Cardiovascular disease is the leading cause of death worldwide. As the unstable and progressive stage of coronary heart disease, acute coronary syndrome (ACS) is characterized by three serious and potentially fatal clinical manifestations: ST-segment elevation myocardial infarction (STEMI), non-STEMI, and unstable angina pectoris (1, 2).

In recent years, consensus has emerged that coronary atherosclerotic plaques with a propensity for thrombosis and a higher likelihood of rapid progression are commonly known as vulnerable or high-risk plaques (HRPs) (3, 4). The CLIMA study [Relationship between Optical Coherence Tomography (OCT) Coronary Plaque Morphology and Clinical Outcome; NCT02883088] introduced the concept of HRP and found that HRPs are associated with a higher risk of major coronary events (5). OCT, a high-resolution intravascular imaging technique, allows for precise identification of coronary plaque characteristics. HRPs are defined by four simultaneous characteristics: minimum lumen area (MLA) <3.5 mm², fibrous cap thickness (FCT) <75 μm, lipid arc circumferential extension >180°, and macrophage infiltration (5). Previous research has found that HRPs are associated with an increased risk of cardiovascular events (6–8). Wang Ying et al. identified 274 patients with acute myocardial infarction using OCT-defined HRP plaques and followed them up for 2.2 years, finding that patients with HRP were 2.05 times more likely to have major adverse cardiovascular events than those without HRPs (7). Early prediction of HRPs formation and appropriate intervention are critical for the prevention and treatment of ACS, but no studies have been conducted on this topic.

Therefore, by analysing clinical data in conjunction with blood coherence indicators of circulation, relevant risk factors were identified, and a rapid early prediction model for HRPs was developed, providing novel insights into the prevention and treatment of ACS diseases.

This study successfully developed and validated a nomogram for accurately predicting HRPs in patients with ACS. The nomogram included age, BMI ≥ 25 kg/m², TG, LDL-C, and Log NT-proBNP levels. It demonstrated good discrimination, calibration, and clinical validity, making it a useful and clinically relevant tool for identifying HRPs defined by OCT and ACS.

ACS is typically caused by the rupture of vulnerable plaques, leading to thrombosis (14–16). Previous research has shown that vulnerable plaques have the following characteristics: TCFA, rich in macrophages, and a large central necrotic core (17, 18). The CLIMA study introduced the concept of HRP and found that HRPs are associated with a higher risk of major coronary events (5). Previous research has also found that HRPs defined by OCT are associated with an increased risk of cardiovascular events (6–8). Recent large-scale studies further support these findings: Matsumura et al. showed that OCT-defined TCFA and a minimal lumen area <3.5 mm² are independent predictors of major adverse cardiovascular events (MACE) in patients with acute myocardial infarction (8), while van Veelen et al. reported that HRPs in nonculprit lesions, even when fractional flow reserve (FFR) negative, are linked to a higher incidence of MACE over 2 years (19). These accumulating data reinforce the clinical importance of early HRP detection and timely intervention. However, current imaging modalities, including OCT and intravascular ultrasonography, are invasive, require specialized equipment, and have limitations in the early detection of plaque stability. In low-middle-income countries and areas, OCT is not feasible due to economic and workforce constraints. Circulating biomarkers can be used as additional tools for predicting the risk of vulnerable plaques, and developing a risk prediction model based on circulating biomarkers has greater utility and potential for widespread use in clinical practice. Our proposed biomarker-based model is intended not to replace OCT but to provide a practical, non-invasive tool for early risk stratification and guiding the need for further invasive assessment.

Due to the crucial role of lipid accumulation and inflammation in atherosclerotic plaque formation, previous research on markers of vulnerable plaque circulation has primarily focused on these processes (20). Mechanistically, elevated levels of specific lipids or inflammatory factors indicate their involvement in initiating harmful events that lead to plaque destabilization, increasing the risk of unstable plaque formation. Conversely, certain molecules released by unstable plaque lesions into the bloodstream are easily detectable and may serve as predictors of vulnerable plaques. Hence, the use of circulating biomarkers has a high potential for detecting plaque vulnerability in patients. Nonetheless, no single reliable biological marker has demonstrated adequate sensitivity and specificity. This emphasizes the importance of identifying a panel of circulating markers for predicting the risk of AS plaque vulnerability, as well as developing a risk factor analysis model with greater utility in clinical practice. The current study analyzed relevant circulating blood indicators in patients, revealing statistically significant differences (P < 0.05) in BMI, Hs-CRP, Hemoglobin A1c, TG, TC, HDL-C, LDL-C, ApoB/ApoA1 ratio, and NT-proBNP levels between high-risk and non-HRP groups. These commonly used laboratory indicators may act as risk factors for HRPs.

In this study, general clinical data and laboratory examination indicators of patients were incorporated into the HRP risk prediction system. The aforementioned variables were used as influencing factors to identify factors affecting plaque stability using the LASSO regression model, which resulted in the creation of a model. Variables included “age”, “BMI ≥ 25 kg/m²”, “TG”, “LDL-C,” and “Log NT-proBNP”. Given the large number of indicators used in this study, there is a high risk of encountering issues such as variable collinearity when selecting variables using the least squares method. This can lead to important variables being overlooked, resulting in inaccuracies in the prediction model. The LASSO regression model addresses these concerns by efficiently selecting variables by compressing regression coefficients to zero (21). In the model development, we used the LASSO–Cox method to estimate the relationship between predictors and HRPs. LASSO regularization is a method for managing overfitting and variable selection that has been widely used in a variety of machine learning algorithms (22). When the LASSO method is applied to the Cox model, the estimation variance is reduced, and a subset of predictors is chosen, resulting in an interpretable Cox model (23). To ensure that the model was accurate, we used a nomogram to simplify the parameters in the model presentation. Furthermore, the nomogram model created using identified HRP risk factors is a simple and intuitive tool. Clinicians can calculate the cumulative scores of risk factors using the nomogram, allowing for quick and easy risk stratification for plaque vulnerability and HRP possibility. Thus, the risk factors evaluated in this study can be easily obtained through patient history collection and routine laboratory examination, allowing for early detection of HRPs. The risk prevention model's AUC of 0.780 (95% CI: 0.705–0.855) developed in this study indicates that assessing HRPs has a high predictive capability.

The results from screening variables in the LASSO regression model found that age, BMI ≥ 25 kg/m², TG, LDL-C, and Log NT-proBNP levels are reliable indicators for predicting HRPs, implying that HRPs can be clinically identified using these indicators alone. Age is well-known as a traditional risk factor for cardiovascular events. Many studies have considered age to be an independent predictor of ACS (24), and in studies focusing on other risk factors, age is typically adjusted (25, 26). Furthermore, BMI ≥ 25 kg/m² was associated with HRP. BMI > 25 kg/m² was recently found to be associated with a significantly increased long-term risk of cardiovascular disease morbidity and mortality (27), and another study linked this epidemiologic evidence to HRP formation (28). Elevated triglyceride levels and triglyceride-rich lipoproteins (TRLs) have been increasingly recognized as important contributors to atherosclerotic cardiovascular disease beyond LDL-C (29). Recent evidence suggests that TRLs may promote the formation of lipid-rich necrotic cores, inflammation, and endothelial dysfunction—features commonly associated with high-risk plaques (HRPs) (30). For instance, the accumulation of TRLs in the arterial wall has been shown to induce macrophage activation and foam cell formation, both of which are implicated in plaque vulnerability. These pathophysiological mechanisms highlight the potential role of elevated TG in the development of OCT-defined HRPs and support its inclusion as a relevant biomarker in risk stratification models. LDL-C has long been recognized as an important risk factor for ASCVD, and numerous studies have consistently shown that LDL-C lowering interventions can effectively reduce plaque vulnerability, regardless of the imaging modality used to assess plaque characteristics (31–33). A recent study on intracoronary imaging using OCT found that high levels of small dense LDL-C are linked to the presence of vulnerable plaques (34). NT-proBNP concentration is regarded as a marker of cardiac function in heart disease, and myocardial ischemia can cause a reversible increase in regional wall stress, potentially leading to increased natriuretic peptide release (35). A previous study found significant associations between NT-proBNP and coronary atherosclerotic plaque parameters, which were consistent with the high-sensitivity cardiac Troponin T results (36).

Although HbA1c and diabetes are recognized cardiovascular risk factors, neither was retained in the final model. HbA1c was excluded during the LASSO regression due to its limited independent predictive value after penalization. Diabetes was initially selected but subsequently removed in the multivariate logistic regression, likely due to collinearity with other glycemic markers and limited statistical significance (P = 0.08 in univariate analysis). In contrast, variables such as age, overweight BMI, TG, LDL-C, and NT-proBNP showed stronger and more consistent associations with high-risk plaques and were prioritized in the final model. This reflects the relative predictive contributions of different variables within our cohort. Importantly, we acknowledge that the exclusion of glycemic variables may also reflect the limited sample size of our study, which could reduce the statistical power to detect their independent associations. Future large-scale, multicenter investigations are warranted to validate these findings and comprehensively clarify the prognostic value of glycemic factors in cardiovascular risk prediction.

The predictive value of circulating biomarkers for high-risk coronary plaque features has been increasingly supported by evidence from non-invasive imaging studies. Russo et al. demonstrated that low HDL-C and elevated levels of leptin and interleukin-6 were independently associated with high-risk coronary anatomy as assessed by coronary CT angiography (CCTA) in patients with stable chest pain (37). Similarly, Nidorf et al. reported that high-sensitivity C-reactive protein (hsCRP) levels were significantly associated with vulnerable plaque features, including low attenuation and positive remodeling, identified by CCTA (38). These findings underscore the role of systemic inflammation and lipid metabolism in plaque vulnerability and support the integration of serological biomarkers into non-invasive risk prediction models. Our current study aligns with this direction, aiming to develop a practical, blood-based tool for identifying high-risk plaques defined by OCT, especially in settings where intracoronary imaging may not be routinely feasible.

It is important to note that the model's strong predictive performance does not imply that other indicators have a weak predictive effect on HRPs. Previous research has shown that individual indicators are closely related to the onset and progression of HRPs, as exemplified by the European Society of Cardiology/European Atherosclerosis Blood Lipid Management Guidelines, which emphasize the causal relationship between LDL-C and all apolipoprotein B lipoproteins in arteriosclerotic cardiovascular disease. It is suggested that the role of apolipoprotein B and lipoprotein in cardiovascular risk stratification should be investigated further (25). More research is needed to understand the underlying mechanisms of plaque instability, and progression and to develop more reliable biomarkers for the early detection of HRPs in ACS.

Therefore, we believe that our model will help patients better understand the disease and doctors make clinical decisions. Particularly for patients with a high risk of ACS, doctors can use this model to determine whether patients would benefit from treatment.

There are some limitations to our study. First, the single-center design and limited sample size reduced the statistical power, restricted the robustness of subgroup analyses, and lacked external validation, which may affect the generalizability of the results. Second, for safety considerations regarding the use of iodinated contrast, patients with an estimated glomerular filtration rate (eGFR) <60 ml/min/1.73 m² were excluded; Additionally, in a subset of patients (n = 26) with severely stenotic or occluded lesions, low-pressure (4–6 atm) balloon pre-dilatation was performed to facilitate OCT catheter passage. These procedures are consistent with standard clinical practice and their impact on plaque parameters (such as MLA and FCT) is minimal or negligible. All images were interpreted by blinded observers, and sensitivity analyses were conducted to minimize potential bias.

Despite these limitations, OCT technology enabled accurate classification of high-risk plaque populations, and rigorous statistical methods were applied for risk prediction, thereby providing the proposed prediction model with high accuracy and reliability.

In this study, an OCT examination was used to accurately identify HRPs associated with risk factors in the ACS cohort. The nomogram risk prediction model developed in response to these findings has high predictive efficacy and clinical applicability, making it critical for identifying, preventing, and treating HRP vulnerability. However, due to the single-center observational cohort design and limited sample size, future research will focus on validating the nomogram model's clinical utility in multi-center studies with larger samples.