Comparative evaluation of serum TGF-β1 and PDGF-BB levels across patients with normal coronary arteries, stable coronary artery disease, and acute coronary syndrome

Introduction

Coronary artery disease (CAD) is one of the most common cardiovascular conditions worldwide and a major cause of morbidity and mortality [1]. It is characterized by atherosclerotic plaque formation in the coronary arteries and presents along a clinical continuum from asymptomatic disease to stable angina and acute coronary syndromes (ACS), including myocardial infarction [2]. These manifestations arise from distinct underlying pathophysiological mechanisms [3].

Atherosclerosis develops through complex pathways involving inflammation, cellular proliferation, apoptosis, and vascular remodeling [2]. Growth factors such as Transforming Growth Factor-β (TGF-β) and Platelet-Derived Growth Factor-BB (PDGF-BB) regulate these processes by influencing inflammation, fibrosis, smooth muscle cell proliferation, and vascular structural changes [4, 5]. Additionally, proteolytic enzymes such as matrix metalloproteinase-8 (MMP-8) actively participate in extracellular matrix degradation and the structural remodeling of the vascular wall, working in tandem with these growth factors to orchestrate plaque progression and destabilization [6]. While both biomarkers have been studied in CAD, comparative data across normal coronary arteries, stable CAD, and ACS remain limited [7, 8]. Understanding these differences is essential, as such markers may aid in elucidating disease mechanisms and distinguishing CAD phenotypes.

This study compared serum TGF-β and PDGF-BB levels among individuals with normal coronary arteries (NCA) and those with stable CAD or ACS, and evaluated their diagnostic performance in identifying stable CAD. The results aim to clarify the potential utility of these growth factors as diagnostic and prognostic biomarkers in CAD.

Methods

Results

The demographic and biochemical characteristics of the symptomatic control (NCA) group and the overall CAD cohort are presented in Table 1. CAD patients were older than those with normal coronaries (61.47 ± 8.6 vs. 58.6 ± 6.8 years; p = 0.032). Positive family history, hypertension, and diabetes mellitus were also more prevalent in the CAD group (p = 0.042, p < 0.001, p < 0.001, respectively), while sex, height, weight, and BMI showed no differences. Neutrophil counts were significantly higher in the CAD group (5.6 ± 1.99 vs. 4.7 ± 1.9; p = 0.004). No significant differences were observed in lymphocyte count, hemoglobin, platelet count, creatinine, lipid profile, or most biochemical markers. Blood urea levels were higher in the CAD group [36 (28–40)] than in the normal group [30 (26–38)] (p = 0.011). Serum TGF-β1 and PDGF-BB levels did not differ significantly (p = 0.299 and p = 0.360).

Table 2 summarizes clinical and biochemical findings across the symptomatic controls NCA group, and the stable CAD, and ACS subgroups. Age, sex, height, weight, and BMI were comparable. While family history did not differ, the prevalence of hypertension and diabetes mellitus varied significantly (p < 0.001 for both). Hypertension was more common in both CAD subgroups versus NCA (p < 0.001 and p = 0.019) and differed between stable CAD and ACS (p < 0.001). Diabetes was more frequent in both CAD groups than in NCA (p < 0.001), with no difference between CAD and ACS (p = 0.572).

Lymphocyte, hemoglobin, and platelet values were similar across groups. Neutrophils differed significantly (p = 0.008), driven by elevated counts in ACS compared with NCA (p = 0.006). Creatinine, total cholesterol, LDL-C, and triglycerides showed no differences. Blood urea and HDL cholesterol varied significantly (p = 0.040 and p = 0.008). Although urea showed no significant pairwise differences, HDL was lower in ACS than in NCA and stable CAD (p = 0.036 and p = 0.033).

TGF-β1 levels were highest in stable CAD [212.9 ng/mL (174.97–252.60)] and lowest in ACS [106.3 ng/mL (102.12–109.25)], with significant group differences (p < 0.001). Post-hoc tests confirmed significant differences among all groups (all p < 0.001). PDGF-BB levels followed a similar pattern, being highest in stable CAD (60.4 ± 10.37 ng/L) and lowest in ACS (39.09 ± 6.07 ng/L) (p < 0.001), with significant pairwise differences (all p < 0.001).

TGF-β1 and PDGF-BB concentrations demonstrated a moderate, positive correlation in all participants (r = 0.597, p < 0.001). The scatter plot in Fig. 2 illustrates this relationship, showing increasing PDGF-BB levels with higher TGF-β1 concentrations.

Multinomial logistic regression showed that TGF-β (p < 0.001, Exp(B) = 1.071; 95% CI: 1.038–1.104) and PDGF-BB (p < 0.001, Exp(B) = 1.108; 95% CI: 1.046–1.174) significantly differentiated stable CAD from NCA, while neither marker distinguished ACS from NCA (p = 0.993 and p = 0.997). These findings indicate that both markers are informative for identifying stable CAD but not ACS.

To identify the independent clinical and biochemical predictors of different coronary phenotypes, separate multivariate logistic regression models were constructed (Table 3). In the logistic regression comparing NCA and stable CAD (Model 1), univariate predictors included hypertension, diabetes mellitus, family history, and both biomarkers. LDL-C was also analyzed in the univariate model but did not emerge as a significant predictor (OR: 0.994, 95%CI: 0.982–1.006, p = 0.323), reflecting the baseline balance of cholesterol levels between the two groups. In the multivariate model incorporating LDL-C as a covariate, hypertension (OR: 88.112; p = 0.005), diabetes (OR: 42.351; p = 0.005), TGF-β (OR: 1.071; p = 0.003), and PDGF-BB (OR: 1.234; p = 0.006) remained independent predictors, with strong model performance (Nagelkerke R² = 0.894). A reduced model incorporating only TGF-β and PDGF-BB still demonstrated robust explanatory capacity (Nagelkerke R² = 0.709). Age, sex, HDL cholesterol, urea, neutrophil count, and family history were not significant in the multivariate analysis. To evaluate the predictors of acute presentation versus chronic stable disease within the coronary artery disease cohort (Model 2: ACS vs. Stable CAD), parameters demonstrating univariate significance including hypertension, HDL, LDL-C, TGF-β, and PDGF-BB were analyzed. Due to the high biological and statistical correlation (multicollinearity) between TGF-β and PDGF-BB when evaluated simultaneously, Model 2 was structured into two separate clinical multivariate sub-models (Model 2 A and Model 2B) to prevent model instability. In Model 2 A (TGF-β focused), after adjusting for hypertension, HDL, and LDL-C, lower serum levels of TGF-β (OR: 0.828, 95% CI: 0.718–0.955, p = 0.010) emerged as a strong independent predictor of acute coronary presentation (ACS) over stable CAD. In Model 2B (PDGF-BB focused), lower serum levels of PDGF-BB (OR: 0.698, 95% CI: 0.592–0.824, p < 0.001) were found to be a powerful independent predictor of ACS over stable CAD. Additionally, within this model, hypertension (OR: 266.409, 95% CI: 6.774–10478.174, p = 0.003), lower HDL levels (OR: 0.829, 95% CI: 0.709–0.969, p = 0.019), and higher LDL-C levels (OR: 1.039, 95% CI: 1.003–1.077, p = 0.034) were also identified as significant independent predictors of acute coronary syndrome. To confirm the reproducibility and stability of these findings, internal validation with 1000-resample bootstrapping was applied, confirming that all key diagnostic estimates remained highly significant and robust.

In the binary logistic regression, both biomarkers were independent predictors of stable CAD (p < 0.001), with PDGF-BB showing a slightly stronger association (Exp(B) = 1.108) than TGF-β (Exp(B) = 1.071), and a high model fit (Nagelkerke R² = 0.709).

ROC analyses demonstrated high diagnostic performance for TGF-β in predicting stable CAD (Fig. 3A), (AUC = 0.847; 95% CI: 0.798–0.950; p < 0.001), with an optimal cutoff of 157.7 ng/dL (sensitivity 85%, specificity 83.3%). PDGF-BB also showed good diagnostic utility (Fig. 3B) (AUC = 0.798; 95% CI: 0.717–0.879; p < 0.001), with a cutoff of 54.43 ng/L (sensitivity 73.3%, specificity 76.7%). To evaluate the synergistic diagnostic performance, a combined biomarker analysis of TGF-β1 and PDGF-BB was performed using predicted probabilities from binary logistic regression. This combined model yielded a joint AUC of 0.924 (95% CI: 0.902–0.960; p < 0.001), with a sensitivity of 90.0% and a specificity of 87.9% at a predicted probability threshold of 0.244, which is represented in Fig. 3C. According to the formal DeLong test, the diagnostic performance of this combined biomarker model was statistically superior to both TGF-β1 alone (p = 0.012) and PDGF-BB alone (p < 0.001). Furthermore, multicollinearity diagnostics showed that the Variance Inflation Factor (VIF) for both TGF-β1 and PDGF-BB was 1.18, confirming that the regression model was not affected by collinearity issues. These findings, as detailed in the accompanying ROC Analysis Table, indicate that the combined use of both biomarkers provides superior diagnostic accuracy for predicting stable CAD compared with using either biomarker alone.

To further evaluate the diagnostic performance of these growth factors in predicting acute plaque destabilization within the coronary artery disease cohort (Model 2: ACS vs. Stable CAD), secondary ROC analyses were performed (Fig. 4). In this clinical scenario, a sharp decline in both biomarkers demonstrated exceptional diagnostic accuracy for discriminating ACS patients from those with stable CAD. Serum TGF-β levels yielded an outstanding diagnostic performance (Fig. 4A) with an AUC of 0.992 (95% CI: 0.976–1.000, p < 0.001) and an optimal cutoff of 128.45 ng/ml (sensitivity 98.3%, specificity 100.0%). Similarly, serum PDGF-BB demonstrated highly robust diagnostic utility (Fig. 4B) with an AUC of 0.958 (95% CI: 0.927–0.989, p < 0.001) and an optimal cutoff of 44.05 ng/L (sensitivity 93.3%, specificity 82.1%). These findings strongly indicate that acute suppression of these repair-associated growth factors serves as a highly precise indicator of acute coronary presentation.

Discussion

Growth factors such as transforming growth factor-beta (TGF-β) and platelet-derived growth factor-BB (PDGF-BB) regulate vascular smooth muscle cell migration and proliferation, supporting extracellular matrix production and plaque stability [8] and matrix-degrading proteolytic enzymes such as MMP-8 [6]. While TGF-β promotes extracellular matrix synthesis to support fibrous cap thickness and plaque stability [14–16], dynamic shifts in these pathways drive distinct clinical phenotypes across the CAD spectrum [17, 18]. Our findings demonstrated highly significant, phenotype-specific differences in these serum biomarkers, particularly showing elevated levels in stable CAD and marked suppression in ACS compared to symptomatic controls, highlighting their distinct roles in chronic versus acute vascular pathologies.

In our cohort, traditional risk factors such as older age, hypertension, diabetes mellitus, and family history were expectedly more prevalent in patients with CAD compared to symptomatic controls with NCA, aligning with well-established cardiovascular epidemiology [19, 20]. Similarly, elevated baseline neutrophil counts in the CAD cohort reflect the background chronic systemic inflammatory state characteristic of atherosclerotic disease [21]. While these classical risk factors underscore the representative nature of our study population, our primary focus remains on the specific pathophysiological roles of TGF-β1 and PDGF-BB in modulating different clinical phenotypes of coronary disease.

Additionally, it is clinically vital to address the subset of patients presenting with typical chest pain but exhibiting normal coronary arteries (NCA) on angiography. In clinical practice, these patients are often diagnosed with Ischemia with Non-Obstructive Coronary Arteries (INOCA). Underlying mechanisms such as microvascular dysfunction (microvascular angina) or dynamic epicardial coronary spasms generate true ischemic symptoms despite the absence of obstructive epicardial stenosis. Our finding that growth factor levels in the NCA group differed significantly from the stable CAD and ACS cohorts further highlights that chronic macroscopic plaque remodeling pathways such as those mediated by TGF-β and PDGF-BB are distinct from the functional microvascular or non-cardiac pathologies driving symptoms in NCA patients.

A notable finding of this study was that although serum TGF-β and PDGF-BB levels were higher in the CAD group than in the NCA group, the differences were not statistically significant. This suggests a potential involvement of these growth factors in CAD pathophysiology, but the lack of significance may reflect limited statistical power, biological variability, or unmeasured confounders. Subgroup analyses also imply differing mechanisms across CAD subtypes. Similar to our observations, Rakoczy et al. reported no significant association between TGF-β levels and atherosclerotic burden or disease severity in early-onset CAD, suggesting that TGF-β may reflect broader metabolic–inflammatory processes rather than directly indicating atherosclerosis progression or CAD severity [22].

Comparisons among the NCA group, and the stable CAD, and ACS subgroups revealed distinct biochemical and clinical differences consistent with their underlying pathophysiology. While the higher prevalence of hypertension and diabetes [20, 23, 24], as well as elevated neutrophil counts in the ACS subgroup [21] expectedly reflect the chronic inflammatory and metabolic burden of CAD, TGF-β findings further emphasized more specific differences between stable and unstable disease: in line with Ahmadi et al., who reported increased latent and active platelet-derived TGF-β1 in stable CAD [25], our study identified the highest TGF-β levels in stable CAD and the lowest in ACS. This dual-directional behavior directly corresponds to the distinct microenvironmental demands of coronary lesions: in stable CAD, elevated systemic TGF-β1 serves as a master regulator of tissue repair by promoting vascular smooth muscle cell (VSMC) migration, extracellular matrix synthesis, and fibrous cap maturation, thereby actively driving plaque stabilization [25]. Conversely, the dramatic drop of TGF-β1 in ACS patients suggests a profound failure or exhaustion of these protective anti-inflammatory and pro-survival pathways, leaving the fibrous cap fragile, depleted of collagen, and highly susceptible to rupture.

Similarly, PDGF-BB levels were highest in the stable CAD group and lowest in the ACS group. As a key mediator of vascular smooth muscle proliferation and vascular remodeling, elevated PDGF-BB in stable CAD aligns with chronic plaque maturation, whereas reduced levels in ACS may reflect suppression of growth factor activity during acute inflammation. Consistent with this, Maaniitty et al. reported significantly lower systemic PDGF-BB levels in ACS than in stable CAD, with levels remaining suppressed for months, suggesting prolonged impairment of vascular repair [26]. Mechanistically, PDGF-BB contributes to fibrotic plaque stabilization and chronic remodeling by stimulating the proliferation of vascular smooth muscle cells (VSMCs) and driving their phenotypic transition from a contractile to a synthetic, matrix-producing state. The pronounced deficiency of PDGF-BB observed during ACS further highlights that the repair capacity of the vascular wall is severely compromised during acute local thrombosis and myocardial ischemia, leading to a temporary collapse of vascular regenerative mechanisms [26]. While the significantly lower HDL cholesterol levels in our ACS group [27] expectedly align with HDL’s well-known anti-inflammatory and anti-atherogenic protective roles, our findings suggest that the clinical transition from stable plaque to rupture is more dynamically reflected by the suppression of active vascular repair mediators like PDGF-BB.

These findings show that TGF-β and PDGF-BB levels vary with both the presence and subtype of CAD. Their markedly lower levels in ACS highlight potential diagnostic and prognostic value and suggest that combined biomarker assessment may aid in distinguishing CAD phenotypes. Multinomial logistic regression identified significant differences only between stable CAD and NCA, indicating greater relevance of these biomarkers for chronic, stable disease. Conversely, the absence of differences between ACS and NCA suggests that the acute inflammatory and thrombotic milieu of ACS may suppress or override pathways involving TGF-β and PDGF-BB.

Overall, the data indicate that these biomarkers predominantly reflect chronic atherosclerotic processes; plaque stabilization, endothelial remodeling, and smooth muscle proliferation, rather than acute pathophysiology. This is consistent with Elshaer et al., who reported that mesenchymal stem cell–derived TGF-β promotes immunomodulation, fibrosis regulation, and vascular repair in ischemic models [28]. Accordingly, while these markers have limited value in ACS, they may aid in identifying stable CAD and provide insight into chronic vascular remodeling and atherosclerotic burden.

Univariate analysis showed that hypertension, diabetes mellitus, and serum TGF-β and PDGF-BB levels were significantly associated with stable CAD. In multivariate logistic regression, both biomarkers remained independent predictors after adjustment for traditional risk factors, indicating added diagnostic utility beyond conventional clinical variables. These results suggest that incorporating TGF-β and PDGF-BB into routine risk assessment may improve the detection of stable CAD. Consistent with this, Aula et al. identified TGF-β1 as a potential biomarker in treatment-related cardiotoxicity [29], supporting the broader relevance of TGF-β–related pathways. Overall, these biomarkers may enhance biologically informed risk stratification when integrated with standard predictors.

Across the cohort, TGF-β and PDGF-BB exhibited a significant positive correlation (R² = 0.297), reflecting their involvement in shared biological pathways. Combined with their elevated levels in the stable CAD group, this finding underscores their joint participation in vascular remodeling, fibrosis, and smooth muscle proliferation. The correlation further suggests that assessing both biomarkers together may offer superior diagnostic value compared with evaluating either alone. Although clinical evidence on their combined use is limited, studies supporting multi-cytokine panels [30, 31], align with this approach, and the present findings reinforce its potential.

Importantly, our validated multivariate regression sub-models (Models 2 A and 2B) further substantiate these biological patterns in the acute clinical setting. Given the high biological and statistical collinearity between these two growth factors, analyzing them in separate, biomarker-focused models allowed us to bypass model instability and evaluate their individual predictive roles accurately. After adjusting for major clinical confounders, including hypertension and lipid parameters, both lower TGF-β (Model 2 A) and lower PDGF-BB (Model 2B) levels retained their highly significant independent predictive value for differentiating ACS from stable phenotypes. Crucially, internal validation via 1000-resample bootstrapping verified that these associations are statistically robust and not artifacts of sample size constraints, reinforcing the clinical relevance and stability of these growth factors in acute plaque destabilization.

In this study, ROC analyses demonstrated that both TGF-β and PDGF-BB individually discriminated stable CAD, although PDGF-BB showed lower diagnostic performance than TGF-β. Importantly, the combined ROC analysis yielded an AUC of 0.924 (95% CI: 0.902–0.960, p < 0.001), with high sensitivity (90.0%) and specificity (87.9%). This indicates that joint evaluation of these biomarkers markedly improves diagnostic accuracy compared with either marker alone. This integrated biomarker approach may be particularly valuable in clinically nonspecific presentations such as stable CAD, supporting earlier detection and improved risk stratification. Overall, the findings suggest that TGF-β and PDGF-BB together constitute a highly effective, non-invasive biomarker panel for identifying stable CAD.

Furthermore, the exceptional diagnostic performance of these growth factors in discriminating acute coronary events within the coronary artery disease cohort (Model 2: ACS vs. Stable CAD) highlights their direct implication in plaque vulnerability. The outstanding AUC values of 0.992 for TGF-β and 0.958 for PDGF-BB demonstrate that a profound drop below our identified cutoff values (128.45 ng/ml and 44.05 ng/L, respectively) serves as a highly sensitive and specific biological signature of acute plaque destabilization (Fig. 4). While their chronic, balanced elevation represents an ongoing vascular healing response in stable phenotypes, their severe acute depletion is strongly coupled with a failure in tissue repair and a high propensity for rupture. Consequently, monitoring the sharp decline in this dual-biomarker panel may hold significant clinical utility not only for detecting stable atherosclerosis but also for identifying patients at imminent risk of acute coronary syndrome.

The potential clinical utility of these findings in real-world practice deserves careful consideration, particularly in navigating the transition from non-invasive evaluation to invasive procedures. While invasive coronary angiography remains the diagnostic gold standard, these biomarkers are not intended to replace it. Rather, they can serve as valuable non-invasive tools for risk stratification and clinical decision-making. First, by reflecting the active molecular pathways of vascular remodeling and fibrotic plaque stabilization, these growth factors could serve as non-invasive “gatekeepers” to refine pre-test probability assessments, potentially reducing the rate of purely diagnostic, negative invasive angiographies in patients with atypical presentations. Second, our findings indicate that a pronounced systemic drop in these growth factors characterizes the transition from stable plaque phenotypes to acute coronary syndromes (ACS). In the clinical setting, this biological signature could assist in the rapid triage of patients presenting with chest pain to emergency departments, facilitating earlier identification of plaque destabilization and optimizing the timing and prioritization of invasive interventions.

In summary, while previous research has investigated the individual roles of TGF-β and PDGF-BB in atherogenesis, the present study offers several key novel insights that address gaps in the existing literature. First, by simultaneously evaluating both growth factors across a comprehensive spectrum of coronary phenotypes (NCA, Stable CAD, and ACS), we have demonstrated a distinct biological transition: a compensatory, balanced upregulation during chronic, stable fibrotic remodeling, followed by a profound, systemic depletion during acute plaque destabilization. Second, rather than examining these markers in isolation, we have shown that their combined evaluation as a dual-biomarker panel yields a synergistic diagnostic accuracy (AUC: 0.924) that far exceeds their individual performance. Finally, by integrating robust multivariate modeling with 1000-resample bootstrapping validation, we have provided a highly stable statistical framework that confirms these diagnostic associations are reproducible and clinically reliable despite sample size constraints. These findings elevate our understanding of growth factor dynamics in coronary artery disease from basic pathophysiological associations to a validated, highly accurate, non-invasive risk stratification tool.

Conclusion

This study demonstrates that serum TGF-β and PDGF-BB levels hold substantial promise as non-invasive biomarkers for differentiating stable CAD from NCA and ACS. These growth factors primarily reflect chronic atherosclerotic remodeling and plaque stability, while their acute, profound depletion serves as a highly sensitive biological signature of plaque vulnerability and acute coronary events. Their combined assessment significantly improves diagnostic accuracy beyond either biomarker alone. Importantly, our multivariate predictive findings remained highly stable and robust after rigorous internal validation using a 1000-resample bootstrapping method, confirming their reliability despite cohort size constraints. Integration of these markers with traditional risk factors may enhance early detection, risk stratification, and management of stable CAD. Larger longitudinal studies are needed to validate their prognostic value and explore their application in personalized cardiovascular care.

Limitations

This study’s limitations include a relatively small sample size, reducing statistical power particularly for ACS biomarker analysis, and a cross-sectional design that prevents causal inference and evaluation of temporal biomarker changes. Potential confounders such as medication use, lifestyle factors, and comorbidities were not fully controlled. Furthermore, although patients with established active rheumatologic and autoimmune diseases were strictly excluded during initial screening to prevent confounding growth factor elevations, the potential influence of subclinical or undiagnosed autoimmune conditions cannot be entirely ruled out. Additionally, active smoking status and serum C-reactive protein (CRP) levels were not systematically collected as part of the primary prospective study design. Single-center recruitment limits generalizability. Additionally, only TGF-β and PDGF-BB were assessed; other cytokines and growth factors relevant to plaque biology were not included. Furthermore, although our study was prospectively designed to evaluate the primary diagnostic roles of these growth factors, detailed angiographic indices such as the exact number of diseased vessels, specific coronary artery distribution, and subsequent percutaneous coronary interventions (PCI) during follow-up were not pre-specified in the original prospective data collection protocol. Consequently, these parameters were not systematically recorded at baseline. Larger, prospective multicenter studies with pre-specified, comprehensive angiographic scoring and laboratory protocols are required to validate these findings, clarify the dynamic roles of these biomarkers, and evaluate their correlation with anatomical lesion complexity and inflammatory markers in coronary artery disease.

Authors’ contributions

V.K., H.E., H.K., B.K. and U.K. contributed to the study conception and design. Data collection and analysis were performed by all authors. The first draft of the manuscript was written by V.K. and H.K., and all authors commented on previous versions of the manuscript. H.K. prepared Figs. 1 and 2. All authors read and approved the final manuscript.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Data availability

The datasets generated and analyzed during the current study are not publicly available due to patient privacy and ethical restrictions but are available from the corresponding author on reasonable request.

Declarations