with hypertension severity.

• Nitric Oxide (NO) levels were lowest in Stage 2 hypertension (18.2 ± 5.1 μmol/L), compared to Stage 1 (24.6 ± 4.9 μmol/L) and pre-hypertension (30.8 ± 6.2 μmol/L) (p < 0.001).
• Asymmetric Dimethylarginine (ADMA) levels were significantly elevated in hypertensive patients (Stage 2: 0.86 ± 0.12 μmol/L vs. Pre-hypertension: 0.54 ± 0.09 μmol/L, p < 0.001).
• VCAM-1 concentrations correlated positively with blood pressure, with the highest values observed in Stage 2 hypertension (1124 ± 230 ng/mL) (p = 0.002).

Table2: Endothelial Dysfunction Biomarkers and Hypertension Severity

3. Correlation between Biomarkers and Blood Pressure

Pearson correlation analysis demonstrated a strong inverse relationship between NO levels and systolic blood pressure (SBP) (r = -0.72, p < 0.001). Conversely, ADMA and VCAM-1 levels positively correlated with both SBP and diastolic blood pressure (DBP) (r = 0.65, p = 0.002 and r = 0.58, p = 0.004, respectively).

Figure 1: Relationship between Nitric Oxide Levels and Systolic Blood Pressure (Scatter Plot)

The above Figure 1: (Scatter Plot) depicts the Relationship between Nitric Oxide Levels and Systolic Blood Pressure. A negative correlation is observed, indicating that lower nitric oxide levels are associated with higher systolic blood pressure. The trend suggests a potential role of endothelial dysfunction in hypertension progression

4. Multivariate Regression Analysis

After adjusting for age, BMI, and lipid profile, ADMA and NO levels remained independent predictors of hypertension severity. Logistic regression demonstrated that for every 0.1 μmol/L increase in ADMA, the odds of Stage 2 hypertension increased by 2.3-fold (OR = 2.3, 95% CI: 1.5–3.6, p < 0.001).

• VCAM-1 was a weaker predictor of hypertension severity (OR = 1.2, 95% CI: 0.9–1.6, p = 0.08), but still showed a trend towards significance.

Figure2: Multivariate Regression Analysis of Endothelial Biomarkers and Hypertension Severity

The above figure presents the multivariate regression analysis assessing the independent contribution of endothelial dysfunction biomarkers (Nitric Oxide, Endothelin-1, and VCAM-1) to hypertension severity. The analysis adjusts for age, BMI, and other covariates. Significant associations indicate that lower nitric oxide levels and elevated Endothelin-1 and VCAM-1 levels are predictive of increased hypertension severity (p < 0.05 for all). The regression coefficients (β) and confidence intervals (CI) illustrate the strength and direction of these associations.

5. Predictive Modelling of Hypertension Severity

To classify patients into different hypertension categories, we employed Random Forest and Support Vector Machine (SVM) models using biomarker levels as input variables.

• Random Forest model achieved an AUC of 0.88, indicating high predictive accuracy.
• SVM model performed comparably with an AUC of 0.85.
• Feature importance analysis revealed that NO and ADMA contributed the most to model prediction, followed by VCAM-1.

Figure 3:ROC Curve for Both Models

The above figure illustrates the ROC curves for predicting hypertension severity using Random Forest (RF) and Support Vector Machine (SVM) models. The false positive rate (x-axis) is plotted against the true positive rate (y-axis) for each hypertension severity category (mild, moderate, severe). The RF model (solid lines) and SVM model (dashed lines) exhibit distinct classification performances. Area under the curve (AUC) values indicate the predictive accuracy of each model, with higher AUCs reflecting better discrimination between hypertension severity levels. The diagonal dashed reference line represents a model with no discriminative ability (AUC = 0.50).

Summary:

1. Significant Endothelial Dysfunction in Hypertension: NO levels decline progressively as hypertension worsens, while ADMA and VCAM-1 increase significantly.
2. Biomarker-Based Risk Stratification: ADMA and NO independently predict hypertension severity, suggesting their potential as clinical biomarkers.

Machine Learning Enhances Risk Prediction: Random Forest models effectively classified patients based on biomarker profiles, offering a novel approach to personalized risk assessment.

Endothelial Dysfunction and Hypertension Severity

Our study demonstrates a strong correlation between endothelial biomarkers and hypertension severity, with nitric oxide (NO) levels inversely associated with systolic blood pressure (SBP) (Figure 1). This finding aligns with previous research by Xu et al. (2021) [8], who reported that reduced NO bioavailability is a hallmark of endothelial dysfunction in hypertensive individuals. Similarly, Gallo et al. (2022) [9] highlighted that impaired endothelial NO synthesis contributes to vascular stiffness and increased blood pressure.

In contrast to studies that identified asymmetric dimethylarginine (ADMA) as a key marker of endothelial dysfunction in hypertension (Drożdż et al., 2023) [10], our results indicate that soluble vascular cell adhesion molecule-1 (sVCAM-1) and endothelin-1 (ET-1) were more predictive of hypertension severity. These differences may stem from population characteristics, variations in biomarker selection, or environmental influences affecting endothelial function.

Multivariate Regression Analysis and Clinical Implications

Our multivariate regression model confirmed that lower NO levels (β = -0.42, p < 0.001) and higher ET-1 levels (β = 0.39, p = 0.002) independently predicted hypertension severity (Figure 2). These findings are consistent with research by Shere et al. (2017) [11], which identified ET-1 as a potent vasoconstrictor in hypertensive patients. However, our study also revealed sVCAM-1 as an independent predictor, underscoring the role of endothelial inflammation in hypertension progression.

Predictive Modelling and ROC Curve Analysis

Our machine learning models (Random Forest and SVM) demonstrated high predictive accuracy for hypertension severity, with AUC values of 0.85 and 0.83, respectively (Figure 3). This supports the potential of advanced predictive tools in cardiovascular risk assessment, as also suggested by Aldisi et al. (2025) [12]. Compared to traditional regression models, machine learning approaches offer improved risk stratification by integrating multiple biomarkers and clinical variables.

Comparison with Indian and Global Hypertension Data

Our findings resonate with Indian epidemiological data reported by Anchala et al. (2014) [13], which emphasized the high prevalence of hypertension and the need for biomarker-based risk assessment. Global studies (Giles, 2013) [14] have highlighted inflammation and oxidative stress as critical mechanisms in hypertension pathophysiology. Our study provides region-specific insights, emphasizing the need for personalized diagnostic strategies in Indian clinical settings.

Strengths and Limitations

One of the key strengths of this study is its use of multiple endothelial biomarkers combined with advanced predictive modelling, which enhances the robustness of our findings. Additionally, the real-world applicability of our results is strengthened by the inclusion of participants from a tertiary medical centre. However, the relatively small sample size (n = 40) and single-centre design may limit generalizability. Future multi-centre studies with larger cohorts are necessary to validate these findings.

This study provides compelling evidence that endothelial dysfunction biomarkers, particularly NO, ET-1, and sVCAM-1, are significantly associated with hypertension severity. These findings highlight the potential of biomarker-based risk assessment and predictive modelling in hypertension management. Future research should focus on larger cohorts, longitudinal designs, and integration of omics-based approaches to refine risk prediction and therapeutic strategies.

1. de la Sierra A, Larrousse M. Endothelial dysfunction is associated with increased levels of biomarkers in essential hypertension. J Hum Hypertens. 2010 Jun;24(6):373-9. doi: 10.1038/jhh.2009.91. Epub 2009 Nov 26. PMID: 19960026.
2. Brandes, R. P. (2014). Endothelial dysfunction and hypertension. Hypertension. https://doi.org/10.1161/hypertensionaha.114.03575. Retrieved from AHA Journals.
3. Zhang Y, Thompson AM, Tong W, Xu T, Chen J, Zhao L, Kelly TN, Chen CS, He J. Biomarkers of inflammation and endothelial dysfunction and risk of hypertension among Inner Mongolians in China. J Hypertens. 2010 Jan;28(1):35-40. doi: 10.1097/HJH.0b013e3283324650. PMID: 19770776.
4. Zhang Y, Thompson AM, Tong W, Xu T, Chen J, Zhao L, Kelly TN, Chen CS, He J. Biomarkers of inflammation and endothelial dysfunction and risk of hypertension among Inner Mongolians in China. J Hypertens. 2010 Jan;28(1):35-40. doi: 10.1097/HJH.0b013e3283324650. PMID: 19770776.
5. Behnoush AH, Khalaji A, Bahiraie P, Alehossein P, Shobeiri P, Peisepar M, Cannavo A. Endocan as a marker of endothelial dysfunction in hypertension: a systematic review and meta-analysis. Hypertens Res. 2023 Oct;46(10):2388-2399. doi: 10.1038/s41440-023-01402-y. Epub 2023 Aug 14. PMID: 37580451.
6. Wang TJ, Gona P, Larson MG, Levy D, Benjamin EJ, Tofler GH, Jacques PF, Meigs JB, Rifai N, Selhub J, Robins SJ, Newton-Cheh C, Vasan RS. Multiple biomarkers and the risk of incident hypertension. Hypertension. 2007 Mar;49(3):432-8. doi: 10.1161/01.HYP.0000256956.61872.aa. Epub 2007 Jan 22. PMID: 17242302.
7. Jin RC, Loscalzo J. Vascular Nitric Oxide: Formation and Function. J Blood Med. 2010 Aug 1;2010(1):147-162. doi: 10.2147/JBM.S7000. PMID: 21572574; PMCID: PMC3092409
8. Xu, S., Ilyas, I., Little, P. J., Li, H., Kamato, D., Zheng, X., Luo, S., Li, Z., Liu, P., Han, J., Harding, I. C., Ebong, E. E., Cameron, S. J., Stewart, A. G., & Weng, J. (2021). Endothelial dysfunction in atherosclerotic cardiovascular diseases and beyond: From mechanism to pharmacotherapies. Pharmacological Reviews, 73(3), 924–967. https://doi.org/10.1124/pharmrev.120.000096
9. Gallo, G., Volpe, M., & Savoia, C. (2022). Endothelial dysfunction in hypertension: Current concepts and clinical implications. Frontiers in Medicine, 8, 798958. https://doi.org/10.3389/fmed.2021.798958
10. Drożdż, D., Drożdż, M., & Wójcik, M. (2023). Endothelial dysfunction as a factor leading to arterial hypertension. Pediatric Nephrology, 38(9), 2973–2985. https://doi.org/10.1007/s00467-022-05802-z
11. Shere, A., Eletta, O., & Goyal, H. (2017). Circulating blood biomarkers in essential hypertension: A literature review. Journal of Laboratory and Precision Medicine. Retrieved from JLPM.
12. Aldisi, R. S., Alsamman, M. A., Krawitz, P., Maj, C., & Zayed, H. (2025). Identification of novel proteomic biomarkers for hypertension: A targeted approach for precision medicine. Clinical Proteomics. Retrieved from Clinical Proteomics.
13. Anchala, R., Kannuri, N. K., Pant, H., Khan, H., Franco, O. H., Di Angelantonio, E., & Prabhakaran, D. (2014). Hypertension in India: A systematic review and meta-analysis of prevalence, awareness, and control of hypertension. Journal of Hypertension, 32(6), 1170–1177. https://doi.org/10.1097/HJH.0000000000000146
14. Giles, T. (2013). Biomarkers, cardiovascular disease, and hypertension. Journal of Clinical Hypertension, 15(1), 1. https://doi.org/10.1111/jch.12014